[AMDGPU] Correct NumLoads in clustering

[rampitec]

Summary

Scheduler sends NumLoads argument into shouldClusterMemOps()
one less the actual cluster length. So for 2 instructions
it will pass just 1. Correct this number.

This is NFC for in tree targets.

[foad]

I tried something similar in D72325.

[rampitec]

In D73292#1837215, @foad wrote:

I tried something similar in D72325.

Comments there argue about how much should we cluster, but regardless I do not think we should use a wrong data. If we want more clustering we need to increase thresholds, but still rely on a correct input.

[rampitec]

I also have an early version of a patch which can reschedule without clustering if no optimal schedule were found for a region.

We may postpone this change, use that patch and then finally correct this number while increasing clustering threshold at the same time. It may be easier in terms of tests update and less prone to performance regressions.

[foad]

In D73292#1837231, @rampitec wrote:

In D73292#1837215, @foad wrote:

I tried something similar in D72325.

Comments there argue about how much should we cluster, but regardless I do not think we should use a wrong data. If we want more clustering we need to increase thresholds, but still rely on a correct input.

I agree. I also think we should fix this properly in MachineScheduler:

--- a/llvm/lib/CodeGen/MachineScheduler.cpp
+++ b/llvm/lib/CodeGen/MachineScheduler.cpp
@@ -1584,7 +1584,7 @@ void BaseMemOpClusterMutation::clusterNeighboringMemOps(
     SUnit *SUb = MemOpRecords[Idx+1].SU;
     if (TII->shouldClusterMemOps(MemOpRecords[Idx].BaseOps,
                                  MemOpRecords[Idx + 1].BaseOps,
-                                 ClusterLength)) {
+                                 ClusterLength + 1)) {
       if (SUa->NodeNum > SUb->NodeNum)
         std::swap(SUa, SUb);
       if (DAG->addEdge(SUb, SDep(SUa, SDep::Cluster))) {

... and adjust any other target implementations of shouldClusterMemOps accordingly.

[rampitec]

In D73292#1838140, @foad wrote:

In D73292#1837231, @rampitec wrote:

In D73292#1837215, @foad wrote:

I tried something similar in D72325.

Comments there argue about how much should we cluster, but regardless I do not think we should use a wrong data. If we want more clustering we need to increase thresholds, but still rely on a correct input.

I agree. I also think we should fix this properly in MachineScheduler:

--- a/llvm/lib/CodeGen/MachineScheduler.cpp
+++ b/llvm/lib/CodeGen/MachineScheduler.cpp
@@ -1584,7 +1584,7 @@ void BaseMemOpClusterMutation::clusterNeighboringMemOps(
     SUnit *SUb = MemOpRecords[Idx+1].SU;
     if (TII->shouldClusterMemOps(MemOpRecords[Idx].BaseOps,
                                  MemOpRecords[Idx + 1].BaseOps,
-                                 ClusterLength)) {
+                                 ClusterLength + 1)) {
       if (SUa->NodeNum > SUb->NodeNum)
         std::swap(SUa, SUb);
       if (DAG->addEdge(SUb, SDep(SUa, SDep::Cluster))) {

... and adjust any other target implementations of shouldClusterMemOps accordingly.

In D73292#1838140, @foad wrote:

In D73292#1837231, @rampitec wrote:

In D73292#1837215, @foad wrote:

I tried something similar in D72325.

Comments there argue about how much should we cluster, but regardless I do not think we should use a wrong data. If we want more clustering we need to increase thresholds, but still rely on a correct input.

I agree. I also think we should fix this properly in MachineScheduler:

--- a/llvm/lib/CodeGen/MachineScheduler.cpp
+++ b/llvm/lib/CodeGen/MachineScheduler.cpp
@@ -1584,7 +1584,7 @@ void BaseMemOpClusterMutation::clusterNeighboringMemOps(
     SUnit *SUb = MemOpRecords[Idx+1].SU;
     if (TII->shouldClusterMemOps(MemOpRecords[Idx].BaseOps,
                                  MemOpRecords[Idx + 1].BaseOps,
-                                 ClusterLength)) {
+                                 ClusterLength + 1)) {
       if (SUa->NodeNum > SUb->NodeNum)
         std::swap(SUa, SUb);
       if (DAG->addEdge(SUb, SDep(SUa, SDep::Cluster))) {

... and adjust any other target implementations of shouldClusterMemOps accordingly.

That I agree too. I just cannot deal with any single performance regression in all of the targets, including out of tree targets. This bug is here at least for 6 years when I first noticed it. I belive every in-tree target need to do the same as in this patch and then we can collectively switch away from this bug. Sounds like the only discussion is about the process how to get there.

[foad]

I just cannot deal with any single performance regression in all of the targets, including out of tree targets.

We can adjust AArch64 (the only other affected in-tree target) so there's no change in behaviour, and out-of-tree target maintainers can adjust their own targets.

[vpykhtin]

LGTM.

[rampitec]

Changed patch to only correct clusterNeighboringMemOps. Threshold logic adjusted accordingly.
We will need to retune our threshold in the AMDGPU separately, likely after scheduler will gain the ability to break clustering under a high pressure.

[foad]

LGTM! Perhaps add a comment in TargetInstrInfo.h documenting the argument? E.g. "the number of loads that will be in the cluster if this hook returns true"...?

[rampitec]

Added operand description to TargetInstrInfo::shouldClusterMemOps()

llvm/lib/Target/AMDGPU/SIInstrInfo.cpp

bool SIInstrInfo::shouldClusterMemOps(ArrayRef<const MachineOperand *> BaseOps1,
   const MachineOperand &BaseOp1 = *BaseOps1.front();
   const MachineOperand &BaseOp2 = *BaseOps2.front();
   const MachineInstr &FirstLdSt = *BaseOp1.getParent();
   const MachineInstr &SecondLdSt = *BaseOp2.getParent();
 
   if (!memOpsHaveSameBasePtr(FirstLdSt, BaseOp1, SecondLdSt, BaseOp2))
     return false;
 
+  // Scheduler sends the cluster length less one.
+  ++NumLoads;
+
   const MachineOperand *FirstDst = nullptr;
   const MachineOperand *SecondDst = nullptr;
 
   if ((isMUBUF(FirstLdSt) && isMUBUF(SecondLdSt)) ||
       (isMTBUF(FirstLdSt) && isMTBUF(SecondLdSt)) ||
       (isFLAT(FirstLdSt) && isFLAT(SecondLdSt))) {
     const unsigned MaxGlobalLoadCluster = 6;
     if (NumLoads > MaxGlobalLoadCluster)

llvm/test/CodeGen/AMDGPU/bitreverse.ll

; FLAT-NEXT:    s_endpgm
   %brev = call i32 @llvm.bitreverse.i32(i32 %val) #1
   store i32 %brev, i32 addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @v_brev_i32(i32 addrspace(1)* noalias %out, i32 addrspace(1)* noalias %valptr) #0 {
 ; SI-LABEL: v_brev_i32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_bfrev_b32_e32 v0, v0
-; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; FLAT-LABEL: v_brev_i32:
 ; FLAT:       ; %bb.0:
-; FLAT-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; FLAT-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; FLAT-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; FLAT-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; FLAT-NEXT:    s_mov_b32 s7, 0xf000
-; FLAT-NEXT:    s_mov_b32 s6, -1
+; FLAT-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; FLAT-NEXT:    s_mov_b32 s3, 0xf000
+; FLAT-NEXT:    s_mov_b32 s2, -1
 ; FLAT-NEXT:    s_waitcnt lgkmcnt(0)
-; FLAT-NEXT:    v_mov_b32_e32 v1, s1
-; FLAT-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; FLAT-NEXT:    v_mov_b32_e32 v1, s5
+; FLAT-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; FLAT-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; FLAT-NEXT:    flat_load_dword v0, v[0:1]
 ; FLAT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; FLAT-NEXT:    v_bfrev_b32_e32 v0, v0
-; FLAT-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; FLAT-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; FLAT-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr i32, i32 addrspace(1)* %valptr, i32 %tid
   %val = load i32, i32 addrspace(1)* %gep
   %brev = call i32 @llvm.bitreverse.i32(i32 %val) #1
   store i32 %brev, i32 addrspace(1)* %out
   ret void
 }
; FLAT-NEXT:    s_endpgm
   %brev = call <2 x i32> @llvm.bitreverse.v2i32(<2 x i32> %val) #1
   store <2 x i32> %brev, <2 x i32> addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @v_brev_v2i32(<2 x i32> addrspace(1)* noalias %out, <2 x i32> addrspace(1)* noalias %valptr) #0 {
 ; SI-LABEL: v_brev_v2i32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dwordx2 v[0:1], v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dwordx2 v[0:1], v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_bfrev_b32_e32 v1, v1
 ; SI-NEXT:    v_bfrev_b32_e32 v0, v0
-; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; FLAT-LABEL: v_brev_v2i32:
 ; FLAT:       ; %bb.0:
-; FLAT-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; FLAT-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; FLAT-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; FLAT-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
-; FLAT-NEXT:    s_mov_b32 s7, 0xf000
-; FLAT-NEXT:    s_mov_b32 s6, -1
+; FLAT-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; FLAT-NEXT:    s_mov_b32 s3, 0xf000
+; FLAT-NEXT:    s_mov_b32 s2, -1
 ; FLAT-NEXT:    s_waitcnt lgkmcnt(0)
-; FLAT-NEXT:    v_mov_b32_e32 v1, s1
-; FLAT-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; FLAT-NEXT:    v_mov_b32_e32 v1, s5
+; FLAT-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; FLAT-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; FLAT-NEXT:    flat_load_dwordx2 v[0:1], v[0:1]
 ; FLAT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; FLAT-NEXT:    v_bfrev_b32_e32 v1, v1
 ; FLAT-NEXT:    v_bfrev_b32_e32 v0, v0
-; FLAT-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
+; FLAT-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; FLAT-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr <2 x i32>, <2 x i32> addrspace(1)* %valptr, i32 %tid
   %val = load <2 x i32>, <2 x i32> addrspace(1)* %gep
   %brev = call <2 x i32> @llvm.bitreverse.v2i32(<2 x i32> %val) #1
   store <2 x i32> %brev, <2 x i32> addrspace(1)* %out
   ret void
 }
; FLAT-NEXT:    s_endpgm
   %brev = call i64 @llvm.bitreverse.i64(i64 %val) #1
   store i64 %brev, i64 addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @v_brev_i64(i64 addrspace(1)* noalias %out, i64 addrspace(1)* noalias %valptr) #0 {
 ; SI-LABEL: v_brev_i64:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dwordx2 v[0:1], v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s0, 0xff00
-; SI-NEXT:    s_mov_b32 s1, 0xf0f0f0f
-; SI-NEXT:    s_mov_b32 s2, 0xf0f0f0f0
-; SI-NEXT:    s_mov_b32 s3, 0x33333333
-; SI-NEXT:    s_mov_b32 s6, 0xcccccccc
-; SI-NEXT:    s_mov_b32 s8, 0x55555555
-; SI-NEXT:    s_mov_b32 s9, 0xaaaaaaaa
+; SI-NEXT:    buffer_load_dwordx2 v[0:1], v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_mov_b32 s4, 0xff00
+; SI-NEXT:    s_mov_b32 s5, 0xf0f0f0f
+; SI-NEXT:    s_mov_b32 s6, 0xf0f0f0f0
+; SI-NEXT:    s_mov_b32 s7, 0x33333333
+; SI-NEXT:    s_mov_b32 s8, 0xcccccccc
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s9, 0x55555555
+; SI-NEXT:    s_mov_b32 s10, 0xaaaaaaaa
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_lshl_b64 v[2:3], v[0:1], 8
 ; SI-NEXT:    v_alignbit_b32 v4, v1, v0, 24
 ; SI-NEXT:    v_alignbit_b32 v5, v1, v0, 8
 ; SI-NEXT:    v_lshrrev_b32_e32 v7, 8, v1
 ; SI-NEXT:    v_lshrrev_b32_e32 v6, 24, v1
 ; SI-NEXT:    v_lshl_b64 v[1:2], v[0:1], 24
 ; SI-NEXT:    v_lshlrev_b32_e32 v1, 24, v0
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 8, v0
 ; SI-NEXT:    v_and_b32_e32 v0, 0xff0000, v0
 ; SI-NEXT:    v_and_b32_e32 v4, 0xff0000, v4
 ; SI-NEXT:    v_and_b32_e32 v5, 0xff000000, v5
-; SI-NEXT:    v_and_b32_e32 v7, s0, v7
+; SI-NEXT:    v_and_b32_e32 v7, s4, v7
 ; SI-NEXT:    v_and_b32_e32 v3, 0xff, v3
-; SI-NEXT:    v_and_b32_e32 v2, s0, v2
+; SI-NEXT:    v_and_b32_e32 v2, s4, v2
 ; SI-NEXT:    v_or_b32_e32 v4, v5, v4
 ; SI-NEXT:    v_or_b32_e32 v5, v7, v6
 ; SI-NEXT:    v_or_b32_e32 v0, v1, v0
 ; SI-NEXT:    v_or_b32_e32 v2, v2, v3
 ; SI-NEXT:    v_or_b32_e32 v1, v4, v5
 ; SI-NEXT:    v_or_b32_e32 v3, v0, v2
-; SI-NEXT:    v_and_b32_e32 v0, s1, v1
-; SI-NEXT:    v_and_b32_e32 v2, s2, v1
-; SI-NEXT:    v_and_b32_e32 v1, s1, v3
-; SI-NEXT:    v_and_b32_e32 v3, s2, v3
+; SI-NEXT:    v_and_b32_e32 v0, s5, v1
+; SI-NEXT:    v_and_b32_e32 v2, s6, v1
+; SI-NEXT:    v_and_b32_e32 v1, s5, v3
+; SI-NEXT:    v_and_b32_e32 v3, s6, v3
 ; SI-NEXT:    v_lshl_b64 v[0:1], v[0:1], 4
 ; SI-NEXT:    v_lshr_b64 v[2:3], v[2:3], 4
 ; SI-NEXT:    v_or_b32_e32 v3, v3, v1
 ; SI-NEXT:    v_or_b32_e32 v2, v2, v0
-; SI-NEXT:    v_and_b32_e32 v1, s3, v3
-; SI-NEXT:    v_and_b32_e32 v0, s3, v2
-; SI-NEXT:    v_and_b32_e32 v3, s6, v3
-; SI-NEXT:    v_and_b32_e32 v2, s6, v2
+; SI-NEXT:    v_and_b32_e32 v1, s7, v3
+; SI-NEXT:    v_and_b32_e32 v0, s7, v2
+; SI-NEXT:    v_and_b32_e32 v3, s8, v3
+; SI-NEXT:    v_and_b32_e32 v2, s8, v2
 ; SI-NEXT:    v_lshl_b64 v[0:1], v[0:1], 2
 ; SI-NEXT:    v_lshr_b64 v[2:3], v[2:3], 2
-; SI-NEXT:    s_mov_b32 s6, -1
 ; SI-NEXT:    v_or_b32_e32 v3, v3, v1
 ; SI-NEXT:    v_or_b32_e32 v2, v2, v0
-; SI-NEXT:    v_and_b32_e32 v1, s8, v3
-; SI-NEXT:    v_and_b32_e32 v0, s8, v2
-; SI-NEXT:    v_and_b32_e32 v3, s9, v3
-; SI-NEXT:    v_and_b32_e32 v2, s9, v2
+; SI-NEXT:    v_and_b32_e32 v1, s9, v3
+; SI-NEXT:    v_and_b32_e32 v0, s9, v2
+; SI-NEXT:    v_and_b32_e32 v3, s10, v3
+; SI-NEXT:    v_and_b32_e32 v2, s10, v2
 ; SI-NEXT:    v_lshl_b64 v[0:1], v[0:1], 1
 ; SI-NEXT:    v_lshr_b64 v[2:3], v[2:3], 1
 ; SI-NEXT:    v_or_b32_e32 v1, v3, v1
 ; SI-NEXT:    v_or_b32_e32 v0, v2, v0
-; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; FLAT-LABEL: v_brev_i64:
 ; FLAT:       ; %bb.0:
-; FLAT-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; FLAT-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; FLAT-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x2c
 ; FLAT-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
 ; FLAT-NEXT:    v_mov_b32_e32 v4, 8
-; FLAT-NEXT:    s_mov_b32 s2, 0xf0f0f0f
-; FLAT-NEXT:    s_mov_b32 s3, 0xf0f0f0f0
+; FLAT-NEXT:    s_mov_b32 s4, 0xf0f0f0f
+; FLAT-NEXT:    s_mov_b32 s5, 0xf0f0f0f0
 ; FLAT-NEXT:    s_waitcnt lgkmcnt(0)
-; FLAT-NEXT:    v_mov_b32_e32 v1, s1
-; FLAT-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; FLAT-NEXT:    v_mov_b32_e32 v1, s3
+; FLAT-NEXT:    v_add_u32_e32 v0, vcc, s2, v0
 ; FLAT-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; FLAT-NEXT:    flat_load_dwordx2 v[0:1], v[0:1]
-; FLAT-NEXT:    s_mov_b32 s0, 0x33333333
-; FLAT-NEXT:    s_mov_b32 s1, 0xcccccccc
-; FLAT-NEXT:    s_mov_b32 s6, 0x55555555
-; FLAT-NEXT:    s_mov_b32 s8, 0xaaaaaaaa
-; FLAT-NEXT:    s_mov_b32 s7, 0xf000
+; FLAT-NEXT:    s_mov_b32 s6, 0x33333333
+; FLAT-NEXT:    s_mov_b32 s7, 0xcccccccc
+; FLAT-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; FLAT-NEXT:    s_mov_b32 s8, 0x55555555
+; FLAT-NEXT:    s_mov_b32 s9, 0xaaaaaaaa
+; FLAT-NEXT:    s_mov_b32 s3, 0xf000
+; FLAT-NEXT:    s_mov_b32 s2, -1
 ; FLAT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; FLAT-NEXT:    v_lshlrev_b64 v[2:3], 24, v[0:1]
 ; FLAT-NEXT:    v_alignbit_b32 v2, v1, v0, 24
 ; FLAT-NEXT:    v_alignbit_b32 v6, v1, v0, 8
 ; FLAT-NEXT:    v_lshlrev_b32_sdwa v7, v4, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_2
 ; FLAT-NEXT:    v_lshlrev_b64 v[4:5], 8, v[0:1]
 ; FLAT-NEXT:    v_lshlrev_b32_e32 v4, 24, v0
 ; FLAT-NEXT:    v_lshlrev_b32_e32 v0, 8, v0
 ; FLAT-NEXT:    v_and_b32_e32 v2, 0xff0000, v2
 ; FLAT-NEXT:    v_and_b32_e32 v6, 0xff000000, v6
 ; FLAT-NEXT:    v_and_b32_e32 v0, 0xff0000, v0
 ; FLAT-NEXT:    v_or_b32_sdwa v1, v7, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
 ; FLAT-NEXT:    v_or_b32_e32 v2, v6, v2
 ; FLAT-NEXT:    v_and_b32_e32 v3, 0xff00, v3
 ; FLAT-NEXT:    v_or_b32_e32 v1, v2, v1
 ; FLAT-NEXT:    v_or_b32_e32 v0, v4, v0
 ; FLAT-NEXT:    v_or_b32_sdwa v2, v3, v5 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_0
 ; FLAT-NEXT:    v_or_b32_e32 v3, v0, v2
-; FLAT-NEXT:    v_and_b32_e32 v0, s2, v1
-; FLAT-NEXT:    v_and_b32_e32 v2, s3, v1
-; FLAT-NEXT:    v_and_b32_e32 v1, s2, v3
-; FLAT-NEXT:    v_and_b32_e32 v3, s3, v3
+; FLAT-NEXT:    v_and_b32_e32 v0, s4, v1
+; FLAT-NEXT:    v_and_b32_e32 v2, s5, v1
+; FLAT-NEXT:    v_and_b32_e32 v1, s4, v3
+; FLAT-NEXT:    v_and_b32_e32 v3, s5, v3
 ; FLAT-NEXT:    v_lshlrev_b64 v[0:1], 4, v[0:1]
 ; FLAT-NEXT:    v_lshrrev_b64 v[2:3], 4, v[2:3]
 ; FLAT-NEXT:    v_or_b32_e32 v3, v3, v1
 ; FLAT-NEXT:    v_or_b32_e32 v2, v2, v0
-; FLAT-NEXT:    v_and_b32_e32 v1, s0, v3
-; FLAT-NEXT:    v_and_b32_e32 v0, s0, v2
-; FLAT-NEXT:    v_and_b32_e32 v3, s1, v3
-; FLAT-NEXT:    v_and_b32_e32 v2, s1, v2
+; FLAT-NEXT:    v_and_b32_e32 v1, s6, v3
+; FLAT-NEXT:    v_and_b32_e32 v0, s6, v2
+; FLAT-NEXT:    v_and_b32_e32 v3, s7, v3
+; FLAT-NEXT:    v_and_b32_e32 v2, s7, v2
 ; FLAT-NEXT:    v_lshlrev_b64 v[0:1], 2, v[0:1]
 ; FLAT-NEXT:    v_lshrrev_b64 v[2:3], 2, v[2:3]
 ; FLAT-NEXT:    v_or_b32_e32 v3, v3, v1
 ; FLAT-NEXT:    v_or_b32_e32 v2, v2, v0
-; FLAT-NEXT:    v_and_b32_e32 v1, s6, v3
-; FLAT-NEXT:    v_and_b32_e32 v0, s6, v2
-; FLAT-NEXT:    v_and_b32_e32 v3, s8, v3
-; FLAT-NEXT:    v_and_b32_e32 v2, s8, v2
+; FLAT-NEXT:    v_and_b32_e32 v1, s8, v3
+; FLAT-NEXT:    v_and_b32_e32 v0, s8, v2
+; FLAT-NEXT:    v_and_b32_e32 v3, s9, v3
+; FLAT-NEXT:    v_and_b32_e32 v2, s9, v2
 ; FLAT-NEXT:    v_lshlrev_b64 v[0:1], 1, v[0:1]
 ; FLAT-NEXT:    v_lshrrev_b64 v[2:3], 1, v[2:3]
-; FLAT-NEXT:    s_mov_b32 s6, -1
 ; FLAT-NEXT:    v_or_b32_e32 v1, v3, v1
 ; FLAT-NEXT:    v_or_b32_e32 v0, v2, v0
-; FLAT-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
+; FLAT-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; FLAT-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr i64, i64 addrspace(1)* %valptr, i32 %tid
   %val = load i64, i64 addrspace(1)* %gep
   %brev = call i64 @llvm.bitreverse.i64(i64 %val) #1
   store i64 %brev, i64 addrspace(1)* %out
   ret void
 }
; FLAT-NEXT:    s_endpgm
   %brev = call <2 x i64> @llvm.bitreverse.v2i64(<2 x i64> %val) #1
   store <2 x i64> %brev, <2 x i64> addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @v_brev_v2i64(<2 x i64> addrspace(1)* noalias %out, <2 x i64> addrspace(1)* noalias %valptr) #0 {
 ; SI-LABEL: v_brev_v2i64:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 4, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dwordx4 v[0:3], v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s0, 0xff00
-; SI-NEXT:    s_mov_b32 s1, 0xf0f0f0f
-; SI-NEXT:    s_mov_b32 s2, 0xf0f0f0f0
-; SI-NEXT:    s_mov_b32 s3, 0x33333333
+; SI-NEXT:    buffer_load_dwordx4 v[0:3], v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_mov_b32 s4, 0xff00
+; SI-NEXT:    s_mov_b32 s5, 0xf0f0f0f
+; SI-NEXT:    s_mov_b32 s6, 0xf0f0f0f0
+; SI-NEXT:    s_mov_b32 s7, 0x33333333
 ; SI-NEXT:    s_mov_b32 s8, 0xcccccccc
 ; SI-NEXT:    s_mov_b32 s9, 0x55555555
 ; SI-NEXT:    s_mov_b32 s10, 0xaaaaaaaa
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_lshl_b64 v[4:5], v[2:3], 8
 ; SI-NEXT:    v_alignbit_b32 v6, v3, v2, 24
 ; SI-NEXT:    v_alignbit_b32 v7, v3, v2, 8
 ; SI-NEXT:    v_lshrrev_b32_e32 v9, 8, v3
 ; SI-NEXT:    v_lshrrev_b32_e32 v8, 24, v3
 ; SI-NEXT:    v_lshl_b64 v[3:4], v[2:3], 24
 ; SI-NEXT:    v_lshlrev_b32_e32 v10, 24, v2
 ; SI-NEXT:    v_lshlrev_b32_e32 v11, 8, v2
 ; SI-NEXT:    v_lshl_b64 v[2:3], v[0:1], 8
 ; SI-NEXT:    v_alignbit_b32 v12, v1, v0, 24
 ; SI-NEXT:    v_alignbit_b32 v13, v1, v0, 8
 ; SI-NEXT:    v_lshrrev_b32_e32 v14, 24, v1
 ; SI-NEXT:    v_lshrrev_b32_e32 v15, 8, v1
 ; SI-NEXT:    v_lshlrev_b32_e32 v16, 24, v0
 ; SI-NEXT:    v_lshlrev_b32_e32 v17, 8, v0
 ; SI-NEXT:    v_lshl_b64 v[0:1], v[0:1], 24
 ; SI-NEXT:    v_and_b32_e32 v6, 0xff0000, v6
 ; SI-NEXT:    v_and_b32_e32 v7, 0xff000000, v7
 ; SI-NEXT:    v_mov_b32_e32 v0, 0xff0000
 ; SI-NEXT:    v_or_b32_e32 v6, v7, v6
 ; SI-NEXT:    v_mov_b32_e32 v7, 0xff00
 ; SI-NEXT:    v_and_b32_e32 v2, v0, v11
 ; SI-NEXT:    v_and_b32_e32 v11, v0, v12
-; SI-NEXT:    v_and_b32_e32 v9, s0, v9
+; SI-NEXT:    v_and_b32_e32 v9, s4, v9
 ; SI-NEXT:    v_and_b32_e32 v12, 0xff000000, v13
 ; SI-NEXT:    v_and_b32_e32 v0, v0, v17
 ; SI-NEXT:    v_and_b32_e32 v13, v7, v15
 ; SI-NEXT:    v_and_b32_e32 v1, v7, v1
 ; SI-NEXT:    v_and_b32_e32 v3, 0xff, v3
 ; SI-NEXT:    v_or_b32_e32 v8, v9, v8
 ; SI-NEXT:    v_or_b32_e32 v2, v10, v2
 ; SI-NEXT:    v_and_b32_e32 v5, 0xff, v5
-; SI-NEXT:    v_and_b32_e32 v4, s0, v4
+; SI-NEXT:    v_and_b32_e32 v4, s4, v4
 ; SI-NEXT:    v_or_b32_e32 v7, v16, v0
 ; SI-NEXT:    v_or_b32_e32 v1, v1, v3
 ; SI-NEXT:    v_or_b32_e32 v9, v12, v11
 ; SI-NEXT:    v_or_b32_e32 v10, v13, v14
 ; SI-NEXT:    v_or_b32_e32 v0, v4, v5
 ; SI-NEXT:    v_or_b32_e32 v5, v9, v10
 ; SI-NEXT:    v_or_b32_e32 v6, v6, v8
 ; SI-NEXT:    v_or_b32_e32 v7, v7, v1
 ; SI-NEXT:    v_or_b32_e32 v3, v2, v0
-; SI-NEXT:    v_and_b32_e32 v0, s1, v6
-; SI-NEXT:    v_and_b32_e32 v2, s2, v6
-; SI-NEXT:    v_and_b32_e32 v4, s1, v5
-; SI-NEXT:    v_and_b32_e32 v6, s2, v5
-; SI-NEXT:    v_and_b32_e32 v5, s1, v7
-; SI-NEXT:    v_and_b32_e32 v7, s2, v7
-; SI-NEXT:    v_and_b32_e32 v1, s1, v3
-; SI-NEXT:    v_and_b32_e32 v3, s2, v3
+; SI-NEXT:    v_and_b32_e32 v0, s5, v6
+; SI-NEXT:    v_and_b32_e32 v2, s6, v6
+; SI-NEXT:    v_and_b32_e32 v4, s5, v5
+; SI-NEXT:    v_and_b32_e32 v6, s6, v5
+; SI-NEXT:    v_and_b32_e32 v5, s5, v7
+; SI-NEXT:    v_and_b32_e32 v7, s6, v7
+; SI-NEXT:    v_and_b32_e32 v1, s5, v3
+; SI-NEXT:    v_and_b32_e32 v3, s6, v3
 ; SI-NEXT:    v_lshl_b64 v[4:5], v[4:5], 4
 ; SI-NEXT:    v_lshr_b64 v[6:7], v[6:7], 4
 ; SI-NEXT:    v_lshl_b64 v[0:1], v[0:1], 4
 ; SI-NEXT:    v_lshr_b64 v[2:3], v[2:3], 4
 ; SI-NEXT:    v_or_b32_e32 v7, v7, v5
 ; SI-NEXT:    v_or_b32_e32 v6, v6, v4
 ; SI-NEXT:    v_or_b32_e32 v3, v3, v1
 ; SI-NEXT:    v_or_b32_e32 v2, v2, v0
-; SI-NEXT:    v_and_b32_e32 v5, s3, v7
-; SI-NEXT:    v_and_b32_e32 v4, s3, v6
+; SI-NEXT:    v_and_b32_e32 v5, s7, v7
+; SI-NEXT:    v_and_b32_e32 v4, s7, v6
 ; SI-NEXT:    v_and_b32_e32 v7, s8, v7
 ; SI-NEXT:    v_and_b32_e32 v6, s8, v6
-; SI-NEXT:    v_and_b32_e32 v1, s3, v3
-; SI-NEXT:    v_and_b32_e32 v0, s3, v2
+; SI-NEXT:    v_and_b32_e32 v1, s7, v3
+; SI-NEXT:    v_and_b32_e32 v0, s7, v2
 ; SI-NEXT:    v_and_b32_e32 v3, s8, v3
 ; SI-NEXT:    v_and_b32_e32 v2, s8, v2
 ; SI-NEXT:    v_lshl_b64 v[4:5], v[4:5], 2
 ; SI-NEXT:    v_lshr_b64 v[6:7], v[6:7], 2
 ; SI-NEXT:    v_lshl_b64 v[0:1], v[0:1], 2
 ; SI-NEXT:    v_lshr_b64 v[2:3], v[2:3], 2
 ; SI-NEXT:    v_or_b32_e32 v7, v7, v5
 ; SI-NEXT:    v_or_b32_e32 v9, v6, v4

 ; SI-NEXT:    v_and_b32_e32 v2, s10, v2
 ; SI-NEXT:    v_lshl_b64 v[0:1], v[0:1], 1
 ; SI-NEXT:    v_lshr_b64 v[2:3], v[2:3], 1
 ; SI-NEXT:    v_lshr_b64 v[4:5], v[5:6], 1
 ; SI-NEXT:    v_or_b32_e32 v3, v3, v1
 ; SI-NEXT:    v_or_b32_e32 v2, v2, v0
 ; SI-NEXT:    v_or_b32_e32 v1, v5, v8
 ; SI-NEXT:    v_or_b32_e32 v0, v4, v7
-; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; FLAT-LABEL: v_brev_v2i64:
 ; FLAT:       ; %bb.0:
-; FLAT-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; FLAT-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; FLAT-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x2c
 ; FLAT-NEXT:    v_lshlrev_b32_e32 v0, 4, v0
 ; FLAT-NEXT:    v_mov_b32_e32 v8, 8
 ; FLAT-NEXT:    v_mov_b32_e32 v10, 0xff0000
-; FLAT-NEXT:    s_mov_b32 s2, 0xf0f0f0f
+; FLAT-NEXT:    s_mov_b32 s4, 0xf0f0f0f
 ; FLAT-NEXT:    s_waitcnt lgkmcnt(0)
-; FLAT-NEXT:    v_mov_b32_e32 v1, s1
-; FLAT-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; FLAT-NEXT:    v_mov_b32_e32 v1, s3
+; FLAT-NEXT:    v_add_u32_e32 v0, vcc, s2, v0
 ; FLAT-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; FLAT-NEXT:    flat_load_dwordx4 v[0:3], v[0:1]
-; FLAT-NEXT:    s_mov_b32 s0, 0xf0f0f0f0
-; FLAT-NEXT:    s_mov_b32 s1, 0x33333333
-; FLAT-NEXT:    s_mov_b32 s3, 0xcccccccc
+; FLAT-NEXT:    s_mov_b32 s5, 0xf0f0f0f0
+; FLAT-NEXT:    s_mov_b32 s6, 0x33333333
+; FLAT-NEXT:    s_mov_b32 s7, 0xcccccccc
+; FLAT-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; FLAT-NEXT:    s_mov_b32 s8, 0x55555555
 ; FLAT-NEXT:    s_mov_b32 s9, 0xaaaaaaaa
-; FLAT-NEXT:    s_mov_b32 s7, 0xf000
-; FLAT-NEXT:    s_mov_b32 s6, -1
+; FLAT-NEXT:    s_mov_b32 s3, 0xf000
+; FLAT-NEXT:    s_mov_b32 s2, -1
 ; FLAT-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; FLAT-NEXT:    v_lshlrev_b64 v[4:5], 24, v[2:3]
 ; FLAT-NEXT:    v_lshlrev_b32_sdwa v12, v8, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_2
 ; FLAT-NEXT:    v_lshlrev_b32_sdwa v15, v8, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_2
 ; FLAT-NEXT:    v_lshlrev_b64 v[8:9], 8, v[0:1]
 ; FLAT-NEXT:    v_lshlrev_b64 v[6:7], 8, v[2:3]
 ; FLAT-NEXT:    v_alignbit_b32 v4, v3, v2, 24
 ; FLAT-NEXT:    v_alignbit_b32 v11, v3, v2, 8

 ; FLAT-NEXT:    v_or_b32_e32 v2, v6, v2
 ; FLAT-NEXT:    v_or_b32_e32 v3, v0, v3
 ; FLAT-NEXT:    v_or_b32_sdwa v0, v4, v7 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_0
 ; FLAT-NEXT:    v_or_b32_e32 v6, v8, v10
 ; FLAT-NEXT:    v_or_b32_sdwa v1, v1, v9 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_0
 ; FLAT-NEXT:    v_or_b32_e32 v7, v2, v0
 ; FLAT-NEXT:    v_or_b32_e32 v5, v5, v12
 ; FLAT-NEXT:    v_or_b32_e32 v8, v6, v1
-; FLAT-NEXT:    v_and_b32_e32 v0, s2, v3
-; FLAT-NEXT:    v_and_b32_e32 v1, s2, v7
-; FLAT-NEXT:    v_and_b32_e32 v2, s0, v3
-; FLAT-NEXT:    v_and_b32_e32 v3, s0, v7
-; FLAT-NEXT:    v_and_b32_e32 v4, s2, v5
-; FLAT-NEXT:    v_and_b32_e32 v6, s0, v5
-; FLAT-NEXT:    v_and_b32_e32 v5, s2, v8
-; FLAT-NEXT:    v_and_b32_e32 v7, s0, v8
+; FLAT-NEXT:    v_and_b32_e32 v0, s4, v3
+; FLAT-NEXT:    v_and_b32_e32 v1, s4, v7
+; FLAT-NEXT:    v_and_b32_e32 v2, s5, v3
+; FLAT-NEXT:    v_and_b32_e32 v3, s5, v7
+; FLAT-NEXT:    v_and_b32_e32 v4, s4, v5
+; FLAT-NEXT:    v_and_b32_e32 v6, s5, v5
+; FLAT-NEXT:    v_and_b32_e32 v5, s4, v8
+; FLAT-NEXT:    v_and_b32_e32 v7, s5, v8
 ; FLAT-NEXT:    v_lshlrev_b64 v[0:1], 4, v[0:1]
 ; FLAT-NEXT:    v_lshrrev_b64 v[2:3], 4, v[2:3]
 ; FLAT-NEXT:    v_lshlrev_b64 v[4:5], 4, v[4:5]
 ; FLAT-NEXT:    v_lshrrev_b64 v[6:7], 4, v[6:7]
 ; FLAT-NEXT:    v_or_b32_e32 v3, v3, v1
 ; FLAT-NEXT:    v_or_b32_e32 v2, v2, v0
 ; FLAT-NEXT:    v_or_b32_e32 v7, v7, v5
 ; FLAT-NEXT:    v_or_b32_e32 v6, v6, v4
-; FLAT-NEXT:    v_and_b32_e32 v1, s1, v3
-; FLAT-NEXT:    v_and_b32_e32 v0, s1, v2
-; FLAT-NEXT:    v_and_b32_e32 v5, s1, v7
-; FLAT-NEXT:    v_and_b32_e32 v4, s1, v6
-; FLAT-NEXT:    v_and_b32_e32 v3, s3, v3
-; FLAT-NEXT:    v_and_b32_e32 v2, s3, v2
-; FLAT-NEXT:    v_and_b32_e32 v7, s3, v7
-; FLAT-NEXT:    v_and_b32_e32 v6, s3, v6
+; FLAT-NEXT:    v_and_b32_e32 v1, s6, v3
+; FLAT-NEXT:    v_and_b32_e32 v0, s6, v2
+; FLAT-NEXT:    v_and_b32_e32 v5, s6, v7
+; FLAT-NEXT:    v_and_b32_e32 v4, s6, v6
+; FLAT-NEXT:    v_and_b32_e32 v3, s7, v3
+; FLAT-NEXT:    v_and_b32_e32 v2, s7, v2
+; FLAT-NEXT:    v_and_b32_e32 v7, s7, v7
+; FLAT-NEXT:    v_and_b32_e32 v6, s7, v6
 ; FLAT-NEXT:    v_lshlrev_b64 v[0:1], 2, v[0:1]
 ; FLAT-NEXT:    v_lshrrev_b64 v[2:3], 2, v[2:3]
 ; FLAT-NEXT:    v_lshlrev_b64 v[4:5], 2, v[4:5]
 ; FLAT-NEXT:    v_lshrrev_b64 v[6:7], 2, v[6:7]
 ; FLAT-NEXT:    v_or_b32_e32 v3, v3, v1
 ; FLAT-NEXT:    v_or_b32_e32 v2, v2, v0
 ; FLAT-NEXT:    v_or_b32_e32 v7, v7, v5
 ; FLAT-NEXT:    v_or_b32_e32 v6, v6, v4
 ; FLAT-NEXT:    v_and_b32_e32 v1, s8, v3
 ; FLAT-NEXT:    v_and_b32_e32 v0, s8, v2
 ; FLAT-NEXT:    v_and_b32_e32 v5, s8, v7
 ; FLAT-NEXT:    v_and_b32_e32 v4, s8, v6
 ; FLAT-NEXT:    v_and_b32_e32 v3, s9, v3
 ; FLAT-NEXT:    v_and_b32_e32 v2, s9, v2
 ; FLAT-NEXT:    v_and_b32_e32 v7, s9, v7
 ; FLAT-NEXT:    v_and_b32_e32 v6, s9, v6
 ; FLAT-NEXT:    v_lshlrev_b64 v[0:1], 1, v[0:1]
 ; FLAT-NEXT:    v_lshrrev_b64 v[2:3], 1, v[2:3]
 ; FLAT-NEXT:    v_lshlrev_b64 v[4:5], 1, v[4:5]
 ; FLAT-NEXT:    v_lshrrev_b64 v[6:7], 1, v[6:7]
 ; FLAT-NEXT:    v_or_b32_e32 v3, v3, v1
 ; FLAT-NEXT:    v_or_b32_e32 v2, v2, v0
 ; FLAT-NEXT:    v_or_b32_e32 v1, v7, v5
 ; FLAT-NEXT:    v_or_b32_e32 v0, v6, v4
-; FLAT-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
+; FLAT-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
 ; FLAT-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr <2 x i64> , <2 x i64> addrspace(1)* %valptr, i32 %tid
   %val = load <2 x i64>, <2 x i64> addrspace(1)* %gep
   %brev = call <2 x i64> @llvm.bitreverse.v2i64(<2 x i64> %val) #1
   store <2 x i64> %brev, <2 x i64> addrspace(1)* %out
   ret void
 }

llvm/test/CodeGen/AMDGPU/call-argument-types.ll

call void @external_void_func_12xv3f32(
       <3 x float><float 9.0, float 9.0, float 9.0>,
       <3 x float><float 10.0, float 11.0, float 12.0>,
       <3 x float><float 13.0, float 14.0, float 15.0>)
   ret void
 }
 
 ; GCN-LABEL: {{^}}stack_8xv5i32:
 
-; GCN: v_mov_b32_e32 [[REG8:v[0-9]+]], 8
-; GCN: v_mov_b32_e32 [[REG9:v[0-9]+]], 9
-; GCN: v_mov_b32_e32 [[REG10:v[0-9]+]], 10
-; GCN: v_mov_b32_e32 [[REG11:v[0-9]+]], 11
-; GCN: v_mov_b32_e32 [[REG12:v[0-9]+]], 12
-; GCN: buffer_store_dword [[REG8]], {{.*$}}
-; GCN: buffer_store_dword [[REG9]], {{.*}} offset:4
-; GCN: buffer_store_dword [[REG10]], {{.*}} offset:8
-; GCN: buffer_store_dword [[REG11]], {{.*}} offset:12
-; GCN: buffer_store_dword [[REG12]], {{.*}} offset:16
-; GCN: v_mov_b32_e32 [[REG13:v[0-9]+]], 13
-; GCN: v_mov_b32_e32 [[REG14:v[0-9]+]], 14
-; GCN: v_mov_b32_e32 [[REG15:v[0-9]+]], 15
-; GCN: buffer_store_dword [[REG13]], {{.*}} offset:20
-; GCN: buffer_store_dword [[REG14]], {{.*}} offset:24
-; GCN: buffer_store_dword [[REG15]], {{.*}} offset:28
+; GCN-DAG: v_mov_b32_e32 [[REG8:v[0-9]+]], 8
+; GCN-DAG: v_mov_b32_e32 [[REG9:v[0-9]+]], 9
+; GCN-DAG: v_mov_b32_e32 [[REG10:v[0-9]+]], 10
+; GCN-DAG: v_mov_b32_e32 [[REG11:v[0-9]+]], 11
+; GCN-DAG: v_mov_b32_e32 [[REG12:v[0-9]+]], 12
+; GCN-DAG: buffer_store_dword [[REG8]], {{.*$}}
+; GCN-DAG: buffer_store_dword [[REG9]], {{.*}} offset:4
+; GCN-DAG: buffer_store_dword [[REG10]], {{.*}} offset:8
+; GCN-DAG: buffer_store_dword [[REG11]], {{.*}} offset:12
+; GCN-DAG: buffer_store_dword [[REG12]], {{.*}} offset:16
+; GCN-DAG: v_mov_b32_e32 [[REG13:v[0-9]+]], 13
+; GCN-DAG: v_mov_b32_e32 [[REG14:v[0-9]+]], 14
+; GCN-DAG: v_mov_b32_e32 [[REG15:v[0-9]+]], 15
+; GCN-DAG: buffer_store_dword [[REG13]], {{.*}} offset:20
+; GCN-DAG: buffer_store_dword [[REG14]], {{.*}} offset:24
+; GCN-DAG: buffer_store_dword [[REG15]], {{.*}} offset:28
 
 ; GCN: v_mov_b32_e32 v31, 7
 ; GCN: s_getpc
 define void @stack_8xv5i32() #0 {
 entry:
   call void @external_void_func_8xv5i32(
       <5 x i32><i32 0, i32 0, i32 0, i32 0, i32 0>,
       <5 x i32><i32 1, i32 1, i32 1, i32 1, i32 1>,
       <5 x i32><i32 2, i32 2, i32 2, i32 2, i32 2>,
       <5 x i32><i32 3, i32 3, i32 3, i32 3, i32 3>,
       <5 x i32><i32 4, i32 4, i32 4, i32 4, i32 4>,
       <5 x i32><i32 5, i32 5, i32 5, i32 5, i32 5>,
       <5 x i32><i32 6, i32 7, i32 8, i32 9, i32 10>,
       <5 x i32><i32 11, i32 12, i32 13, i32 14, i32 15>)
   ret void
 }
 
 ; GCN-LABEL: {{^}}stack_8xv5f32:
-; GCN: v_mov_b32_e32 [[REG8:v[0-9]+]], 0x41000000
-; GCN: v_mov_b32_e32 [[REG9:v[0-9]+]], 0x41100000
-; GCN: v_mov_b32_e32 [[REG10:v[0-9]+]], 0x41200000
-; GCN: v_mov_b32_e32 [[REG11:v[0-9]+]], 0x41300000
-; GCN: v_mov_b32_e32 [[REG12:v[0-9]+]], 0x41400000
-; GCN: buffer_store_dword [[REG8]], {{.*$}}
-; GCN: buffer_store_dword [[REG9]], {{.*}} offset:4
-; GCN: buffer_store_dword [[REG10]], {{.*}} offset:8
-; GCN: buffer_store_dword [[REG11]], {{.*}} offset:12
-; GCN: buffer_store_dword [[REG12]], {{.*}} offset:16
-; GCN: v_mov_b32_e32 [[REG13:v[0-9]+]], 0x41500000
-; GCN: v_mov_b32_e32 [[REG14:v[0-9]+]], 0x41600000
-; GCN: v_mov_b32_e32 [[REG15:v[0-9]+]], 0x41700000
-; GCN: buffer_store_dword [[REG13]], {{.*}} offset:20
-; GCN: buffer_store_dword [[REG14]], {{.*}} offset:24
-; GCN: buffer_store_dword [[REG15]], {{.*}} offset:28
+; GCN-DAG: v_mov_b32_e32 [[REG8:v[0-9]+]], 0x41000000
+; GCN-DAG: v_mov_b32_e32 [[REG9:v[0-9]+]], 0x41100000
+; GCN-DAG: v_mov_b32_e32 [[REG10:v[0-9]+]], 0x41200000
+; GCN-DAG: v_mov_b32_e32 [[REG11:v[0-9]+]], 0x41300000
+; GCN-DAG: v_mov_b32_e32 [[REG12:v[0-9]+]], 0x41400000
+; GCN-DAG: buffer_store_dword [[REG8]], {{.*$}}
+; GCN-DAG: buffer_store_dword [[REG9]], {{.*}} offset:4
+; GCN-DAG: buffer_store_dword [[REG10]], {{.*}} offset:8
+; GCN-DAG: buffer_store_dword [[REG11]], {{.*}} offset:12
+; GCN-DAG: buffer_store_dword [[REG12]], {{.*}} offset:16
+; GCN-DAG: v_mov_b32_e32 [[REG13:v[0-9]+]], 0x41500000
+; GCN-DAG: v_mov_b32_e32 [[REG14:v[0-9]+]], 0x41600000
+; GCN-DAG: v_mov_b32_e32 [[REG15:v[0-9]+]], 0x41700000
+; GCN-DAG: buffer_store_dword [[REG13]], {{.*}} offset:20
+; GCN-DAG: buffer_store_dword [[REG14]], {{.*}} offset:24
+; GCN-DAG: buffer_store_dword [[REG15]], {{.*}} offset:28
 
 ; GCN: v_mov_b32_e32 v31, 0x40e00000
 ; GCN: s_getpc
 define void @stack_8xv5f32() #0 {
 entry:
   call void @external_void_func_8xv5f32(
       <5 x float><float 0.0, float 0.0, float 0.0, float 0.0, float 0.0>,
       <5 x float><float 1.0, float 1.0, float 1.0, float 1.0, float 1.0>,

llvm/test/CodeGen/AMDGPU/copy-illegal-type.ll

; VI-NEXT:    s_endpgm
   %val = load <4 x i8>, <4 x i8> addrspace(1)* %gep, align 4
   store <4 x i8> %val, <4 x i8> addrspace(1)* %out, align 4
   ret void
 }
 
 define amdgpu_kernel void @test_copy_v4i8_x2(<4 x i8> addrspace(1)* %out0, <4 x i8> addrspace(1)* %out1, <4 x i8> addrspace(1)* %in) nounwind {
 ; SI-LABEL: test_copy_v4i8_x2:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xd
-; SI-NEXT:    s_mov_b32 s11, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s11
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s10, -1
-; SI-NEXT:    s_mov_b32 s8, s6
-; SI-NEXT:    s_mov_b32 s9, s7
-; SI-NEXT:    s_mov_b32 s6, s10
-; SI-NEXT:    s_mov_b32 s7, s11
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    s_mov_b32 s0, s6
+; SI-NEXT:    s_mov_b32 s1, s7
+; SI-NEXT:    s_mov_b32 s6, s2
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
-; SI-NEXT:    buffer_store_dword v0, off, s[8:11], 0
+; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: test_copy_v4i8_x2:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0x34
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; VI-NEXT:    s_mov_b32 s3, 0xf000
 ; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_mov_b32 s0, s6
-; VI-NEXT:    v_mov_b32_e32 v1, s9
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s8, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v0, v[0:1]
+; VI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    s_mov_b32 s0, s6
 ; VI-NEXT:    s_mov_b32 s1, s7
 ; VI-NEXT:    s_mov_b32 s6, s2
 ; VI-NEXT:    s_mov_b32 s7, s3
-; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
+; VI-NEXT:    s_waitcnt vmcnt(0)
 ; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; VI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
   %tid.x = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %in, i32 %tid.x
   %val = load <4 x i8>, <4 x i8> addrspace(1)* %gep, align 4
   store <4 x i8> %val, <4 x i8> addrspace(1)* %out0, align 4
   store <4 x i8> %val, <4 x i8> addrspace(1)* %out1, align 4
; VI-NEXT:    s_endpgm
   store <4 x i8> %val, <4 x i8> addrspace(1)* %out2, align 4
   store <4 x i8> %val, <4 x i8> addrspace(1)* %out3, align 4
   ret void
 }
 
 define amdgpu_kernel void @test_copy_v4i8_extra_use(<4 x i8> addrspace(1)* %out0, <4 x i8> addrspace(1)* %out1, <4 x i8> addrspace(1)* %in) nounwind {
 ; SI-LABEL: test_copy_v4i8_extra_use:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xd
-; SI-NEXT:    s_mov_b32 s11, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s11
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s10, -1
-; SI-NEXT:    s_mov_b32 s0, 0xff00
-; SI-NEXT:    s_mov_b32 s8, s6
-; SI-NEXT:    s_mov_b32 s9, s7
-; SI-NEXT:    s_mov_b32 s6, s10
-; SI-NEXT:    s_mov_b32 s7, s11
-; SI-NEXT:    s_movk_i32 s1, 0xff
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
+; SI-NEXT:    s_mov_b32 s8, 0xff00
+; SI-NEXT:    s_movk_i32 s9, 0xff
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    s_mov_b32 s0, s6
+; SI-NEXT:    s_mov_b32 s1, s7
+; SI-NEXT:    s_mov_b32 s6, s2
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_lshrrev_b32_e32 v1, 16, v0
 ; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
-; SI-NEXT:    v_and_b32_e32 v2, s0, v0
+; SI-NEXT:    v_and_b32_e32 v2, s8, v0
 ; SI-NEXT:    s_waitcnt expcnt(0)
 ; SI-NEXT:    v_add_i32_e32 v0, vcc, 9, v0
-; SI-NEXT:    v_and_b32_e32 v0, s1, v0
-; SI-NEXT:    v_and_b32_e32 v3, s0, v1
+; SI-NEXT:    v_and_b32_e32 v0, s9, v0
+; SI-NEXT:    v_and_b32_e32 v3, s8, v1
 ; SI-NEXT:    v_add_i32_e32 v1, vcc, 9, v1
 ; SI-NEXT:    v_or_b32_e32 v0, v2, v0
-; SI-NEXT:    v_and_b32_e32 v1, s1, v1
+; SI-NEXT:    v_and_b32_e32 v1, s9, v1
 ; SI-NEXT:    v_add_i32_e32 v0, vcc, 0x900, v0
 ; SI-NEXT:    v_or_b32_e32 v1, v3, v1
 ; SI-NEXT:    v_and_b32_e32 v0, 0xffff, v0
 ; SI-NEXT:    v_lshlrev_b32_e32 v1, 16, v1
 ; SI-NEXT:    v_or_b32_e32 v0, v1, v0
 ; SI-NEXT:    v_add_i32_e32 v0, vcc, 0x9000000, v0
-; SI-NEXT:    buffer_store_dword v0, off, s[8:11], 0
+; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: test_copy_v4i8_extra_use:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; VI-NEXT:    s_movk_i32 s8, 0xff00
-; VI-NEXT:    s_mov_b32 s3, 0xf000
-; VI-NEXT:    s_mov_b32 s2, -1
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    s_movk_i32 s9, 0xff
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v0, v[0:1]
-; VI-NEXT:    s_mov_b32 s0, s6
-; VI-NEXT:    s_mov_b32 s1, s7
-; VI-NEXT:    s_movk_i32 s9, 0xff
-; VI-NEXT:    s_mov_b32 s6, s2
-; VI-NEXT:    s_mov_b32 s7, s3
 ; VI-NEXT:    s_movk_i32 s10, 0x900
+; VI-NEXT:    s_mov_b32 s7, 0xf000
+; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_mov_b32 s4, s2
+; VI-NEXT:    s_mov_b32 s5, s3
+; VI-NEXT:    s_mov_b32 s2, s6
+; VI-NEXT:    s_mov_b32 s3, s7
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_lshrrev_b32_e32 v1, 16, v0
-; VI-NEXT:    v_and_b32_e32 v3, s8, v1
+; VI-NEXT:    v_and_b32_e32 v4, s8, v1
 ; VI-NEXT:    v_add_u16_e32 v1, 9, v1
+; VI-NEXT:    v_add_u16_e32 v3, 9, v0
 ; VI-NEXT:    v_and_b32_e32 v1, s9, v1
-; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    v_or_b32_e32 v1, v4, v1
 ; VI-NEXT:    v_and_b32_e32 v2, s8, v0
-; VI-NEXT:    v_add_u16_e32 v0, 9, v0
-; VI-NEXT:    v_and_b32_e32 v0, s9, v0
-; VI-NEXT:    v_or_b32_e32 v1, v3, v1
-; VI-NEXT:    v_or_b32_e32 v0, v2, v0
+; VI-NEXT:    v_and_b32_e32 v3, s9, v3
+; VI-NEXT:    v_or_b32_e32 v2, v2, v3
 ; VI-NEXT:    v_add_u16_e32 v1, s10, v1
-; VI-NEXT:    v_add_u16_e32 v0, s10, v0
+; VI-NEXT:    v_add_u16_e32 v2, s10, v2
 ; VI-NEXT:    v_lshlrev_b32_e32 v1, 16, v1
-; VI-NEXT:    v_or_b32_e32 v0, v0, v1
+; VI-NEXT:    v_or_b32_e32 v1, v2, v1
 ; VI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
+; VI-NEXT:    buffer_store_dword v1, off, s[4:7], 0
 ; VI-NEXT:    s_endpgm
   %tid.x = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %in, i32 %tid.x
   %val = load <4 x i8>, <4 x i8> addrspace(1)* %gep, align 4
   %add = add <4 x i8> %val, <i8 9, i8 9, i8 9, i8 9>
   store <4 x i8> %val, <4 x i8> addrspace(1)* %out0, align 4
   store <4 x i8> %add, <4 x i8> addrspace(1)* %out1, align 4
   ret void

llvm/test/CodeGen/AMDGPU/ctlz.ll

; EG-NEXT:    32(4.484155e-44), 2(2.802597e-45)
   %ctlz = call i32 @llvm.ctlz.i32(i32 %val, i1 false) nounwind readnone
   store i32 %ctlz, i32 addrspace(1)* %out, align 4
   ret void
 }
 
 define amdgpu_kernel void @v_ctlz_i32(i32 addrspace(1)* noalias %out, i32 addrspace(1)* noalias %valptr) nounwind {
 ; SI-LABEL: v_ctlz_i32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s10, 0
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
-; SI-NEXT:    s_mov_b32 s11, s7
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_ffbh_u32_e32 v1, v0
 ; SI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v0
 ; SI-NEXT:    v_cndmask_b32_e32 v0, 32, v1, vcc
-; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: v_ctlz_i32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_ffbh_u32_e32 v1, v0
 ; VI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v0
 ; VI-NEXT:    v_cndmask_b32_e32 v0, 32, v1, vcc
-; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
 ;
 ; EG-LABEL: v_ctlz_i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 2, @8, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    TEX 0 @6
 ; EG-NEXT:    ALU 3, @11, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.X, T1.X, 1
; EG-NEXT:    32(4.484155e-44), 2(2.802597e-45)
   %ctlz = call i32 @llvm.ctlz.i32(i32 %val, i1 false) nounwind readnone
   store i32 %ctlz, i32 addrspace(1)* %out, align 4
   ret void
 }
 
 define amdgpu_kernel void @v_ctlz_v2i32(<2 x i32> addrspace(1)* noalias %out, <2 x i32> addrspace(1)* noalias %valptr) nounwind {
 ; SI-LABEL: v_ctlz_v2i32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s10, 0
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
-; SI-NEXT:    s_mov_b32 s11, s7
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dwordx2 v[0:1], v[0:1], s[8:11], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dwordx2 v[0:1], v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_ffbh_u32_e32 v2, v1
 ; SI-NEXT:    v_ffbh_u32_e32 v3, v0
 ; SI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v1
 ; SI-NEXT:    v_cndmask_b32_e32 v1, 32, v2, vcc
 ; SI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v0
 ; SI-NEXT:    v_cndmask_b32_e32 v0, 32, v3, vcc
-; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: v_ctlz_v2i32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dwordx2 v[0:1], v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_ffbh_u32_e32 v2, v1
 ; VI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v1
 ; VI-NEXT:    v_cndmask_b32_e32 v1, 32, v2, vcc
 ; VI-NEXT:    v_ffbh_u32_e32 v3, v0
 ; VI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v0
 ; VI-NEXT:    v_cndmask_b32_e32 v0, 32, v3, vcc
-; VI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
+; VI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
 ;
 ; EG-LABEL: v_ctlz_v2i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 2, @8, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    TEX 0 @6
 ; EG-NEXT:    ALU 6, @11, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.XY, T1.X, 1
; EG-NEXT:    32(4.484155e-44), 2(2.802597e-45)
   %ctlz = call <2 x i32> @llvm.ctlz.v2i32(<2 x i32> %val, i1 false) nounwind readnone
   store <2 x i32> %ctlz, <2 x i32> addrspace(1)* %out, align 8
   ret void
 }
 
 define amdgpu_kernel void @v_ctlz_v4i32(<4 x i32> addrspace(1)* noalias %out, <4 x i32> addrspace(1)* noalias %valptr) nounwind {
 ; SI-LABEL: v_ctlz_v4i32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s10, 0
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 4, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
-; SI-NEXT:    s_mov_b32 s11, s7
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dwordx4 v[0:3], v[0:1], s[8:11], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dwordx4 v[0:3], v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_ffbh_u32_e32 v4, v3
 ; SI-NEXT:    v_ffbh_u32_e32 v5, v2
 ; SI-NEXT:    v_ffbh_u32_e32 v6, v1
 ; SI-NEXT:    v_ffbh_u32_e32 v7, v0
 ; SI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v3
 ; SI-NEXT:    v_cndmask_b32_e32 v3, 32, v4, vcc
 ; SI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v2
 ; SI-NEXT:    v_cndmask_b32_e32 v2, 32, v5, vcc
 ; SI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v1
 ; SI-NEXT:    v_cndmask_b32_e32 v1, 32, v6, vcc
 ; SI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v0
 ; SI-NEXT:    v_cndmask_b32_e32 v0, 32, v7, vcc
-; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: v_ctlz_v4i32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 4, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dwordx4 v[0:3], v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_ffbh_u32_e32 v4, v3
 ; VI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v3
 ; VI-NEXT:    v_cndmask_b32_e32 v3, 32, v4, vcc
 ; VI-NEXT:    v_ffbh_u32_e32 v5, v2
 ; VI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v2
 ; VI-NEXT:    v_cndmask_b32_e32 v2, 32, v5, vcc
 ; VI-NEXT:    v_ffbh_u32_e32 v6, v1
 ; VI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v1
 ; VI-NEXT:    v_cndmask_b32_e32 v1, 32, v6, vcc
 ; VI-NEXT:    v_ffbh_u32_e32 v7, v0
 ; VI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v0
 ; VI-NEXT:    v_cndmask_b32_e32 v0, 32, v7, vcc
-; VI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
+; VI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
 ;
 ; EG-LABEL: v_ctlz_v4i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 2, @8, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    TEX 0 @6
 ; EG-NEXT:    ALU 12, @11, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.XYZW, T1.X, 1
; EG-NEXT:    32(4.484155e-44), 2(2.802597e-45)
   %ctlz = call <4 x i32> @llvm.ctlz.v4i32(<4 x i32> %val, i1 false) nounwind readnone
   store <4 x i32> %ctlz, <4 x i32> addrspace(1)* %out, align 16
   ret void
 }
 
 define amdgpu_kernel void @v_ctlz_i8(i8 addrspace(1)* noalias %out, i8 addrspace(1)* noalias %valptr) nounwind {
 ; SI-LABEL: v_ctlz_i8:
 ; SI:       ; %bb.0:
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
 ; SI-NEXT:    s_mov_b32 s3, 0xf000
 ; SI-NEXT:    s_mov_b32 s2, -1
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
 ; SI-NEXT:    s_mov_b32 s6, s2
 ; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
 ; SI-NEXT:    buffer_load_ubyte v0, off, s[4:7], 0
 ; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_ffbh_u32_e32 v1, v0
 ; SI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v0
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
   %trunc = trunc i64 %ctlz to i32
   store i32 %trunc, i32 addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @v_ctlz_i64(i64 addrspace(1)* noalias %out, i64 addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: v_ctlz_i64:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0x9
 ; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
 ; SI-NEXT:    s_mov_b32 s7, 0xf000
 ; SI-NEXT:    s_mov_b32 s6, 0
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
 ; SI-NEXT:    buffer_load_dwordx2 v[2:3], v[0:1], s[4:7], 0 addr64
-; SI-NEXT:    s_mov_b64 s[10:11], s[6:7]
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_ffbh_u32_e32 v4, v2
 ; SI-NEXT:    v_ffbh_u32_e32 v5, v3
 ; SI-NEXT:    v_or_b32_e32 v2, v2, v3
 ; SI-NEXT:    v_add_i32_e32 v4, vcc, 32, v4
 ; SI-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v3
 ; SI-NEXT:    v_cndmask_b32_e32 v3, v5, v4, vcc
 ; SI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v2
 ; SI-NEXT:    v_cndmask_b32_e32 v2, 64, v3, vcc
 ; SI-NEXT:    v_mov_b32_e32 v3, v1
-; SI-NEXT:    buffer_store_dwordx2 v[2:3], v[0:1], s[8:11], 0 addr64
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dwordx2 v[2:3], v[0:1], s[4:7], 0 addr64
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: v_ctlz_i64:
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x24
 ; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
 ; VI-NEXT:    v_mov_b32_e32 v5, 0
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
   %ctlz = call i64 @llvm.ctlz.i64(i64 %val, i1 false)
   store i64 %ctlz, i64 addrspace(1)* %out.gep
   ret void
 }
 
 define amdgpu_kernel void @v_ctlz_i64_trunc(i32 addrspace(1)* noalias %out, i64 addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: v_ctlz_i64_trunc:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0x9
 ; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
 ; SI-NEXT:    s_mov_b32 s7, 0xf000
 ; SI-NEXT:    s_mov_b32 s6, 0
 ; SI-NEXT:    v_lshlrev_b32_e32 v1, 3, v0
 ; SI-NEXT:    v_mov_b32_e32 v2, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
 ; SI-NEXT:    buffer_load_dwordx2 v[3:4], v[1:2], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
 ; SI-NEXT:    v_lshlrev_b32_e32 v1, 2, v0
-; SI-NEXT:    s_mov_b64 s[10:11], s[6:7]
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_ffbh_u32_e32 v0, v3
 ; SI-NEXT:    v_ffbh_u32_e32 v5, v4
 ; SI-NEXT:    v_or_b32_e32 v3, v3, v4
 ; SI-NEXT:    v_add_i32_e32 v0, vcc, 32, v0
 ; SI-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v4
 ; SI-NEXT:    v_cndmask_b32_e32 v0, v5, v0, vcc
 ; SI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v3
 ; SI-NEXT:    v_cndmask_b32_e32 v0, 64, v0, vcc
-; SI-NEXT:    buffer_store_dword v0, v[1:2], s[8:11], 0 addr64
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v0, v[1:2], s[4:7], 0 addr64
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: v_ctlz_i64_trunc:
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x24
 ; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v1, 3, v0
 ; VI-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
   %trunc = trunc i64 %ctlz to i32
   store i32 %trunc, i32 addrspace(1)* %out.gep
   ret void
 }
 
 define amdgpu_kernel void @v_ctlz_i32_sel_eq_neg1(i32 addrspace(1)* noalias %out, i32 addrspace(1)* noalias %valptr) nounwind {
 ; SI-LABEL: v_ctlz_i32_sel_eq_neg1:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s10, 0
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
-; SI-NEXT:    s_mov_b32 s11, s7
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_ffbh_u32_e32 v0, v0
-; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: v_ctlz_i32_sel_eq_neg1:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_ffbh_u32_e32 v0, v0
-; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
 ;
 ; EG-LABEL: v_ctlz_i32_sel_eq_neg1:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 2, @8, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    TEX 0 @6
 ; EG-NEXT:    ALU 5, @11, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.X, T1.X, 1
; EG-NEXT:    -1(nan), 2(2.802597e-45)
   %sel = select i1 %cmp, i32 -1, i32 %ctlz
   store i32 %sel, i32 addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @v_ctlz_i32_sel_ne_neg1(i32 addrspace(1)* noalias %out, i32 addrspace(1)* noalias %valptr) nounwind {
 ; SI-LABEL: v_ctlz_i32_sel_ne_neg1:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s10, 0
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
-; SI-NEXT:    s_mov_b32 s11, s7
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_ffbh_u32_e32 v0, v0
-; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: v_ctlz_i32_sel_ne_neg1:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_ffbh_u32_e32 v0, v0
-; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
 ;
 ; EG-LABEL: v_ctlz_i32_sel_ne_neg1:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 2, @8, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    TEX 0 @6
 ; EG-NEXT:    ALU 5, @11, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.X, T1.X, 1
; EG-NEXT:    -1(nan), 2(2.802597e-45)
   store i32 %sel, i32 addrspace(1)* %out
   ret void
 }
 
 ; TODO: Should be able to eliminate select here as well.
 define amdgpu_kernel void @v_ctlz_i32_sel_eq_bitwidth(i32 addrspace(1)* noalias %out, i32 addrspace(1)* noalias %valptr) nounwind {
 ; SI-LABEL: v_ctlz_i32_sel_eq_bitwidth:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s10, 0
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
-; SI-NEXT:    s_mov_b32 s11, s7
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_ffbh_u32_e32 v1, v0
 ; SI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v0
 ; SI-NEXT:    v_cndmask_b32_e32 v0, 32, v1, vcc
 ; SI-NEXT:    v_cmp_ne_u32_e32 vcc, 32, v0
 ; SI-NEXT:    v_cndmask_b32_e32 v0, -1, v0, vcc
-; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: v_ctlz_i32_sel_eq_bitwidth:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_ffbh_u32_e32 v1, v0
 ; VI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v0
 ; VI-NEXT:    v_cndmask_b32_e32 v0, 32, v1, vcc
 ; VI-NEXT:    v_cmp_ne_u32_e32 vcc, 32, v0
 ; VI-NEXT:    v_cndmask_b32_e32 v0, -1, v0, vcc
-; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
 ;
 ; EG-LABEL: v_ctlz_i32_sel_eq_bitwidth:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 2, @8, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    TEX 0 @6
 ; EG-NEXT:    ALU 7, @11, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.X, T1.X, 1
; EG-NEXT:    -1(nan), 2(2.802597e-45)
   %sel = select i1 %cmp, i32 -1, i32 %ctlz
   store i32 %sel, i32 addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @v_ctlz_i32_sel_ne_bitwidth(i32 addrspace(1)* noalias %out, i32 addrspace(1)* noalias %valptr) nounwind {
 ; SI-LABEL: v_ctlz_i32_sel_ne_bitwidth:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s10, 0
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
-; SI-NEXT:    s_mov_b32 s11, s7
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_ffbh_u32_e32 v1, v0
 ; SI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v0
 ; SI-NEXT:    v_cndmask_b32_e32 v0, 32, v1, vcc
 ; SI-NEXT:    v_cmp_ne_u32_e32 vcc, 32, v0
 ; SI-NEXT:    v_cndmask_b32_e32 v0, -1, v0, vcc
-; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: v_ctlz_i32_sel_ne_bitwidth:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_ffbh_u32_e32 v1, v0
 ; VI-NEXT:    v_cmp_ne_u32_e32 vcc, 0, v0
 ; VI-NEXT:    v_cndmask_b32_e32 v0, 32, v1, vcc
 ; VI-NEXT:    v_cmp_ne_u32_e32 vcc, 32, v0
 ; VI-NEXT:    v_cndmask_b32_e32 v0, -1, v0, vcc
-; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
 ;
 ; EG-LABEL: v_ctlz_i32_sel_ne_bitwidth:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 2, @8, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    TEX 0 @6
 ; EG-NEXT:    ALU 7, @11, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.X, T1.X, 1
; EG-NEXT:    -1(nan), 2(2.802597e-45)
   %sel = select i1 %cmp, i32 %ctlz, i32 -1
   store i32 %sel, i32 addrspace(1)* %out
   ret void
 }
 
  define amdgpu_kernel void @v_ctlz_i8_sel_eq_neg1(i8 addrspace(1)* noalias %out, i8 addrspace(1)* noalias %valptr) nounwind {
 ; SI-LABEL: v_ctlz_i8_sel_eq_neg1:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
-; SI-NEXT:    s_mov_b32 s10, 0
-; SI-NEXT:    s_mov_b32 s11, s7
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_ubyte v0, v[0:1], s[8:11], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_ffbh_u32_e32 v0, v0
-; SI-NEXT:    buffer_store_byte v0, off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_byte v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: v_ctlz_i8_sel_eq_neg1:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_ubyte v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_ffbh_u32_e32 v0, v0
-; VI-NEXT:    buffer_store_byte v0, off, s[4:7], 0
+; VI-NEXT:    buffer_store_byte v0, off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
 ;
 ; EG-LABEL: v_ctlz_i8_sel_eq_neg1:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 0, @8, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    TEX 0 @6
 ; EG-NEXT:    ALU 12, @9, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    MEM_RAT MSKOR T0.XW, T1.X
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
   %sel = select i1 %cmp, i8 -1, i8 %ctlz
   store i8 %sel, i8 addrspace(1)* %out
   ret void
 }
 
  define amdgpu_kernel void @v_ctlz_i16_sel_eq_neg1(i16 addrspace(1)* noalias %out, i16 addrspace(1)* noalias %valptr) nounwind {
 ; SI-LABEL: v_ctlz_i16_sel_eq_neg1:
 ; SI:       ; %bb.0:
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
 ; SI-NEXT:    s_mov_b32 s3, 0xf000
 ; SI-NEXT:    s_mov_b32 s2, -1
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
 ; SI-NEXT:    s_mov_b32 s6, s2
 ; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
 ; SI-NEXT:    buffer_load_ushort v0, off, s[4:7], 0
 ; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_ffbh_u32_e32 v0, v0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
   store i16 %sel, i16 addrspace(1)* %out
   ret void
 }
 
 ; FIXME: Need to handle non-uniform case for function below (load without gep).
 define amdgpu_kernel void @v_ctlz_i7_sel_eq_neg1(i7 addrspace(1)* noalias %out, i7 addrspace(1)* noalias %valptr) nounwind {
 ; SI-LABEL: v_ctlz_i7_sel_eq_neg1:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
-; SI-NEXT:    s_mov_b32 s10, 0
-; SI-NEXT:    s_mov_b32 s11, s7
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_ubyte v0, v[0:1], s[8:11], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_ffbh_u32_e32 v0, v0
 ; SI-NEXT:    v_and_b32_e32 v0, 0x7f, v0
-; SI-NEXT:    buffer_store_byte v0, off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_byte v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: v_ctlz_i7_sel_eq_neg1:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_ubyte v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_ffbh_u32_e32 v0, v0
 ; VI-NEXT:    v_and_b32_e32 v0, 0x7f, v0
-; VI-NEXT:    buffer_store_byte v0, off, s[4:7], 0
+; VI-NEXT:    buffer_store_byte v0, off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
 ;
 ; EG-LABEL: v_ctlz_i7_sel_eq_neg1:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 0, @8, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    TEX 0 @6
 ; EG-NEXT:    ALU 12, @9, KC0[CB0:0-32], KC1[]
 ; EG-NEXT:    MEM_RAT MSKOR T0.XW, T1.X

llvm/test/CodeGen/AMDGPU/ctpop64.ll

define amdgpu_kernel void @v_ctpop_i64(i32 addrspace(1)* noalias %out, i64 addrspace(1)* noalias %in) nounwind {
   %ctpop = call i64 @llvm.ctpop.i64(i64 %val) nounwind readnone
   %truncctpop = trunc i64 %ctpop to i32
   store i32 %truncctpop, i32 addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v_ctpop_i64_user:
 ; GCN: {{buffer|flat}}_load_dwordx2 v{{\[}}[[LOVAL:[0-9]+]]:[[HIVAL:[0-9]+]]{{\]}},
-; GCN: v_bcnt_u32_b32{{(_e64)*}} [[MIDRESULT:v[0-9]+]], v[[LOVAL]], 0
-; SI-NEXT: v_bcnt_u32_b32_e32 [[RESULT:v[0-9]+]], v[[HIVAL]], [[MIDRESULT]]
-; VI-NEXT: v_bcnt_u32_b32 [[RESULT:v[0-9]+]], v[[HIVAL]], [[MIDRESULT]]
+; GCN-DAG: v_bcnt_u32_b32{{(_e64)*}} [[MIDRESULT:v[0-9]+]], v[[LOVAL]], 0
+; SI-DAG: v_bcnt_u32_b32_e32 [[RESULT:v[0-9]+]], v[[HIVAL]], [[MIDRESULT]]
+; VI-DAG: v_bcnt_u32_b32 [[RESULT:v[0-9]+]], v[[HIVAL]], [[MIDRESULT]]
 ; GCN-DAG: v_or_b32_e32 v[[RESULT_LO:[0-9]+]], s{{[0-9]+}}, [[RESULT]]
 ; GCN-DAG: v_mov_b32_e32 v[[RESULT_HI:[0-9]+]], s{{[0-9]+}}
 ; GCN: buffer_store_dwordx2 v{{\[}}[[RESULT_LO]]:[[RESULT_HI]]{{\]}}
 ; GCN: s_endpgm
 define amdgpu_kernel void @v_ctpop_i64_user(i64 addrspace(1)* noalias %out, i64 addrspace(1)* noalias %in, i64 %s.val) nounwind {
   %tid = call i32 @llvm.r600.read.tidig.x()
   %in.gep = getelementptr i64, i64 addrspace(1)* %in, i32 %tid
   %val = load i64, i64 addrspace(1)* %in.gep, align 8

llvm/test/CodeGen/AMDGPU/cvt_f32_ubyte.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=amdgcn-- -mcpu=tahiti -verify-machineinstrs < %s | FileCheck %s -allow-deprecated-dag-overlap -check-prefixes=GCN,SI
 ; RUN: llc -mtriple=amdgcn-- -mcpu=tonga -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck %s -allow-deprecated-dag-overlap -check-prefixes=GCN,VI
 
 declare i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
 declare i32 @llvm.amdgcn.workitem.id.y() nounwind readnone
 
 define amdgpu_kernel void @load_i8_to_f32(float addrspace(1)* noalias %out, i8 addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: load_i8_to_f32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_ubyte v0, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: load_i8_to_f32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_ubyte v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr i8, i8 addrspace(1)* %in, i32 %tid
   %load = load i8, i8 addrspace(1)* %gep, align 1
   %cvt = uitofp i8 %load to float
   store float %cvt, float addrspace(1)* %out, align 4
   ret void
 }
 
 define amdgpu_kernel void @load_v2i8_to_v2f32(<2 x float> addrspace(1)* noalias %out, <2 x i8> addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: load_v2i8_to_v2f32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 1, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_ushort v0, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_ushort v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_cvt_f32_ubyte1_e32 v1, v0
 ; SI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: load_v2i8_to_v2f32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 1, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_ushort v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_cvt_f32_ubyte1_e32 v1, v0
 ; VI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; VI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
+; VI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr <2 x i8>, <2 x i8> addrspace(1)* %in, i32 %tid
   %load = load <2 x i8>, <2 x i8> addrspace(1)* %gep, align 2
   %cvt = uitofp <2 x i8> %load to <2 x float>
   store <2 x float> %cvt, <2 x float> addrspace(1)* %out, align 16
   ret void
 }
 
 define amdgpu_kernel void @load_v3i8_to_v3f32(<3 x float> addrspace(1)* noalias %out, <3 x i8> addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: load_v3i8_to_v3f32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v2, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_cvt_f32_ubyte1_e32 v1, v2
 ; SI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v2
 ; SI-NEXT:    v_cvt_f32_ubyte2_e32 v2, v2
-; SI-NEXT:    buffer_store_dword v2, off, s[4:7], 0 offset:8
-; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v2, off, s[0:3], 0 offset:8
+; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: load_v3i8_to_v3f32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_cvt_f32_ubyte2_e32 v2, v0
 ; VI-NEXT:    v_cvt_f32_ubyte1_e32 v1, v0
 ; VI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; VI-NEXT:    buffer_store_dwordx3 v[0:2], off, s[4:7], 0
+; VI-NEXT:    buffer_store_dwordx3 v[0:2], off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr <3 x i8>, <3 x i8> addrspace(1)* %in, i32 %tid
   %load = load <3 x i8>, <3 x i8> addrspace(1)* %gep, align 4
   %cvt = uitofp <3 x i8> %load to <3 x float>
   store <3 x float> %cvt, <3 x float> addrspace(1)* %out, align 16
   ret void
 }
 
 define amdgpu_kernel void @load_v4i8_to_v4f32(<4 x float> addrspace(1)* noalias %out, <4 x i8> addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: load_v4i8_to_v4f32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_cvt_f32_ubyte3_e32 v3, v0
 ; SI-NEXT:    v_cvt_f32_ubyte2_e32 v2, v0
 ; SI-NEXT:    v_cvt_f32_ubyte1_e32 v1, v0
 ; SI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: load_v4i8_to_v4f32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_cvt_f32_ubyte3_e32 v3, v0
 ; VI-NEXT:    v_cvt_f32_ubyte2_e32 v2, v0
 ; VI-NEXT:    v_cvt_f32_ubyte1_e32 v1, v0
 ; VI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; VI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
+; VI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %in, i32 %tid
   %load = load <4 x i8>, <4 x i8> addrspace(1)* %gep, align 4
   %cvt = uitofp <4 x i8> %load to <4 x float>
   store <4 x float> %cvt, <4 x float> addrspace(1)* %out, align 16
   ret void
 }
 
 ; This should not be adding instructions to shift into the correct
 ; position in the word for the component.
 
 ; FIXME: Packing bytes
 define amdgpu_kernel void @load_v4i8_to_v4f32_unaligned(<4 x float> addrspace(1)* noalias %out, <4 x i8> addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: load_v4i8_to_v4f32_unaligned:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_ubyte v2, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    buffer_load_ubyte v3, v[0:1], s[0:3], 0 addr64 offset:1
-; SI-NEXT:    buffer_load_ubyte v4, v[0:1], s[0:3], 0 addr64 offset:2
-; SI-NEXT:    buffer_load_ubyte v0, v[0:1], s[0:3], 0 addr64 offset:3
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_ubyte v2, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    buffer_load_ubyte v3, v[0:1], s[4:7], 0 addr64 offset:1
+; SI-NEXT:    buffer_load_ubyte v4, v[0:1], s[4:7], 0 addr64 offset:2
+; SI-NEXT:    buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:3
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(2)
 ; SI-NEXT:    v_lshlrev_b32_e32 v1, 8, v3
 ; SI-NEXT:    v_or_b32_e32 v1, v1, v2
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 8, v0
 ; SI-NEXT:    v_or_b32_e32 v0, v0, v4
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 16, v0
 ; SI-NEXT:    v_or_b32_e32 v0, v0, v1
 ; SI-NEXT:    v_cvt_f32_ubyte3_e32 v3, v0
 ; SI-NEXT:    v_cvt_f32_ubyte2_e32 v2, v0
 ; SI-NEXT:    v_cvt_f32_ubyte1_e32 v1, v0
 ; SI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: load_v4i8_to_v4f32_unaligned:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    v_add_u32_e32 v2, vcc, 3, v0
 ; VI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v1, vcc
 ; VI-NEXT:    v_add_u32_e32 v4, vcc, 2, v0
 ; VI-NEXT:    v_addc_u32_e32 v5, vcc, 0, v1, vcc
 ; VI-NEXT:    v_add_u32_e32 v6, vcc, 1, v0
 ; VI-NEXT:    v_addc_u32_e32 v7, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_ubyte v2, v[2:3]
 ; VI-NEXT:    flat_load_ubyte v3, v[4:5]
 ; VI-NEXT:    flat_load_ubyte v4, v[6:7]
 ; VI-NEXT:    flat_load_ubyte v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(3) lgkmcnt(3)
 ; VI-NEXT:    v_lshlrev_b32_e32 v1, 8, v2
 ; VI-NEXT:    s_waitcnt vmcnt(2) lgkmcnt(2)
 ; VI-NEXT:    v_or_b32_e32 v1, v1, v3
 ; VI-NEXT:    s_waitcnt vmcnt(1) lgkmcnt(1)
 ; VI-NEXT:    v_lshlrev_b32_e32 v2, 8, v4
 ; VI-NEXT:    v_lshlrev_b32_e32 v3, 16, v1
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_or_b32_e32 v0, v2, v0
 ; VI-NEXT:    v_cvt_f32_ubyte0_e32 v2, v1
 ; VI-NEXT:    v_cvt_f32_ubyte1_e32 v1, v0
 ; VI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
 ; VI-NEXT:    v_cvt_f32_ubyte3_e32 v3, v3
-; VI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
+; VI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %in, i32 %tid
   %load = load <4 x i8>, <4 x i8> addrspace(1)* %gep, align 1
   %cvt = uitofp <4 x i8> %load to <4 x float>
   store <4 x float> %cvt, <4 x float> addrspace(1)* %out, align 16
   ret void
 }
; VI-NEXT:    s_endpgm
   store <4 x i8> %add, <4 x i8> addrspace(1)* %out2, align 4
   ret void
 }
 
 ; Make sure this doesn't crash.
 define amdgpu_kernel void @load_v7i8_to_v7f32(<7 x float> addrspace(1)* noalias %out, <7 x i8> addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: load_v7i8_to_v7f32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_ubyte v2, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    buffer_load_ubyte v3, v[0:1], s[0:3], 0 addr64 offset:1
-; SI-NEXT:    buffer_load_ubyte v4, v[0:1], s[0:3], 0 addr64 offset:2
-; SI-NEXT:    buffer_load_ubyte v5, v[0:1], s[0:3], 0 addr64 offset:3
-; SI-NEXT:    buffer_load_ubyte v6, v[0:1], s[0:3], 0 addr64 offset:4
-; SI-NEXT:    buffer_load_ubyte v7, v[0:1], s[0:3], 0 addr64 offset:5
-; SI-NEXT:    buffer_load_ubyte v0, v[0:1], s[0:3], 0 addr64 offset:6
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_ubyte v2, v[0:1], s[4:7], 0 addr64 offset:4
+; SI-NEXT:    buffer_load_ubyte v3, v[0:1], s[4:7], 0 addr64 offset:5
+; SI-NEXT:    buffer_load_ubyte v4, v[0:1], s[4:7], 0 addr64 offset:6
+; SI-NEXT:    buffer_load_ubyte v5, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    buffer_load_ubyte v6, v[0:1], s[4:7], 0 addr64 offset:1
+; SI-NEXT:    buffer_load_ubyte v7, v[0:1], s[4:7], 0 addr64 offset:2
+; SI-NEXT:    buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:3
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(5)
-; SI-NEXT:    v_lshlrev_b32_e32 v1, 8, v3
-; SI-NEXT:    v_or_b32_e32 v1, v1, v2
-; SI-NEXT:    s_waitcnt vmcnt(3)
-; SI-NEXT:    v_lshlrev_b32_e32 v2, 8, v5
-; SI-NEXT:    v_or_b32_e32 v2, v2, v4
-; SI-NEXT:    s_waitcnt vmcnt(1)
-; SI-NEXT:    v_lshlrev_b32_e32 v3, 8, v7
+; SI-NEXT:    v_lshlrev_b32_e32 v3, 8, v3
+; SI-NEXT:    s_waitcnt vmcnt(4)
+; SI-NEXT:    v_cvt_f32_ubyte0_e32 v4, v4
+; SI-NEXT:    v_or_b32_e32 v2, v3, v2
+; SI-NEXT:    s_waitcnt vmcnt(2)
+; SI-NEXT:    v_lshlrev_b32_e32 v1, 8, v6
+; SI-NEXT:    v_or_b32_e32 v1, v1, v5
 ; SI-NEXT:    s_waitcnt vmcnt(0)
-; SI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0 offset:24
-; SI-NEXT:    s_waitcnt expcnt(0)
-; SI-NEXT:    v_lshlrev_b32_e32 v0, 16, v2
+; SI-NEXT:    v_lshlrev_b32_e32 v0, 8, v0
+; SI-NEXT:    v_or_b32_e32 v0, v0, v7
+; SI-NEXT:    v_lshlrev_b32_e32 v0, 16, v0
 ; SI-NEXT:    v_or_b32_e32 v0, v0, v1
-; SI-NEXT:    v_or_b32_e32 v3, v3, v6
-; SI-NEXT:    v_cvt_f32_ubyte1_e32 v5, v3
-; SI-NEXT:    v_cvt_f32_ubyte0_e32 v4, v3
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v4, off, s[0:3], 0 offset:24
+; SI-NEXT:    v_cvt_f32_ubyte1_e32 v5, v2
+; SI-NEXT:    s_waitcnt expcnt(0)
+; SI-NEXT:    v_cvt_f32_ubyte0_e32 v4, v2
 ; SI-NEXT:    v_cvt_f32_ubyte3_e32 v3, v0
 ; SI-NEXT:    v_cvt_f32_ubyte2_e32 v2, v0
 ; SI-NEXT:    v_cvt_f32_ubyte1_e32 v1, v0
 ; SI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; SI-NEXT:    buffer_store_dwordx2 v[4:5], off, s[4:7], 0 offset:16
-; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
+; SI-NEXT:    buffer_store_dwordx2 v[4:5], off, s[0:3], 0 offset:16
+; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: load_v7i8_to_v7f32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    v_add_u32_e32 v2, vcc, 1, v0
 ; VI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v1, vcc
 ; VI-NEXT:    v_add_u32_e32 v4, vcc, 3, v0
 ; VI-NEXT:    v_addc_u32_e32 v5, vcc, 0, v1, vcc
+; VI-NEXT:    v_add_u32_e32 v6, vcc, 2, v0
+; VI-NEXT:    v_addc_u32_e32 v7, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_ubyte v9, v[2:3]
 ; VI-NEXT:    flat_load_ubyte v10, v[4:5]
-; VI-NEXT:    v_add_u32_e32 v2, vcc, 2, v0
+; VI-NEXT:    flat_load_ubyte v7, v[6:7]
+; VI-NEXT:    v_add_u32_e32 v2, vcc, 5, v0
 ; VI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v1, vcc
-; VI-NEXT:    v_add_u32_e32 v4, vcc, 5, v0
+; VI-NEXT:    v_add_u32_e32 v4, vcc, 4, v0
 ; VI-NEXT:    v_addc_u32_e32 v5, vcc, 0, v1, vcc
-; VI-NEXT:    v_add_u32_e32 v6, vcc, 4, v0
-; VI-NEXT:    v_addc_u32_e32 v7, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_ubyte v8, v[0:1]
 ; VI-NEXT:    v_add_u32_e32 v0, vcc, 6, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_ubyte v2, v[2:3]
 ; VI-NEXT:    flat_load_ubyte v3, v[4:5]
-; VI-NEXT:    flat_load_ubyte v4, v[6:7]
 ; VI-NEXT:    flat_load_ubyte v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(6) lgkmcnt(6)
 ; VI-NEXT:    v_lshlrev_b32_e32 v1, 8, v9
 ; VI-NEXT:    s_waitcnt vmcnt(5) lgkmcnt(5)
-; VI-NEXT:    v_lshlrev_b32_e32 v5, 8, v10
+; VI-NEXT:    v_lshlrev_b32_e32 v4, 8, v10
 ; VI-NEXT:    s_waitcnt vmcnt(2) lgkmcnt(2)
-; VI-NEXT:    v_lshlrev_b32_e32 v3, 8, v3
+; VI-NEXT:    v_lshlrev_b32_e32 v2, 8, v2
+; VI-NEXT:    s_waitcnt vmcnt(1) lgkmcnt(1)
+; VI-NEXT:    v_or_b32_e32 v2, v2, v3
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_cvt_f32_ubyte0_e32 v6, v0
 ; VI-NEXT:    v_or_b32_e32 v0, v1, v8
-; VI-NEXT:    v_or_b32_sdwa v1, v5, v2 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
-; VI-NEXT:    v_or_b32_e32 v2, v3, v4
+; VI-NEXT:    v_or_b32_sdwa v1, v4, v7 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI-NEXT:    v_or_b32_e32 v0, v1, v0
 ; VI-NEXT:    v_cvt_f32_ubyte1_e32 v5, v2
 ; VI-NEXT:    v_cvt_f32_ubyte0_e32 v4, v2
 ; VI-NEXT:    v_cvt_f32_ubyte3_e32 v3, v0
 ; VI-NEXT:    v_cvt_f32_ubyte2_e32 v2, v0
 ; VI-NEXT:    v_cvt_f32_ubyte1_e32 v1, v0
 ; VI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; VI-NEXT:    buffer_store_dwordx3 v[4:6], off, s[4:7], 0 offset:16
-; VI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
+; VI-NEXT:    buffer_store_dwordx3 v[4:6], off, s[0:3], 0 offset:16
+; VI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr <7 x i8>, <7 x i8> addrspace(1)* %in, i32 %tid
   %load = load <7 x i8>, <7 x i8> addrspace(1)* %gep, align 1
   %cvt = uitofp <7 x i8> %load to <7 x float>
   store <7 x float> %cvt, <7 x float> addrspace(1)* %out, align 16
   ret void
 }
 
 define amdgpu_kernel void @load_v8i8_to_v8f32(<8 x float> addrspace(1)* noalias %out, <8 x i8> addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: load_v8i8_to_v8f32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dwordx2 v[7:8], v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dwordx2 v[7:8], v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_cvt_f32_ubyte3_e32 v3, v7
 ; SI-NEXT:    v_cvt_f32_ubyte2_e32 v2, v7
 ; SI-NEXT:    v_cvt_f32_ubyte1_e32 v1, v7
 ; SI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v7
 ; SI-NEXT:    v_cvt_f32_ubyte3_e32 v7, v8
 ; SI-NEXT:    v_cvt_f32_ubyte2_e32 v6, v8
 ; SI-NEXT:    v_cvt_f32_ubyte1_e32 v5, v8
 ; SI-NEXT:    v_cvt_f32_ubyte0_e32 v4, v8
-; SI-NEXT:    buffer_store_dwordx4 v[4:7], off, s[4:7], 0 offset:16
-; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dwordx4 v[4:7], off, s[0:3], 0 offset:16
+; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: load_v8i8_to_v8f32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dwordx2 v[7:8], v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_cvt_f32_ubyte3_e32 v3, v7
 ; VI-NEXT:    v_cvt_f32_ubyte2_e32 v2, v7
 ; VI-NEXT:    v_cvt_f32_ubyte1_e32 v1, v7
 ; VI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v7
 ; VI-NEXT:    v_cvt_f32_ubyte3_e32 v7, v8
 ; VI-NEXT:    v_cvt_f32_ubyte2_e32 v6, v8
 ; VI-NEXT:    v_cvt_f32_ubyte1_e32 v5, v8
 ; VI-NEXT:    v_cvt_f32_ubyte0_e32 v4, v8
-; VI-NEXT:    buffer_store_dwordx4 v[4:7], off, s[4:7], 0 offset:16
-; VI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
+; VI-NEXT:    buffer_store_dwordx4 v[4:7], off, s[0:3], 0 offset:16
+; VI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr <8 x i8>, <8 x i8> addrspace(1)* %in, i32 %tid
   %load = load <8 x i8>, <8 x i8> addrspace(1)* %gep, align 8
   %cvt = uitofp <8 x i8> %load to <8 x float>
   store <8 x float> %cvt, <8 x float> addrspace(1)* %out, align 16
   ret void
 }
 
 define amdgpu_kernel void @i8_zext_inreg_i32_to_f32(float addrspace(1)* noalias %out, i32 addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: i8_zext_inreg_i32_to_f32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_add_i32_e32 v0, vcc, 2, v0
 ; SI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: i8_zext_inreg_i32_to_f32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_add_u32_e32 v0, vcc, 2, v0
 ; VI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr i32, i32 addrspace(1)* %in, i32 %tid
   %load = load i32, i32 addrspace(1)* %gep, align 4
   %add = add i32 %load, 2
   %inreg = and i32 %add, 255
   %cvt = uitofp i32 %inreg to float
   store float %cvt, float addrspace(1)* %out, align 4
   ret void
 }
 
 define amdgpu_kernel void @i8_zext_inreg_hi1_to_f32(float addrspace(1)* noalias %out, i32 addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: i8_zext_inreg_hi1_to_f32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_cvt_f32_ubyte1_e32 v0, v0
-; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: i8_zext_inreg_hi1_to_f32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_cvt_f32_ubyte1_e32 v0, v0
-; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr i32, i32 addrspace(1)* %in, i32 %tid
   %load = load i32, i32 addrspace(1)* %gep, align 4
   %inreg = and i32 %load, 65280
   %shr = lshr i32 %inreg, 8
   %cvt = uitofp i32 %shr to float
   store float %cvt, float addrspace(1)* %out, align 4
   ret void
 }
 
 ; We don't get these ones because of the zext, but instcombine removes
 ; them so it shouldn't really matter.
 define amdgpu_kernel void @i8_zext_i32_to_f32(float addrspace(1)* noalias %out, i8 addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: i8_zext_i32_to_f32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_ubyte v0, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: i8_zext_i32_to_f32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_ubyte v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr i8, i8 addrspace(1)* %in, i32 %tid
   %load = load i8, i8 addrspace(1)* %gep, align 1
   %ext = zext i8 %load to i32
   %cvt = uitofp i32 %ext to float
   store float %cvt, float addrspace(1)* %out, align 4
   ret void
 }
 
 define amdgpu_kernel void @v4i8_zext_v4i32_to_v4f32(<4 x float> addrspace(1)* noalias %out, <4 x i8> addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: v4i8_zext_v4i32_to_v4f32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_ubyte v2, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    buffer_load_ubyte v3, v[0:1], s[0:3], 0 addr64 offset:1
-; SI-NEXT:    buffer_load_ubyte v4, v[0:1], s[0:3], 0 addr64 offset:2
-; SI-NEXT:    buffer_load_ubyte v0, v[0:1], s[0:3], 0 addr64 offset:3
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_ubyte v2, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    buffer_load_ubyte v3, v[0:1], s[4:7], 0 addr64 offset:1
+; SI-NEXT:    buffer_load_ubyte v4, v[0:1], s[4:7], 0 addr64 offset:2
+; SI-NEXT:    buffer_load_ubyte v0, v[0:1], s[4:7], 0 addr64 offset:3
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(2)
 ; SI-NEXT:    v_lshlrev_b32_e32 v1, 8, v3
 ; SI-NEXT:    v_or_b32_e32 v1, v1, v2
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 8, v0
 ; SI-NEXT:    v_or_b32_e32 v0, v0, v4
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 16, v0
 ; SI-NEXT:    v_or_b32_e32 v0, v0, v1
 ; SI-NEXT:    v_cvt_f32_ubyte3_e32 v3, v0
 ; SI-NEXT:    v_cvt_f32_ubyte2_e32 v2, v0
 ; SI-NEXT:    v_cvt_f32_ubyte1_e32 v1, v0
 ; SI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: v4i8_zext_v4i32_to_v4f32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    v_add_u32_e32 v2, vcc, 1, v0
 ; VI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v1, vcc
 ; VI-NEXT:    v_add_u32_e32 v4, vcc, 3, v0
 ; VI-NEXT:    v_addc_u32_e32 v5, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_ubyte v6, v[0:1]
 ; VI-NEXT:    v_add_u32_e32 v0, vcc, 2, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_ubyte v2, v[2:3]
 ; VI-NEXT:    flat_load_ubyte v3, v[4:5]
 ; VI-NEXT:    flat_load_ubyte v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(2) lgkmcnt(2)
 ; VI-NEXT:    v_lshlrev_b32_e32 v1, 8, v2
 ; VI-NEXT:    s_waitcnt vmcnt(1) lgkmcnt(1)
 ; VI-NEXT:    v_lshlrev_b32_e32 v2, 8, v3
 ; VI-NEXT:    v_or_b32_e32 v1, v1, v6
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_or_b32_sdwa v0, v2, v0 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI-NEXT:    v_or_b32_e32 v0, v0, v1
 ; VI-NEXT:    v_cvt_f32_ubyte3_e32 v3, v0
 ; VI-NEXT:    v_cvt_f32_ubyte2_e32 v2, v0
 ; VI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
 ; VI-NEXT:    v_cvt_f32_ubyte1_e32 v1, v1
-; VI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
+; VI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr <4 x i8>, <4 x i8> addrspace(1)* %in, i32 %tid
   %load = load <4 x i8>, <4 x i8> addrspace(1)* %gep, align 1
   %ext = zext <4 x i8> %load to <4 x i32>
   %cvt = uitofp <4 x i32> %ext to <4 x float>
   store <4 x float> %cvt, <4 x float> addrspace(1)* %out, align 16
   ret void
 }
 
 define amdgpu_kernel void @extract_byte0_to_f32(float addrspace(1)* noalias %out, i32 addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: extract_byte0_to_f32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: extract_byte0_to_f32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_cvt_f32_ubyte0_e32 v0, v0
-; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr i32, i32 addrspace(1)* %in, i32 %tid
   %val = load i32, i32 addrspace(1)* %gep
   %and = and i32 %val, 255
   %cvt = uitofp i32 %and to float
   store float %cvt, float addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @extract_byte1_to_f32(float addrspace(1)* noalias %out, i32 addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: extract_byte1_to_f32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_cvt_f32_ubyte1_e32 v0, v0
-; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: extract_byte1_to_f32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_cvt_f32_ubyte1_e32 v0, v0
-; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr i32, i32 addrspace(1)* %in, i32 %tid
   %val = load i32, i32 addrspace(1)* %gep
   %srl = lshr i32 %val, 8
   %and = and i32 %srl, 255
   %cvt = uitofp i32 %and to float
   store float %cvt, float addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @extract_byte2_to_f32(float addrspace(1)* noalias %out, i32 addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: extract_byte2_to_f32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_cvt_f32_ubyte2_e32 v0, v0
-; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: extract_byte2_to_f32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_cvt_f32_ubyte2_e32 v0, v0
-; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr i32, i32 addrspace(1)* %in, i32 %tid
   %val = load i32, i32 addrspace(1)* %gep
   %srl = lshr i32 %val, 16
   %and = and i32 %srl, 255
   %cvt = uitofp i32 %and to float
   store float %cvt, float addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @extract_byte3_to_f32(float addrspace(1)* noalias %out, i32 addrspace(1)* noalias %in) nounwind {
 ; SI-LABEL: extract_byte3_to_f32:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xb
-; SI-NEXT:    s_mov_b32 s7, 0xf000
-; SI-NEXT:    s_mov_b32 s2, 0
-; SI-NEXT:    s_mov_b32 s3, s7
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xb
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
-; SI-NEXT:    s_mov_b32 s6, -1
+; SI-NEXT:    buffer_load_dword v0, v[0:1], s[4:7], 0 addr64
+; SI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_waitcnt vmcnt(0)
 ; SI-NEXT:    v_cvt_f32_ubyte3_e32 v0, v0
-; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: extract_byte3_to_f32:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
-; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x2c
+; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x2c
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; VI-NEXT:    s_mov_b32 s7, 0xf000
-; VI-NEXT:    s_mov_b32 s6, -1
+; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; VI-NEXT:    s_mov_b32 s3, 0xf000
+; VI-NEXT:    s_mov_b32 s2, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
+; VI-NEXT:    v_mov_b32_e32 v1, s5
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v0
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v0, v[0:1]
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    v_cvt_f32_ubyte3_e32 v0, v0
-; VI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; VI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; VI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep = getelementptr i32, i32 addrspace(1)* %in, i32 %tid
   %val = load i32, i32 addrspace(1)* %gep
   %srl = lshr i32 %val, 24
   %and = and i32 %srl, 255
   %cvt = uitofp i32 %and to float
   store float %cvt, float addrspace(1)* %out

llvm/test/CodeGen/AMDGPU/fneg-combines.ll

define amdgpu_kernel void @free_fold_src_code_size_cost_use_f32(float addrspace(1)* %out, float addrspace(1)* %a.ptr, float addrspace(1)* %b.ptr, float addrspace(1)* %c.ptr, float addrspace(1)* %d.ptr) #0 {
   store volatile float %mul1, float addrspace(1)* %out
   store volatile float %mul2, float addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: {{^}}free_fold_src_code_size_cost_use_f64:
 ; GCN: {{buffer|flat}}_load_dwordx2 [[A:v\[[0-9]+:[0-9]+\]]]
 ; GCN: {{buffer|flat}}_load_dwordx2 [[B:v\[[0-9]+:[0-9]+\]]]
-; GCN: {{buffer|flat}}_load_dwordx2 [[C:v\[[0-9]+:[0-9]+\]]]
-; GCN: {{buffer|flat}}_load_dwordx2 [[D:v\[[0-9]+:[0-9]+\]]]
+; GCN-DAG: {{buffer|flat}}_load_dwordx2 [[C:v\[[0-9]+:[0-9]+\]]]
+; GCN-DAG: {{buffer|flat}}_load_dwordx2 [[D:v\[[0-9]+:[0-9]+\]]]
 
-; GCN: v_fma_f64 [[FMA0:v\[[0-9]+:[0-9]+\]]], [[A]], [[B]], 2.0
+; GCN-DAG: v_fma_f64 [[FMA0:v\[[0-9]+:[0-9]+\]]], [[A]], [[B]], 2.0
 ; GCN-DAG: v_mul_f64 [[MUL0:v\[[0-9]+:[0-9]+\]]], -[[FMA0]], [[C]]
 ; GCN-DAG: v_mul_f64 [[MUL1:v\[[0-9]+:[0-9]+\]]], -[[FMA0]], [[D]]
 
 ; GCN: flat_store_dwordx2 v{{\[[0-9]+:[0-9]+\]}}, [[MUL0]]
 ; GCN-NEXT: flat_store_dwordx2 v{{\[[0-9]+:[0-9]+\]}}, [[MUL1]]
 define amdgpu_kernel void @free_fold_src_code_size_cost_use_f64(double addrspace(1)* %out, double addrspace(1)* %a.ptr, double addrspace(1)* %b.ptr, double addrspace(1)* %c.ptr, double addrspace(1)* %d.ptr) #0 {
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64

llvm/test/CodeGen/AMDGPU/idot2.ll

 ; GFX9-DL-NEXT:    v_dot2_u32_u16 v2, s4, v0, v1
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX9-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot2:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    s_load_dword s2, s[0:1], 0x0
-; GFX10-DL-NEXT:    s_load_dword s3, s[4:5], 0x0
-; GFX10-DL-NEXT:    s_load_dword s4, s[6:7], 0x0
+; GFX10-DL-NEXT:    s_load_dword s6, s[4:5], 0x0
+; GFX10-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX10-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
-; GFX10-DL-NEXT:    v_dot2_u32_u16 v2, s4, s3, v0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s6
+; GFX10-DL-NEXT:    v_dot2_u32_u16 v2, s1, s0, v0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX10-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX10-DL-NEXT:    s_endpgm
                                  <2 x i16> addrspace(1)* %src2,
                                  i32 addrspace(1)* nocapture %dst) {
 entry:
   %vec1 = load <2 x i16>, <2 x i16> addrspace(1)* %src1
   %vec2 = load <2 x i16>, <2 x i16> addrspace(1)* %src2
 

 ; GFX9-DL-NEXT:    v_dot2_i32_i16 v2, s4, v0, v1
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX9-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: idot2:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    s_load_dword s2, s[0:1], 0x0
-; GFX10-DL-NEXT:    s_load_dword s3, s[4:5], 0x0
-; GFX10-DL-NEXT:    s_load_dword s4, s[6:7], 0x0
+; GFX10-DL-NEXT:    s_load_dword s6, s[4:5], 0x0
+; GFX10-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX10-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
-; GFX10-DL-NEXT:    v_dot2_i32_i16 v2, s4, s3, v0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s6
+; GFX10-DL-NEXT:    v_dot2_i32_i16 v2, s1, s0, v0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX10-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX10-DL-NEXT:    s_endpgm
                                  <2 x i16> addrspace(1)* %src2,
                                  i32 addrspace(1)* nocapture %dst) {
 entry:
   %vec1 = load <2 x i16>, <2 x i16> addrspace(1)* %src1
   %vec2 = load <2 x i16>, <2 x i16> addrspace(1)* %src2
 

 ; GFX9-DL-NEXT:    v_dot2_u32_u16 v2, s4, v0, v1
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX9-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot2_alt_AddOperands:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    s_load_dword s2, s[0:1], 0x0
-; GFX10-DL-NEXT:    s_load_dword s3, s[4:5], 0x0
-; GFX10-DL-NEXT:    s_load_dword s4, s[6:7], 0x0
+; GFX10-DL-NEXT:    s_load_dword s6, s[4:5], 0x0
+; GFX10-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX10-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
-; GFX10-DL-NEXT:    v_dot2_u32_u16 v2, s4, s3, v0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s6
+; GFX10-DL-NEXT:    v_dot2_u32_u16 v2, s1, s0, v0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX10-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX10-DL-NEXT:    s_endpgm
                                                  <2 x i16> addrspace(1)* %src2,
                                                  i32 addrspace(1)* nocapture %dst) {
 entry:
   %vec1 = load <2 x i16>, <2 x i16> addrspace(1)* %src1
   %vec2 = load <2 x i16>, <2 x i16> addrspace(1)* %src2
 

 ; GFX8-LABEL: notudot2_SameVec:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
 ; GFX8-NEXT:    s_load_dword s3, s[0:1], 0x0
 ; GFX8-NEXT:    s_load_dword s4, s[4:5], 0x0
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    s_lshr_b32 s2, s2, 16
+; GFX8-NEXT:    v_mov_b32_e32 v0, s3
+; GFX8-NEXT:    s_and_b32 s4, s4, 0xffff
+; GFX8-NEXT:    v_mad_u32_u24 v0, s2, s2, v0
+; GFX8-NEXT:    v_mad_u32_u24 v2, s4, s4, v0
 ; GFX8-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX8-NEXT:    v_mov_b32_e32 v1, s1
-; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_lshr_b32 s0, s2, 16
-; GFX8-NEXT:    v_mov_b32_e32 v2, s3
-; GFX8-NEXT:    s_and_b32 s1, s4, 0xffff
-; GFX8-NEXT:    v_mad_u32_u24 v2, s0, s0, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s1, s1, v2
 ; GFX8-NEXT:    flat_store_dword v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-NODL-LABEL: notudot2_SameVec:
 ; GFX9-NODL:       ; %bb.0: ; %entry
 ; GFX9-NODL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-NODL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NODL-NEXT:    s_load_dword s2, s[6:7], 0x0
 ; GFX9-NODL-NEXT:    s_load_dword s3, s[0:1], 0x0
 ; GFX9-NODL-NEXT:    s_load_dword s4, s[4:5], 0x0
+; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NODL-NEXT:    s_lshr_b32 s2, s2, 16
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s3
+; GFX9-NODL-NEXT:    s_and_b32 s4, s4, 0xffff
+; GFX9-NODL-NEXT:    v_mad_u32_u24 v0, s2, s2, v0
+; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s4, s4, v0
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s1
-; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_lshr_b32 s0, s2, 16
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v2, s3
-; GFX9-NODL-NEXT:    s_and_b32 s1, s4, 0xffff
-; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s0, s0, v2
-; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s1, s1, v2
 ; GFX9-NODL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NODL-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: notudot2_SameVec:
 ; GFX9-DL:       ; %bb.0: ; %entry
 ; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_load_dword s3, s[0:1], 0x0
 ; GFX9-DL-NEXT:    s_load_dword s4, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-DL-NEXT:    s_lshr_b32 s2, s2, 16
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s3
+; GFX9-DL-NEXT:    s_and_b32 s4, s4, 0xffff
+; GFX9-DL-NEXT:    v_mad_u32_u24 v0, s2, s2, v0
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s4, s4, v0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
-; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_lshr_b32 s0, s2, 16
-; GFX9-DL-NEXT:    v_mov_b32_e32 v2, s3
-; GFX9-DL-NEXT:    s_and_b32 s1, s4, 0xffff
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s0, s0, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s1, s1, v2
 ; GFX9-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: notudot2_SameVec:
 ; GFX10-DL:       ; %bb.0: ; %entry
 ; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi

 ; GFX9-DL-NEXT:    v_dot2_u32_u16 v2, s4, v0, v1
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX9-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot2_v4i16:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    s_load_dword s2, s[0:1], 0x0
-; GFX10-DL-NEXT:    s_load_dword s3, s[4:5], 0x0
-; GFX10-DL-NEXT:    s_load_dword s4, s[6:7], 0x0
+; GFX10-DL-NEXT:    s_load_dword s6, s[4:5], 0x0
+; GFX10-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX10-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
-; GFX10-DL-NEXT:    v_dot2_u32_u16 v2, s4, s3, v0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s6
+; GFX10-DL-NEXT:    v_dot2_u32_u16 v2, s1, s0, v0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX10-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX10-DL-NEXT:    s_endpgm
                                        <4 x i16> addrspace(1)* %src2,
                                        i32 addrspace(1)* nocapture %dst) {
 entry:
   %vec1 = load <4 x i16>, <4 x i16> addrspace(1)* %src1
   %vec2 = load <4 x i16>, <4 x i16> addrspace(1)* %src2
 

 ; GFX9-DL-NEXT:    v_dot2_u32_u16 v2, s4, v0, v1
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX9-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot2_v4i16_Hi:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    s_load_dword s2, s[0:1], 0x0
-; GFX10-DL-NEXT:    s_load_dword s3, s[4:5], 0x4
-; GFX10-DL-NEXT:    s_load_dword s4, s[6:7], 0x4
+; GFX10-DL-NEXT:    s_load_dword s6, s[4:5], 0x0
+; GFX10-DL-NEXT:    s_load_dword s0, s[0:1], 0x4
+; GFX10-DL-NEXT:    s_load_dword s1, s[2:3], 0x4
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
-; GFX10-DL-NEXT:    v_dot2_u32_u16 v2, s4, s3, v0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s6
+; GFX10-DL-NEXT:    v_dot2_u32_u16 v2, s1, s0, v0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX10-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX10-DL-NEXT:    s_endpgm
                                           <4 x i16> addrspace(1)* %src2,
                                           i32 addrspace(1)* nocapture %dst) {
 entry:
   %vec1 = load <4 x i16>, <4 x i16> addrspace(1)* %src1
   %vec2 = load <4 x i16>, <4 x i16> addrspace(1)* %src2
 
entry:
 
   store i32 %add2, i32 addrspace(1)* %dst, align 4
   ret void
 }
 
 define amdgpu_kernel void @udot2_acc16(<2 x i16> addrspace(1)* %src1,
 ; GFX7-LABEL: udot2_acc16:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
 ; GFX7-NEXT:    s_mov_b32 s8, 0xffff
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s4, s[4:5], 0x0
-; GFX7-NEXT:    buffer_load_ushort v0, off, s[0:3], 0
-; GFX7-NEXT:    s_load_dword s5, s[6:7], 0x0
+; GFX7-NEXT:    buffer_load_ushort v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX7-NEXT:    s_load_dword s1, s[2:3], 0x0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_lshr_b32 s6, s4, 16
-; GFX7-NEXT:    s_and_b32 s4, s4, s8
-; GFX7-NEXT:    s_lshr_b32 s7, s5, 16
-; GFX7-NEXT:    v_mov_b32_e32 v1, s7
-; GFX7-NEXT:    s_and_b32 s5, s5, s8
+; GFX7-NEXT:    s_lshr_b32 s2, s0, 16
+; GFX7-NEXT:    s_lshr_b32 s3, s1, 16
+; GFX7-NEXT:    v_mov_b32_e32 v1, s3
+; GFX7-NEXT:    s_and_b32 s1, s1, s8
+; GFX7-NEXT:    s_and_b32 s0, s0, s8
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v0, s6, v1, v0
-; GFX7-NEXT:    v_mov_b32_e32 v1, s5
-; GFX7-NEXT:    v_mad_u32_u24 v0, s4, v1, v0
-; GFX7-NEXT:    buffer_store_short v0, off, s[0:3], 0
+; GFX7-NEXT:    v_mad_u32_u24 v0, s2, v1, v0
+; GFX7-NEXT:    v_mov_b32_e32 v1, s1
+; GFX7-NEXT:    v_mad_u32_u24 v0, s0, v1, v0
+; GFX7-NEXT:    buffer_store_short v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot2_acc16:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX8-NEXT:    v_mov_b32_e32 v0, s4
+; GFX8-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s2, s[4:5], 0x0
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX8-NEXT:    flat_load_ushort v2, v[0:1]
-; GFX8-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX8-NEXT:    s_mov_b32 s0, 0xffff
+; GFX8-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX8-NEXT:    s_mov_b32 s1, 0xffff
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_and_b32 s3, s1, s0
-; GFX8-NEXT:    s_lshr_b32 s1, s1, 16
-; GFX8-NEXT:    s_and_b32 s0, s2, s0
+; GFX8-NEXT:    s_and_b32 s3, s2, s1
 ; GFX8-NEXT:    s_lshr_b32 s2, s2, 16
-; GFX8-NEXT:    v_mov_b32_e32 v3, s1
+; GFX8-NEXT:    s_and_b32 s1, s0, s1
+; GFX8-NEXT:    s_lshr_b32 s0, s0, 16
+; GFX8-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
-; GFX8-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s3
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX8-NEXT:    v_mov_b32_e32 v3, s3
+; GFX8-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
 ; GFX8-NEXT:    flat_store_short v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-NODL-LABEL: udot2_acc16:
 ; GFX9-NODL:       ; %bb.0: ; %entry
-; GFX9-NODL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NODL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_load_dword s2, s[4:5], 0x0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NODL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-NODL-NEXT:    global_load_ushort v2, v[0:1], off
-; GFX9-NODL-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX9-NODL-NEXT:    s_mov_b32 s0, 0xffff
+; GFX9-NODL-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX9-NODL-NEXT:    s_mov_b32 s1, 0xffff
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_and_b32 s3, s1, s0
-; GFX9-NODL-NEXT:    s_lshr_b32 s1, s1, 16
-; GFX9-NODL-NEXT:    s_and_b32 s0, s2, s0
+; GFX9-NODL-NEXT:    s_and_b32 s3, s2, s1
 ; GFX9-NODL-NEXT:    s_lshr_b32 s2, s2, 16
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s1
+; GFX9-NODL-NEXT:    s_and_b32 s1, s0, s1
+; GFX9-NODL-NEXT:    s_lshr_b32 s0, s0, 16
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX9-NODL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s3
 ; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s3
+; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
 ; GFX9-NODL-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-NODL-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: udot2_acc16:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX9-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-DL-NEXT:    global_load_ushort v2, v[0:1], off
-; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s1
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-DL-NEXT:    v_dot2_u32_u16 v2, s0, v3, v2
 ; GFX9-DL-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot2_acc16:
 ; GFX10-DL:       ; %bb.0: ; %entry
 ; GFX10-DL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34

 ; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xd
 ; GFX7-NEXT:    s_mov_b32 s10, s2
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    s_mov_b32 s8, s6
 ; GFX7-NEXT:    s_mov_b32 s9, s7
 ; GFX7-NEXT:    s_mov_b32 s11, s3
 ; GFX7-NEXT:    s_mov_b32 s6, s2
 ; GFX7-NEXT:    s_mov_b32 s7, s3
-; GFX7-NEXT:    buffer_load_ushort v0, off, s[4:7], 0
-; GFX7-NEXT:    buffer_load_ushort v1, off, s[8:11], 0
+; GFX7-NEXT:    buffer_load_ushort v0, off, s[8:11], 0
+; GFX7-NEXT:    buffer_load_ushort v1, off, s[4:7], 0
 ; GFX7-NEXT:    s_load_dword s4, s[0:1], 0x0
 ; GFX7-NEXT:    s_waitcnt vmcnt(1)
-; GFX7-NEXT:    v_bfe_i32 v2, v0, 0, 8
+; GFX7-NEXT:    v_bfe_i32 v3, v0, 0, 8
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
-; GFX7-NEXT:    v_bfe_i32 v3, v1, 0, 8
-; GFX7-NEXT:    v_bfe_i32 v0, v0, 8, 8
+; GFX7-NEXT:    v_bfe_i32 v2, v1, 0, 8
 ; GFX7-NEXT:    v_bfe_i32 v1, v1, 8, 8
+; GFX7-NEXT:    v_bfe_i32 v0, v0, 8, 8
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    v_mad_i32_i24 v0, v1, v0, s4
+; GFX7-NEXT:    v_mad_i32_i24 v0, v0, v1, s4
 ; GFX7-NEXT:    v_mad_i32_i24 v0, v3, v2, v0
 ; GFX7-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: notsdot2_sext8:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s2, s[0:1], 0x0
 ; GFX8-NEXT:    v_mov_b32_e32 v0, s6
 ; GFX8-NEXT:    v_mov_b32_e32 v1, s7
 ; GFX8-NEXT:    v_mov_b32_e32 v2, s4
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s5
-; GFX8-NEXT:    flat_load_ushort v2, v[2:3]
 ; GFX8-NEXT:    flat_load_ushort v0, v[0:1]
+; GFX8-NEXT:    flat_load_ushort v1, v[2:3]
+; GFX8-NEXT:    s_load_dword s2, s[0:1], 0x0
 ; GFX8-NEXT:    s_waitcnt vmcnt(1) lgkmcnt(0)
-; GFX8-NEXT:    v_bfe_i32 v1, v2, 0, 8
-; GFX8-NEXT:    v_lshrrev_b16_e32 v2, 8, v2
-; GFX8-NEXT:    s_waitcnt vmcnt(0)
 ; GFX8-NEXT:    v_bfe_i32 v3, v0, 0, 8
 ; GFX8-NEXT:    v_lshrrev_b16_e32 v0, 8, v0
-; GFX8-NEXT:    v_bfe_i32 v2, v2, 0, 8
+; GFX8-NEXT:    s_waitcnt vmcnt(0)
+; GFX8-NEXT:    v_bfe_i32 v2, v1, 0, 8
+; GFX8-NEXT:    v_lshrrev_b16_e32 v1, 8, v1
+; GFX8-NEXT:    v_bfe_i32 v1, v1, 0, 8
 ; GFX8-NEXT:    v_bfe_i32 v0, v0, 0, 8
-; GFX8-NEXT:    v_mad_i32_i24 v0, v0, v2, s2
-; GFX8-NEXT:    v_mad_i32_i24 v2, v3, v1, v0
+; GFX8-NEXT:    v_mad_i32_i24 v0, v0, v1, s2
+; GFX8-NEXT:    v_mad_i32_i24 v2, v3, v2, v0
 ; GFX8-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX8-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX8-NEXT:    flat_store_dword v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-NODL-LABEL: notsdot2_sext8:
 ; GFX9-NODL:       ; %bb.0: ; %entry
 ; GFX9-NODL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-NODL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_load_dword s2, s[0:1], 0x0
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s6
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s7
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v2, s4
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s5
-; GFX9-NODL-NEXT:    global_load_ushort v2, v[2:3], off
 ; GFX9-NODL-NEXT:    global_load_ushort v0, v[0:1], off
+; GFX9-NODL-NEXT:    global_load_ushort v1, v[2:3], off
+; GFX9-NODL-NEXT:    s_load_dword s2, s[0:1], 0x0
 ; GFX9-NODL-NEXT:    s_waitcnt vmcnt(1)
-; GFX9-NODL-NEXT:    v_bfe_i32 v1, v2, 0, 8
-; GFX9-NODL-NEXT:    v_lshrrev_b16_e32 v2, 8, v2
-; GFX9-NODL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NODL-NEXT:    v_bfe_i32 v3, v0, 0, 8
 ; GFX9-NODL-NEXT:    v_lshrrev_b16_e32 v0, 8, v0
-; GFX9-NODL-NEXT:    v_bfe_i32 v2, v2, 0, 8
+; GFX9-NODL-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NODL-NEXT:    v_bfe_i32 v2, v1, 0, 8
+; GFX9-NODL-NEXT:    v_lshrrev_b16_e32 v1, 8, v1
+; GFX9-NODL-NEXT:    v_bfe_i32 v1, v1, 0, 8
 ; GFX9-NODL-NEXT:    v_bfe_i32 v0, v0, 0, 8
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    v_mad_i32_i24 v0, v0, v2, s2
-; GFX9-NODL-NEXT:    v_mad_i32_i24 v2, v3, v1, v0
+; GFX9-NODL-NEXT:    v_mad_i32_i24 v0, v0, v1, s2
+; GFX9-NODL-NEXT:    v_mad_i32_i24 v2, v3, v2, v0
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX9-NODL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NODL-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: notsdot2_sext8:
 ; GFX9-DL:       ; %bb.0: ; %entry
 ; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s2, s[0:1], 0x0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s6
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s7
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v2, s4
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s5
-; GFX9-DL-NEXT:    global_load_ushort v2, v[2:3], off
 ; GFX9-DL-NEXT:    global_load_ushort v0, v[0:1], off
+; GFX9-DL-NEXT:    global_load_ushort v1, v[2:3], off
+; GFX9-DL-NEXT:    s_load_dword s2, s[0:1], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(1)
-; GFX9-DL-NEXT:    v_bfe_i32 v1, v2, 0, 8
-; GFX9-DL-NEXT:    v_lshrrev_b16_e32 v2, 8, v2
-; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-DL-NEXT:    v_bfe_i32 v3, v0, 0, 8
 ; GFX9-DL-NEXT:    v_lshrrev_b16_e32 v0, 8, v0
-; GFX9-DL-NEXT:    v_bfe_i32 v2, v2, 0, 8
+; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-DL-NEXT:    v_bfe_i32 v2, v1, 0, 8
+; GFX9-DL-NEXT:    v_lshrrev_b16_e32 v1, 8, v1
+; GFX9-DL-NEXT:    v_bfe_i32 v1, v1, 0, 8
 ; GFX9-DL-NEXT:    v_bfe_i32 v0, v0, 0, 8
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mad_i32_i24 v0, v0, v2, s2
-; GFX9-DL-NEXT:    v_mad_i32_i24 v2, v3, v1, v0
+; GFX9-DL-NEXT:    v_mad_i32_i24 v0, v0, v1, s2
+; GFX9-DL-NEXT:    v_mad_i32_i24 v2, v3, v2, v0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX9-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: notsdot2_sext8:
 ; GFX10-DL:       ; %bb.0: ; %entry
 ; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24

llvm/test/CodeGen/AMDGPU/idot4s.ll

 ; GFX9-DL-LABEL: idot4_acc32:
 ; GFX9-DL:       ; %bb.0: ; %entry
 ; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_load_dword s3, s[0:1], 0x0
 ; GFX9-DL-NEXT:    s_load_dword s4, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s3
+; GFX9-DL-NEXT:    v_dot4_i32_i8 v2, s4, v0, v1
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
-; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mov_b32_e32 v2, s2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s3
-; GFX9-DL-NEXT:    v_dot4_i32_i8 v2, s4, v2, v3
 ; GFX9-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: idot4_acc32:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    s_load_dword s2, s[0:1], 0x0
-; GFX10-DL-NEXT:    s_load_dword s3, s[4:5], 0x0
-; GFX10-DL-NEXT:    s_load_dword s4, s[6:7], 0x0
+; GFX10-DL-NEXT:    s_load_dword s6, s[4:5], 0x0
+; GFX10-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX10-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
-; GFX10-DL-NEXT:    v_dot4_i32_i8 v2, s3, s4, v0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s6
+; GFX10-DL-NEXT:    v_dot4_i32_i8 v2, s0, s1, v0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX10-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX10-DL-NEXT:    s_endpgm
                                        <4 x i8> addrspace(1)* %src2,
                                        i32 addrspace(1)* nocapture %dst) {
 entry:
   %vec1 = load <4 x i8>, <4 x i8> addrspace(1)* %src1
   %vec2 = load <4 x i8>, <4 x i8> addrspace(1)* %src2
 
entry:
   ret void
 }
 
 ; TODO: Currently, vector elements{0 and 3} get zero_extended from i16 to i32 which should
 ; be sign_extended directly to i32; prevents the pattern recognizer to recognize this pattern.
 define amdgpu_kernel void @idot4_acc16(<4 x i8> addrspace(1)* %src1,
 ; GFX7-LABEL: idot4_acc16:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
 ; GFX7-NEXT:    s_mov_b32 s8, 0xffff
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s4, s[4:5], 0x0
-; GFX7-NEXT:    buffer_load_ushort v0, off, s[0:3], 0
-; GFX7-NEXT:    s_load_dword s5, s[6:7], 0x0
+; GFX7-NEXT:    buffer_load_ushort v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX7-NEXT:    s_load_dword s1, s[2:3], 0x0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_sext_i32_i8 s6, s4
-; GFX7-NEXT:    s_bfe_i32 s9, s4, 0x80008
-; GFX7-NEXT:    s_sext_i32_i8 s7, s5
-; GFX7-NEXT:    s_bfe_i32 s10, s5, 0x80008
-; GFX7-NEXT:    s_and_b32 s7, s7, s8
-; GFX7-NEXT:    s_bfe_i32 s12, s5, 0x80010
+; GFX7-NEXT:    s_sext_i32_i8 s2, s0
+; GFX7-NEXT:    s_sext_i32_i8 s3, s1
+; GFX7-NEXT:    s_bfe_i32 s10, s1, 0x80008
+; GFX7-NEXT:    s_and_b32 s3, s3, s8
+; GFX7-NEXT:    s_bfe_i32 s12, s1, 0x80010
+; GFX7-NEXT:    s_bfe_i32 s9, s0, 0x80008
 ; GFX7-NEXT:    s_and_b32 s10, s10, s8
-; GFX7-NEXT:    s_and_b32 s6, s6, s8
-; GFX7-NEXT:    v_mov_b32_e32 v1, s7
-; GFX7-NEXT:    s_bfe_i32 s11, s4, 0x80010
-; GFX7-NEXT:    s_ashr_i32 s5, s5, 24
+; GFX7-NEXT:    s_and_b32 s2, s2, s8
+; GFX7-NEXT:    v_mov_b32_e32 v1, s3
+; GFX7-NEXT:    s_bfe_i32 s11, s0, 0x80010
+; GFX7-NEXT:    s_ashr_i32 s1, s1, 24
 ; GFX7-NEXT:    s_and_b32 s12, s12, s8
 ; GFX7-NEXT:    s_and_b32 s9, s9, s8
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s10
-; GFX7-NEXT:    s_ashr_i32 s4, s4, 24
+; GFX7-NEXT:    s_ashr_i32 s0, s0, 24
 ; GFX7-NEXT:    s_and_b32 s11, s11, s8
-; GFX7-NEXT:    s_and_b32 s5, s5, s8
+; GFX7-NEXT:    s_and_b32 s1, s1, s8
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s12
-; GFX7-NEXT:    s_and_b32 s4, s4, s8
+; GFX7-NEXT:    s_and_b32 s0, s0, s8
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v0, s6, v1, v0
+; GFX7-NEXT:    v_mad_u32_u24 v0, s2, v1, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s9, v2, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s11, v3, v0
-; GFX7-NEXT:    v_mov_b32_e32 v1, s5
-; GFX7-NEXT:    v_mad_u32_u24 v0, s4, v1, v0
-; GFX7-NEXT:    buffer_store_short v0, off, s[0:3], 0
+; GFX7-NEXT:    v_mov_b32_e32 v1, s1
+; GFX7-NEXT:    v_mad_u32_u24 v0, s0, v1, v0
+; GFX7-NEXT:    buffer_store_short v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: idot4_acc16:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX8-NEXT:    v_mov_b32_e32 v0, s4
+; GFX8-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX8-NEXT:    flat_load_ushort v2, v[0:1]
-; GFX8-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX8-NEXT:    s_load_dword s0, s[0:1], 0x0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_sext_i32_i8 s1, s2
-; GFX8-NEXT:    s_bfe_i32 s3, s2, 0x80008
-; GFX8-NEXT:    v_mov_b32_e32 v3, s1
-; GFX8-NEXT:    s_bfe_i32 s5, s2, 0x80010
+; GFX8-NEXT:    s_sext_i32_i8 s3, s2
+; GFX8-NEXT:    s_bfe_i32 s5, s2, 0x80008
+; GFX8-NEXT:    v_mov_b32_e32 v3, s3
+; GFX8-NEXT:    s_bfe_i32 s7, s2, 0x80010
 ; GFX8-NEXT:    s_sext_i32_i8 s1, s0
-; GFX8-NEXT:    v_mov_b32_e32 v4, s3
 ; GFX8-NEXT:    s_bfe_i32 s4, s0, 0x80008
-; GFX8-NEXT:    s_bfe_i32 s3, s0, 0x80010
+; GFX8-NEXT:    v_mov_b32_e32 v4, s5
+; GFX8-NEXT:    s_bfe_i32 s6, s0, 0x80010
 ; GFX8-NEXT:    s_ashr_i32 s2, s2, 24
-; GFX8-NEXT:    v_mov_b32_e32 v5, s5
+; GFX8-NEXT:    v_mov_b32_e32 v5, s7
 ; GFX8-NEXT:    s_ashr_i32 s0, s0, 24
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
 ; GFX8-NEXT:    v_mad_i32_i24 v2, s1, v3, v2
 ; GFX8-NEXT:    v_mad_i32_i24 v2, s4, v4, v2
-; GFX8-NEXT:    v_mad_i32_i24 v2, s3, v5, v2
+; GFX8-NEXT:    v_mad_i32_i24 v2, s6, v5, v2
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX8-NEXT:    v_mad_i32_i24 v2, s0, v3, v2
 ; GFX8-NEXT:    flat_store_short v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-NODL-LABEL: idot4_acc16:
 ; GFX9-NODL:       ; %bb.0: ; %entry
-; GFX9-NODL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NODL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NODL-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-NODL-NEXT:    global_load_ushort v2, v[0:1], off
-; GFX9-NODL-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-NODL-NEXT:    s_load_dword s0, s[0:1], 0x0
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_sext_i32_i8 s1, s2
-; GFX9-NODL-NEXT:    s_bfe_i32 s3, s2, 0x80008
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-NODL-NEXT:    s_bfe_i32 s5, s2, 0x80010
+; GFX9-NODL-NEXT:    s_sext_i32_i8 s3, s2
+; GFX9-NODL-NEXT:    s_bfe_i32 s5, s2, 0x80008
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s3
+; GFX9-NODL-NEXT:    s_bfe_i32 s7, s2, 0x80010
 ; GFX9-NODL-NEXT:    s_sext_i32_i8 s1, s0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v4, s3
 ; GFX9-NODL-NEXT:    s_bfe_i32 s4, s0, 0x80008
-; GFX9-NODL-NEXT:    s_bfe_i32 s3, s0, 0x80010
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v4, s5
+; GFX9-NODL-NEXT:    s_bfe_i32 s6, s0, 0x80010
 ; GFX9-NODL-NEXT:    s_ashr_i32 s2, s2, 24
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v5, s5
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v5, s7
 ; GFX9-NODL-NEXT:    s_ashr_i32 s0, s0, 24
 ; GFX9-NODL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NODL-NEXT:    v_mad_i32_i24 v2, s1, v3, v2
 ; GFX9-NODL-NEXT:    v_mad_i32_i24 v2, s4, v4, v2
-; GFX9-NODL-NEXT:    v_mad_i32_i24 v2, s3, v5, v2
+; GFX9-NODL-NEXT:    v_mad_i32_i24 v2, s6, v5, v2
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX9-NODL-NEXT:    v_mad_i32_i24 v2, s0, v3, v2
 ; GFX9-NODL-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-NODL-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: idot4_acc16:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX9-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-DL-NEXT:    global_load_ushort v2, v[0:1], off
-; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s1
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-DL-NEXT:    v_dot4_i32_i8 v2, s0, v3, v2
 ; GFX9-DL-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: idot4_acc16:
 ; GFX10-DL:       ; %bb.0: ; %entry
 ; GFX10-DL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
entry:
   %add4 = add i16 %add3, %mul4
   store i16 %add4, i16 addrspace(1)* %dst, align 2
   ret void
 }
 
 define amdgpu_kernel void @idot4_acc8(<4 x i8> addrspace(1)* %src1,
 ; GFX7-LABEL: idot4_acc8:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s4, s[4:5], 0x0
-; GFX7-NEXT:    buffer_load_ubyte v0, off, s[0:3], 0
-; GFX7-NEXT:    s_load_dword s6, s[6:7], 0x0
-; GFX7-NEXT:    s_movk_i32 s5, 0xff
+; GFX7-NEXT:    buffer_load_ubyte v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX7-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX7-NEXT:    s_movk_i32 s1, 0xff
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_bfe_u32 s9, s4, 0x80008
-; GFX7-NEXT:    s_bfe_u32 s11, s4, 0x80010
-; GFX7-NEXT:    s_and_b32 s7, s6, s5
-; GFX7-NEXT:    s_and_b32 s5, s4, s5
-; GFX7-NEXT:    s_bfe_u32 s8, s6, 0x80008
-; GFX7-NEXT:    v_mov_b32_e32 v1, s7
-; GFX7-NEXT:    s_bfe_u32 s10, s6, 0x80010
+; GFX7-NEXT:    s_bfe_u32 s9, s0, 0x80008
+; GFX7-NEXT:    s_and_b32 s3, s2, s1
+; GFX7-NEXT:    s_bfe_u32 s8, s2, 0x80008
+; GFX7-NEXT:    s_and_b32 s1, s0, s1
+; GFX7-NEXT:    v_mov_b32_e32 v1, s3
+; GFX7-NEXT:    s_bfe_u32 s10, s2, 0x80010
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s8
-; GFX7-NEXT:    s_lshr_b32 s6, s6, 24
+; GFX7-NEXT:    s_bfe_u32 s11, s0, 0x80010
+; GFX7-NEXT:    s_lshr_b32 s2, s2, 24
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s10
-; GFX7-NEXT:    s_lshr_b32 s4, s4, 24
+; GFX7-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v0, s5, v1, v0
+; GFX7-NEXT:    v_mad_u32_u24 v0, s1, v1, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s9, v2, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s11, v3, v0
-; GFX7-NEXT:    v_mov_b32_e32 v1, s6
-; GFX7-NEXT:    v_mad_u32_u24 v0, s4, v1, v0
-; GFX7-NEXT:    buffer_store_byte v0, off, s[0:3], 0
+; GFX7-NEXT:    v_mov_b32_e32 v1, s2
+; GFX7-NEXT:    v_mad_u32_u24 v0, s0, v1, v0
+; GFX7-NEXT:    buffer_store_byte v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: idot4_acc8:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX8-NEXT:    v_mov_b32_e32 v0, s4
+; GFX8-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s2, s[4:5], 0x0
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX8-NEXT:    flat_load_ubyte v2, v[0:1]
-; GFX8-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX8-NEXT:    s_movk_i32 s0, 0xff
+; GFX8-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX8-NEXT:    s_movk_i32 s1, 0xff
+; GFX8-NEXT:    s_bfe_u32 s5, s0, 0x80008
+; GFX8-NEXT:    s_bfe_u32 s7, s0, 0x80010
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_bfe_u32 s5, s2, 0x80008
-; GFX8-NEXT:    s_bfe_u32 s7, s2, 0x80010
-; GFX8-NEXT:    s_and_b32 s3, s1, s0
-; GFX8-NEXT:    s_and_b32 s0, s2, s0
-; GFX8-NEXT:    s_bfe_u32 s4, s1, 0x80008
+; GFX8-NEXT:    s_and_b32 s3, s2, s1
+; GFX8-NEXT:    s_and_b32 s1, s0, s1
+; GFX8-NEXT:    s_bfe_u32 s4, s2, 0x80008
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s3
-; GFX8-NEXT:    s_bfe_u32 s6, s1, 0x80010
+; GFX8-NEXT:    s_bfe_u32 s6, s2, 0x80010
 ; GFX8-NEXT:    v_mov_b32_e32 v4, s4
-; GFX8-NEXT:    s_lshr_b32 s1, s1, 24
-; GFX8-NEXT:    v_mov_b32_e32 v5, s6
 ; GFX8-NEXT:    s_lshr_b32 s2, s2, 24
+; GFX8-NEXT:    v_mov_b32_e32 v5, s6
+; GFX8-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
-; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s7, v5, v2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s1
-; GFX8-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
+; GFX8-NEXT:    v_mov_b32_e32 v3, s2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
 ; GFX8-NEXT:    flat_store_byte v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-NODL-LABEL: idot4_acc8:
 ; GFX9-NODL:       ; %bb.0: ; %entry
-; GFX9-NODL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NODL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_load_dword s2, s[4:5], 0x0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NODL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-NODL-NEXT:    global_load_ubyte v2, v[0:1], off
-; GFX9-NODL-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX9-NODL-NEXT:    s_movk_i32 s0, 0xff
+; GFX9-NODL-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX9-NODL-NEXT:    s_movk_i32 s1, 0xff
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_bfe_u32 s5, s2, 0x80008
-; GFX9-NODL-NEXT:    s_bfe_u32 s7, s2, 0x80010
-; GFX9-NODL-NEXT:    s_and_b32 s3, s1, s0
-; GFX9-NODL-NEXT:    s_and_b32 s0, s2, s0
-; GFX9-NODL-NEXT:    s_bfe_u32 s4, s1, 0x80008
+; GFX9-NODL-NEXT:    s_bfe_u32 s5, s0, 0x80008
+; GFX9-NODL-NEXT:    s_bfe_u32 s7, s0, 0x80010
+; GFX9-NODL-NEXT:    s_and_b32 s3, s2, s1
+; GFX9-NODL-NEXT:    s_and_b32 s1, s0, s1
+; GFX9-NODL-NEXT:    s_bfe_u32 s4, s2, 0x80008
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s3
-; GFX9-NODL-NEXT:    s_bfe_u32 s6, s1, 0x80010
+; GFX9-NODL-NEXT:    s_bfe_u32 s6, s2, 0x80010
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v4, s4
-; GFX9-NODL-NEXT:    s_lshr_b32 s1, s1, 24
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v5, s6
 ; GFX9-NODL-NEXT:    s_lshr_b32 s2, s2, 24
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v5, s6
+; GFX9-NODL-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX9-NODL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
 ; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
 ; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s7, v5, v2
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s2
+; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
 ; GFX9-NODL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-NODL-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: idot4_acc8:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX9-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-DL-NEXT:    global_load_ubyte v2, v[0:1], off
-; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s1
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-DL-NEXT:    v_dot4_u32_u8 v2, s0, v3, v2
 ; GFX9-DL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: idot4_acc8:
 ; GFX10-DL:       ; %bb.0: ; %entry
 ; GFX10-DL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
entry:
 
   store i32 %add4, i32 addrspace(1)* %dst, align 4
   ret void
 }
 
 define amdgpu_kernel void @idot4_acc16_vecMul(<4 x i8> addrspace(1)* %src1,
 ; GFX7-LABEL: idot4_acc16_vecMul:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s4, s[4:5], 0x0
-; GFX7-NEXT:    buffer_load_ushort v0, off, s[0:3], 0
-; GFX7-NEXT:    s_load_dword s5, s[6:7], 0x0
+; GFX7-NEXT:    buffer_load_ushort v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX7-NEXT:    s_load_dword s1, s[2:3], 0x0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_ashr_i32 s6, s4, 24
-; GFX7-NEXT:    s_bfe_i32 s7, s4, 0x80010
-; GFX7-NEXT:    s_bfe_i32 s10, s5, 0x80010
-; GFX7-NEXT:    s_bfe_i32 s11, s5, 0x80008
-; GFX7-NEXT:    s_ashr_i32 s9, s5, 24
-; GFX7-NEXT:    s_sext_i32_i8 s5, s5
-; GFX7-NEXT:    s_bfe_i32 s8, s4, 0x80008
-; GFX7-NEXT:    s_sext_i32_i8 s4, s4
-; GFX7-NEXT:    v_mov_b32_e32 v1, s5
+; GFX7-NEXT:    s_ashr_i32 s2, s0, 24
+; GFX7-NEXT:    s_bfe_i32 s10, s1, 0x80010
+; GFX7-NEXT:    s_bfe_i32 s11, s1, 0x80008
+; GFX7-NEXT:    s_ashr_i32 s9, s1, 24
+; GFX7-NEXT:    s_sext_i32_i8 s1, s1
+; GFX7-NEXT:    s_bfe_i32 s3, s0, 0x80010
+; GFX7-NEXT:    s_bfe_i32 s8, s0, 0x80008
+; GFX7-NEXT:    s_sext_i32_i8 s0, s0
+; GFX7-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s11
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s10
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
-; GFX7-NEXT:    v_mad_i32_i24 v0, s4, v1, v0
+; GFX7-NEXT:    v_mad_i32_i24 v0, s0, v1, v0
 ; GFX7-NEXT:    v_mad_i32_i24 v0, s8, v2, v0
-; GFX7-NEXT:    v_mad_i32_i24 v0, s7, v3, v0
+; GFX7-NEXT:    v_mad_i32_i24 v0, s3, v3, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s9
-; GFX7-NEXT:    v_mad_i32_i24 v0, s6, v1, v0
-; GFX7-NEXT:    buffer_store_short v0, off, s[0:3], 0
+; GFX7-NEXT:    v_mad_i32_i24 v0, s2, v1, v0
+; GFX7-NEXT:    buffer_store_short v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: idot4_acc16_vecMul:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
-; GFX8-NEXT:    s_mov_b32 s2, 0xffff
+; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX8-NEXT:    s_mov_b32 s6, 0xffff
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s3, s[6:7], 0x0
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    v_mov_b32_e32 v0, s4
+; GFX8-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX8-NEXT:    flat_load_ushort v2, v[0:1]
-; GFX8-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX8-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX8-NEXT:    s_load_dword s0, s[0:1], 0x0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_bfe_i32 s6, s3, 0x80000
-; GFX8-NEXT:    s_lshr_b32 s4, s3, 16
-; GFX8-NEXT:    v_ashrrev_i16_e64 v3, 8, s3
-; GFX8-NEXT:    s_bfe_i32 s3, s4, 0x80000
+; GFX8-NEXT:    s_bfe_i32 s5, s2, 0x80000
 ; GFX8-NEXT:    s_lshr_b32 s1, s0, 16
-; GFX8-NEXT:    s_bfe_i32 s5, s0, 0x80000
+; GFX8-NEXT:    s_lshr_b32 s3, s2, 16
+; GFX8-NEXT:    s_bfe_i32 s4, s0, 0x80000
 ; GFX8-NEXT:    v_ashrrev_i16_e64 v4, 8, s0
 ; GFX8-NEXT:    s_bfe_i32 s0, s1, 0x80000
 ; GFX8-NEXT:    v_ashrrev_i16_e64 v6, 8, s1
-; GFX8-NEXT:    s_and_b32 s1, s2, s6
-; GFX8-NEXT:    v_ashrrev_i16_e64 v5, 8, s4
-; GFX8-NEXT:    s_and_b32 s4, s2, s5
+; GFX8-NEXT:    s_and_b32 s1, s6, s5
+; GFX8-NEXT:    v_ashrrev_i16_e64 v3, 8, s2
+; GFX8-NEXT:    s_bfe_i32 s2, s3, 0x80000
+; GFX8-NEXT:    v_ashrrev_i16_e64 v5, 8, s3
+; GFX8-NEXT:    s_and_b32 s3, s6, s4
 ; GFX8-NEXT:    v_mov_b32_e32 v7, s1
-; GFX8-NEXT:    s_and_b32 s3, s2, s3
-; GFX8-NEXT:    s_and_b32 s0, s2, s0
+; GFX8-NEXT:    s_and_b32 s2, s6, s2
+; GFX8-NEXT:    s_and_b32 s0, s6, s0
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
-; GFX8-NEXT:    v_mad_u32_u24 v2, s4, v7, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s3, v7, v2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, v4, v3, v2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s3
+; GFX8-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, v6, v5, v2
 ; GFX8-NEXT:    flat_store_short v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-NODL-LABEL: idot4_acc16_vecMul:
 ; GFX9-NODL:       ; %bb.0: ; %entry
 ; GFX9-NODL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24

 ; GFX9-DL-NEXT:    v_add_u32_sdwa v3, v4, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
 ; GFX9-DL-NEXT:    v_add_u32_e32 v3, v3, v2
 ; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v3, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
 ; GFX9-DL-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: idot4_acc16_vecMul:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
 ; GFX10-DL-NEXT:    v_mov_b32_e32 v2, 0xffff
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
-; GFX10-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s3
+; GFX10-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX10-DL-NEXT:    global_load_ushort v3, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX10-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_bfe_i32 s4, s0, 0x80000
 ; GFX10-DL-NEXT:    s_bfe_i32 s3, s1, 0x80000
 ; GFX10-DL-NEXT:    s_lshr_b32 s2, s0, 16
 ; GFX10-DL-NEXT:    v_ashrrev_i16_e64 v4, 8, s0
 ; GFX10-DL-NEXT:    s_lshr_b32 s5, s1, 16
 ; GFX10-DL-NEXT:    v_ashrrev_i16_e64 v5, 8, s1
 ; GFX10-DL-NEXT:    v_and_b32_e32 v6, s3, v2
 ; GFX10-DL-NEXT:    v_and_b32_e32 v7, s4, v2

llvm/test/CodeGen/AMDGPU/idot4u.ll

 ; GFX9-DL-LABEL: udot4_acc32:
 ; GFX9-DL:       ; %bb.0: ; %entry
 ; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_load_dword s3, s[0:1], 0x0
 ; GFX9-DL-NEXT:    s_load_dword s4, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s3
+; GFX9-DL-NEXT:    v_dot4_u32_u8 v2, s4, v0, v1
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
-; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mov_b32_e32 v2, s2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s3
-; GFX9-DL-NEXT:    v_dot4_u32_u8 v2, s4, v2, v3
 ; GFX9-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot4_acc32:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    s_load_dword s2, s[0:1], 0x0
-; GFX10-DL-NEXT:    s_load_dword s3, s[4:5], 0x0
-; GFX10-DL-NEXT:    s_load_dword s4, s[6:7], 0x0
+; GFX10-DL-NEXT:    s_load_dword s6, s[4:5], 0x0
+; GFX10-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX10-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
-; GFX10-DL-NEXT:    v_dot4_u32_u8 v2, s3, s4, v0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s6
+; GFX10-DL-NEXT:    v_dot4_u32_u8 v2, s0, s1, v0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX10-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX10-DL-NEXT:    s_endpgm
                                        <4 x i8> addrspace(1)* %src2,
                                        i32 addrspace(1)* nocapture %dst) {
 entry:
   %vec1 = load <4 x i8>, <4 x i8> addrspace(1)* %src1
   %vec2 = load <4 x i8>, <4 x i8> addrspace(1)* %src2
 
entry:
 
   store i32 %mad4, i32 addrspace(1)* %dst, align 4
   ret void
 }
 
 define amdgpu_kernel void @udot4_acc16(<4 x i8> addrspace(1)* %src1,
 ; GFX7-LABEL: udot4_acc16:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s2, -1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s4, s[4:5], 0x0
-; GFX7-NEXT:    buffer_load_ushort v0, off, s[0:3], 0
-; GFX7-NEXT:    s_load_dword s6, s[6:7], 0x0
-; GFX7-NEXT:    s_movk_i32 s5, 0xff
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_bfe_u32 s9, s4, 0x80008
-; GFX7-NEXT:    s_bfe_u32 s11, s4, 0x80010
-; GFX7-NEXT:    s_and_b32 s7, s6, s5
-; GFX7-NEXT:    s_and_b32 s5, s4, s5
-; GFX7-NEXT:    s_bfe_u32 s8, s6, 0x80008
-; GFX7-NEXT:    v_mov_b32_e32 v1, s7
-; GFX7-NEXT:    s_bfe_u32 s10, s6, 0x80010
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX7-NEXT:    buffer_load_ushort v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX7-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX7-NEXT:    s_movk_i32 s1, 0xff
+; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX7-NEXT:    s_bfe_u32 s9, s0, 0x80008
+; GFX7-NEXT:    s_and_b32 s3, s2, s1
+; GFX7-NEXT:    s_bfe_u32 s8, s2, 0x80008
+; GFX7-NEXT:    s_and_b32 s1, s0, s1
+; GFX7-NEXT:    v_mov_b32_e32 v1, s3
+; GFX7-NEXT:    s_bfe_u32 s10, s2, 0x80010
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s8
-; GFX7-NEXT:    s_lshr_b32 s6, s6, 24
+; GFX7-NEXT:    s_bfe_u32 s11, s0, 0x80010
+; GFX7-NEXT:    s_lshr_b32 s2, s2, 24
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s10
-; GFX7-NEXT:    s_lshr_b32 s4, s4, 24
+; GFX7-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v0, s5, v1, v0
+; GFX7-NEXT:    v_mad_u32_u24 v0, s1, v1, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s9, v2, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s11, v3, v0
-; GFX7-NEXT:    v_mov_b32_e32 v1, s6
-; GFX7-NEXT:    v_mad_u32_u24 v0, s4, v1, v0
-; GFX7-NEXT:    buffer_store_short v0, off, s[0:3], 0
+; GFX7-NEXT:    v_mov_b32_e32 v1, s2
+; GFX7-NEXT:    v_mad_u32_u24 v0, s0, v1, v0
+; GFX7-NEXT:    buffer_store_short v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot4_acc16:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX8-NEXT:    v_mov_b32_e32 v0, s4
+; GFX8-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX8-NEXT:    flat_load_ushort v2, v[0:1]
-; GFX8-NEXT:    s_load_dword s1, s[4:5], 0x0
-; GFX8-NEXT:    s_movk_i32 s0, 0xff
+; GFX8-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX8-NEXT:    s_movk_i32 s3, 0xff
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    s_bfe_u32 s5, s2, 0x80008
 ; GFX8-NEXT:    s_bfe_u32 s7, s2, 0x80010
 ; GFX8-NEXT:    v_mov_b32_e32 v4, s5
-; GFX8-NEXT:    s_and_b32 s3, s1, s0
-; GFX8-NEXT:    s_and_b32 s0, s2, s0
-; GFX8-NEXT:    v_mov_b32_e32 v3, s0
-; GFX8-NEXT:    s_bfe_u32 s4, s1, 0x80008
-; GFX8-NEXT:    s_bfe_u32 s6, s1, 0x80010
+; GFX8-NEXT:    s_and_b32 s1, s0, s3
+; GFX8-NEXT:    s_and_b32 s3, s2, s3
+; GFX8-NEXT:    v_mov_b32_e32 v3, s3
+; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x80008
+; GFX8-NEXT:    s_bfe_u32 s6, s0, 0x80010
 ; GFX8-NEXT:    s_lshr_b32 s2, s2, 24
 ; GFX8-NEXT:    v_mov_b32_e32 v5, s7
-; GFX8-NEXT:    s_lshr_b32 s1, s1, 24
+; GFX8-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
-; GFX8-NEXT:    v_mad_u32_u24 v2, s3, v3, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s4, v4, v2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s6, v5, v2
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
 ; GFX8-NEXT:    flat_store_short v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-NODL-LABEL: udot4_acc16:
 ; GFX9-NODL:       ; %bb.0: ; %entry
-; GFX9-NODL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NODL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NODL-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-NODL-NEXT:    global_load_ushort v2, v[0:1], off
-; GFX9-NODL-NEXT:    s_load_dword s1, s[4:5], 0x0
-; GFX9-NODL-NEXT:    s_movk_i32 s0, 0xff
+; GFX9-NODL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX9-NODL-NEXT:    s_movk_i32 s3, 0xff
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NODL-NEXT:    s_bfe_u32 s5, s2, 0x80008
 ; GFX9-NODL-NEXT:    s_bfe_u32 s7, s2, 0x80010
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v4, s5
-; GFX9-NODL-NEXT:    s_and_b32 s3, s1, s0
-; GFX9-NODL-NEXT:    s_and_b32 s0, s2, s0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s0
-; GFX9-NODL-NEXT:    s_bfe_u32 s4, s1, 0x80008
-; GFX9-NODL-NEXT:    s_bfe_u32 s6, s1, 0x80010
+; GFX9-NODL-NEXT:    s_and_b32 s1, s0, s3
+; GFX9-NODL-NEXT:    s_and_b32 s3, s2, s3
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s3
+; GFX9-NODL-NEXT:    s_bfe_u32 s4, s0, 0x80008
+; GFX9-NODL-NEXT:    s_bfe_u32 s6, s0, 0x80010
 ; GFX9-NODL-NEXT:    s_lshr_b32 s2, s2, 24
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v5, s7
-; GFX9-NODL-NEXT:    s_lshr_b32 s1, s1, 24
+; GFX9-NODL-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX9-NODL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s3, v3, v2
+; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
 ; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s4, v4, v2
 ; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s6, v5, v2
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s2
-; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
+; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
 ; GFX9-NODL-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-NODL-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: udot4_acc16:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX9-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-DL-NEXT:    global_load_ushort v2, v[0:1], off
-; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s1
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-DL-NEXT:    v_dot4_u32_u8 v2, s0, v3, v2
 ; GFX9-DL-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot4_acc16:
 ; GFX10-DL:       ; %bb.0: ; %entry
 ; GFX10-DL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
entry:
 
   store i16 %mad4, i16 addrspace(1)* %dst, align 2
   ret void
 }
 
 define amdgpu_kernel void @udot4_acc8(<4 x i8> addrspace(1)* %src1,
 ; GFX7-LABEL: udot4_acc8:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s2, -1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s4, s[4:5], 0x0
-; GFX7-NEXT:    buffer_load_ubyte v0, off, s[0:3], 0
-; GFX7-NEXT:    s_load_dword s6, s[6:7], 0x0
-; GFX7-NEXT:    s_movk_i32 s5, 0xff
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_bfe_u32 s9, s4, 0x80008
-; GFX7-NEXT:    s_bfe_u32 s11, s4, 0x80010
-; GFX7-NEXT:    s_and_b32 s7, s6, s5
-; GFX7-NEXT:    s_and_b32 s5, s4, s5
-; GFX7-NEXT:    s_bfe_u32 s8, s6, 0x80008
-; GFX7-NEXT:    v_mov_b32_e32 v1, s7
-; GFX7-NEXT:    s_bfe_u32 s10, s6, 0x80010
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX7-NEXT:    buffer_load_ubyte v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX7-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX7-NEXT:    s_movk_i32 s1, 0xff
+; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX7-NEXT:    s_bfe_u32 s9, s0, 0x80008
+; GFX7-NEXT:    s_and_b32 s3, s2, s1
+; GFX7-NEXT:    s_bfe_u32 s8, s2, 0x80008
+; GFX7-NEXT:    s_and_b32 s1, s0, s1
+; GFX7-NEXT:    v_mov_b32_e32 v1, s3
+; GFX7-NEXT:    s_bfe_u32 s10, s2, 0x80010
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s8
-; GFX7-NEXT:    s_lshr_b32 s6, s6, 24
+; GFX7-NEXT:    s_bfe_u32 s11, s0, 0x80010
+; GFX7-NEXT:    s_lshr_b32 s2, s2, 24
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s10
-; GFX7-NEXT:    s_lshr_b32 s4, s4, 24
+; GFX7-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v0, s5, v1, v0
+; GFX7-NEXT:    v_mad_u32_u24 v0, s1, v1, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s9, v2, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s11, v3, v0
-; GFX7-NEXT:    v_mov_b32_e32 v1, s6
-; GFX7-NEXT:    v_mad_u32_u24 v0, s4, v1, v0
-; GFX7-NEXT:    buffer_store_byte v0, off, s[0:3], 0
+; GFX7-NEXT:    v_mov_b32_e32 v1, s2
+; GFX7-NEXT:    v_mad_u32_u24 v0, s0, v1, v0
+; GFX7-NEXT:    buffer_store_byte v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot4_acc8:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s2, s[4:5], 0x0
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
-; GFX8-NEXT:    flat_load_ubyte v2, v[0:1]
-; GFX8-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX8-NEXT:    s_movk_i32 s0, 0xff
+; GFX8-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX8-NEXT:    v_mov_b32_e32 v0, s4
+; GFX8-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_bfe_u32 s5, s2, 0x80008
-; GFX8-NEXT:    s_bfe_u32 s7, s2, 0x80010
-; GFX8-NEXT:    s_and_b32 s3, s1, s0
-; GFX8-NEXT:    s_and_b32 s0, s2, s0
-; GFX8-NEXT:    s_bfe_u32 s4, s1, 0x80008
+; GFX8-NEXT:    flat_load_ubyte v2, v[0:1]
+; GFX8-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX8-NEXT:    s_movk_i32 s1, 0xff
+; GFX8-NEXT:    s_bfe_u32 s5, s0, 0x80008
+; GFX8-NEXT:    s_bfe_u32 s7, s0, 0x80010
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    s_and_b32 s3, s2, s1
+; GFX8-NEXT:    s_and_b32 s1, s0, s1
+; GFX8-NEXT:    s_bfe_u32 s4, s2, 0x80008
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s3
-; GFX8-NEXT:    s_bfe_u32 s6, s1, 0x80010
+; GFX8-NEXT:    s_bfe_u32 s6, s2, 0x80010
 ; GFX8-NEXT:    v_mov_b32_e32 v4, s4
-; GFX8-NEXT:    s_lshr_b32 s1, s1, 24
-; GFX8-NEXT:    v_mov_b32_e32 v5, s6
 ; GFX8-NEXT:    s_lshr_b32 s2, s2, 24
+; GFX8-NEXT:    v_mov_b32_e32 v5, s6
+; GFX8-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
-; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s7, v5, v2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s1
-; GFX8-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
+; GFX8-NEXT:    v_mov_b32_e32 v3, s2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
 ; GFX8-NEXT:    flat_store_byte v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-NODL-LABEL: udot4_acc8:
 ; GFX9-NODL:       ; %bb.0: ; %entry
-; GFX9-NODL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NODL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_load_dword s2, s[4:5], 0x0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NODL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-NODL-NEXT:    global_load_ubyte v2, v[0:1], off
-; GFX9-NODL-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX9-NODL-NEXT:    s_movk_i32 s0, 0xff
+; GFX9-NODL-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX9-NODL-NEXT:    s_movk_i32 s1, 0xff
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_bfe_u32 s5, s2, 0x80008
-; GFX9-NODL-NEXT:    s_bfe_u32 s7, s2, 0x80010
-; GFX9-NODL-NEXT:    s_and_b32 s3, s1, s0
-; GFX9-NODL-NEXT:    s_and_b32 s0, s2, s0
-; GFX9-NODL-NEXT:    s_bfe_u32 s4, s1, 0x80008
+; GFX9-NODL-NEXT:    s_bfe_u32 s5, s0, 0x80008
+; GFX9-NODL-NEXT:    s_bfe_u32 s7, s0, 0x80010
+; GFX9-NODL-NEXT:    s_and_b32 s3, s2, s1
+; GFX9-NODL-NEXT:    s_and_b32 s1, s0, s1
+; GFX9-NODL-NEXT:    s_bfe_u32 s4, s2, 0x80008
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s3
-; GFX9-NODL-NEXT:    s_bfe_u32 s6, s1, 0x80010
+; GFX9-NODL-NEXT:    s_bfe_u32 s6, s2, 0x80010
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v4, s4
-; GFX9-NODL-NEXT:    s_lshr_b32 s1, s1, 24
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v5, s6
 ; GFX9-NODL-NEXT:    s_lshr_b32 s2, s2, 24
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v5, s6
+; GFX9-NODL-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX9-NODL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
 ; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
 ; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s7, v5, v2
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s2
+; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
 ; GFX9-NODL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-NODL-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: udot4_acc8:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX9-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-DL-NEXT:    global_load_ubyte v2, v[0:1], off
-; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s1
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-DL-NEXT:    v_dot4_u32_u8 v2, s0, v3, v2
 ; GFX9-DL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot4_acc8:
 ; GFX10-DL:       ; %bb.0: ; %entry
 ; GFX10-DL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
entry:
   store i8 %mad4, i8 addrspace(1)* %dst, align 2
   ret void
 }
 
 ; TODO: Generate udot4?
 define amdgpu_kernel void @udot2_8(<4 x i8> addrspace(1)* %src1,
 ; GFX7-LABEL: udot2_8:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
 ; GFX7-NEXT:    s_movk_i32 s8, 0xff
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s4, s[4:5], 0x0
-; GFX7-NEXT:    buffer_load_ubyte v0, off, s[0:3], 0
-; GFX7-NEXT:    s_load_dword s5, s[6:7], 0x0
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_and_b32 s7, s4, s8
-; GFX7-NEXT:    s_bfe_u32 s4, s4, 0x80008
-; GFX7-NEXT:    s_and_b32 s6, s5, s8
-; GFX7-NEXT:    v_mov_b32_e32 v1, s6
-; GFX7-NEXT:    s_bfe_u32 s5, s5, 0x80008
+; GFX7-NEXT:    buffer_load_ubyte v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX7-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX7-NEXT:    s_and_b32 s3, s0, s8
+; GFX7-NEXT:    s_and_b32 s2, s1, s8
+; GFX7-NEXT:    v_mov_b32_e32 v1, s2
+; GFX7-NEXT:    s_bfe_u32 s1, s1, 0x80008
+; GFX7-NEXT:    s_bfe_u32 s0, s0, 0x80008
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v0, s7, v1, v0
-; GFX7-NEXT:    v_mov_b32_e32 v1, s5
-; GFX7-NEXT:    v_mad_u32_u24 v0, s4, v1, v0
-; GFX7-NEXT:    buffer_store_byte v0, off, s[0:3], 0
+; GFX7-NEXT:    v_mad_u32_u24 v0, s3, v1, v0
+; GFX7-NEXT:    v_mov_b32_e32 v1, s1
+; GFX7-NEXT:    v_mad_u32_u24 v0, s0, v1, v0
+; GFX7-NEXT:    buffer_store_byte v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot2_8:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
+; GFX8-NEXT:    s_movk_i32 s4, 0xff
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    v_mov_b32_e32 v0, s2
+; GFX8-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX8-NEXT:    flat_load_ubyte v2, v[0:1]
-; GFX8-NEXT:    s_load_dword s1, s[4:5], 0x0
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    s_movk_i32 s0, 0xff
+; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_and_b32 s3, s2, s0
-; GFX8-NEXT:    s_and_b32 s0, s1, s0
-; GFX8-NEXT:    v_mov_b32_e32 v3, s3
-; GFX8-NEXT:    s_bfe_u32 s2, s2, 0x80008
+; GFX8-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX8-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    s_and_b32 s3, s0, s4
+; GFX8-NEXT:    s_and_b32 s2, s1, s4
+; GFX8-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX8-NEXT:    s_bfe_u32 s1, s1, 0x80008
+; GFX8-NEXT:    s_bfe_u32 s0, s0, 0x80008
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
+; GFX8-NEXT:    v_mad_u32_u24 v2, s3, v3, v2
+; GFX8-NEXT:    v_mov_b32_e32 v3, s1
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
 ; GFX8-NEXT:    flat_store_byte v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-NODL-LABEL: udot2_8:
 ; GFX9-NODL:       ; %bb.0: ; %entry
-; GFX9-NODL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NODL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_movk_i32 s4, 0xff
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s2
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX9-NODL-NEXT:    global_load_ubyte v2, v[0:1], off
-; GFX9-NODL-NEXT:    s_load_dword s1, s[4:5], 0x0
-; GFX9-NODL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-NODL-NEXT:    s_movk_i32 s0, 0xff
+; GFX9-NODL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_and_b32 s3, s2, s0
-; GFX9-NODL-NEXT:    s_and_b32 s0, s1, s0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s3
-; GFX9-NODL-NEXT:    s_bfe_u32 s2, s2, 0x80008
+; GFX9-NODL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX9-NODL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NODL-NEXT:    s_and_b32 s3, s0, s4
+; GFX9-NODL-NEXT:    s_and_b32 s2, s1, s4
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX9-NODL-NEXT:    s_bfe_u32 s1, s1, 0x80008
+; GFX9-NODL-NEXT:    s_bfe_u32 s0, s0, 0x80008
 ; GFX9-NODL-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s3, v3, v2
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s1
 ; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s2
-; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
 ; GFX9-NODL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-NODL-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: udot2_8:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_movk_i32 s4, 0xff
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX9-DL-NEXT:    global_load_ubyte v2, v[0:1], off
-; GFX9-DL-NEXT:    s_load_dword s1, s[4:5], 0x0
-; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-DL-NEXT:    s_movk_i32 s0, 0xff
+; GFX9-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_and_b32 s3, s2, s0
-; GFX9-DL-NEXT:    s_and_b32 s0, s1, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s3
-; GFX9-DL-NEXT:    s_bfe_u32 s2, s2, 0x80008
+; GFX9-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX9-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-DL-NEXT:    s_and_b32 s3, s0, s4
+; GFX9-DL-NEXT:    s_and_b32 s2, s1, s4
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX9-DL-NEXT:    s_bfe_u32 s1, s1, 0x80008
+; GFX9-DL-NEXT:    s_bfe_u32 s0, s0, 0x80008
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s3, v3, v2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s1
 ; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
 ; GFX9-DL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot2_8:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
-; GFX10-DL-NEXT:    s_movk_i32 s2, 0xff
+; GFX10-DL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_movk_i32 s4, 0xff
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-DL-NEXT:    s_load_dword s0, s[6:7], 0x0
-; GFX10-DL-NEXT:    s_load_dword s1, s[4:5], 0x0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s3
+; GFX10-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    s_and_b32 s3, s0, s2
-; GFX10-DL-NEXT:    s_and_b32 s2, s1, s2
+; GFX10-DL-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX10-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-DL-NEXT:    s_and_b32 s1, s2, s4
+; GFX10-DL-NEXT:    s_and_b32 s3, s0, s4
+; GFX10-DL-NEXT:    s_bfe_u32 s2, s2, 0x80008
 ; GFX10-DL-NEXT:    s_bfe_u32 s0, s0, 0x80008
-; GFX10-DL-NEXT:    s_bfe_u32 s1, s1, 0x80008
 ; GFX10-DL-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-DL-NEXT:    v_mad_u32_u24 v2, s2, s3, v2
-; GFX10-DL-NEXT:    v_mad_u32_u24 v2, s1, s0, v2
+; GFX10-DL-NEXT:    v_mad_u32_u24 v2, s3, s1, v2
+; GFX10-DL-NEXT:    v_mad_u32_u24 v2, s0, s2, v2
 ; GFX10-DL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX10-DL-NEXT:    s_endpgm
                                    <4 x i8> addrspace(1)* %src2,
                                    i8 addrspace(1)* nocapture %dst) {
 entry:
   %vec1 = load <4 x i8>, <4 x i8> addrspace(1)* %src1
   %vec2 = load <4 x i8>, <4 x i8> addrspace(1)* %src2
 
entry:
   %mad2 = add i8 %mad1, %mul2
   store i8 %mad2, i8 addrspace(1)* %dst, align 2
   ret void
 }
 
 define amdgpu_kernel void @udot4_CommutationInsideMAD(<4 x i8> addrspace(1)* %src1,
 ; GFX7-LABEL: udot4_CommutationInsideMAD:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
 ; GFX7-NEXT:    s_movk_i32 s8, 0xff
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s4, s[4:5], 0x0
-; GFX7-NEXT:    buffer_load_ubyte v0, off, s[0:3], 0
-; GFX7-NEXT:    s_load_dword s5, s[6:7], 0x0
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_and_b32 s6, s4, s8
-; GFX7-NEXT:    v_mov_b32_e32 v1, s6
-; GFX7-NEXT:    s_and_b32 s7, s5, s8
-; GFX7-NEXT:    s_bfe_u32 s8, s4, 0x80008
-; GFX7-NEXT:    s_bfe_u32 s10, s4, 0x80010
-; GFX7-NEXT:    s_bfe_u32 s9, s5, 0x80008
+; GFX7-NEXT:    buffer_load_ubyte v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX7-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX7-NEXT:    s_and_b32 s2, s0, s8
+; GFX7-NEXT:    s_and_b32 s3, s1, s8
+; GFX7-NEXT:    s_bfe_u32 s8, s0, 0x80008
+; GFX7-NEXT:    v_mov_b32_e32 v1, s2
+; GFX7-NEXT:    s_bfe_u32 s10, s0, 0x80010
+; GFX7-NEXT:    s_bfe_u32 s9, s1, 0x80008
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s8
-; GFX7-NEXT:    s_bfe_u32 s11, s5, 0x80010
-; GFX7-NEXT:    s_lshr_b32 s4, s4, 24
+; GFX7-NEXT:    s_bfe_u32 s11, s1, 0x80010
+; GFX7-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s10
-; GFX7-NEXT:    s_lshr_b32 s5, s5, 24
+; GFX7-NEXT:    s_lshr_b32 s1, s1, 24
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v0, s7, v1, v0
+; GFX7-NEXT:    v_mad_u32_u24 v0, s3, v1, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s9, v2, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s11, v3, v0
-; GFX7-NEXT:    v_mov_b32_e32 v1, s4
-; GFX7-NEXT:    v_mad_u32_u24 v0, s5, v1, v0
-; GFX7-NEXT:    buffer_store_byte v0, off, s[0:3], 0
+; GFX7-NEXT:    v_mov_b32_e32 v1, s0
+; GFX7-NEXT:    v_mad_u32_u24 v0, s1, v1, v0
+; GFX7-NEXT:    buffer_store_byte v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot4_CommutationInsideMAD:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
+; GFX8-NEXT:    s_movk_i32 s4, 0xff
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    v_mov_b32_e32 v0, s2
+; GFX8-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX8-NEXT:    flat_load_ubyte v2, v[0:1]
-; GFX8-NEXT:    s_load_dword s1, s[4:5], 0x0
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    s_movk_i32 s0, 0xff
-; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_and_b32 s3, s1, s0
-; GFX8-NEXT:    s_bfe_u32 s4, s1, 0x80008
-; GFX8-NEXT:    s_and_b32 s0, s2, s0
-; GFX8-NEXT:    v_mov_b32_e32 v3, s3
-; GFX8-NEXT:    s_bfe_u32 s6, s1, 0x80010
-; GFX8-NEXT:    s_bfe_u32 s5, s2, 0x80008
+; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX8-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    s_and_b32 s2, s0, s4
+; GFX8-NEXT:    s_and_b32 s3, s1, s4
+; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x80008
+; GFX8-NEXT:    v_mov_b32_e32 v3, s2
+; GFX8-NEXT:    s_bfe_u32 s6, s0, 0x80010
+; GFX8-NEXT:    s_bfe_u32 s5, s1, 0x80008
 ; GFX8-NEXT:    v_mov_b32_e32 v4, s4
-; GFX8-NEXT:    s_bfe_u32 s7, s2, 0x80010
-; GFX8-NEXT:    s_lshr_b32 s1, s1, 24
+; GFX8-NEXT:    s_bfe_u32 s7, s1, 0x80010
+; GFX8-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX8-NEXT:    v_mov_b32_e32 v5, s6
-; GFX8-NEXT:    s_lshr_b32 s2, s2, 24
+; GFX8-NEXT:    s_lshr_b32 s1, s1, 24
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
-; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s3, v3, v2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s7, v5, v2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s1
-; GFX8-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
+; GFX8-NEXT:    v_mov_b32_e32 v3, s0
+; GFX8-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
 ; GFX8-NEXT:    flat_store_byte v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-NODL-LABEL: udot4_CommutationInsideMAD:
 ; GFX9-NODL:       ; %bb.0: ; %entry
-; GFX9-NODL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NODL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_movk_i32 s4, 0xff
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s2
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX9-NODL-NEXT:    global_load_ubyte v2, v[0:1], off
-; GFX9-NODL-NEXT:    s_load_dword s1, s[4:5], 0x0
-; GFX9-NODL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-NODL-NEXT:    s_movk_i32 s0, 0xff
-; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_and_b32 s3, s1, s0
-; GFX9-NODL-NEXT:    s_bfe_u32 s4, s1, 0x80008
-; GFX9-NODL-NEXT:    s_and_b32 s0, s2, s0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s3
-; GFX9-NODL-NEXT:    s_bfe_u32 s6, s1, 0x80010
-; GFX9-NODL-NEXT:    s_bfe_u32 s5, s2, 0x80008
+; GFX9-NODL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NODL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX9-NODL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NODL-NEXT:    s_and_b32 s2, s0, s4
+; GFX9-NODL-NEXT:    s_and_b32 s3, s1, s4
+; GFX9-NODL-NEXT:    s_bfe_u32 s4, s0, 0x80008
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s2
+; GFX9-NODL-NEXT:    s_bfe_u32 s6, s0, 0x80010
+; GFX9-NODL-NEXT:    s_bfe_u32 s5, s1, 0x80008
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v4, s4
-; GFX9-NODL-NEXT:    s_bfe_u32 s7, s2, 0x80010
-; GFX9-NODL-NEXT:    s_lshr_b32 s1, s1, 24
+; GFX9-NODL-NEXT:    s_bfe_u32 s7, s1, 0x80010
+; GFX9-NODL-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v5, s6
-; GFX9-NODL-NEXT:    s_lshr_b32 s2, s2, 24
+; GFX9-NODL-NEXT:    s_lshr_b32 s1, s1, 24
 ; GFX9-NODL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s3, v3, v2
 ; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
 ; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s7, v5, v2
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s0
+; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
 ; GFX9-NODL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-NODL-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: udot4_CommutationInsideMAD:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s2, s[4:5], 0x0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX9-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-DL-NEXT:    global_load_ubyte v2, v[0:1], off
-; GFX9-DL-NEXT:    s_load_dword s0, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s0
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-DL-NEXT:    v_dot4_u32_u8 v2, s0, v3, v2
+; GFX9-DL-NEXT:    v_dot4_u32_u8 v2, s1, v3, v2
 ; GFX9-DL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot4_CommutationInsideMAD:
 ; GFX10-DL:       ; %bb.0: ; %entry
 ; GFX10-DL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
entry:
   store i8 %mad4, i8 addrspace(1)* %dst, align 2
   ret void
 }
 
 ; TODO: Support commutation accross the adds.
 define amdgpu_kernel void @udot4_CommutationAccrossMADs(<4 x i8> addrspace(1)* %src1,
 ; GFX7-LABEL: udot4_CommutationAccrossMADs:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
 ; GFX7-NEXT:    s_movk_i32 s8, 0xff
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s4, s[4:5], 0x0
-; GFX7-NEXT:    buffer_load_ubyte v0, off, s[0:3], 0
-; GFX7-NEXT:    s_load_dword s5, s[6:7], 0x0
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_and_b32 s6, s4, s8
-; GFX7-NEXT:    s_bfe_u32 s10, s4, 0x80010
-; GFX7-NEXT:    s_and_b32 s7, s5, s8
-; GFX7-NEXT:    s_bfe_u32 s8, s4, 0x80008
-; GFX7-NEXT:    s_bfe_u32 s9, s5, 0x80008
+; GFX7-NEXT:    buffer_load_ubyte v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX7-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX7-NEXT:    s_and_b32 s2, s0, s8
+; GFX7-NEXT:    s_and_b32 s3, s1, s8
+; GFX7-NEXT:    s_bfe_u32 s8, s0, 0x80008
+; GFX7-NEXT:    s_bfe_u32 s9, s1, 0x80008
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s8
-; GFX7-NEXT:    v_mov_b32_e32 v2, s6
-; GFX7-NEXT:    s_bfe_u32 s11, s5, 0x80010
-; GFX7-NEXT:    s_lshr_b32 s4, s4, 24
+; GFX7-NEXT:    s_bfe_u32 s10, s0, 0x80010
+; GFX7-NEXT:    v_mov_b32_e32 v2, s2
+; GFX7-NEXT:    s_bfe_u32 s11, s1, 0x80010
+; GFX7-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s10
-; GFX7-NEXT:    s_lshr_b32 s5, s5, 24
+; GFX7-NEXT:    s_lshr_b32 s1, s1, 24
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s9, v1, v0
-; GFX7-NEXT:    v_mad_u32_u24 v0, s7, v2, v0
+; GFX7-NEXT:    v_mad_u32_u24 v0, s3, v2, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s11, v3, v0
-; GFX7-NEXT:    v_mov_b32_e32 v1, s4
-; GFX7-NEXT:    v_mad_u32_u24 v0, s5, v1, v0
-; GFX7-NEXT:    buffer_store_byte v0, off, s[0:3], 0
+; GFX7-NEXT:    v_mov_b32_e32 v1, s0
+; GFX7-NEXT:    v_mad_u32_u24 v0, s1, v1, v0
+; GFX7-NEXT:    buffer_store_byte v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot4_CommutationAccrossMADs:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
+; GFX8-NEXT:    s_movk_i32 s4, 0xff
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    v_mov_b32_e32 v0, s2
+; GFX8-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX8-NEXT:    flat_load_ubyte v2, v[0:1]
-; GFX8-NEXT:    s_load_dword s1, s[4:5], 0x0
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    s_movk_i32 s0, 0xff
+; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_bfe_u32 s4, s1, 0x80008
-; GFX8-NEXT:    s_and_b32 s3, s1, s0
-; GFX8-NEXT:    s_bfe_u32 s5, s2, 0x80008
+; GFX8-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX8-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    s_and_b32 s2, s0, s4
+; GFX8-NEXT:    s_and_b32 s3, s1, s4
+; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x80008
+; GFX8-NEXT:    s_bfe_u32 s5, s1, 0x80008
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s4
-; GFX8-NEXT:    s_bfe_u32 s6, s1, 0x80010
-; GFX8-NEXT:    s_and_b32 s0, s2, s0
-; GFX8-NEXT:    v_mov_b32_e32 v4, s3
-; GFX8-NEXT:    s_bfe_u32 s7, s2, 0x80010
-; GFX8-NEXT:    s_lshr_b32 s1, s1, 24
+; GFX8-NEXT:    s_bfe_u32 s6, s0, 0x80010
+; GFX8-NEXT:    v_mov_b32_e32 v4, s2
+; GFX8-NEXT:    s_bfe_u32 s7, s1, 0x80010
+; GFX8-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX8-NEXT:    v_mov_b32_e32 v5, s6
-; GFX8-NEXT:    s_lshr_b32 s2, s2, 24
+; GFX8-NEXT:    s_lshr_b32 s1, s1, 24
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v3, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v4, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s3, v4, v2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s7, v5, v2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s1
-; GFX8-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
+; GFX8-NEXT:    v_mov_b32_e32 v3, s0
+; GFX8-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
 ; GFX8-NEXT:    flat_store_byte v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-NODL-LABEL: udot4_CommutationAccrossMADs:
 ; GFX9-NODL:       ; %bb.0: ; %entry
-; GFX9-NODL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NODL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_movk_i32 s4, 0xff
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s2
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX9-NODL-NEXT:    global_load_ubyte v2, v[0:1], off
-; GFX9-NODL-NEXT:    s_load_dword s1, s[4:5], 0x0
-; GFX9-NODL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-NODL-NEXT:    s_movk_i32 s0, 0xff
+; GFX9-NODL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_bfe_u32 s4, s1, 0x80008
-; GFX9-NODL-NEXT:    s_and_b32 s3, s1, s0
-; GFX9-NODL-NEXT:    s_bfe_u32 s5, s2, 0x80008
+; GFX9-NODL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX9-NODL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NODL-NEXT:    s_and_b32 s2, s0, s4
+; GFX9-NODL-NEXT:    s_and_b32 s3, s1, s4
+; GFX9-NODL-NEXT:    s_bfe_u32 s4, s0, 0x80008
+; GFX9-NODL-NEXT:    s_bfe_u32 s5, s1, 0x80008
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s4
-; GFX9-NODL-NEXT:    s_bfe_u32 s6, s1, 0x80010
-; GFX9-NODL-NEXT:    s_and_b32 s0, s2, s0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v4, s3
-; GFX9-NODL-NEXT:    s_bfe_u32 s7, s2, 0x80010
-; GFX9-NODL-NEXT:    s_lshr_b32 s1, s1, 24
+; GFX9-NODL-NEXT:    s_bfe_u32 s6, s0, 0x80010
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v4, s2
+; GFX9-NODL-NEXT:    s_bfe_u32 s7, s1, 0x80010
+; GFX9-NODL-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v5, s6
-; GFX9-NODL-NEXT:    s_lshr_b32 s2, s2, 24
+; GFX9-NODL-NEXT:    s_lshr_b32 s1, s1, 24
 ; GFX9-NODL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s5, v3, v2
-; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s0, v4, v2
+; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s3, v4, v2
 ; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s7, v5, v2
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s0
+; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
 ; GFX9-NODL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-NODL-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: udot4_CommutationAccrossMADs:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_movk_i32 s4, 0xff
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX9-DL-NEXT:    global_load_ubyte v2, v[0:1], off
-; GFX9-DL-NEXT:    s_load_dword s1, s[4:5], 0x0
-; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-DL-NEXT:    s_movk_i32 s0, 0xff
+; GFX9-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_bfe_u32 s4, s1, 0x80008
-; GFX9-DL-NEXT:    s_and_b32 s3, s1, s0
-; GFX9-DL-NEXT:    s_bfe_u32 s5, s2, 0x80008
+; GFX9-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX9-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-DL-NEXT:    s_and_b32 s2, s0, s4
+; GFX9-DL-NEXT:    s_and_b32 s3, s1, s4
+; GFX9-DL-NEXT:    s_bfe_u32 s4, s0, 0x80008
+; GFX9-DL-NEXT:    s_bfe_u32 s5, s1, 0x80008
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s4
-; GFX9-DL-NEXT:    s_bfe_u32 s6, s1, 0x80010
-; GFX9-DL-NEXT:    s_and_b32 s0, s2, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s3
-; GFX9-DL-NEXT:    s_bfe_u32 s7, s2, 0x80010
-; GFX9-DL-NEXT:    s_lshr_b32 s1, s1, 24
+; GFX9-DL-NEXT:    s_bfe_u32 s6, s0, 0x80010
+; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s2
+; GFX9-DL-NEXT:    s_bfe_u32 s7, s1, 0x80010
+; GFX9-DL-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v5, s6
-; GFX9-DL-NEXT:    s_lshr_b32 s2, s2, 24
+; GFX9-DL-NEXT:    s_lshr_b32 s1, s1, 24
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s5, v3, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s0, v4, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s3, v4, v2
 ; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s7, v5, v2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s0
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
 ; GFX9-DL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot4_CommutationAccrossMADs:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
-; GFX10-DL-NEXT:    s_movk_i32 s2, 0xff
+; GFX10-DL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_movk_i32 s4, 0xff
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
-; GFX10-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s3
+; GFX10-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    s_bfe_u32 s3, s0, 0x80008
-; GFX10-DL-NEXT:    s_bfe_u32 s4, s1, 0x80008
-; GFX10-DL-NEXT:    s_and_b32 s5, s0, s2
-; GFX10-DL-NEXT:    s_and_b32 s2, s1, s2
+; GFX10-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX10-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-DL-NEXT:    s_bfe_u32 s2, s0, 0x80008
+; GFX10-DL-NEXT:    s_bfe_u32 s3, s1, 0x80008
+; GFX10-DL-NEXT:    s_and_b32 s5, s0, s4
+; GFX10-DL-NEXT:    s_and_b32 s4, s1, s4
 ; GFX10-DL-NEXT:    s_waitcnt vmcnt(0)
-; GFX10-DL-NEXT:    v_mad_u32_u24 v2, s4, s3, v2
-; GFX10-DL-NEXT:    s_bfe_u32 s3, s0, 0x80010
-; GFX10-DL-NEXT:    s_bfe_u32 s4, s1, 0x80010
+; GFX10-DL-NEXT:    v_mad_u32_u24 v2, s3, s2, v2
+; GFX10-DL-NEXT:    s_bfe_u32 s2, s0, 0x80010
+; GFX10-DL-NEXT:    s_bfe_u32 s3, s1, 0x80010
 ; GFX10-DL-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX10-DL-NEXT:    s_lshr_b32 s1, s1, 24
-; GFX10-DL-NEXT:    v_mad_u32_u24 v2, s2, s5, v2
-; GFX10-DL-NEXT:    v_mad_u32_u24 v2, s4, s3, v2
+; GFX10-DL-NEXT:    v_mad_u32_u24 v2, s4, s5, v2
+; GFX10-DL-NEXT:    v_mad_u32_u24 v2, s3, s2, v2
 ; GFX10-DL-NEXT:    v_mad_u32_u24 v2, s1, s0, v2
 ; GFX10-DL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX10-DL-NEXT:    s_endpgm
                                                         <4 x i8> addrspace(1)* %src2,
                                                         i8 addrspace(1)* nocapture %dst) {
 entry:
   %vec1 = load <4 x i8>, <4 x i8> addrspace(1)* %src1
   %vec2 = load <4 x i8>, <4 x i8> addrspace(1)* %src2
entry:
   %res = add i32 %add4, %add
   store i32 %res, i32 addrspace(1)* %dst, align 4
   ret void
 }
 
 define amdgpu_kernel void @notdot4_mixedtypes(<4 x i8> addrspace(1)* %src1,
 ; GFX7-LABEL: notdot4_mixedtypes:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
 ; GFX7-NEXT:    s_mov_b32 s8, 0xffff
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s6, s[6:7], 0x0
-; GFX7-NEXT:    buffer_load_ushort v0, off, s[0:3], 0
-; GFX7-NEXT:    s_load_dword s4, s[4:5], 0x0
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_sext_i32_i8 s7, s6
-; GFX7-NEXT:    s_bfe_u32 s9, s6, 0x80008
-; GFX7-NEXT:    s_sext_i32_i8 s5, s4
-; GFX7-NEXT:    s_and_b32 s7, s7, s8
-; GFX7-NEXT:    s_bfe_u32 s10, s4, 0x80008
+; GFX7-NEXT:    buffer_load_ushort v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX7-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX7-NEXT:    s_sext_i32_i8 s3, s2
+; GFX7-NEXT:    s_bfe_u32 s9, s2, 0x80008
+; GFX7-NEXT:    s_sext_i32_i8 s1, s0
+; GFX7-NEXT:    s_and_b32 s3, s3, s8
+; GFX7-NEXT:    s_bfe_u32 s10, s0, 0x80008
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s9
-; GFX7-NEXT:    s_bfe_u32 s11, s6, 0x80010
-; GFX7-NEXT:    s_and_b32 s5, s5, s8
-; GFX7-NEXT:    v_mov_b32_e32 v3, s7
-; GFX7-NEXT:    s_bfe_u32 s12, s4, 0x80010
-; GFX7-NEXT:    s_lshr_b32 s6, s6, 24
+; GFX7-NEXT:    s_bfe_u32 s11, s2, 0x80010
+; GFX7-NEXT:    s_and_b32 s1, s1, s8
+; GFX7-NEXT:    v_mov_b32_e32 v3, s3
+; GFX7-NEXT:    s_bfe_u32 s12, s0, 0x80010
+; GFX7-NEXT:    s_lshr_b32 s2, s2, 24
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s11
-; GFX7-NEXT:    s_lshr_b32 s4, s4, 24
+; GFX7-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s10, v1, v0
-; GFX7-NEXT:    v_mad_u32_u24 v0, s5, v3, v0
+; GFX7-NEXT:    v_mad_u32_u24 v0, s1, v3, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s12, v2, v0
-; GFX7-NEXT:    v_mov_b32_e32 v1, s6
-; GFX7-NEXT:    v_mad_u32_u24 v0, s4, v1, v0
-; GFX7-NEXT:    buffer_store_short v0, off, s[0:3], 0
+; GFX7-NEXT:    v_mov_b32_e32 v1, s2
+; GFX7-NEXT:    v_mad_u32_u24 v0, s0, v1, v0
+; GFX7-NEXT:    buffer_store_short v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: notdot4_mixedtypes:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX8-NEXT:    v_mov_b32_e32 v0, s4
+; GFX8-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX8-NEXT:    flat_load_ushort v2, v[0:1]
-; GFX8-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX8-NEXT:    s_load_dword s0, s[0:1], 0x0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_bfe_u32 s1, s2, 0x80008
+; GFX8-NEXT:    s_bfe_u32 s5, s2, 0x80008
 ; GFX8-NEXT:    s_sext_i32_i8 s3, s2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s1
-; GFX8-NEXT:    s_bfe_u32 s5, s2, 0x80010
-; GFX8-NEXT:    s_bfe_u32 s1, s0, 0x80008
+; GFX8-NEXT:    v_mov_b32_e32 v3, s5
+; GFX8-NEXT:    s_bfe_u32 s7, s2, 0x80010
+; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x80008
+; GFX8-NEXT:    s_sext_i32_i8 s1, s0
 ; GFX8-NEXT:    v_mov_b32_e32 v4, s3
-; GFX8-NEXT:    s_sext_i32_i8 s4, s0
-; GFX8-NEXT:    s_bfe_u32 s3, s0, 0x80010
+; GFX8-NEXT:    s_bfe_u32 s6, s0, 0x80010
 ; GFX8-NEXT:    s_lshr_b32 s2, s2, 24
-; GFX8-NEXT:    v_mov_b32_e32 v5, s5
+; GFX8-NEXT:    v_mov_b32_e32 v5, s7
 ; GFX8-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
-; GFX8-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX8-NEXT:    v_mad_i32_i24 v2, s4, v4, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s3, v5, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s4, v3, v2
+; GFX8-NEXT:    v_mad_i32_i24 v2, s1, v4, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s6, v5, v2
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
 ; GFX8-NEXT:    flat_store_short v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-NODL-LABEL: notdot4_mixedtypes:
 ; GFX9-NODL:       ; %bb.0: ; %entry
-; GFX9-NODL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NODL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX9-NODL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NODL-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-NODL-NEXT:    global_load_ushort v2, v[0:1], off
-; GFX9-NODL-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-NODL-NEXT:    s_load_dword s0, s[0:1], 0x0
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_bfe_u32 s1, s2, 0x80008
+; GFX9-NODL-NEXT:    s_bfe_u32 s5, s2, 0x80008
 ; GFX9-NODL-NEXT:    s_sext_i32_i8 s3, s2
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-NODL-NEXT:    s_bfe_u32 s5, s2, 0x80010
-; GFX9-NODL-NEXT:    s_bfe_u32 s1, s0, 0x80008
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s5
+; GFX9-NODL-NEXT:    s_bfe_u32 s7, s2, 0x80010
+; GFX9-NODL-NEXT:    s_bfe_u32 s4, s0, 0x80008
+; GFX9-NODL-NEXT:    s_sext_i32_i8 s1, s0
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v4, s3
-; GFX9-NODL-NEXT:    s_sext_i32_i8 s4, s0
-; GFX9-NODL-NEXT:    s_bfe_u32 s3, s0, 0x80010
+; GFX9-NODL-NEXT:    s_bfe_u32 s6, s0, 0x80010
 ; GFX9-NODL-NEXT:    s_lshr_b32 s2, s2, 24
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v5, s5
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v5, s7
 ; GFX9-NODL-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX9-NODL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX9-NODL-NEXT:    v_mad_i32_i24 v2, s4, v4, v2
-; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s3, v5, v2
+; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s4, v3, v2
+; GFX9-NODL-NEXT:    v_mad_i32_i24 v2, s1, v4, v2
+; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s6, v5, v2
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX9-NODL-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
 ; GFX9-NODL-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-NODL-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: notdot4_mixedtypes:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-DL-NEXT:    global_load_ushort v2, v[0:1], off
-; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_bfe_u32 s1, s2, 0x80008
+; GFX9-DL-NEXT:    s_bfe_u32 s5, s2, 0x80008
 ; GFX9-DL-NEXT:    s_sext_i32_i8 s3, s2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-DL-NEXT:    s_bfe_u32 s5, s2, 0x80010
-; GFX9-DL-NEXT:    s_bfe_u32 s1, s0, 0x80008
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s5
+; GFX9-DL-NEXT:    s_bfe_u32 s7, s2, 0x80010
+; GFX9-DL-NEXT:    s_bfe_u32 s4, s0, 0x80008
+; GFX9-DL-NEXT:    s_sext_i32_i8 s1, s0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s3
-; GFX9-DL-NEXT:    s_sext_i32_i8 s4, s0
-; GFX9-DL-NEXT:    s_bfe_u32 s3, s0, 0x80010
+; GFX9-DL-NEXT:    s_bfe_u32 s6, s0, 0x80010
 ; GFX9-DL-NEXT:    s_lshr_b32 s2, s2, 24
-; GFX9-DL-NEXT:    v_mov_b32_e32 v5, s5
+; GFX9-DL-NEXT:    v_mov_b32_e32 v5, s7
 ; GFX9-DL-NEXT:    s_lshr_b32 s0, s0, 24
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s4, v4, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s3, v5, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s4, v3, v2
+; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s1, v4, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s6, v5, v2
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
 ; GFX9-DL-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: notdot4_mixedtypes:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
-; GFX10-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s3
+; GFX10-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX10-DL-NEXT:    global_load_ushort v2, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX10-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_bfe_u32 s2, s0, 0x80008
 ; GFX10-DL-NEXT:    s_bfe_u32 s3, s1, 0x80008
 ; GFX10-DL-NEXT:    s_sext_i32_i8 s4, s0
 ; GFX10-DL-NEXT:    s_sext_i32_i8 s5, s1
 ; GFX10-DL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX10-DL-NEXT:    v_mad_u32_u24 v2, s2, s3, v2
 ; GFX10-DL-NEXT:    s_bfe_u32 s2, s0, 0x80010
 ; GFX10-DL-NEXT:    s_bfe_u32 s3, s1, 0x80010
entry:
   store i32 %add4, i32 addrspace(1)* %dst, align 4
   ret void
 }
 
 ; TODO: This pattern should be recognized.
 define amdgpu_kernel void @udot4_acc16_vecMul(<4 x i8> addrspace(1)* %src1,
 ; GFX7-LABEL: udot4_acc16_vecMul:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s2, -1
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s6, s[6:7], 0x0
-; GFX7-NEXT:    buffer_load_ushort v0, off, s[0:3], 0
-; GFX7-NEXT:    s_load_dword s4, s[4:5], 0x0
-; GFX7-NEXT:    s_movk_i32 s7, 0xff
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_bfe_u32 s10, s6, 0x80008
-; GFX7-NEXT:    s_bfe_u32 s12, s6, 0x80010
-; GFX7-NEXT:    s_lshr_b32 s9, s6, 24
-; GFX7-NEXT:    s_and_b32 s6, s6, s7
-; GFX7-NEXT:    s_lshr_b32 s5, s4, 24
-; GFX7-NEXT:    s_bfe_u32 s8, s4, 0x80008
-; GFX7-NEXT:    s_bfe_u32 s11, s4, 0x80010
-; GFX7-NEXT:    s_and_b32 s4, s4, s7
-; GFX7-NEXT:    v_mov_b32_e32 v1, s6
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX7-NEXT:    buffer_load_ushort v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s2, s[2:3], 0x0
+; GFX7-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX7-NEXT:    s_movk_i32 s3, 0xff
+; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX7-NEXT:    s_bfe_u32 s10, s2, 0x80008
+; GFX7-NEXT:    s_bfe_u32 s12, s2, 0x80010
+; GFX7-NEXT:    s_lshr_b32 s9, s2, 24
+; GFX7-NEXT:    s_and_b32 s2, s2, s3
+; GFX7-NEXT:    s_lshr_b32 s1, s0, 24
+; GFX7-NEXT:    s_bfe_u32 s8, s0, 0x80008
+; GFX7-NEXT:    s_bfe_u32 s11, s0, 0x80010
+; GFX7-NEXT:    s_and_b32 s0, s0, s3
+; GFX7-NEXT:    v_mov_b32_e32 v1, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s10
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s12
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
-; GFX7-NEXT:    v_mad_u32_u24 v0, s4, v1, v0
+; GFX7-NEXT:    v_mad_u32_u24 v0, s0, v1, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s8, v2, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s11, v3, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s9
-; GFX7-NEXT:    v_mad_u32_u24 v0, s5, v1, v0
-; GFX7-NEXT:    buffer_store_short v0, off, s[0:3], 0
+; GFX7-NEXT:    v_mad_u32_u24 v0, s1, v1, v0
+; GFX7-NEXT:    buffer_store_short v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot4_acc16_vecMul:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
+; GFX8-NEXT:    s_movk_i32 s4, 0xff
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    v_mov_b32_e32 v0, s2
+; GFX8-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX8-NEXT:    flat_load_ushort v2, v[0:1]
-; GFX8-NEXT:    s_load_dword s1, s[4:5], 0x0
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    s_movk_i32 s0, 0xff
-; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_and_b32 s6, s1, s0
-; GFX8-NEXT:    s_and_b32 s0, s2, s0
-; GFX8-NEXT:    v_mov_b32_e32 v5, s0
-; GFX8-NEXT:    s_bfe_u32 s7, s2, 0x80010
-; GFX8-NEXT:    v_lshrrev_b16_e64 v3, 8, s2
-; GFX8-NEXT:    v_lshrrev_b16_e64 v4, 8, s1
-; GFX8-NEXT:    s_lshr_b32 s4, s2, 24
-; GFX8-NEXT:    s_bfe_u32 s5, s1, 0x80010
-; GFX8-NEXT:    v_mov_b32_e32 v6, s7
+; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX8-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    s_and_b32 s6, s0, s4
+; GFX8-NEXT:    s_and_b32 s4, s1, s4
+; GFX8-NEXT:    v_mov_b32_e32 v5, s4
+; GFX8-NEXT:    s_bfe_u32 s7, s1, 0x80010
+; GFX8-NEXT:    v_lshrrev_b16_e64 v3, 8, s1
+; GFX8-NEXT:    v_lshrrev_b16_e64 v4, 8, s0
 ; GFX8-NEXT:    s_lshr_b32 s3, s1, 24
+; GFX8-NEXT:    s_bfe_u32 s5, s0, 0x80010
+; GFX8-NEXT:    v_mov_b32_e32 v6, s7
+; GFX8-NEXT:    s_lshr_b32 s2, s0, 24
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s6, v5, v2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, v4, v3, v2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v6, v2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s4
-; GFX8-NEXT:    v_mad_u32_u24 v2, s3, v3, v2
+; GFX8-NEXT:    v_mov_b32_e32 v3, s3
+; GFX8-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX8-NEXT:    flat_store_short v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-NODL-LABEL: udot4_acc16_vecMul:
 ; GFX9-NODL:       ; %bb.0: ; %entry
 ; GFX9-NODL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-NODL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, 0xffff

 ; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v4, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
 ; GFX9-DL-NEXT:    v_add_u32_e32 v2, v2, v3
 ; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
 ; GFX9-DL-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot4_acc16_vecMul:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
 ; GFX10-DL-NEXT:    v_mov_b32_e32 v2, 0xffff
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
-; GFX10-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s3
+; GFX10-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX10-DL-NEXT:    global_load_ushort v3, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX10-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    v_lshrrev_b16_e64 v4, 8, s0
 ; GFX10-DL-NEXT:    v_and_b32_sdwa v7, v2, s0 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_0
 ; GFX10-DL-NEXT:    v_lshrrev_b16_e64 v5, 8, s1
 ; GFX10-DL-NEXT:    v_and_b32_sdwa v6, v2, s1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_0
 ; GFX10-DL-NEXT:    s_lshr_b32 s2, s0, 16
 ; GFX10-DL-NEXT:    s_lshr_b32 s3, s1, 16
 ; GFX10-DL-NEXT:    v_lshl_or_b32 v4, v4, 16, v7
 ; GFX10-DL-NEXT:    s_lshr_b32 s0, s0, 24
entry:
   store i16 %add4, i16 addrspace(1)* %dst, align 4
   ret void
 }
 
 ; TODO: Support this pattern.
 define amdgpu_kernel void @udot4_acc8_vecMul(<4 x i8> addrspace(1)* %src1,
 ; GFX7-LABEL: udot4_acc8_vecMul:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; GFX7-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0xd
-; GFX7-NEXT:    s_mov_b32 s3, 0xf000
-; GFX7-NEXT:    s_mov_b32 s2, -1
+; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
+; GFX7-NEXT:    s_mov_b32 s7, 0xf000
+; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
 ; GFX7-NEXT:    s_movk_i32 s8, 0xff
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s4, s[4:5], 0x0
-; GFX7-NEXT:    buffer_load_ubyte v0, off, s[0:3], 0
-; GFX7-NEXT:    s_load_dword s5, s[6:7], 0x0
-; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_bfe_u32 s6, s4, 0x80008
-; GFX7-NEXT:    s_lshr_b32 s7, s4, 16
-; GFX7-NEXT:    s_bfe_u32 s10, s5, 0x80008
-; GFX7-NEXT:    s_lshr_b32 s11, s5, 16
-; GFX7-NEXT:    s_lshr_b32 s12, s5, 24
-; GFX7-NEXT:    v_mov_b32_e32 v2, s11
+; GFX7-NEXT:    buffer_load_ubyte v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX7-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX7-NEXT:    s_bfe_u32 s2, s0, 0x80008
+; GFX7-NEXT:    s_bfe_u32 s10, s1, 0x80008
+; GFX7-NEXT:    s_lshr_b32 s11, s1, 16
+; GFX7-NEXT:    s_lshr_b32 s12, s1, 24
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s10
-; GFX7-NEXT:    s_lshr_b32 s9, s4, 24
+; GFX7-NEXT:    s_lshr_b32 s3, s0, 16
+; GFX7-NEXT:    v_mov_b32_e32 v2, s11
+; GFX7-NEXT:    s_lshr_b32 s9, s0, 24
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s12
-; GFX7-NEXT:    s_mul_i32 s4, s4, s5
+; GFX7-NEXT:    s_mul_i32 s0, s0, s1
 ; GFX7-NEXT:    v_mul_u32_u24_e32 v1, s9, v1
-; GFX7-NEXT:    v_mul_u32_u24_e32 v2, s7, v2
-; GFX7-NEXT:    v_mul_u32_u24_e32 v3, s6, v3
-; GFX7-NEXT:    s_and_b32 s5, s4, s8
+; GFX7-NEXT:    v_mul_u32_u24_e32 v2, s3, v2
+; GFX7-NEXT:    v_mul_u32_u24_e32 v3, s2, v3
+; GFX7-NEXT:    s_and_b32 s1, s0, s8
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v1, 8, v1
 ; GFX7-NEXT:    v_and_b32_e32 v2, s8, v2
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v3, 8, v3
 ; GFX7-NEXT:    v_or_b32_e32 v1, v2, v1
-; GFX7-NEXT:    v_or_b32_e32 v2, s5, v3
+; GFX7-NEXT:    v_or_b32_e32 v2, s1, v3
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v1, 16, v1
 ; GFX7-NEXT:    v_and_b32_e32 v2, 0xffff, v2
 ; GFX7-NEXT:    v_or_b32_e32 v1, v2, v1
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v2, 8, v1
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v3, 16, v1
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v1, 24, v1
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
-; GFX7-NEXT:    v_add_i32_e32 v0, vcc, s4, v0
+; GFX7-NEXT:    v_add_i32_e32 v0, vcc, s0, v0
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v0, v2
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v0, v3
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v0, v1
-; GFX7-NEXT:    buffer_store_byte v0, off, s[0:3], 0
+; GFX7-NEXT:    buffer_store_byte v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot4_acc8_vecMul:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
-; GFX8-NEXT:    s_movk_i32 s8, 0xff
+; GFX8-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
+; GFX8-NEXT:    s_movk_i32 s7, 0xff
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    v_mov_b32_e32 v0, s2
+; GFX8-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX8-NEXT:    flat_load_ubyte v2, v[0:1]
-; GFX8-NEXT:    s_load_dword s0, s[4:5], 0x0
-; GFX8-NEXT:    s_load_dword s1, s[6:7], 0x0
+; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX8-NEXT:    s_load_dword s1, s[2:3], 0x0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    s_and_b32 s5, s0, s7
+; GFX8-NEXT:    s_and_b32 s7, s1, s7
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s0
 ; GFX8-NEXT:    v_mov_b32_e32 v4, s1
-; GFX8-NEXT:    s_and_b32 s7, s1, s8
 ; GFX8-NEXT:    s_lshr_b32 s2, s0, 24
 ; GFX8-NEXT:    s_lshr_b32 s3, s1, 24
 ; GFX8-NEXT:    s_bfe_u32 s6, s1, 0x80010
 ; GFX8-NEXT:    v_mul_u32_u24_sdwa v3, v3, v4 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:BYTE_1
-; GFX8-NEXT:    s_and_b32 s5, s0, s8
 ; GFX8-NEXT:    v_mov_b32_e32 v4, s7
 ; GFX8-NEXT:    v_mul_u32_u24_e32 v4, s5, v4
 ; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x80010
 ; GFX8-NEXT:    v_mov_b32_e32 v5, s6
 ; GFX8-NEXT:    v_mov_b32_e32 v6, s3
 ; GFX8-NEXT:    v_mov_b32_e32 v7, s2
 ; GFX8-NEXT:    v_or_b32_sdwa v3, v4, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
 ; GFX8-NEXT:    v_mul_u32_u24_e32 v5, s4, v5

 ; GFX8-NEXT:    flat_store_byte v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-NODL-LABEL: udot4_acc8_vecMul:
 ; GFX9-NODL:       ; %bb.0: ; %entry
 ; GFX9-NODL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-NODL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    s_load_dword s2, s[6:7], 0x0
+; GFX9-NODL-NEXT:    s_load_dword s2, s[4:5], 0x0
+; GFX9-NODL-NEXT:    s_load_dword s3, s[6:7], 0x0
+; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NODL-NEXT:    s_lshr_b32 s4, s2, 16
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s3
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s3
+; GFX9-NODL-NEXT:    v_mul_lo_u16_e32 v0, s2, v0
+; GFX9-NODL-NEXT:    v_mul_lo_u16_sdwa v1, s2, v1 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:BYTE_1
+; GFX9-NODL-NEXT:    v_or_b32_sdwa v0, v0, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
+; GFX9-NODL-NEXT:    s_lshr_b32 s6, s3, 16
+; GFX9-NODL-NEXT:    s_lshr_b32 s7, s3, 24
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s6
+; GFX9-NODL-NEXT:    v_and_b32_e32 v2, 0xffff, v0
+; GFX9-NODL-NEXT:    s_lshr_b32 s5, s2, 24
+; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s7
+; GFX9-NODL-NEXT:    v_mul_lo_u16_sdwa v0, s5, v0 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; GFX9-NODL-NEXT:    v_mul_lo_u16_e32 v1, s4, v1
+; GFX9-NODL-NEXT:    v_or_b32_sdwa v0, v1, v0 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
+; GFX9-NODL-NEXT:    v_or_b32_e32 v3, v2, v0
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-NODL-NEXT:    v_mov_b32_e32 v1, s1
-; GFX9-NODL-NEXT:    global_load_ubyte v2, v[0:1], off
-; GFX9-NODL-NEXT:    s_load_dword s0, s[4:5], 0x0
-; GFX9-NODL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v3, s2
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v4, s2
-; GFX9-NODL-NEXT:    s_lshr_b32 s1, s2, 16
-; GFX9-NODL-NEXT:    s_lshr_b32 s3, s2, 24
-; GFX9-NODL-NEXT:    v_mul_lo_u16_e32 v3, s0, v3
-; GFX9-NODL-NEXT:    v_mul_lo_u16_sdwa v4, s0, v4 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:BYTE_1
-; GFX9-NODL-NEXT:    v_or_b32_sdwa v3, v3, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
-; GFX9-NODL-NEXT:    s_lshr_b32 s2, s0, 16
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v5, s1
-; GFX9-NODL-NEXT:    s_lshr_b32 s4, s0, 24
-; GFX9-NODL-NEXT:    v_mov_b32_e32 v4, s3
-; GFX9-NODL-NEXT:    v_mul_lo_u16_sdwa v4, s4, v4 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
-; GFX9-NODL-NEXT:    v_mul_lo_u16_e32 v5, s2, v5
-; GFX9-NODL-NEXT:    v_and_b32_e32 v3, 0xffff, v3
-; GFX9-NODL-NEXT:    v_or_b32_sdwa v4, v5, v4 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
-; GFX9-NODL-NEXT:    v_or_b32_e32 v4, v3, v4
-; GFX9-NODL-NEXT:    v_lshrrev_b32_e32 v5, 8, v4
+; GFX9-NODL-NEXT:    global_load_ubyte v5, v[0:1], off
+; GFX9-NODL-NEXT:    v_lshrrev_b32_e32 v4, 8, v3
 ; GFX9-NODL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NODL-NEXT:    v_add_u32_e32 v2, v3, v2
 ; GFX9-NODL-NEXT:    v_add_u32_e32 v2, v2, v5
-; GFX9-NODL-NEXT:    v_add_u32_sdwa v2, v2, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
-; GFX9-NODL-NEXT:    v_add_u32_sdwa v2, v2, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
+; GFX9-NODL-NEXT:    v_add_u32_e32 v2, v2, v4
+; GFX9-NODL-NEXT:    v_add_u32_sdwa v2, v2, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; GFX9-NODL-NEXT:    v_add_u32_sdwa v2, v2, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
 ; GFX9-NODL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-NODL-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: udot4_acc8_vecMul:
 ; GFX9-DL:       ; %bb.0: ; %entry
 ; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
+; GFX9-DL-NEXT:    s_load_dword s2, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_load_dword s3, s[6:7], 0x0
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-DL-NEXT:    s_lshr_b32 s4, s2, 16
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s3
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s3
+; GFX9-DL-NEXT:    v_mul_lo_u16_e32 v0, s2, v0
+; GFX9-DL-NEXT:    v_mul_lo_u16_sdwa v1, s2, v1 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:BYTE_1
+; GFX9-DL-NEXT:    v_or_b32_sdwa v0, v0, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
+; GFX9-DL-NEXT:    s_lshr_b32 s6, s3, 16
+; GFX9-DL-NEXT:    s_lshr_b32 s7, s3, 24
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s6
+; GFX9-DL-NEXT:    v_and_b32_e32 v2, 0xffff, v0
+; GFX9-DL-NEXT:    s_lshr_b32 s5, s2, 24
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s7
+; GFX9-DL-NEXT:    v_mul_lo_u16_sdwa v0, s5, v0 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; GFX9-DL-NEXT:    v_mul_lo_u16_e32 v1, s4, v1
+; GFX9-DL-NEXT:    v_or_b32_sdwa v0, v1, v0 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
+; GFX9-DL-NEXT:    v_or_b32_e32 v3, v2, v0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
-; GFX9-DL-NEXT:    global_load_ubyte v2, v[0:1], off
-; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
-; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s2
-; GFX9-DL-NEXT:    s_lshr_b32 s1, s2, 16
-; GFX9-DL-NEXT:    s_lshr_b32 s3, s2, 24
-; GFX9-DL-NEXT:    v_mul_lo_u16_e32 v3, s0, v3
-; GFX9-DL-NEXT:    v_mul_lo_u16_sdwa v4, s0, v4 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:BYTE_1
-; GFX9-DL-NEXT:    v_or_b32_sdwa v3, v3, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
-; GFX9-DL-NEXT:    s_lshr_b32 s2, s0, 16
-; GFX9-DL-NEXT:    v_mov_b32_e32 v5, s1
-; GFX9-DL-NEXT:    s_lshr_b32 s4, s0, 24
-; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s3
-; GFX9-DL-NEXT:    v_mul_lo_u16_sdwa v4, s4, v4 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
-; GFX9-DL-NEXT:    v_mul_lo_u16_e32 v5, s2, v5
-; GFX9-DL-NEXT:    v_and_b32_e32 v3, 0xffff, v3
-; GFX9-DL-NEXT:    v_or_b32_sdwa v4, v5, v4 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
-; GFX9-DL-NEXT:    v_or_b32_e32 v4, v3, v4
-; GFX9-DL-NEXT:    v_lshrrev_b32_e32 v5, 8, v4
+; GFX9-DL-NEXT:    global_load_ubyte v5, v[0:1], off
+; GFX9-DL-NEXT:    v_lshrrev_b32_e32 v4, 8, v3
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-DL-NEXT:    v_add_u32_e32 v2, v3, v2
 ; GFX9-DL-NEXT:    v_add_u32_e32 v2, v2, v5
-; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
-; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
+; GFX9-DL-NEXT:    v_add_u32_e32 v2, v2, v4
+; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
 ; GFX9-DL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot4_acc8_vecMul:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x34
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
-; GFX10-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
-; GFX10-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s3
+; GFX10-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX10-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    v_lshrrev_b16_e64 v3, 8, s0
 ; GFX10-DL-NEXT:    v_lshrrev_b16_e64 v4, 8, s1
 ; GFX10-DL-NEXT:    s_lshr_b32 s2, s0, 24
 ; GFX10-DL-NEXT:    s_lshr_b32 s3, s1, 24
 ; GFX10-DL-NEXT:    s_lshr_b32 s4, s0, 16
 ; GFX10-DL-NEXT:    v_mul_lo_u16_e64 v3, v3, v4
 ; GFX10-DL-NEXT:    v_mul_lo_u16_e64 v4, s0, s1
 ; GFX10-DL-NEXT:    v_mul_lo_u16_e64 v5, s2, s3

llvm/test/CodeGen/AMDGPU/idot8s.ll

 ; GFX9-DL-LABEL: idot8_acc32:
 ; GFX9-DL:       ; %bb.0: ; %entry
 ; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_load_dword s6, s[0:1], 0x0
 ; GFX9-DL-NEXT:    s_load_dword s4, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s6
+; GFX9-DL-NEXT:    v_dot8_i32_i4 v2, s4, v0, v1
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
-; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mov_b32_e32 v2, s2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s6
-; GFX9-DL-NEXT:    v_dot8_i32_i4 v2, s4, v2, v3
 ; GFX9-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: idot8_acc32:
 ; GFX10-DL:       ; %bb.0: ; %entry
+; GFX10-DL-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0x34
 ; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    s_load_dword s2, s[0:1], 0x0
-; GFX10-DL-NEXT:    s_load_dword s4, s[4:5], 0x0
-; GFX10-DL-NEXT:    s_load_dword s5, s[6:7], 0x0
+; GFX10-DL-NEXT:    s_load_dword s0, s[8:9], 0x0
+; GFX10-DL-NEXT:    s_load_dword s1, s[4:5], 0x0
+; GFX10-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
-; GFX10-DL-NEXT:    v_dot8_i32_i4 v2, s4, s5, v0
 ; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_dot8_i32_i4 v2, s1, s2, v0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s8
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s9
 ; GFX10-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX10-DL-NEXT:    s_endpgm
                                        <8 x i4> addrspace(1)* %src2,
                                        i32 addrspace(1)* nocapture %dst) {
 entry:
   %vec1 = load <8 x i4>, <8 x i4> addrspace(1)* %src1
   %vec2 = load <8 x i4>, <8 x i4> addrspace(1)* %src2
 
entry:
   ret void
 }
 
 ; TODO: Once the unnecessary zero extentions of the elements are removed;
 ; pattern recognizer will kick in.
 define amdgpu_kernel void @idot8_acc16(<8 x i4> addrspace(1)* %src1,
 ; GFX7-LABEL: idot8_acc16:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
 ; GFX7-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX7-NEXT:    s_mov_b32 s6, -1
-; GFX7-NEXT:    s_mov_b32 s0, 0xffff
+; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
+; GFX7-NEXT:    s_mov_b32 s2, 0xffff
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s1, s[8:9], 0x0
 ; GFX7-NEXT:    buffer_load_ushort v0, off, s[4:7], 0
-; GFX7-NEXT:    s_load_dword s2, s[10:11], 0x0
+; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
+; GFX7-NEXT:    s_load_dword s1, s[10:11], 0x0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_bfe_i32 s8, s1, 0x40000
-; GFX7-NEXT:    s_bfe_i32 s10, s1, 0x40004
-; GFX7-NEXT:    s_bfe_i32 s9, s2, 0x40000
-; GFX7-NEXT:    s_bfe_i32 s11, s2, 0x40004
-; GFX7-NEXT:    s_and_b32 s9, s9, s0
-; GFX7-NEXT:    s_bfe_i32 s13, s2, 0x40008
-; GFX7-NEXT:    s_and_b32 s11, s11, s0
-; GFX7-NEXT:    s_and_b32 s8, s8, s0
+; GFX7-NEXT:    s_bfe_i32 s8, s0, 0x40000
+; GFX7-NEXT:    s_bfe_i32 s9, s1, 0x40000
+; GFX7-NEXT:    s_bfe_i32 s11, s1, 0x40004
+; GFX7-NEXT:    s_and_b32 s9, s9, s2
+; GFX7-NEXT:    s_bfe_i32 s10, s0, 0x40004
+; GFX7-NEXT:    s_bfe_i32 s13, s1, 0x40008
+; GFX7-NEXT:    s_and_b32 s11, s11, s2
+; GFX7-NEXT:    s_and_b32 s8, s8, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s9
-; GFX7-NEXT:    s_bfe_i32 s12, s1, 0x40008
-; GFX7-NEXT:    s_bfe_i32 s15, s2, 0x4000c
-; GFX7-NEXT:    s_and_b32 s13, s13, s0
-; GFX7-NEXT:    s_and_b32 s10, s10, s0
+; GFX7-NEXT:    s_bfe_i32 s12, s0, 0x40008
+; GFX7-NEXT:    s_bfe_i32 s15, s1, 0x4000c
+; GFX7-NEXT:    s_and_b32 s13, s13, s2
+; GFX7-NEXT:    s_and_b32 s10, s10, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s11
-; GFX7-NEXT:    s_bfe_i32 s14, s1, 0x4000c
-; GFX7-NEXT:    s_bfe_i32 s17, s2, 0x40010
-; GFX7-NEXT:    s_and_b32 s15, s15, s0
-; GFX7-NEXT:    s_and_b32 s12, s12, s0
+; GFX7-NEXT:    s_bfe_i32 s14, s0, 0x4000c
+; GFX7-NEXT:    s_bfe_i32 s17, s1, 0x40010
+; GFX7-NEXT:    s_and_b32 s15, s15, s2
+; GFX7-NEXT:    s_and_b32 s12, s12, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s13
-; GFX7-NEXT:    s_bfe_i32 s16, s1, 0x40010
-; GFX7-NEXT:    s_bfe_i32 s19, s2, 0x40014
-; GFX7-NEXT:    s_and_b32 s17, s17, s0
-; GFX7-NEXT:    s_and_b32 s14, s14, s0
+; GFX7-NEXT:    s_bfe_i32 s16, s0, 0x40010
+; GFX7-NEXT:    s_bfe_i32 s19, s1, 0x40014
+; GFX7-NEXT:    s_and_b32 s17, s17, s2
+; GFX7-NEXT:    s_and_b32 s14, s14, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v4, s15
-; GFX7-NEXT:    s_bfe_i32 s21, s2, 0x40018
-; GFX7-NEXT:    s_bfe_i32 s18, s1, 0x40014
-; GFX7-NEXT:    s_and_b32 s19, s19, s0
-; GFX7-NEXT:    s_and_b32 s16, s16, s0
+; GFX7-NEXT:    s_bfe_i32 s21, s1, 0x40018
+; GFX7-NEXT:    s_bfe_i32 s18, s0, 0x40014
+; GFX7-NEXT:    s_and_b32 s19, s19, s2
+; GFX7-NEXT:    s_and_b32 s16, s16, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v5, s17
-; GFX7-NEXT:    s_bfe_i32 s20, s1, 0x40018
-; GFX7-NEXT:    s_ashr_i32 s2, s2, 28
-; GFX7-NEXT:    s_and_b32 s21, s21, s0
-; GFX7-NEXT:    s_and_b32 s18, s18, s0
-; GFX7-NEXT:    v_mov_b32_e32 v6, s19
+; GFX7-NEXT:    s_bfe_i32 s20, s0, 0x40018
 ; GFX7-NEXT:    s_ashr_i32 s1, s1, 28
-; GFX7-NEXT:    s_and_b32 s20, s20, s0
-; GFX7-NEXT:    s_and_b32 s2, s2, s0
+; GFX7-NEXT:    s_and_b32 s21, s21, s2
+; GFX7-NEXT:    s_and_b32 s18, s18, s2
+; GFX7-NEXT:    v_mov_b32_e32 v6, s19
+; GFX7-NEXT:    s_ashr_i32 s0, s0, 28
+; GFX7-NEXT:    s_and_b32 s20, s20, s2
+; GFX7-NEXT:    s_and_b32 s1, s1, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v7, s21
-; GFX7-NEXT:    s_and_b32 s0, s1, s0
+; GFX7-NEXT:    s_and_b32 s0, s0, s2
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s8, v1, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s10, v2, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s12, v3, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s14, v4, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s16, v5, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s18, v6, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s20, v7, v0
-; GFX7-NEXT:    v_mov_b32_e32 v1, s2
+; GFX7-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s0, v1, v0
 ; GFX7-NEXT:    buffer_store_short v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: idot8_acc16:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    v_mov_b32_e32 v0, s4
+; GFX8-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX8-NEXT:    flat_load_ushort v2, v[0:1]
+; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX8-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_bfe_i32 s1, s2, 0x40000
-; GFX8-NEXT:    s_bfe_i32 s4, s2, 0x40004
-; GFX8-NEXT:    s_bfe_i32 s5, s2, 0x40008
-; GFX8-NEXT:    v_mov_b32_e32 v3, s1
-; GFX8-NEXT:    s_bfe_i32 s6, s0, 0x40000
-; GFX8-NEXT:    s_lshr_b32 s1, s0, 12
-; GFX8-NEXT:    s_lshr_b32 s7, s2, 12
-; GFX8-NEXT:    v_mov_b32_e32 v4, s5
+; GFX8-NEXT:    s_bfe_i32 s5, s0, 0x40000
+; GFX8-NEXT:    s_bfe_i32 s6, s1, 0x40000
+; GFX8-NEXT:    s_bfe_i32 s8, s1, 0x40004
+; GFX8-NEXT:    s_bfe_i32 s10, s1, 0x40008
+; GFX8-NEXT:    v_mov_b32_e32 v6, s6
+; GFX8-NEXT:    s_lshr_b32 s2, s0, 12
+; GFX8-NEXT:    s_lshr_b32 s4, s1, 12
+; GFX8-NEXT:    s_bfe_i32 s7, s0, 0x40004
 ; GFX8-NEXT:    s_bfe_i32 s9, s0, 0x40008
-; GFX8-NEXT:    v_mov_b32_e32 v5, s4
-; GFX8-NEXT:    s_bfe_i32 s8, s0, 0x40004
-; GFX8-NEXT:    v_lshlrev_b16_e64 v6, 12, s1
-; GFX8-NEXT:    v_lshlrev_b16_e64 v7, 12, s7
-; GFX8-NEXT:    v_mul_i32_i24_e32 v4, s9, v4
-; GFX8-NEXT:    s_bfe_i32 s1, s2, 0x40010
-; GFX8-NEXT:    v_ashrrev_i16_e32 v6, 12, v6
-; GFX8-NEXT:    v_ashrrev_i16_e32 v7, 12, v7
-; GFX8-NEXT:    s_bfe_i32 s5, s2, 0x40014
-; GFX8-NEXT:    v_mov_b32_e32 v8, s1
-; GFX8-NEXT:    s_bfe_i32 s4, s0, 0x40010
-; GFX8-NEXT:    s_bfe_i32 s7, s2, 0x40018
-; GFX8-NEXT:    v_mov_b32_e32 v9, s5
-; GFX8-NEXT:    s_bfe_i32 s1, s0, 0x40014
-; GFX8-NEXT:    s_bfe_i32 s5, s0, 0x40018
-; GFX8-NEXT:    s_ashr_i32 s2, s2, 28
-; GFX8-NEXT:    v_mov_b32_e32 v10, s7
+; GFX8-NEXT:    v_mov_b32_e32 v3, s10
+; GFX8-NEXT:    v_mov_b32_e32 v7, s8
+; GFX8-NEXT:    v_lshlrev_b16_e64 v4, 12, s2
+; GFX8-NEXT:    v_lshlrev_b16_e64 v5, 12, s4
+; GFX8-NEXT:    v_mul_i32_i24_e32 v3, s9, v3
+; GFX8-NEXT:    s_bfe_i32 s12, s1, 0x40010
+; GFX8-NEXT:    v_ashrrev_i16_e32 v4, 12, v4
+; GFX8-NEXT:    v_ashrrev_i16_e32 v5, 12, v5
+; GFX8-NEXT:    s_bfe_i32 s14, s1, 0x40014
+; GFX8-NEXT:    s_bfe_i32 s11, s0, 0x40010
+; GFX8-NEXT:    v_mov_b32_e32 v8, s12
+; GFX8-NEXT:    s_bfe_i32 s16, s1, 0x40018
+; GFX8-NEXT:    s_bfe_i32 s13, s0, 0x40014
+; GFX8-NEXT:    v_mov_b32_e32 v9, s14
+; GFX8-NEXT:    s_bfe_i32 s15, s0, 0x40018
+; GFX8-NEXT:    s_ashr_i32 s1, s1, 28
+; GFX8-NEXT:    v_mov_b32_e32 v10, s16
 ; GFX8-NEXT:    s_ashr_i32 s0, s0, 28
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
-; GFX8-NEXT:    v_mad_i32_i24 v2, s6, v3, v2
-; GFX8-NEXT:    v_mad_i32_i24 v2, s8, v5, v2
-; GFX8-NEXT:    v_add_u32_sdwa v2, vcc, v4, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:WORD_0
-; GFX8-NEXT:    v_mad_u32_u24 v2, v6, v7, v2
-; GFX8-NEXT:    v_mad_i32_i24 v2, s4, v8, v2
-; GFX8-NEXT:    v_mad_i32_i24 v2, s1, v9, v2
-; GFX8-NEXT:    v_mad_i32_i24 v2, s5, v10, v2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s2
+; GFX8-NEXT:    v_mad_i32_i24 v2, s5, v6, v2
+; GFX8-NEXT:    v_mad_i32_i24 v2, s7, v7, v2
+; GFX8-NEXT:    v_add_u32_sdwa v2, vcc, v3, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:WORD_0
+; GFX8-NEXT:    v_mad_u32_u24 v2, v4, v5, v2
+; GFX8-NEXT:    v_mad_i32_i24 v2, s11, v8, v2
+; GFX8-NEXT:    v_mad_i32_i24 v2, s13, v9, v2
+; GFX8-NEXT:    v_mad_i32_i24 v2, s15, v10, v2
+; GFX8-NEXT:    v_mov_b32_e32 v3, s1
 ; GFX8-NEXT:    v_mad_i32_i24 v2, s0, v3, v2
 ; GFX8-NEXT:    flat_store_short v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: idot8_acc16:
 ; GFX9:       ; %bb.0: ; %entry
-; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-NEXT:    global_load_ushort v2, v[0:1], off
+; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_bfe_i32 s1, s2, 0x40000
-; GFX9-NEXT:    s_bfe_i32 s4, s2, 0x40004
-; GFX9-NEXT:    s_bfe_i32 s5, s2, 0x40008
-; GFX9-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-NEXT:    s_bfe_i32 s6, s0, 0x40000
-; GFX9-NEXT:    s_lshr_b32 s1, s0, 12
-; GFX9-NEXT:    s_lshr_b32 s7, s2, 12
-; GFX9-NEXT:    v_mov_b32_e32 v4, s5
+; GFX9-NEXT:    s_bfe_i32 s5, s0, 0x40000
+; GFX9-NEXT:    s_bfe_i32 s6, s1, 0x40000
+; GFX9-NEXT:    s_bfe_i32 s8, s1, 0x40004
+; GFX9-NEXT:    s_bfe_i32 s10, s1, 0x40008
+; GFX9-NEXT:    v_mov_b32_e32 v6, s6
+; GFX9-NEXT:    s_lshr_b32 s2, s0, 12
+; GFX9-NEXT:    s_lshr_b32 s4, s1, 12
+; GFX9-NEXT:    s_bfe_i32 s7, s0, 0x40004
 ; GFX9-NEXT:    s_bfe_i32 s9, s0, 0x40008
-; GFX9-NEXT:    v_mov_b32_e32 v5, s4
-; GFX9-NEXT:    s_bfe_i32 s8, s0, 0x40004
-; GFX9-NEXT:    v_lshlrev_b16_e64 v6, 12, s1
-; GFX9-NEXT:    v_lshlrev_b16_e64 v7, 12, s7
-; GFX9-NEXT:    v_mul_i32_i24_e32 v4, s9, v4
-; GFX9-NEXT:    s_bfe_i32 s1, s2, 0x40010
-; GFX9-NEXT:    v_ashrrev_i16_e32 v6, 12, v6
-; GFX9-NEXT:    v_ashrrev_i16_e32 v7, 12, v7
-; GFX9-NEXT:    s_bfe_i32 s5, s2, 0x40014
-; GFX9-NEXT:    v_mov_b32_e32 v8, s1
-; GFX9-NEXT:    s_bfe_i32 s4, s0, 0x40010
-; GFX9-NEXT:    s_bfe_i32 s7, s2, 0x40018
-; GFX9-NEXT:    v_mov_b32_e32 v9, s5
-; GFX9-NEXT:    s_bfe_i32 s1, s0, 0x40014
-; GFX9-NEXT:    s_bfe_i32 s5, s0, 0x40018
-; GFX9-NEXT:    s_ashr_i32 s2, s2, 28
-; GFX9-NEXT:    v_mov_b32_e32 v10, s7
+; GFX9-NEXT:    v_mov_b32_e32 v3, s10
+; GFX9-NEXT:    v_mov_b32_e32 v7, s8
+; GFX9-NEXT:    v_lshlrev_b16_e64 v4, 12, s2
+; GFX9-NEXT:    v_lshlrev_b16_e64 v5, 12, s4
+; GFX9-NEXT:    v_mul_i32_i24_e32 v3, s9, v3
+; GFX9-NEXT:    s_bfe_i32 s12, s1, 0x40010
+; GFX9-NEXT:    v_ashrrev_i16_e32 v4, 12, v4
+; GFX9-NEXT:    v_ashrrev_i16_e32 v5, 12, v5
+; GFX9-NEXT:    s_bfe_i32 s14, s1, 0x40014
+; GFX9-NEXT:    s_bfe_i32 s11, s0, 0x40010
+; GFX9-NEXT:    v_mov_b32_e32 v8, s12
+; GFX9-NEXT:    s_bfe_i32 s16, s1, 0x40018
+; GFX9-NEXT:    s_bfe_i32 s13, s0, 0x40014
+; GFX9-NEXT:    v_mov_b32_e32 v9, s14
+; GFX9-NEXT:    s_bfe_i32 s15, s0, 0x40018
+; GFX9-NEXT:    s_ashr_i32 s1, s1, 28
+; GFX9-NEXT:    v_mov_b32_e32 v10, s16
 ; GFX9-NEXT:    s_ashr_i32 s0, s0, 28
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mad_i32_i24 v2, s6, v3, v2
-; GFX9-NEXT:    v_mad_i32_i24 v2, s8, v5, v2
-; GFX9-NEXT:    v_add_u32_sdwa v2, v2, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:WORD_0
-; GFX9-NEXT:    v_mad_u32_u24 v2, v6, v7, v2
-; GFX9-NEXT:    v_mad_i32_i24 v2, s4, v8, v2
-; GFX9-NEXT:    v_mad_i32_i24 v2, s1, v9, v2
-; GFX9-NEXT:    v_mad_i32_i24 v2, s5, v10, v2
-; GFX9-NEXT:    v_mov_b32_e32 v3, s2
+; GFX9-NEXT:    v_mad_i32_i24 v2, s5, v6, v2
+; GFX9-NEXT:    v_mad_i32_i24 v2, s7, v7, v2
+; GFX9-NEXT:    v_add_u32_sdwa v2, v2, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:WORD_0
+; GFX9-NEXT:    v_mad_u32_u24 v2, v4, v5, v2
+; GFX9-NEXT:    v_mad_i32_i24 v2, s11, v8, v2
+; GFX9-NEXT:    v_mad_i32_i24 v2, s13, v9, v2
+; GFX9-NEXT:    v_mad_i32_i24 v2, s15, v10, v2
+; GFX9-NEXT:    v_mov_b32_e32 v3, s1
 ; GFX9-NEXT:    v_mad_i32_i24 v2, s0, v3, v2
 ; GFX9-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: idot8_acc16:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-DL-NEXT:    global_load_ushort v2, v[0:1], off
+; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_bfe_i32 s1, s2, 0x40000
-; GFX9-DL-NEXT:    s_bfe_i32 s4, s2, 0x40004
-; GFX9-DL-NEXT:    s_bfe_i32 s5, s2, 0x40008
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-DL-NEXT:    s_bfe_i32 s6, s0, 0x40000
-; GFX9-DL-NEXT:    s_lshr_b32 s1, s0, 12
-; GFX9-DL-NEXT:    s_lshr_b32 s7, s2, 12
-; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s5
+; GFX9-DL-NEXT:    s_bfe_i32 s5, s0, 0x40000
+; GFX9-DL-NEXT:    s_bfe_i32 s6, s1, 0x40000
+; GFX9-DL-NEXT:    s_bfe_i32 s8, s1, 0x40004
+; GFX9-DL-NEXT:    s_bfe_i32 s10, s1, 0x40008
+; GFX9-DL-NEXT:    v_mov_b32_e32 v6, s6
+; GFX9-DL-NEXT:    s_lshr_b32 s2, s0, 12
+; GFX9-DL-NEXT:    s_lshr_b32 s4, s1, 12
+; GFX9-DL-NEXT:    s_bfe_i32 s7, s0, 0x40004
 ; GFX9-DL-NEXT:    s_bfe_i32 s9, s0, 0x40008
-; GFX9-DL-NEXT:    v_mov_b32_e32 v5, s4
-; GFX9-DL-NEXT:    s_bfe_i32 s8, s0, 0x40004
-; GFX9-DL-NEXT:    v_lshlrev_b16_e64 v6, 12, s1
-; GFX9-DL-NEXT:    v_lshlrev_b16_e64 v7, 12, s7
-; GFX9-DL-NEXT:    v_mul_i32_i24_e32 v4, s9, v4
-; GFX9-DL-NEXT:    s_bfe_i32 s1, s2, 0x40010
-; GFX9-DL-NEXT:    v_ashrrev_i16_e32 v6, 12, v6
-; GFX9-DL-NEXT:    v_ashrrev_i16_e32 v7, 12, v7
-; GFX9-DL-NEXT:    s_bfe_i32 s5, s2, 0x40014
-; GFX9-DL-NEXT:    v_mov_b32_e32 v8, s1
-; GFX9-DL-NEXT:    s_bfe_i32 s4, s0, 0x40010
-; GFX9-DL-NEXT:    s_bfe_i32 s7, s2, 0x40018
-; GFX9-DL-NEXT:    v_mov_b32_e32 v9, s5
-; GFX9-DL-NEXT:    s_bfe_i32 s1, s0, 0x40014
-; GFX9-DL-NEXT:    s_bfe_i32 s5, s0, 0x40018
-; GFX9-DL-NEXT:    s_ashr_i32 s2, s2, 28
-; GFX9-DL-NEXT:    v_mov_b32_e32 v10, s7
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s10
+; GFX9-DL-NEXT:    v_mov_b32_e32 v7, s8
+; GFX9-DL-NEXT:    v_lshlrev_b16_e64 v4, 12, s2
+; GFX9-DL-NEXT:    v_lshlrev_b16_e64 v5, 12, s4
+; GFX9-DL-NEXT:    v_mul_i32_i24_e32 v3, s9, v3
+; GFX9-DL-NEXT:    s_bfe_i32 s12, s1, 0x40010
+; GFX9-DL-NEXT:    v_ashrrev_i16_e32 v4, 12, v4
+; GFX9-DL-NEXT:    v_ashrrev_i16_e32 v5, 12, v5
+; GFX9-DL-NEXT:    s_bfe_i32 s14, s1, 0x40014
+; GFX9-DL-NEXT:    s_bfe_i32 s11, s0, 0x40010
+; GFX9-DL-NEXT:    v_mov_b32_e32 v8, s12
+; GFX9-DL-NEXT:    s_bfe_i32 s16, s1, 0x40018
+; GFX9-DL-NEXT:    s_bfe_i32 s13, s0, 0x40014
+; GFX9-DL-NEXT:    v_mov_b32_e32 v9, s14
+; GFX9-DL-NEXT:    s_bfe_i32 s15, s0, 0x40018
+; GFX9-DL-NEXT:    s_ashr_i32 s1, s1, 28
+; GFX9-DL-NEXT:    v_mov_b32_e32 v10, s16
 ; GFX9-DL-NEXT:    s_ashr_i32 s0, s0, 28
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s6, v3, v2
-; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s8, v5, v2
-; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:WORD_0
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, v6, v7, v2
-; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s4, v8, v2
-; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s1, v9, v2
-; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s5, v10, v2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s2
+; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s5, v6, v2
+; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s7, v7, v2
+; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:WORD_0
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, v4, v5, v2
+; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s11, v8, v2
+; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s13, v9, v2
+; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s15, v10, v2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s1
 ; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s0, v3, v2
 ; GFX9-DL-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: idot8_acc16:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX10-DL-NEXT:    s_mov_b32 s2, 0xffff
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
+; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX10-DL-NEXT:    global_load_ushort v2, v[0:1], off
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX10-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX10-DL-NEXT:    global_load_ushort v2, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_lshr_b32 s4, s0, 12
 ; GFX10-DL-NEXT:    s_lshr_b32 s5, s1, 12
 ; GFX10-DL-NEXT:    s_bfe_i32 s6, s0, 0x40000
 ; GFX10-DL-NEXT:    s_bfe_i32 s7, s1, 0x40000
 ; GFX10-DL-NEXT:    s_bfe_i32 s8, s0, 0x40004
 ; GFX10-DL-NEXT:    v_lshlrev_b16_e64 v3, 12, s4
 ; GFX10-DL-NEXT:    v_lshlrev_b16_e64 v4, 12, s5
entry:
   store i16 %add8, i16 addrspace(1)* %dst, align 4
   ret void
 }
 
 ; TODO: Support this pattern.
 define amdgpu_kernel void @idot8_acc8(<8 x i4> addrspace(1)* %src1,
 ; GFX7-LABEL: idot8_acc8:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
 ; GFX7-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX7-NEXT:    s_mov_b32 s6, -1
-; GFX7-NEXT:    s_movk_i32 s0, 0xff
+; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
+; GFX7-NEXT:    s_movk_i32 s2, 0xff
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s1, s[8:9], 0x0
 ; GFX7-NEXT:    buffer_load_ubyte v0, off, s[4:7], 0
-; GFX7-NEXT:    s_load_dword s2, s[10:11], 0x0
+; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
+; GFX7-NEXT:    s_load_dword s1, s[10:11], 0x0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_bfe_i32 s8, s1, 0x40000
-; GFX7-NEXT:    s_bfe_i32 s10, s1, 0x40004
-; GFX7-NEXT:    s_bfe_i32 s9, s2, 0x40000
-; GFX7-NEXT:    s_bfe_i32 s11, s2, 0x40004
-; GFX7-NEXT:    s_and_b32 s9, s9, s0
-; GFX7-NEXT:    s_bfe_i32 s13, s2, 0x40008
-; GFX7-NEXT:    s_and_b32 s11, s11, s0
-; GFX7-NEXT:    s_and_b32 s8, s8, s0
+; GFX7-NEXT:    s_bfe_i32 s8, s0, 0x40000
+; GFX7-NEXT:    s_bfe_i32 s9, s1, 0x40000
+; GFX7-NEXT:    s_bfe_i32 s11, s1, 0x40004
+; GFX7-NEXT:    s_and_b32 s9, s9, s2
+; GFX7-NEXT:    s_bfe_i32 s10, s0, 0x40004
+; GFX7-NEXT:    s_bfe_i32 s13, s1, 0x40008
+; GFX7-NEXT:    s_and_b32 s11, s11, s2
+; GFX7-NEXT:    s_and_b32 s8, s8, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s9
-; GFX7-NEXT:    s_bfe_i32 s12, s1, 0x40008
-; GFX7-NEXT:    s_bfe_i32 s15, s2, 0x4000c
-; GFX7-NEXT:    s_and_b32 s13, s13, s0
-; GFX7-NEXT:    s_and_b32 s10, s10, s0
+; GFX7-NEXT:    s_bfe_i32 s12, s0, 0x40008
+; GFX7-NEXT:    s_bfe_i32 s15, s1, 0x4000c
+; GFX7-NEXT:    s_and_b32 s13, s13, s2
+; GFX7-NEXT:    s_and_b32 s10, s10, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s11
-; GFX7-NEXT:    s_bfe_i32 s14, s1, 0x4000c
-; GFX7-NEXT:    s_bfe_i32 s17, s2, 0x40010
-; GFX7-NEXT:    s_and_b32 s15, s15, s0
-; GFX7-NEXT:    s_and_b32 s12, s12, s0
+; GFX7-NEXT:    s_bfe_i32 s14, s0, 0x4000c
+; GFX7-NEXT:    s_bfe_i32 s17, s1, 0x40010
+; GFX7-NEXT:    s_and_b32 s15, s15, s2
+; GFX7-NEXT:    s_and_b32 s12, s12, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s13
-; GFX7-NEXT:    s_bfe_i32 s16, s1, 0x40010
-; GFX7-NEXT:    s_bfe_i32 s19, s2, 0x40014
-; GFX7-NEXT:    s_and_b32 s17, s17, s0
-; GFX7-NEXT:    s_and_b32 s14, s14, s0
+; GFX7-NEXT:    s_bfe_i32 s16, s0, 0x40010
+; GFX7-NEXT:    s_bfe_i32 s19, s1, 0x40014
+; GFX7-NEXT:    s_and_b32 s17, s17, s2
+; GFX7-NEXT:    s_and_b32 s14, s14, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v4, s15
-; GFX7-NEXT:    s_bfe_i32 s21, s2, 0x40018
-; GFX7-NEXT:    s_bfe_i32 s18, s1, 0x40014
-; GFX7-NEXT:    s_and_b32 s19, s19, s0
-; GFX7-NEXT:    s_and_b32 s16, s16, s0
+; GFX7-NEXT:    s_bfe_i32 s21, s1, 0x40018
+; GFX7-NEXT:    s_bfe_i32 s18, s0, 0x40014
+; GFX7-NEXT:    s_and_b32 s19, s19, s2
+; GFX7-NEXT:    s_and_b32 s16, s16, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v5, s17
-; GFX7-NEXT:    s_bfe_i32 s20, s1, 0x40018
-; GFX7-NEXT:    s_ashr_i32 s2, s2, 28
-; GFX7-NEXT:    s_and_b32 s21, s21, s0
-; GFX7-NEXT:    s_and_b32 s18, s18, s0
-; GFX7-NEXT:    v_mov_b32_e32 v6, s19
+; GFX7-NEXT:    s_bfe_i32 s20, s0, 0x40018
 ; GFX7-NEXT:    s_ashr_i32 s1, s1, 28
-; GFX7-NEXT:    s_and_b32 s20, s20, s0
-; GFX7-NEXT:    s_and_b32 s2, s2, s0
+; GFX7-NEXT:    s_and_b32 s21, s21, s2
+; GFX7-NEXT:    s_and_b32 s18, s18, s2
+; GFX7-NEXT:    v_mov_b32_e32 v6, s19
+; GFX7-NEXT:    s_ashr_i32 s0, s0, 28
+; GFX7-NEXT:    s_and_b32 s20, s20, s2
+; GFX7-NEXT:    s_and_b32 s1, s1, s2
 ; GFX7-NEXT:    v_mov_b32_e32 v7, s21
-; GFX7-NEXT:    s_and_b32 s0, s1, s0
+; GFX7-NEXT:    s_and_b32 s0, s0, s2
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s8, v1, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s10, v2, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s12, v3, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s14, v4, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s16, v5, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s18, v6, v0
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s20, v7, v0
-; GFX7-NEXT:    v_mov_b32_e32 v1, s2
+; GFX7-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s0, v1, v0
 ; GFX7-NEXT:    buffer_store_byte v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: idot8_acc8:
 ; GFX8:       ; %bb.0: ; %entry
+; GFX8-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0x34
 ; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX8-NEXT:    s_movk_i32 s2, 0xff
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s6, s[6:7], 0x0
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    v_mov_b32_e32 v0, s8
+; GFX8-NEXT:    v_mov_b32_e32 v1, s9
 ; GFX8-NEXT:    flat_load_ubyte v2, v[0:1]
-; GFX8-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX8-NEXT:    s_load_dword s0, s[6:7], 0x0
+; GFX8-NEXT:    s_load_dword s1, s[4:5], 0x0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_bfe_i32 s7, s6, 0x40000
-; GFX8-NEXT:    s_lshr_b32 s4, s6, 12
-; GFX8-NEXT:    s_bfe_i32 s9, s6, 0x40004
-; GFX8-NEXT:    s_bfe_i32 s11, s6, 0x40008
-; GFX8-NEXT:    s_lshr_b32 s1, s0, 12
-; GFX8-NEXT:    s_bfe_i32 s5, s0, 0x40000
+; GFX8-NEXT:    s_bfe_i32 s7, s0, 0x40000
+; GFX8-NEXT:    s_lshr_b32 s4, s1, 12
+; GFX8-NEXT:    s_lshr_b32 s5, s0, 12
+; GFX8-NEXT:    s_bfe_i32 s9, s0, 0x40004
+; GFX8-NEXT:    s_bfe_i32 s11, s0, 0x40008
+; GFX8-NEXT:    s_bfe_i32 s6, s1, 0x40000
 ; GFX8-NEXT:    v_mov_b32_e32 v6, s7
-; GFX8-NEXT:    v_lshlrev_b16_e64 v4, 12, s1
-; GFX8-NEXT:    v_lshlrev_b16_e64 v5, 12, s4
-; GFX8-NEXT:    s_bfe_i32 s8, s0, 0x40004
-; GFX8-NEXT:    s_bfe_i32 s10, s0, 0x40008
+; GFX8-NEXT:    v_lshlrev_b16_e64 v4, 12, s4
+; GFX8-NEXT:    v_lshlrev_b16_e64 v5, 12, s5
+; GFX8-NEXT:    s_bfe_i32 s8, s1, 0x40004
+; GFX8-NEXT:    s_bfe_i32 s10, s1, 0x40008
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s11
 ; GFX8-NEXT:    v_mov_b32_e32 v7, s9
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v4, 12, v4
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v5, 12, v5
 ; GFX8-NEXT:    v_mul_i32_i24_e32 v3, s10, v3
-; GFX8-NEXT:    s_bfe_i32 s13, s6, 0x40010
+; GFX8-NEXT:    s_bfe_i32 s13, s0, 0x40010
 ; GFX8-NEXT:    v_and_b32_e32 v4, s2, v4
 ; GFX8-NEXT:    v_and_b32_e32 v5, s2, v5
-; GFX8-NEXT:    s_bfe_i32 s15, s6, 0x40014
-; GFX8-NEXT:    s_bfe_i32 s12, s0, 0x40010
+; GFX8-NEXT:    s_bfe_i32 s15, s0, 0x40014
+; GFX8-NEXT:    s_bfe_i32 s12, s1, 0x40010
 ; GFX8-NEXT:    v_mov_b32_e32 v8, s13
-; GFX8-NEXT:    s_bfe_i32 s17, s6, 0x40018
-; GFX8-NEXT:    s_bfe_i32 s14, s0, 0x40014
+; GFX8-NEXT:    s_bfe_i32 s17, s0, 0x40018
+; GFX8-NEXT:    s_bfe_i32 s14, s1, 0x40014
 ; GFX8-NEXT:    v_mov_b32_e32 v9, s15
-; GFX8-NEXT:    s_bfe_i32 s16, s0, 0x40018
-; GFX8-NEXT:    s_ashr_i32 s6, s6, 28
-; GFX8-NEXT:    v_mov_b32_e32 v10, s17
+; GFX8-NEXT:    s_bfe_i32 s16, s1, 0x40018
 ; GFX8-NEXT:    s_ashr_i32 s0, s0, 28
+; GFX8-NEXT:    v_mov_b32_e32 v10, s17
+; GFX8-NEXT:    s_ashr_i32 s1, s1, 28
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
-; GFX8-NEXT:    v_mad_i32_i24 v2, s5, v6, v2
+; GFX8-NEXT:    v_mad_i32_i24 v2, s6, v6, v2
 ; GFX8-NEXT:    v_mad_i32_i24 v2, s8, v7, v2
 ; GFX8-NEXT:    v_add_u32_sdwa v2, vcc, v3, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_0
 ; GFX8-NEXT:    v_mad_u32_u24 v2, v4, v5, v2
 ; GFX8-NEXT:    v_mad_i32_i24 v2, s12, v8, v2
 ; GFX8-NEXT:    v_mad_i32_i24 v2, s14, v9, v2
 ; GFX8-NEXT:    v_mad_i32_i24 v2, s16, v10, v2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s6
-; GFX8-NEXT:    v_mad_i32_i24 v2, s0, v3, v2
+; GFX8-NEXT:    v_mov_b32_e32 v3, s0
+; GFX8-NEXT:    v_mad_i32_i24 v2, s1, v3, v2
 ; GFX8-NEXT:    flat_store_byte v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: idot8_acc8:
 ; GFX9:       ; %bb.0: ; %entry
+; GFX9-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0x34
 ; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-NEXT:    s_movk_i32 s2, 0xff
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_load_dword s6, s[6:7], 0x0
-; GFX9-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_mov_b32_e32 v0, s8
+; GFX9-NEXT:    v_mov_b32_e32 v1, s9
 ; GFX9-NEXT:    global_load_ubyte v2, v[0:1], off
-; GFX9-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-NEXT:    s_load_dword s0, s[6:7], 0x0
+; GFX9-NEXT:    s_load_dword s1, s[4:5], 0x0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_bfe_i32 s7, s6, 0x40000
-; GFX9-NEXT:    s_lshr_b32 s4, s6, 12
-; GFX9-NEXT:    s_bfe_i32 s9, s6, 0x40004
-; GFX9-NEXT:    s_bfe_i32 s11, s6, 0x40008
-; GFX9-NEXT:    s_lshr_b32 s1, s0, 12
-; GFX9-NEXT:    s_bfe_i32 s5, s0, 0x40000
+; GFX9-NEXT:    s_bfe_i32 s7, s0, 0x40000
+; GFX9-NEXT:    s_lshr_b32 s4, s1, 12
+; GFX9-NEXT:    s_lshr_b32 s5, s0, 12
+; GFX9-NEXT:    s_bfe_i32 s9, s0, 0x40004
+; GFX9-NEXT:    s_bfe_i32 s11, s0, 0x40008
+; GFX9-NEXT:    s_bfe_i32 s6, s1, 0x40000
 ; GFX9-NEXT:    v_mov_b32_e32 v6, s7
-; GFX9-NEXT:    v_lshlrev_b16_e64 v4, 12, s1
-; GFX9-NEXT:    v_lshlrev_b16_e64 v5, 12, s4
-; GFX9-NEXT:    s_bfe_i32 s8, s0, 0x40004
-; GFX9-NEXT:    s_bfe_i32 s10, s0, 0x40008
+; GFX9-NEXT:    v_lshlrev_b16_e64 v4, 12, s4
+; GFX9-NEXT:    v_lshlrev_b16_e64 v5, 12, s5
+; GFX9-NEXT:    s_bfe_i32 s8, s1, 0x40004
+; GFX9-NEXT:    s_bfe_i32 s10, s1, 0x40008
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s11
 ; GFX9-NEXT:    v_mov_b32_e32 v7, s9
 ; GFX9-NEXT:    v_ashrrev_i16_e32 v4, 12, v4
 ; GFX9-NEXT:    v_ashrrev_i16_e32 v5, 12, v5
 ; GFX9-NEXT:    v_mul_i32_i24_e32 v3, s10, v3
-; GFX9-NEXT:    s_bfe_i32 s13, s6, 0x40010
+; GFX9-NEXT:    s_bfe_i32 s13, s0, 0x40010
 ; GFX9-NEXT:    v_and_b32_e32 v4, s2, v4
 ; GFX9-NEXT:    v_and_b32_e32 v5, s2, v5
-; GFX9-NEXT:    s_bfe_i32 s15, s6, 0x40014
-; GFX9-NEXT:    s_bfe_i32 s12, s0, 0x40010
+; GFX9-NEXT:    s_bfe_i32 s15, s0, 0x40014
+; GFX9-NEXT:    s_bfe_i32 s12, s1, 0x40010
 ; GFX9-NEXT:    v_mov_b32_e32 v8, s13
-; GFX9-NEXT:    s_bfe_i32 s17, s6, 0x40018
-; GFX9-NEXT:    s_bfe_i32 s14, s0, 0x40014
+; GFX9-NEXT:    s_bfe_i32 s17, s0, 0x40018
+; GFX9-NEXT:    s_bfe_i32 s14, s1, 0x40014
 ; GFX9-NEXT:    v_mov_b32_e32 v9, s15
-; GFX9-NEXT:    s_bfe_i32 s16, s0, 0x40018
-; GFX9-NEXT:    s_ashr_i32 s6, s6, 28
-; GFX9-NEXT:    v_mov_b32_e32 v10, s17
+; GFX9-NEXT:    s_bfe_i32 s16, s1, 0x40018
 ; GFX9-NEXT:    s_ashr_i32 s0, s0, 28
+; GFX9-NEXT:    v_mov_b32_e32 v10, s17
+; GFX9-NEXT:    s_ashr_i32 s1, s1, 28
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mad_i32_i24 v2, s5, v6, v2
+; GFX9-NEXT:    v_mad_i32_i24 v2, s6, v6, v2
 ; GFX9-NEXT:    v_mad_i32_i24 v2, s8, v7, v2
 ; GFX9-NEXT:    v_add_u32_sdwa v2, v2, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_0
 ; GFX9-NEXT:    v_mad_u32_u24 v2, v4, v5, v2
 ; GFX9-NEXT:    v_mad_i32_i24 v2, s12, v8, v2
 ; GFX9-NEXT:    v_mad_i32_i24 v2, s14, v9, v2
 ; GFX9-NEXT:    v_mad_i32_i24 v2, s16, v10, v2
-; GFX9-NEXT:    v_mov_b32_e32 v3, s6
-; GFX9-NEXT:    v_mad_i32_i24 v2, s0, v3, v2
+; GFX9-NEXT:    v_mov_b32_e32 v3, s0
+; GFX9-NEXT:    v_mad_i32_i24 v2, s1, v3, v2
 ; GFX9-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: idot8_acc8:
 ; GFX9-DL:       ; %bb.0: ; %entry
+; GFX9-DL-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_movk_i32 s2, 0xff
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s6, s[6:7], 0x0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s8
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s9
 ; GFX9-DL-NEXT:    global_load_ubyte v2, v[0:1], off
-; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_load_dword s0, s[6:7], 0x0
+; GFX9-DL-NEXT:    s_load_dword s1, s[4:5], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_bfe_i32 s7, s6, 0x40000
-; GFX9-DL-NEXT:    s_lshr_b32 s4, s6, 12
-; GFX9-DL-NEXT:    s_bfe_i32 s9, s6, 0x40004
-; GFX9-DL-NEXT:    s_bfe_i32 s11, s6, 0x40008
-; GFX9-DL-NEXT:    s_lshr_b32 s1, s0, 12
-; GFX9-DL-NEXT:    s_bfe_i32 s5, s0, 0x40000
+; GFX9-DL-NEXT:    s_bfe_i32 s7, s0, 0x40000
+; GFX9-DL-NEXT:    s_lshr_b32 s4, s1, 12
+; GFX9-DL-NEXT:    s_lshr_b32 s5, s0, 12
+; GFX9-DL-NEXT:    s_bfe_i32 s9, s0, 0x40004
+; GFX9-DL-NEXT:    s_bfe_i32 s11, s0, 0x40008
+; GFX9-DL-NEXT:    s_bfe_i32 s6, s1, 0x40000
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v6, s7
-; GFX9-DL-NEXT:    v_lshlrev_b16_e64 v4, 12, s1
-; GFX9-DL-NEXT:    v_lshlrev_b16_e64 v5, 12, s4
-; GFX9-DL-NEXT:    s_bfe_i32 s8, s0, 0x40004
-; GFX9-DL-NEXT:    s_bfe_i32 s10, s0, 0x40008
+; GFX9-DL-NEXT:    v_lshlrev_b16_e64 v4, 12, s4
+; GFX9-DL-NEXT:    v_lshlrev_b16_e64 v5, 12, s5
+; GFX9-DL-NEXT:    s_bfe_i32 s8, s1, 0x40004
+; GFX9-DL-NEXT:    s_bfe_i32 s10, s1, 0x40008
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s11
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v7, s9
 ; GFX9-DL-NEXT:    v_ashrrev_i16_e32 v4, 12, v4
 ; GFX9-DL-NEXT:    v_ashrrev_i16_e32 v5, 12, v5
 ; GFX9-DL-NEXT:    v_mul_i32_i24_e32 v3, s10, v3
-; GFX9-DL-NEXT:    s_bfe_i32 s13, s6, 0x40010
+; GFX9-DL-NEXT:    s_bfe_i32 s13, s0, 0x40010
 ; GFX9-DL-NEXT:    v_and_b32_e32 v4, s2, v4
 ; GFX9-DL-NEXT:    v_and_b32_e32 v5, s2, v5
-; GFX9-DL-NEXT:    s_bfe_i32 s15, s6, 0x40014
-; GFX9-DL-NEXT:    s_bfe_i32 s12, s0, 0x40010
+; GFX9-DL-NEXT:    s_bfe_i32 s15, s0, 0x40014
+; GFX9-DL-NEXT:    s_bfe_i32 s12, s1, 0x40010
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v8, s13
-; GFX9-DL-NEXT:    s_bfe_i32 s17, s6, 0x40018
-; GFX9-DL-NEXT:    s_bfe_i32 s14, s0, 0x40014
+; GFX9-DL-NEXT:    s_bfe_i32 s17, s0, 0x40018
+; GFX9-DL-NEXT:    s_bfe_i32 s14, s1, 0x40014
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v9, s15
-; GFX9-DL-NEXT:    s_bfe_i32 s16, s0, 0x40018
-; GFX9-DL-NEXT:    s_ashr_i32 s6, s6, 28
-; GFX9-DL-NEXT:    v_mov_b32_e32 v10, s17
+; GFX9-DL-NEXT:    s_bfe_i32 s16, s1, 0x40018
 ; GFX9-DL-NEXT:    s_ashr_i32 s0, s0, 28
+; GFX9-DL-NEXT:    v_mov_b32_e32 v10, s17
+; GFX9-DL-NEXT:    s_ashr_i32 s1, s1, 28
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s5, v6, v2
+; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s6, v6, v2
 ; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s8, v7, v2
 ; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_0
 ; GFX9-DL-NEXT:    v_mad_u32_u24 v2, v4, v5, v2
 ; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s12, v8, v2
 ; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s14, v9, v2
 ; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s16, v10, v2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s6
-; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s0, v3, v2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s0
+; GFX9-DL-NEXT:    v_mad_i32_i24 v2, s1, v3, v2
 ; GFX9-DL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: idot8_acc8:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX10-DL-NEXT:    s_movk_i32 s2, 0xff
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
+; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX10-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_lshr_b32 s4, s0, 12
 ; GFX10-DL-NEXT:    s_lshr_b32 s5, s1, 12
 ; GFX10-DL-NEXT:    s_bfe_i32 s6, s0, 0x40000
 ; GFX10-DL-NEXT:    s_bfe_i32 s7, s1, 0x40000
 ; GFX10-DL-NEXT:    s_bfe_i32 s8, s0, 0x40004
 ; GFX10-DL-NEXT:    v_lshlrev_b16_e64 v3, 12, s4
 ; GFX10-DL-NEXT:    v_lshlrev_b16_e64 v4, 12, s5

 ; GFX7-NEXT:    v_mad_i32_i24 v0, s9, v7, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s15
 ; GFX7-NEXT:    v_mad_i32_i24 v0, s8, v1, v0
 ; GFX7-NEXT:    buffer_store_short v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: idot8_acc16_vecMul:
 ; GFX8:       ; %bb.0: ; %entry
+; GFX8-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0x34
 ; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
-; GFX8-NEXT:    flat_load_ushort v2, v[0:1]
-; GFX8-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX8-NEXT:    s_load_dword s0, s[6:7], 0x0
+; GFX8-NEXT:    v_mov_b32_e32 v0, s8
+; GFX8-NEXT:    v_mov_b32_e32 v1, s9
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    v_lshlrev_b16_e64 v4, 12, s2
-; GFX8-NEXT:    s_lshr_b32 s15, s2, 4
-; GFX8-NEXT:    s_lshr_b32 s16, s2, 8
+; GFX8-NEXT:    flat_load_ushort v2, v[0:1]
+; GFX8-NEXT:    s_load_dword s1, s[4:5], 0x0
+; GFX8-NEXT:    v_lshlrev_b16_e64 v4, 12, s0
+; GFX8-NEXT:    s_lshr_b32 s15, s0, 4
+; GFX8-NEXT:    s_lshr_b32 s16, s0, 8
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v12, 12, s16
-; GFX8-NEXT:    v_lshlrev_b16_e64 v3, 12, s0
-; GFX8-NEXT:    s_lshr_b32 s8, s0, 4
-; GFX8-NEXT:    s_lshr_b32 s9, s0, 8
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    v_lshlrev_b16_e64 v3, 12, s1
+; GFX8-NEXT:    s_lshr_b32 s8, s1, 4
+; GFX8-NEXT:    s_lshr_b32 s9, s1, 8
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v5, 12, s9
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v6, 12, s8
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v13, 12, s15
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v3, 12, v3
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v4, 12, v4
-; GFX8-NEXT:    s_lshr_b32 s7, s0, 12
-; GFX8-NEXT:    s_lshr_b32 s14, s2, 12
+; GFX8-NEXT:    s_lshr_b32 s7, s1, 12
+; GFX8-NEXT:    s_lshr_b32 s14, s0, 12
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v6, 12, v6
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v5, 12, v5
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v12, 12, v12
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v13, 12, v13
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v7, 12, s7
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v14, 12, s14
-; GFX8-NEXT:    s_lshr_b32 s6, s0, 16
-; GFX8-NEXT:    s_lshr_b32 s13, s2, 16
+; GFX8-NEXT:    s_lshr_b32 s6, s1, 16
+; GFX8-NEXT:    s_lshr_b32 s13, s0, 16
 ; GFX8-NEXT:    v_mul_u32_u24_e32 v5, v5, v12
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v8, 12, s6
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v15, 12, s13
-; GFX8-NEXT:    s_lshr_b32 s5, s0, 20
-; GFX8-NEXT:    s_lshr_b32 s12, s2, 20
+; GFX8-NEXT:    s_lshr_b32 s5, s1, 20
+; GFX8-NEXT:    s_lshr_b32 s12, s0, 20
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v7, 12, v7
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v14, 12, v14
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v9, 12, s5
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v16, 12, s12
-; GFX8-NEXT:    s_lshr_b32 s4, s0, 24
-; GFX8-NEXT:    s_lshr_b32 s11, s2, 24
+; GFX8-NEXT:    s_lshr_b32 s4, s1, 24
+; GFX8-NEXT:    s_lshr_b32 s11, s0, 24
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v8, 12, v8
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v15, 12, v15
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v10, 12, s4
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v17, 12, s11
-; GFX8-NEXT:    s_lshr_b32 s1, s0, 28
-; GFX8-NEXT:    s_lshr_b32 s10, s2, 28
+; GFX8-NEXT:    s_lshr_b32 s2, s1, 28
+; GFX8-NEXT:    s_lshr_b32 s10, s0, 28
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v9, 12, v9
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v16, 12, v16
-; GFX8-NEXT:    v_lshlrev_b16_e64 v11, 12, s1
+; GFX8-NEXT:    v_lshlrev_b16_e64 v11, 12, s2
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v18, 12, s10
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v10, 12, v10
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v17, 12, v17
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v11, 12, v11
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v18, 12, v18
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
 ; GFX8-NEXT:    v_mad_u32_u24 v2, v3, v4, v2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, v6, v13, v2

 ; GFX9-LABEL: idot8_acc16_vecMul:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_load_dword s2, s[4:5], 0x0
 ; GFX9-NEXT:    s_load_dword s6, s[6:7], 0x0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_and_b32 s4, s2, 15
-; GFX9-NEXT:    s_bfe_u32 s5, s2, 0x40004
-; GFX9-NEXT:    s_bfe_u32 s8, s2, 0x40008
-; GFX9-NEXT:    s_bfe_u32 s9, s2, 0x4000c
-; GFX9-NEXT:    s_bfe_u32 s10, s2, 0x40010
-; GFX9-NEXT:    s_bfe_u32 s11, s2, 0x40018
-; GFX9-NEXT:    s_lshr_b32 s12, s2, 28
-; GFX9-NEXT:    s_bfe_u32 s2, s2, 0x40014
-; GFX9-NEXT:    s_pack_ll_b32_b16 s2, s10, s2
+; GFX9-NEXT:    s_bfe_u32 s4, s2, 0x40018
+; GFX9-NEXT:    s_lshr_b32 s5, s2, 28
+; GFX9-NEXT:    s_bfe_u32 s8, s2, 0x40010
+; GFX9-NEXT:    s_bfe_u32 s9, s2, 0x40014
+; GFX9-NEXT:    s_bfe_u32 s10, s2, 0x40008
+; GFX9-NEXT:    s_bfe_u32 s11, s2, 0x4000c
+; GFX9-NEXT:    s_and_b32 s12, s2, 15
+; GFX9-NEXT:    s_bfe_u32 s2, s2, 0x40004
+; GFX9-NEXT:    s_pack_ll_b32_b16 s2, s12, s2
+; GFX9-NEXT:    v_pk_lshlrev_b16 v0, 12, s2 op_sel_hi:[0,1]
+; GFX9-NEXT:    s_pack_ll_b32_b16 s2, s10, s11
+; GFX9-NEXT:    v_pk_lshlrev_b16 v1, 12, s2 op_sel_hi:[0,1]
+; GFX9-NEXT:    s_pack_ll_b32_b16 s2, s8, s9
 ; GFX9-NEXT:    v_pk_lshlrev_b16 v2, 12, s2 op_sel_hi:[0,1]
-; GFX9-NEXT:    s_pack_ll_b32_b16 s2, s11, s12
+; GFX9-NEXT:    s_pack_ll_b32_b16 s2, s4, s5
 ; GFX9-NEXT:    s_bfe_u32 s7, s6, 0x40018
 ; GFX9-NEXT:    s_lshr_b32 s13, s6, 28
 ; GFX9-NEXT:    s_bfe_u32 s14, s6, 0x40010
 ; GFX9-NEXT:    s_bfe_u32 s15, s6, 0x40014
 ; GFX9-NEXT:    s_bfe_u32 s16, s6, 0x40008
 ; GFX9-NEXT:    s_bfe_u32 s17, s6, 0x4000c
 ; GFX9-NEXT:    s_and_b32 s18, s6, 15
 ; GFX9-NEXT:    s_bfe_u32 s6, s6, 0x40004
 ; GFX9-NEXT:    v_pk_lshlrev_b16 v3, 12, s2 op_sel_hi:[0,1]
-; GFX9-NEXT:    s_pack_ll_b32_b16 s4, s4, s5
 ; GFX9-NEXT:    s_pack_ll_b32_b16 s2, s18, s6
-; GFX9-NEXT:    v_pk_lshlrev_b16 v0, 12, s4 op_sel_hi:[0,1]
 ; GFX9-NEXT:    v_pk_lshlrev_b16 v4, 12, s2 op_sel_hi:[0,1]
-; GFX9-NEXT:    s_pack_ll_b32_b16 s4, s8, s9
 ; GFX9-NEXT:    s_pack_ll_b32_b16 s2, s16, s17
 ; GFX9-NEXT:    v_pk_lshlrev_b16 v5, 12, s2 op_sel_hi:[0,1]
-; GFX9-NEXT:    v_pk_lshlrev_b16 v1, 12, s4 op_sel_hi:[0,1]
 ; GFX9-NEXT:    s_pack_ll_b32_b16 s2, s14, s15
 ; GFX9-NEXT:    v_pk_ashrrev_i16 v0, 12, v0 op_sel_hi:[0,1]
 ; GFX9-NEXT:    v_pk_ashrrev_i16 v4, 12, v4 op_sel_hi:[0,1]
 ; GFX9-NEXT:    v_pk_ashrrev_i16 v1, 12, v1 op_sel_hi:[0,1]
 ; GFX9-NEXT:    v_pk_ashrrev_i16 v5, 12, v5 op_sel_hi:[0,1]
 ; GFX9-NEXT:    v_pk_lshlrev_b16 v6, 12, s2 op_sel_hi:[0,1]
 ; GFX9-NEXT:    v_pk_mul_lo_u16 v5, v1, v5
 ; GFX9-NEXT:    v_pk_mul_lo_u16 v4, v0, v4

 ; GFX9-DL-LABEL: idot8_acc16_vecMul:
 ; GFX9-DL:       ; %bb.0: ; %entry
 ; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_load_dword s2, s[4:5], 0x0
 ; GFX9-DL-NEXT:    s_load_dword s6, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_and_b32 s4, s2, 15
-; GFX9-DL-NEXT:    s_bfe_u32 s5, s2, 0x40004
-; GFX9-DL-NEXT:    s_bfe_u32 s8, s2, 0x40008
-; GFX9-DL-NEXT:    s_bfe_u32 s9, s2, 0x4000c
-; GFX9-DL-NEXT:    s_bfe_u32 s10, s2, 0x40010
-; GFX9-DL-NEXT:    s_bfe_u32 s11, s2, 0x40018
-; GFX9-DL-NEXT:    s_lshr_b32 s12, s2, 28
-; GFX9-DL-NEXT:    s_bfe_u32 s2, s2, 0x40014
-; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s2, s10, s2
+; GFX9-DL-NEXT:    s_bfe_u32 s4, s2, 0x40018
+; GFX9-DL-NEXT:    s_lshr_b32 s5, s2, 28
+; GFX9-DL-NEXT:    s_bfe_u32 s8, s2, 0x40010
+; GFX9-DL-NEXT:    s_bfe_u32 s9, s2, 0x40014
+; GFX9-DL-NEXT:    s_bfe_u32 s10, s2, 0x40008
+; GFX9-DL-NEXT:    s_bfe_u32 s11, s2, 0x4000c
+; GFX9-DL-NEXT:    s_and_b32 s12, s2, 15
+; GFX9-DL-NEXT:    s_bfe_u32 s2, s2, 0x40004
+; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s2, s12, s2
+; GFX9-DL-NEXT:    v_pk_lshlrev_b16 v0, 12, s2 op_sel_hi:[0,1]
+; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s2, s10, s11
+; GFX9-DL-NEXT:    v_pk_lshlrev_b16 v1, 12, s2 op_sel_hi:[0,1]
+; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s2, s8, s9
 ; GFX9-DL-NEXT:    v_pk_lshlrev_b16 v2, 12, s2 op_sel_hi:[0,1]
-; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s2, s11, s12
+; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s2, s4, s5
 ; GFX9-DL-NEXT:    s_bfe_u32 s7, s6, 0x40018
 ; GFX9-DL-NEXT:    s_lshr_b32 s13, s6, 28
 ; GFX9-DL-NEXT:    s_bfe_u32 s14, s6, 0x40010
 ; GFX9-DL-NEXT:    s_bfe_u32 s15, s6, 0x40014
 ; GFX9-DL-NEXT:    s_bfe_u32 s16, s6, 0x40008
 ; GFX9-DL-NEXT:    s_bfe_u32 s17, s6, 0x4000c
 ; GFX9-DL-NEXT:    s_and_b32 s18, s6, 15
 ; GFX9-DL-NEXT:    s_bfe_u32 s6, s6, 0x40004
 ; GFX9-DL-NEXT:    v_pk_lshlrev_b16 v3, 12, s2 op_sel_hi:[0,1]
-; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s4, s4, s5
 ; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s2, s18, s6
-; GFX9-DL-NEXT:    v_pk_lshlrev_b16 v0, 12, s4 op_sel_hi:[0,1]
 ; GFX9-DL-NEXT:    v_pk_lshlrev_b16 v4, 12, s2 op_sel_hi:[0,1]
-; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s4, s8, s9
 ; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s2, s16, s17
 ; GFX9-DL-NEXT:    v_pk_lshlrev_b16 v5, 12, s2 op_sel_hi:[0,1]
-; GFX9-DL-NEXT:    v_pk_lshlrev_b16 v1, 12, s4 op_sel_hi:[0,1]
 ; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s2, s14, s15
 ; GFX9-DL-NEXT:    v_pk_ashrrev_i16 v0, 12, v0 op_sel_hi:[0,1]
 ; GFX9-DL-NEXT:    v_pk_ashrrev_i16 v4, 12, v4 op_sel_hi:[0,1]
 ; GFX9-DL-NEXT:    v_pk_ashrrev_i16 v1, 12, v1 op_sel_hi:[0,1]
 ; GFX9-DL-NEXT:    v_pk_ashrrev_i16 v5, 12, v5 op_sel_hi:[0,1]
 ; GFX9-DL-NEXT:    v_pk_lshlrev_b16 v6, 12, s2 op_sel_hi:[0,1]
 ; GFX9-DL-NEXT:    v_pk_mul_lo_u16 v5, v1, v5
 ; GFX9-DL-NEXT:    v_pk_mul_lo_u16 v4, v0, v4

 ; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v4, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
 ; GFX9-DL-NEXT:    v_add_u32_e32 v2, v2, v3
 ; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
 ; GFX9-DL-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: idot8_acc16_vecMul:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
+; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX10-DL-NEXT:    global_load_ushort v2, v[0:1], off
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX10-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX10-DL-NEXT:    global_load_ushort v2, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_and_b32 s5, s0, 15
 ; GFX10-DL-NEXT:    s_bfe_u32 s6, s0, 0x40004
 ; GFX10-DL-NEXT:    s_and_b32 s7, s1, 15
 ; GFX10-DL-NEXT:    s_bfe_u32 s8, s1, 0x40004
 ; GFX10-DL-NEXT:    s_bfe_u32 s2, s0, 0x40018
 ; GFX10-DL-NEXT:    s_lshr_b32 s4, s0, 28
 ; GFX10-DL-NEXT:    s_pack_ll_b32_b16 s5, s5, s6
entry:
   store i16 %add8, i16 addrspace(1)* %dst, align 4
   ret void
 }
 
 ; TODO: Support this pattern.
 define amdgpu_kernel void @idot8_acc8_vecMul(<8 x i4> addrspace(1)* %src1,
 ; GFX7-LABEL: idot8_acc8_vecMul:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
 ; GFX7-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX7-NEXT:    s_mov_b32 s6, -1
-; GFX7-NEXT:    s_movk_i32 s0, 0xff
+; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
+; GFX7-NEXT:    s_movk_i32 s2, 0xff
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s2, s[8:9], 0x0
 ; GFX7-NEXT:    buffer_load_ubyte v0, off, s[4:7], 0
-; GFX7-NEXT:    s_load_dword s8, s[10:11], 0x0
-; GFX7-NEXT:    s_mov_b32 s1, 0xffff
+; GFX7-NEXT:    s_mov_b32 s12, 0xffff
+; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
+; GFX7-NEXT:    s_load_dword s1, s[10:11], 0x0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_bfe_i32 s9, s2, 0x40000
-; GFX7-NEXT:    s_bfe_i32 s10, s2, 0x40004
-; GFX7-NEXT:    s_bfe_i32 s16, s8, 0x40000
-; GFX7-NEXT:    s_bfe_i32 s17, s8, 0x40004
-; GFX7-NEXT:    s_bfe_i32 s18, s8, 0x40008
-; GFX7-NEXT:    s_bfe_i32 s19, s8, 0x4000c
-; GFX7-NEXT:    s_bfe_i32 s20, s8, 0x40010
-; GFX7-NEXT:    s_bfe_i32 s21, s8, 0x40014
-; GFX7-NEXT:    s_bfe_i32 s22, s8, 0x40018
-; GFX7-NEXT:    s_ashr_i32 s8, s8, 28
-; GFX7-NEXT:    v_mov_b32_e32 v7, s17
+; GFX7-NEXT:    s_bfe_i32 s8, s0, 0x40000
+; GFX7-NEXT:    s_bfe_i32 s16, s1, 0x40000
+; GFX7-NEXT:    s_bfe_i32 s17, s1, 0x40004
+; GFX7-NEXT:    s_bfe_i32 s18, s1, 0x40008
+; GFX7-NEXT:    s_bfe_i32 s19, s1, 0x4000c
+; GFX7-NEXT:    s_bfe_i32 s20, s1, 0x40010
+; GFX7-NEXT:    s_bfe_i32 s21, s1, 0x40014
+; GFX7-NEXT:    s_bfe_i32 s22, s1, 0x40018
+; GFX7-NEXT:    s_ashr_i32 s1, s1, 28
 ; GFX7-NEXT:    v_mov_b32_e32 v8, s16
-; GFX7-NEXT:    s_bfe_i32 s11, s2, 0x40008
+; GFX7-NEXT:    s_bfe_i32 s9, s0, 0x40004
+; GFX7-NEXT:    v_mov_b32_e32 v7, s17
+; GFX7-NEXT:    s_bfe_i32 s10, s0, 0x40008
 ; GFX7-NEXT:    v_mov_b32_e32 v6, s18
-; GFX7-NEXT:    s_bfe_i32 s12, s2, 0x4000c
+; GFX7-NEXT:    s_bfe_i32 s11, s0, 0x4000c
 ; GFX7-NEXT:    v_mov_b32_e32 v5, s19
-; GFX7-NEXT:    s_bfe_i32 s13, s2, 0x40010
+; GFX7-NEXT:    s_bfe_i32 s13, s0, 0x40010
 ; GFX7-NEXT:    v_mov_b32_e32 v4, s20
-; GFX7-NEXT:    s_bfe_i32 s14, s2, 0x40014
+; GFX7-NEXT:    s_bfe_i32 s14, s0, 0x40014
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s21
-; GFX7-NEXT:    s_bfe_i32 s15, s2, 0x40018
+; GFX7-NEXT:    s_bfe_i32 s15, s0, 0x40018
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s22
-; GFX7-NEXT:    s_ashr_i32 s2, s2, 28
-; GFX7-NEXT:    v_mov_b32_e32 v1, s8
-; GFX7-NEXT:    v_mul_i32_i24_e32 v1, s2, v1
+; GFX7-NEXT:    s_ashr_i32 s0, s0, 28
+; GFX7-NEXT:    v_mov_b32_e32 v1, s1
+; GFX7-NEXT:    v_mul_i32_i24_e32 v1, s0, v1
 ; GFX7-NEXT:    v_mul_i32_i24_e32 v2, s15, v2
 ; GFX7-NEXT:    v_mul_i32_i24_e32 v3, s14, v3
 ; GFX7-NEXT:    v_mul_i32_i24_e32 v9, s13, v4
-; GFX7-NEXT:    v_mul_i32_i24_e32 v5, s12, v5
-; GFX7-NEXT:    v_mul_i32_i24_e32 v6, s11, v6
-; GFX7-NEXT:    v_mul_i32_i24_e32 v7, s10, v7
-; GFX7-NEXT:    v_mul_i32_i24_e32 v8, s9, v8
+; GFX7-NEXT:    v_mul_i32_i24_e32 v5, s11, v5
+; GFX7-NEXT:    v_mul_i32_i24_e32 v6, s10, v6
+; GFX7-NEXT:    v_mul_i32_i24_e32 v7, s9, v7
+; GFX7-NEXT:    v_mul_i32_i24_e32 v8, s8, v8
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v1, 8, v1
-; GFX7-NEXT:    v_and_b32_e32 v2, s0, v2
+; GFX7-NEXT:    v_and_b32_e32 v2, s2, v2
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v3, 8, v3
-; GFX7-NEXT:    v_and_b32_e32 v9, s0, v9
+; GFX7-NEXT:    v_and_b32_e32 v9, s2, v9
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v5, 8, v5
-; GFX7-NEXT:    v_and_b32_e32 v6, s0, v6
+; GFX7-NEXT:    v_and_b32_e32 v6, s2, v6
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v7, 8, v7
-; GFX7-NEXT:    v_and_b32_e32 v8, s0, v8
+; GFX7-NEXT:    v_and_b32_e32 v8, s2, v8
 ; GFX7-NEXT:    v_or_b32_e32 v1, v2, v1
 ; GFX7-NEXT:    v_or_b32_e32 v2, v9, v3
 ; GFX7-NEXT:    v_or_b32_e32 v3, v6, v5
 ; GFX7-NEXT:    v_or_b32_e32 v5, v8, v7
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v1, 16, v1
-; GFX7-NEXT:    v_and_b32_e32 v2, s1, v2
+; GFX7-NEXT:    v_and_b32_e32 v2, s12, v2
 ; GFX7-NEXT:    v_lshlrev_b32_e32 v3, 16, v3
-; GFX7-NEXT:    v_and_b32_e32 v5, s1, v5
+; GFX7-NEXT:    v_and_b32_e32 v5, s12, v5
 ; GFX7-NEXT:    v_or_b32_e32 v1, v2, v1
 ; GFX7-NEXT:    v_or_b32_e32 v2, v5, v3
 ; GFX7-NEXT:    v_alignbit_b32 v3, v1, v2, 8
 ; GFX7-NEXT:    v_alignbit_b32 v5, v1, v2, 16
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v6, 24, v2
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v7, 8, v1
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v8, 16, v1
 ; GFX7-NEXT:    v_lshrrev_b32_e32 v1, 24, v1
 ; GFX7-NEXT:    s_waitcnt vmcnt(0)
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v0, v2
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v3, v0
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v5, v0
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v6, v0
 ; GFX7-NEXT:    v_mad_i32_i24 v0, s13, v4, v0
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v0, v7
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v0, v8
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v0, v1
 ; GFX7-NEXT:    buffer_store_byte v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: idot8_acc8_vecMul:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX8-NEXT:    s_mov_b32 s2, 0xffff
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    v_mov_b32_e32 v0, s4
+; GFX8-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX8-NEXT:    flat_load_ubyte v2, v[0:1]
-; GFX8-NEXT:    s_load_dword s1, s[4:5], 0x0
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    s_mov_b32 s0, 0xffff
+; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_lshr_b32 s8, s1, 4
-; GFX8-NEXT:    s_lshr_b32 s9, s1, 12
-; GFX8-NEXT:    s_lshr_b32 s10, s1, 8
-; GFX8-NEXT:    s_lshr_b32 s15, s2, 4
-; GFX8-NEXT:    s_lshr_b32 s16, s2, 12
-; GFX8-NEXT:    s_lshr_b32 s17, s2, 8
-; GFX8-NEXT:    v_lshlrev_b16_e64 v3, 12, s1
-; GFX8-NEXT:    v_lshlrev_b16_e64 v4, 12, s2
+; GFX8-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX8-NEXT:    s_load_dword s1, s[6:7], 0x0
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    s_lshr_b32 s8, s0, 4
+; GFX8-NEXT:    s_lshr_b32 s9, s0, 12
+; GFX8-NEXT:    s_lshr_b32 s10, s0, 8
+; GFX8-NEXT:    s_lshr_b32 s15, s1, 4
+; GFX8-NEXT:    s_lshr_b32 s16, s1, 12
+; GFX8-NEXT:    s_lshr_b32 s17, s1, 8
+; GFX8-NEXT:    v_lshlrev_b16_e64 v3, 12, s0
+; GFX8-NEXT:    v_lshlrev_b16_e64 v4, 12, s1
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v5, 12, s10
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v6, 12, s9
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v7, 12, s8
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v12, 12, s17
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v13, 12, s16
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v14, 12, s15
-; GFX8-NEXT:    s_lshr_b32 s4, s1, 20
-; GFX8-NEXT:    s_lshr_b32 s5, s1, 16
-; GFX8-NEXT:    s_lshr_b32 s6, s1, 28
-; GFX8-NEXT:    s_lshr_b32 s7, s1, 24
-; GFX8-NEXT:    s_lshr_b32 s11, s2, 20
-; GFX8-NEXT:    s_lshr_b32 s12, s2, 16
-; GFX8-NEXT:    s_lshr_b32 s13, s2, 28
-; GFX8-NEXT:    s_lshr_b32 s14, s2, 24
+; GFX8-NEXT:    s_lshr_b32 s4, s0, 20
+; GFX8-NEXT:    s_lshr_b32 s5, s0, 16
+; GFX8-NEXT:    s_lshr_b32 s6, s0, 28
+; GFX8-NEXT:    s_lshr_b32 s7, s0, 24
+; GFX8-NEXT:    s_lshr_b32 s11, s1, 20
+; GFX8-NEXT:    s_lshr_b32 s12, s1, 16
+; GFX8-NEXT:    s_lshr_b32 s13, s1, 28
+; GFX8-NEXT:    s_lshr_b32 s14, s1, 24
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v8, 12, s7
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v9, 12, s6
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v10, 12, s5
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v11, 12, s4
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v15, 12, s14
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v16, 12, s13
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v17, 12, s12
 ; GFX8-NEXT:    v_lshlrev_b16_e64 v18, 12, s11

 ; GFX8-NEXT:    v_ashrrev_i16_e32 v16, 12, v16
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v17, 12, v17
 ; GFX8-NEXT:    v_ashrrev_i16_e32 v18, 12, v18
 ; GFX8-NEXT:    v_or_b32_sdwa v3, v3, v6 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
 ; GFX8-NEXT:    v_mul_u32_u24_sdwa v7, v8, v15 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_0
 ; GFX8-NEXT:    v_mul_u32_u24_sdwa v8, v9, v16 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_0
 ; GFX8-NEXT:    v_mul_u32_u24_sdwa v9, v10, v17 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_0
 ; GFX8-NEXT:    v_mul_u32_u24_sdwa v10, v11, v18 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:BYTE_0
-; GFX8-NEXT:    v_and_b32_e32 v3, s0, v3
+; GFX8-NEXT:    v_and_b32_e32 v3, s2, v3
 ; GFX8-NEXT:    v_or_b32_sdwa v4, v4, v5 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
 ; GFX8-NEXT:    v_or_b32_sdwa v9, v9, v10 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
 ; GFX8-NEXT:    v_or_b32_sdwa v7, v7, v8 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
-; GFX8-NEXT:    v_and_b32_e32 v5, s0, v9
+; GFX8-NEXT:    v_and_b32_e32 v5, s2, v9
 ; GFX8-NEXT:    v_or_b32_e32 v4, v3, v4
 ; GFX8-NEXT:    v_or_b32_e32 v6, v5, v7
 ; GFX8-NEXT:    v_lshrrev_b32_e32 v7, 8, v4
 ; GFX8-NEXT:    v_lshrrev_b32_e32 v8, 8, v6
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
 ; GFX8-NEXT:    v_add_u32_e32 v2, vcc, v2, v3
 ; GFX8-NEXT:    v_add_u32_e32 v2, vcc, v7, v2
 ; GFX8-NEXT:    v_add_u32_sdwa v2, vcc, v4, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_2 src1_sel:BYTE_0
 ; GFX8-NEXT:    v_add_u32_sdwa v2, vcc, v4, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:DWORD
 ; GFX8-NEXT:    v_add_u32_e32 v2, vcc, v2, v5
 ; GFX8-NEXT:    v_add_u32_e32 v2, vcc, v8, v2
 ; GFX8-NEXT:    v_add_u32_sdwa v2, vcc, v6, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; GFX8-NEXT:    v_add_u32_sdwa v2, vcc, v6, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:DWORD
 ; GFX8-NEXT:    flat_store_byte v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: idot8_acc8_vecMul:
 ; GFX9:       ; %bb.0: ; %entry
-; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX9-NEXT:    s_mov_b32 s2, 0xffff
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX9-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_lshr_b32 s8, s0, 4
 ; GFX9-NEXT:    s_lshr_b32 s15, s1, 4
 ; GFX9-NEXT:    v_lshlrev_b16_e64 v3, 12, s0
 ; GFX9-NEXT:    v_lshlrev_b16_e64 v4, 12, s1
 ; GFX9-NEXT:    v_lshlrev_b16_e64 v7, 12, s8

 ; GFX9-NEXT:    v_add_u32_e32 v2, v2, v3
 ; GFX9-NEXT:    v_add_u32_sdwa v2, v2, v6 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
 ; GFX9-NEXT:    v_add_u32_sdwa v2, v2, v6 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
 ; GFX9-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: idot8_acc8_vecMul:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_mov_b32 s2, 0xffff
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-DL-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX9-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_lshr_b32 s8, s0, 4
 ; GFX9-DL-NEXT:    s_lshr_b32 s15, s1, 4
 ; GFX9-DL-NEXT:    v_lshlrev_b16_e64 v3, 12, s0
 ; GFX9-DL-NEXT:    v_lshlrev_b16_e64 v4, 12, s1
 ; GFX9-DL-NEXT:    v_lshlrev_b16_e64 v7, 12, s8

 ; GFX9-DL-NEXT:    v_add_u32_e32 v2, v2, v3
 ; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v6 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
 ; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v6 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
 ; GFX9-DL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: idot8_acc8_vecMul:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX10-DL-NEXT:    s_mov_b32 s2, 0xffff
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
+; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX10-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_lshr_b32 s8, s0, 4
 ; GFX10-DL-NEXT:    s_lshr_b32 s15, s1, 4
 ; GFX10-DL-NEXT:    s_lshr_b32 s9, s0, 12
 ; GFX10-DL-NEXT:    s_lshr_b32 s16, s1, 12
 ; GFX10-DL-NEXT:    v_lshlrev_b16_e64 v3, 12, s0
 ; GFX10-DL-NEXT:    v_lshlrev_b16_e64 v7, 12, s8
 ; GFX10-DL-NEXT:    v_lshlrev_b16_e64 v12, 12, s15

llvm/test/CodeGen/AMDGPU/idot8u.ll

 ; GFX9-DL-LABEL: udot8_acc32:
 ; GFX9-DL:       ; %bb.0: ; %entry
 ; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_load_dword s6, s[0:1], 0x0
 ; GFX9-DL-NEXT:    s_load_dword s4, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s6
+; GFX9-DL-NEXT:    v_dot8_u32_u4 v2, s4, v0, v1
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
-; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mov_b32_e32 v2, s2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s6
-; GFX9-DL-NEXT:    v_dot8_u32_u4 v2, s4, v2, v3
 ; GFX9-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot8_acc32:
 ; GFX10-DL:       ; %bb.0: ; %entry
+; GFX10-DL-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0x34
 ; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    s_load_dword s2, s[0:1], 0x0
-; GFX10-DL-NEXT:    s_load_dword s4, s[4:5], 0x0
-; GFX10-DL-NEXT:    s_load_dword s5, s[6:7], 0x0
+; GFX10-DL-NEXT:    s_load_dword s0, s[8:9], 0x0
+; GFX10-DL-NEXT:    s_load_dword s1, s[4:5], 0x0
+; GFX10-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
-; GFX10-DL-NEXT:    v_dot8_u32_u4 v2, s4, s5, v0
 ; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_dot8_u32_u4 v2, s1, s2, v0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s8
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s9
 ; GFX10-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX10-DL-NEXT:    s_endpgm
                                        <8 x i4> addrspace(1)* %src2,
                                        i32 addrspace(1)* nocapture %dst) {
 entry:
   %vec1 = load <8 x i4>, <8 x i4> addrspace(1)* %src1
   %vec2 = load <8 x i4>, <8 x i4> addrspace(1)* %src2
 
entry:
   ret void
 }
 
 ; TODO: Remove the unnecessary instruction(that is zero-extending the
 ; 2nd MAD) to have the pattern-recognizer to kick in.
 define amdgpu_kernel void @udot8_acc16(<8 x i4> addrspace(1)* %src1,
 ; GFX7-LABEL: udot8_acc16:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
 ; GFX7-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
 ; GFX7-NEXT:    buffer_load_ushort v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
 ; GFX7-NEXT:    s_load_dword s1, s[10:11], 0x0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    s_lshr_b32 s2, s0, 28
-; GFX7-NEXT:    s_bfe_u32 s8, s0, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s15, s1, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s16, s1, 0x40014
 ; GFX7-NEXT:    s_bfe_u32 s17, s1, 0x40010
 ; GFX7-NEXT:    s_bfe_u32 s18, s1, 0x4000c
 ; GFX7-NEXT:    s_bfe_u32 s19, s1, 0x40008
 ; GFX7-NEXT:    s_bfe_u32 s20, s1, 0x40004
 ; GFX7-NEXT:    s_lshr_b32 s14, s1, 28
 ; GFX7-NEXT:    s_and_b32 s1, s1, 15
+; GFX7-NEXT:    s_bfe_u32 s8, s0, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s9, s0, 0x40014
 ; GFX7-NEXT:    s_bfe_u32 s10, s0, 0x40010
 ; GFX7-NEXT:    s_bfe_u32 s11, s0, 0x4000c
 ; GFX7-NEXT:    s_bfe_u32 s12, s0, 0x40008
 ; GFX7-NEXT:    s_bfe_u32 s13, s0, 0x40004
 ; GFX7-NEXT:    s_and_b32 s0, s0, 15
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s20

 ; GFX7-NEXT:    v_mad_u32_u24 v0, s8, v7, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s14
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s2, v1, v0
 ; GFX7-NEXT:    buffer_store_short v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot8_acc16:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    v_mov_b32_e32 v0, s4
+; GFX8-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX8-NEXT:    flat_load_ushort v2, v[0:1]
+; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX8-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_and_b32 s1, s2, 15
-; GFX8-NEXT:    s_bfe_u32 s4, s2, 0x40004
+; GFX8-NEXT:    s_lshr_b32 s2, s0, 28
+; GFX8-NEXT:    s_bfe_u32 s11, s1, 0x40018
+; GFX8-NEXT:    s_bfe_u32 s12, s1, 0x40014
+; GFX8-NEXT:    s_bfe_u32 s13, s1, 0x40010
+; GFX8-NEXT:    s_bfe_u32 s14, s1, 0x4000c
+; GFX8-NEXT:    s_bfe_u32 s15, s1, 0x40008
+; GFX8-NEXT:    s_bfe_u32 s16, s1, 0x40004
+; GFX8-NEXT:    s_lshr_b32 s10, s1, 28
+; GFX8-NEXT:    s_and_b32 s1, s1, 15
+; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x40018
+; GFX8-NEXT:    s_bfe_u32 s5, s0, 0x40014
+; GFX8-NEXT:    s_bfe_u32 s6, s0, 0x40010
+; GFX8-NEXT:    s_bfe_u32 s7, s0, 0x4000c
+; GFX8-NEXT:    s_bfe_u32 s8, s0, 0x40008
+; GFX8-NEXT:    s_bfe_u32 s9, s0, 0x40004
+; GFX8-NEXT:    s_and_b32 s0, s0, 15
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s1
-; GFX8-NEXT:    v_mov_b32_e32 v4, s4
-; GFX8-NEXT:    s_and_b32 s1, s0, 15
-; GFX8-NEXT:    s_bfe_u32 s5, s0, 0x40004
-; GFX8-NEXT:    s_bfe_u32 s4, s2, 0x40008
-; GFX8-NEXT:    s_bfe_u32 s7, s2, 0x4000c
-; GFX8-NEXT:    v_mov_b32_e32 v5, s4
-; GFX8-NEXT:    s_bfe_u32 s6, s0, 0x40008
-; GFX8-NEXT:    s_bfe_u32 s8, s2, 0x40010
-; GFX8-NEXT:    v_mov_b32_e32 v6, s7
-; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x4000c
-; GFX8-NEXT:    s_bfe_u32 s9, s2, 0x40014
-; GFX8-NEXT:    s_bfe_u32 s7, s0, 0x40010
-; GFX8-NEXT:    v_mov_b32_e32 v7, s8
-; GFX8-NEXT:    s_bfe_u32 s11, s2, 0x40018
-; GFX8-NEXT:    s_bfe_u32 s10, s0, 0x40014
-; GFX8-NEXT:    v_mov_b32_e32 v8, s9
-; GFX8-NEXT:    s_bfe_u32 s12, s0, 0x40018
-; GFX8-NEXT:    s_lshr_b32 s2, s2, 28
+; GFX8-NEXT:    v_mov_b32_e32 v4, s16
+; GFX8-NEXT:    v_mov_b32_e32 v5, s15
+; GFX8-NEXT:    v_mov_b32_e32 v6, s14
+; GFX8-NEXT:    v_mov_b32_e32 v7, s13
+; GFX8-NEXT:    v_mov_b32_e32 v8, s12
 ; GFX8-NEXT:    v_mov_b32_e32 v9, s11
-; GFX8-NEXT:    s_lshr_b32 s0, s0, 28
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
-; GFX8-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
-; GFX8-NEXT:    v_and_b32_e32 v2, 0xffff, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s6, v5, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s4, v6, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s7, v7, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s10, v8, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s12, v9, v2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s9, v4, v2
+; GFX8-NEXT:    v_and_b32_e32 v2, 0xffff, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s8, v5, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s7, v6, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s6, v7, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v8, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s4, v9, v2
+; GFX8-NEXT:    v_mov_b32_e32 v3, s10
+; GFX8-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX8-NEXT:    flat_store_short v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: udot8_acc16:
 ; GFX9:       ; %bb.0: ; %entry
-; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-NEXT:    global_load_ushort v2, v[0:1], off
+; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_and_b32 s1, s2, 15
-; GFX9-NEXT:    s_bfe_u32 s4, s2, 0x40004
+; GFX9-NEXT:    s_lshr_b32 s2, s0, 28
+; GFX9-NEXT:    s_bfe_u32 s11, s1, 0x40018
+; GFX9-NEXT:    s_bfe_u32 s12, s1, 0x40014
+; GFX9-NEXT:    s_bfe_u32 s13, s1, 0x40010
+; GFX9-NEXT:    s_bfe_u32 s14, s1, 0x4000c
+; GFX9-NEXT:    s_bfe_u32 s15, s1, 0x40008
+; GFX9-NEXT:    s_bfe_u32 s16, s1, 0x40004
+; GFX9-NEXT:    s_lshr_b32 s10, s1, 28
+; GFX9-NEXT:    s_and_b32 s1, s1, 15
+; GFX9-NEXT:    s_bfe_u32 s4, s0, 0x40018
+; GFX9-NEXT:    s_bfe_u32 s5, s0, 0x40014
+; GFX9-NEXT:    s_bfe_u32 s6, s0, 0x40010
+; GFX9-NEXT:    s_bfe_u32 s7, s0, 0x4000c
+; GFX9-NEXT:    s_bfe_u32 s8, s0, 0x40008
+; GFX9-NEXT:    s_bfe_u32 s9, s0, 0x40004
+; GFX9-NEXT:    s_and_b32 s0, s0, 15
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-NEXT:    v_mov_b32_e32 v4, s4
-; GFX9-NEXT:    s_and_b32 s1, s0, 15
-; GFX9-NEXT:    s_bfe_u32 s5, s0, 0x40004
-; GFX9-NEXT:    s_bfe_u32 s4, s2, 0x40008
-; GFX9-NEXT:    s_bfe_u32 s7, s2, 0x4000c
-; GFX9-NEXT:    v_mov_b32_e32 v5, s4
-; GFX9-NEXT:    s_bfe_u32 s6, s0, 0x40008
-; GFX9-NEXT:    s_bfe_u32 s8, s2, 0x40010
-; GFX9-NEXT:    v_mov_b32_e32 v6, s7
-; GFX9-NEXT:    s_bfe_u32 s4, s0, 0x4000c
-; GFX9-NEXT:    s_bfe_u32 s9, s2, 0x40014
-; GFX9-NEXT:    s_bfe_u32 s7, s0, 0x40010
-; GFX9-NEXT:    v_mov_b32_e32 v7, s8
-; GFX9-NEXT:    s_bfe_u32 s11, s2, 0x40018
-; GFX9-NEXT:    s_bfe_u32 s10, s0, 0x40014
-; GFX9-NEXT:    v_mov_b32_e32 v8, s9
-; GFX9-NEXT:    s_bfe_u32 s12, s0, 0x40018
-; GFX9-NEXT:    s_lshr_b32 s2, s2, 28
+; GFX9-NEXT:    v_mov_b32_e32 v4, s16
+; GFX9-NEXT:    v_mov_b32_e32 v5, s15
+; GFX9-NEXT:    v_mov_b32_e32 v6, s14
+; GFX9-NEXT:    v_mov_b32_e32 v7, s13
+; GFX9-NEXT:    v_mov_b32_e32 v8, s12
 ; GFX9-NEXT:    v_mov_b32_e32 v9, s11
-; GFX9-NEXT:    s_lshr_b32 s0, s0, 28
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
-; GFX9-NEXT:    v_and_b32_e32 v2, 0xffff, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s6, v5, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s4, v6, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s7, v7, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s10, v8, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s12, v9, v2
-; GFX9-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX9-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s9, v4, v2
+; GFX9-NEXT:    v_and_b32_e32 v2, 0xffff, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s8, v5, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s7, v6, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s6, v7, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s5, v8, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s4, v9, v2
+; GFX9-NEXT:    v_mov_b32_e32 v3, s10
+; GFX9-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX9-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: udot8_acc16:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-DL-NEXT:    global_load_ushort v2, v[0:1], off
+; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_and_b32 s1, s2, 15
-; GFX9-DL-NEXT:    s_bfe_u32 s4, s2, 0x40004
+; GFX9-DL-NEXT:    s_lshr_b32 s2, s0, 28
+; GFX9-DL-NEXT:    s_bfe_u32 s11, s1, 0x40018
+; GFX9-DL-NEXT:    s_bfe_u32 s12, s1, 0x40014
+; GFX9-DL-NEXT:    s_bfe_u32 s13, s1, 0x40010
+; GFX9-DL-NEXT:    s_bfe_u32 s14, s1, 0x4000c
+; GFX9-DL-NEXT:    s_bfe_u32 s15, s1, 0x40008
+; GFX9-DL-NEXT:    s_bfe_u32 s16, s1, 0x40004
+; GFX9-DL-NEXT:    s_lshr_b32 s10, s1, 28
+; GFX9-DL-NEXT:    s_and_b32 s1, s1, 15
+; GFX9-DL-NEXT:    s_bfe_u32 s4, s0, 0x40018
+; GFX9-DL-NEXT:    s_bfe_u32 s5, s0, 0x40014
+; GFX9-DL-NEXT:    s_bfe_u32 s6, s0, 0x40010
+; GFX9-DL-NEXT:    s_bfe_u32 s7, s0, 0x4000c
+; GFX9-DL-NEXT:    s_bfe_u32 s8, s0, 0x40008
+; GFX9-DL-NEXT:    s_bfe_u32 s9, s0, 0x40004
+; GFX9-DL-NEXT:    s_and_b32 s0, s0, 15
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s4
-; GFX9-DL-NEXT:    s_and_b32 s1, s0, 15
-; GFX9-DL-NEXT:    s_bfe_u32 s5, s0, 0x40004
-; GFX9-DL-NEXT:    s_bfe_u32 s4, s2, 0x40008
-; GFX9-DL-NEXT:    s_bfe_u32 s7, s2, 0x4000c
-; GFX9-DL-NEXT:    v_mov_b32_e32 v5, s4
-; GFX9-DL-NEXT:    s_bfe_u32 s6, s0, 0x40008
-; GFX9-DL-NEXT:    s_bfe_u32 s8, s2, 0x40010
-; GFX9-DL-NEXT:    v_mov_b32_e32 v6, s7
-; GFX9-DL-NEXT:    s_bfe_u32 s4, s0, 0x4000c
-; GFX9-DL-NEXT:    s_bfe_u32 s9, s2, 0x40014
-; GFX9-DL-NEXT:    s_bfe_u32 s7, s0, 0x40010
-; GFX9-DL-NEXT:    v_mov_b32_e32 v7, s8
-; GFX9-DL-NEXT:    s_bfe_u32 s11, s2, 0x40018
-; GFX9-DL-NEXT:    s_bfe_u32 s10, s0, 0x40014
-; GFX9-DL-NEXT:    v_mov_b32_e32 v8, s9
-; GFX9-DL-NEXT:    s_bfe_u32 s12, s0, 0x40018
-; GFX9-DL-NEXT:    s_lshr_b32 s2, s2, 28
+; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s16
+; GFX9-DL-NEXT:    v_mov_b32_e32 v5, s15
+; GFX9-DL-NEXT:    v_mov_b32_e32 v6, s14
+; GFX9-DL-NEXT:    v_mov_b32_e32 v7, s13
+; GFX9-DL-NEXT:    v_mov_b32_e32 v8, s12
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v9, s11
-; GFX9-DL-NEXT:    s_lshr_b32 s0, s0, 28
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
-; GFX9-DL-NEXT:    v_and_b32_e32 v2, 0xffff, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s6, v5, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s4, v6, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s7, v7, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s10, v8, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s12, v9, v2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s9, v4, v2
+; GFX9-DL-NEXT:    v_and_b32_e32 v2, 0xffff, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s8, v5, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s7, v6, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s6, v7, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s5, v8, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s4, v9, v2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s10
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX9-DL-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot8_acc16:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
+; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX10-DL-NEXT:    global_load_ushort v2, v[0:1], off
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX10-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX10-DL-NEXT:    global_load_ushort v2, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_and_b32 s2, s0, 15
 ; GFX10-DL-NEXT:    s_and_b32 s4, s1, 15
 ; GFX10-DL-NEXT:    s_bfe_u32 s5, s0, 0x40004
 ; GFX10-DL-NEXT:    s_bfe_u32 s6, s1, 0x40004
 ; GFX10-DL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX10-DL-NEXT:    v_mad_u32_u24 v2, s2, s4, v2
 ; GFX10-DL-NEXT:    s_bfe_u32 s2, s0, 0x40008
entry:
   ret void
 }
 
 ; TODO: Remove the unnecessary instruction(that is zero-extending the
 ; 2nd MAD) to have the pattern-recognizer to kick in.
 define amdgpu_kernel void @udot8_acc8(<8 x i4> addrspace(1)* %src1,
 ; GFX7-LABEL: udot8_acc8:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
 ; GFX7-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
 ; GFX7-NEXT:    buffer_load_ubyte v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
 ; GFX7-NEXT:    s_load_dword s1, s[10:11], 0x0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    s_lshr_b32 s2, s0, 28
-; GFX7-NEXT:    s_bfe_u32 s8, s0, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s15, s1, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s16, s1, 0x40014
 ; GFX7-NEXT:    s_bfe_u32 s17, s1, 0x40010
 ; GFX7-NEXT:    s_bfe_u32 s18, s1, 0x4000c
 ; GFX7-NEXT:    s_bfe_u32 s19, s1, 0x40008
 ; GFX7-NEXT:    s_bfe_u32 s20, s1, 0x40004
 ; GFX7-NEXT:    s_lshr_b32 s14, s1, 28
 ; GFX7-NEXT:    s_and_b32 s1, s1, 15
+; GFX7-NEXT:    s_bfe_u32 s8, s0, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s9, s0, 0x40014
 ; GFX7-NEXT:    s_bfe_u32 s10, s0, 0x40010
 ; GFX7-NEXT:    s_bfe_u32 s11, s0, 0x4000c
 ; GFX7-NEXT:    s_bfe_u32 s12, s0, 0x40008
 ; GFX7-NEXT:    s_bfe_u32 s13, s0, 0x40004
 ; GFX7-NEXT:    s_and_b32 s0, s0, 15
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s20

 ; GFX7-NEXT:    v_mad_u32_u24 v0, s8, v7, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s14
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s2, v1, v0
 ; GFX7-NEXT:    buffer_store_byte v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot8_acc8:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    v_mov_b32_e32 v0, s4
+; GFX8-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX8-NEXT:    flat_load_ubyte v2, v[0:1]
+; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX8-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_and_b32 s1, s2, 15
-; GFX8-NEXT:    s_bfe_u32 s4, s2, 0x40004
+; GFX8-NEXT:    s_lshr_b32 s2, s0, 28
+; GFX8-NEXT:    s_bfe_u32 s11, s1, 0x40018
+; GFX8-NEXT:    s_bfe_u32 s12, s1, 0x40014
+; GFX8-NEXT:    s_bfe_u32 s13, s1, 0x40010
+; GFX8-NEXT:    s_bfe_u32 s14, s1, 0x4000c
+; GFX8-NEXT:    s_bfe_u32 s15, s1, 0x40008
+; GFX8-NEXT:    s_bfe_u32 s16, s1, 0x40004
+; GFX8-NEXT:    s_lshr_b32 s10, s1, 28
+; GFX8-NEXT:    s_and_b32 s1, s1, 15
+; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x40018
+; GFX8-NEXT:    s_bfe_u32 s5, s0, 0x40014
+; GFX8-NEXT:    s_bfe_u32 s6, s0, 0x40010
+; GFX8-NEXT:    s_bfe_u32 s7, s0, 0x4000c
+; GFX8-NEXT:    s_bfe_u32 s8, s0, 0x40008
+; GFX8-NEXT:    s_bfe_u32 s9, s0, 0x40004
+; GFX8-NEXT:    s_and_b32 s0, s0, 15
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s1
-; GFX8-NEXT:    v_mov_b32_e32 v4, s4
-; GFX8-NEXT:    s_and_b32 s1, s0, 15
-; GFX8-NEXT:    s_bfe_u32 s5, s0, 0x40004
-; GFX8-NEXT:    s_bfe_u32 s4, s2, 0x40008
-; GFX8-NEXT:    s_bfe_u32 s7, s2, 0x4000c
-; GFX8-NEXT:    v_mov_b32_e32 v5, s4
-; GFX8-NEXT:    s_bfe_u32 s6, s0, 0x40008
-; GFX8-NEXT:    s_bfe_u32 s8, s2, 0x40010
-; GFX8-NEXT:    v_mov_b32_e32 v6, s7
-; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x4000c
-; GFX8-NEXT:    s_bfe_u32 s9, s2, 0x40014
-; GFX8-NEXT:    s_bfe_u32 s7, s0, 0x40010
-; GFX8-NEXT:    v_mov_b32_e32 v7, s8
-; GFX8-NEXT:    s_bfe_u32 s11, s2, 0x40018
-; GFX8-NEXT:    s_bfe_u32 s10, s0, 0x40014
-; GFX8-NEXT:    v_mov_b32_e32 v8, s9
-; GFX8-NEXT:    s_bfe_u32 s12, s0, 0x40018
-; GFX8-NEXT:    s_lshr_b32 s2, s2, 28
+; GFX8-NEXT:    v_mov_b32_e32 v4, s16
+; GFX8-NEXT:    v_mov_b32_e32 v5, s15
+; GFX8-NEXT:    v_mov_b32_e32 v6, s14
+; GFX8-NEXT:    v_mov_b32_e32 v7, s13
+; GFX8-NEXT:    v_mov_b32_e32 v8, s12
 ; GFX8-NEXT:    v_mov_b32_e32 v9, s11
-; GFX8-NEXT:    s_lshr_b32 s0, s0, 28
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
-; GFX8-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
-; GFX8-NEXT:    v_and_b32_e32 v2, 0xff, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s6, v5, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s4, v6, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s7, v7, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s10, v8, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s12, v9, v2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s9, v4, v2
+; GFX8-NEXT:    v_and_b32_e32 v2, 0xff, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s8, v5, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s7, v6, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s6, v7, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v8, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s4, v9, v2
+; GFX8-NEXT:    v_mov_b32_e32 v3, s10
+; GFX8-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX8-NEXT:    flat_store_byte v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: udot8_acc8:
 ; GFX9:       ; %bb.0: ; %entry
-; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_and_b32 s1, s2, 15
-; GFX9-NEXT:    s_bfe_u32 s4, s2, 0x40004
+; GFX9-NEXT:    s_lshr_b32 s2, s0, 28
+; GFX9-NEXT:    s_bfe_u32 s11, s1, 0x40018
+; GFX9-NEXT:    s_bfe_u32 s12, s1, 0x40014
+; GFX9-NEXT:    s_bfe_u32 s13, s1, 0x40010
+; GFX9-NEXT:    s_bfe_u32 s14, s1, 0x4000c
+; GFX9-NEXT:    s_bfe_u32 s15, s1, 0x40008
+; GFX9-NEXT:    s_bfe_u32 s16, s1, 0x40004
+; GFX9-NEXT:    s_lshr_b32 s10, s1, 28
+; GFX9-NEXT:    s_and_b32 s1, s1, 15
+; GFX9-NEXT:    s_bfe_u32 s4, s0, 0x40018
+; GFX9-NEXT:    s_bfe_u32 s5, s0, 0x40014
+; GFX9-NEXT:    s_bfe_u32 s6, s0, 0x40010
+; GFX9-NEXT:    s_bfe_u32 s7, s0, 0x4000c
+; GFX9-NEXT:    s_bfe_u32 s8, s0, 0x40008
+; GFX9-NEXT:    s_bfe_u32 s9, s0, 0x40004
+; GFX9-NEXT:    s_and_b32 s0, s0, 15
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-NEXT:    v_mov_b32_e32 v4, s4
-; GFX9-NEXT:    s_and_b32 s1, s0, 15
-; GFX9-NEXT:    s_bfe_u32 s5, s0, 0x40004
-; GFX9-NEXT:    s_bfe_u32 s4, s2, 0x40008
-; GFX9-NEXT:    s_bfe_u32 s7, s2, 0x4000c
-; GFX9-NEXT:    v_mov_b32_e32 v5, s4
-; GFX9-NEXT:    s_bfe_u32 s6, s0, 0x40008
-; GFX9-NEXT:    s_bfe_u32 s8, s2, 0x40010
-; GFX9-NEXT:    v_mov_b32_e32 v6, s7
-; GFX9-NEXT:    s_bfe_u32 s4, s0, 0x4000c
-; GFX9-NEXT:    s_bfe_u32 s9, s2, 0x40014
-; GFX9-NEXT:    s_bfe_u32 s7, s0, 0x40010
-; GFX9-NEXT:    v_mov_b32_e32 v7, s8
-; GFX9-NEXT:    s_bfe_u32 s11, s2, 0x40018
-; GFX9-NEXT:    s_bfe_u32 s10, s0, 0x40014
-; GFX9-NEXT:    v_mov_b32_e32 v8, s9
-; GFX9-NEXT:    s_bfe_u32 s12, s0, 0x40018
-; GFX9-NEXT:    s_lshr_b32 s2, s2, 28
+; GFX9-NEXT:    v_mov_b32_e32 v4, s16
+; GFX9-NEXT:    v_mov_b32_e32 v5, s15
+; GFX9-NEXT:    v_mov_b32_e32 v6, s14
+; GFX9-NEXT:    v_mov_b32_e32 v7, s13
+; GFX9-NEXT:    v_mov_b32_e32 v8, s12
 ; GFX9-NEXT:    v_mov_b32_e32 v9, s11
-; GFX9-NEXT:    s_lshr_b32 s0, s0, 28
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
-; GFX9-NEXT:    v_and_b32_e32 v2, 0xff, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s6, v5, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s4, v6, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s7, v7, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s10, v8, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s12, v9, v2
-; GFX9-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX9-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s9, v4, v2
+; GFX9-NEXT:    v_and_b32_e32 v2, 0xff, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s8, v5, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s7, v6, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s6, v7, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s5, v8, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s4, v9, v2
+; GFX9-NEXT:    v_mov_b32_e32 v3, s10
+; GFX9-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX9-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: udot8_acc8:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-DL-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_and_b32 s1, s2, 15
-; GFX9-DL-NEXT:    s_bfe_u32 s4, s2, 0x40004
+; GFX9-DL-NEXT:    s_lshr_b32 s2, s0, 28
+; GFX9-DL-NEXT:    s_bfe_u32 s11, s1, 0x40018
+; GFX9-DL-NEXT:    s_bfe_u32 s12, s1, 0x40014
+; GFX9-DL-NEXT:    s_bfe_u32 s13, s1, 0x40010
+; GFX9-DL-NEXT:    s_bfe_u32 s14, s1, 0x4000c
+; GFX9-DL-NEXT:    s_bfe_u32 s15, s1, 0x40008
+; GFX9-DL-NEXT:    s_bfe_u32 s16, s1, 0x40004
+; GFX9-DL-NEXT:    s_lshr_b32 s10, s1, 28
+; GFX9-DL-NEXT:    s_and_b32 s1, s1, 15
+; GFX9-DL-NEXT:    s_bfe_u32 s4, s0, 0x40018
+; GFX9-DL-NEXT:    s_bfe_u32 s5, s0, 0x40014
+; GFX9-DL-NEXT:    s_bfe_u32 s6, s0, 0x40010
+; GFX9-DL-NEXT:    s_bfe_u32 s7, s0, 0x4000c
+; GFX9-DL-NEXT:    s_bfe_u32 s8, s0, 0x40008
+; GFX9-DL-NEXT:    s_bfe_u32 s9, s0, 0x40004
+; GFX9-DL-NEXT:    s_and_b32 s0, s0, 15
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s4
-; GFX9-DL-NEXT:    s_and_b32 s1, s0, 15
-; GFX9-DL-NEXT:    s_bfe_u32 s5, s0, 0x40004
-; GFX9-DL-NEXT:    s_bfe_u32 s4, s2, 0x40008
-; GFX9-DL-NEXT:    s_bfe_u32 s7, s2, 0x4000c
-; GFX9-DL-NEXT:    v_mov_b32_e32 v5, s4
-; GFX9-DL-NEXT:    s_bfe_u32 s6, s0, 0x40008
-; GFX9-DL-NEXT:    s_bfe_u32 s8, s2, 0x40010
-; GFX9-DL-NEXT:    v_mov_b32_e32 v6, s7
-; GFX9-DL-NEXT:    s_bfe_u32 s4, s0, 0x4000c
-; GFX9-DL-NEXT:    s_bfe_u32 s9, s2, 0x40014
-; GFX9-DL-NEXT:    s_bfe_u32 s7, s0, 0x40010
-; GFX9-DL-NEXT:    v_mov_b32_e32 v7, s8
-; GFX9-DL-NEXT:    s_bfe_u32 s11, s2, 0x40018
-; GFX9-DL-NEXT:    s_bfe_u32 s10, s0, 0x40014
-; GFX9-DL-NEXT:    v_mov_b32_e32 v8, s9
-; GFX9-DL-NEXT:    s_bfe_u32 s12, s0, 0x40018
-; GFX9-DL-NEXT:    s_lshr_b32 s2, s2, 28
+; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s16
+; GFX9-DL-NEXT:    v_mov_b32_e32 v5, s15
+; GFX9-DL-NEXT:    v_mov_b32_e32 v6, s14
+; GFX9-DL-NEXT:    v_mov_b32_e32 v7, s13
+; GFX9-DL-NEXT:    v_mov_b32_e32 v8, s12
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v9, s11
-; GFX9-DL-NEXT:    s_lshr_b32 s0, s0, 28
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
-; GFX9-DL-NEXT:    v_and_b32_e32 v2, 0xff, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s6, v5, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s4, v6, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s7, v7, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s10, v8, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s12, v9, v2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s9, v4, v2
+; GFX9-DL-NEXT:    v_and_b32_e32 v2, 0xff, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s8, v5, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s7, v6, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s6, v7, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s5, v8, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s4, v9, v2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s10
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX9-DL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot8_acc8:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
+; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX10-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_and_b32 s2, s0, 15
 ; GFX10-DL-NEXT:    s_and_b32 s4, s1, 15
 ; GFX10-DL-NEXT:    s_bfe_u32 s5, s0, 0x40004
 ; GFX10-DL-NEXT:    s_bfe_u32 s6, s1, 0x40004
 ; GFX10-DL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX10-DL-NEXT:    v_mad_u32_u24 v2, s2, s4, v2
 ; GFX10-DL-NEXT:    s_bfe_u32 s2, s0, 0x40008
entry:
   ret void
 }
 
 ; TODO: Remove the two unnecessary instructions(and+add after 2nd MAD)
 ; to have the pattern-recognizer to kick in.
 define amdgpu_kernel void @udot8_acc4(<8 x i4> addrspace(1)* %src1,
 ; GFX7-LABEL: udot8_acc4:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
 ; GFX7-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
 ; GFX7-NEXT:    buffer_load_ubyte v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
 ; GFX7-NEXT:    s_load_dword s1, s[10:11], 0x0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    s_lshr_b32 s2, s0, 28
-; GFX7-NEXT:    s_bfe_u32 s8, s0, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s15, s1, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s16, s1, 0x40014
 ; GFX7-NEXT:    s_bfe_u32 s17, s1, 0x40010
 ; GFX7-NEXT:    s_bfe_u32 s18, s1, 0x4000c
 ; GFX7-NEXT:    s_bfe_u32 s19, s1, 0x40008
 ; GFX7-NEXT:    s_bfe_u32 s20, s1, 0x40004
 ; GFX7-NEXT:    s_lshr_b32 s14, s1, 28
 ; GFX7-NEXT:    s_and_b32 s1, s1, 15
+; GFX7-NEXT:    s_bfe_u32 s8, s0, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s9, s0, 0x40014
 ; GFX7-NEXT:    s_bfe_u32 s10, s0, 0x40010
 ; GFX7-NEXT:    s_bfe_u32 s11, s0, 0x4000c
 ; GFX7-NEXT:    s_bfe_u32 s12, s0, 0x40008
 ; GFX7-NEXT:    s_bfe_u32 s13, s0, 0x40004
 ; GFX7-NEXT:    s_and_b32 s0, s0, 15
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s20

 ; GFX7-NEXT:    v_mov_b32_e32 v1, s14
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s2, v1, v0
 ; GFX7-NEXT:    v_and_b32_e32 v0, 15, v0
 ; GFX7-NEXT:    buffer_store_byte v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot8_acc4:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    v_mov_b32_e32 v0, s4
+; GFX8-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX8-NEXT:    flat_load_ubyte v2, v[0:1]
+; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX8-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_and_b32 s1, s2, 15
-; GFX8-NEXT:    s_bfe_u32 s4, s2, 0x40004
+; GFX8-NEXT:    s_and_b32 s9, s0, 15
+; GFX8-NEXT:    s_and_b32 s16, s1, 15
+; GFX8-NEXT:    s_bfe_u32 s15, s1, 0x40004
+; GFX8-NEXT:    v_mov_b32_e32 v4, s16
+; GFX8-NEXT:    s_bfe_u32 s11, s1, 0x40018
+; GFX8-NEXT:    s_bfe_u32 s12, s1, 0x40014
+; GFX8-NEXT:    s_bfe_u32 s13, s1, 0x40010
+; GFX8-NEXT:    s_bfe_u32 s14, s1, 0x40008
+; GFX8-NEXT:    s_lshr_b32 s10, s1, 28
+; GFX8-NEXT:    s_bfe_u32 s1, s1, 0x4000c
+; GFX8-NEXT:    s_bfe_u32 s8, s0, 0x40004
+; GFX8-NEXT:    v_mov_b32_e32 v5, s15
+; GFX8-NEXT:    s_lshr_b32 s2, s0, 28
+; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x40018
+; GFX8-NEXT:    s_bfe_u32 s5, s0, 0x40014
+; GFX8-NEXT:    s_bfe_u32 s6, s0, 0x40010
+; GFX8-NEXT:    s_bfe_u32 s7, s0, 0x40008
+; GFX8-NEXT:    s_bfe_u32 s0, s0, 0x4000c
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s1
-; GFX8-NEXT:    s_bfe_u32 s6, s2, 0x40008
-; GFX8-NEXT:    s_and_b32 s1, s0, 15
-; GFX8-NEXT:    s_bfe_u32 s7, s2, 0x4000c
-; GFX8-NEXT:    v_mov_b32_e32 v4, s4
-; GFX8-NEXT:    s_bfe_u32 s5, s0, 0x40004
-; GFX8-NEXT:    v_mov_b32_e32 v5, s7
-; GFX8-NEXT:    s_bfe_u32 s8, s0, 0x4000c
-; GFX8-NEXT:    v_mov_b32_e32 v6, s6
-; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x40008
-; GFX8-NEXT:    v_mul_u32_u24_e32 v5, s8, v5
-; GFX8-NEXT:    s_bfe_u32 s6, s2, 0x40010
-; GFX8-NEXT:    v_and_b32_e32 v5, 15, v5
-; GFX8-NEXT:    s_bfe_u32 s8, s2, 0x40014
-; GFX8-NEXT:    s_bfe_u32 s7, s0, 0x40010
-; GFX8-NEXT:    v_mov_b32_e32 v7, s6
-; GFX8-NEXT:    s_lshr_b32 s11, s2, 28
-; GFX8-NEXT:    s_bfe_u32 s2, s2, 0x40018
-; GFX8-NEXT:    s_bfe_u32 s9, s0, 0x40014
-; GFX8-NEXT:    v_mov_b32_e32 v8, s8
-; GFX8-NEXT:    s_bfe_u32 s10, s0, 0x40018
-; GFX8-NEXT:    v_mov_b32_e32 v9, s2
-; GFX8-NEXT:    s_lshr_b32 s0, s0, 28
+; GFX8-NEXT:    v_mov_b32_e32 v6, s14
+; GFX8-NEXT:    v_mul_u32_u24_e32 v3, s0, v3
+; GFX8-NEXT:    v_and_b32_e32 v3, 15, v3
+; GFX8-NEXT:    v_mov_b32_e32 v7, s13
+; GFX8-NEXT:    v_mov_b32_e32 v8, s12
+; GFX8-NEXT:    v_mov_b32_e32 v9, s11
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
-; GFX8-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s4, v6, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s9, v4, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s8, v5, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s7, v6, v2
 ; GFX8-NEXT:    v_and_b32_e32 v2, 15, v2
-; GFX8-NEXT:    v_add_u32_e32 v2, vcc, v5, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s7, v7, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s9, v8, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s10, v9, v2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s11
-; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX8-NEXT:    v_add_u32_e32 v2, vcc, v3, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s6, v7, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v8, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s4, v9, v2
+; GFX8-NEXT:    v_mov_b32_e32 v3, s10
+; GFX8-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX8-NEXT:    v_and_b32_e32 v2, 15, v2
 ; GFX8-NEXT:    flat_store_byte v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: udot8_acc4:
 ; GFX9:       ; %bb.0: ; %entry
-; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_and_b32 s1, s2, 15
-; GFX9-NEXT:    s_bfe_u32 s4, s2, 0x40004
+; GFX9-NEXT:    s_and_b32 s9, s0, 15
+; GFX9-NEXT:    s_and_b32 s16, s1, 15
+; GFX9-NEXT:    s_bfe_u32 s15, s1, 0x40004
+; GFX9-NEXT:    v_mov_b32_e32 v4, s16
+; GFX9-NEXT:    s_bfe_u32 s11, s1, 0x40018
+; GFX9-NEXT:    s_bfe_u32 s12, s1, 0x40014
+; GFX9-NEXT:    s_bfe_u32 s13, s1, 0x40010
+; GFX9-NEXT:    s_bfe_u32 s14, s1, 0x40008
+; GFX9-NEXT:    s_lshr_b32 s10, s1, 28
+; GFX9-NEXT:    s_bfe_u32 s1, s1, 0x4000c
+; GFX9-NEXT:    s_bfe_u32 s8, s0, 0x40004
+; GFX9-NEXT:    v_mov_b32_e32 v5, s15
+; GFX9-NEXT:    s_lshr_b32 s2, s0, 28
+; GFX9-NEXT:    s_bfe_u32 s4, s0, 0x40018
+; GFX9-NEXT:    s_bfe_u32 s5, s0, 0x40014
+; GFX9-NEXT:    s_bfe_u32 s6, s0, 0x40010
+; GFX9-NEXT:    s_bfe_u32 s7, s0, 0x40008
+; GFX9-NEXT:    s_bfe_u32 s0, s0, 0x4000c
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-NEXT:    s_bfe_u32 s6, s2, 0x40008
-; GFX9-NEXT:    s_and_b32 s1, s0, 15
-; GFX9-NEXT:    s_bfe_u32 s7, s2, 0x4000c
-; GFX9-NEXT:    v_mov_b32_e32 v4, s4
-; GFX9-NEXT:    s_bfe_u32 s5, s0, 0x40004
-; GFX9-NEXT:    v_mov_b32_e32 v5, s7
-; GFX9-NEXT:    s_bfe_u32 s8, s0, 0x4000c
-; GFX9-NEXT:    v_mov_b32_e32 v6, s6
-; GFX9-NEXT:    s_bfe_u32 s4, s0, 0x40008
-; GFX9-NEXT:    v_mul_u32_u24_e32 v5, s8, v5
-; GFX9-NEXT:    s_bfe_u32 s6, s2, 0x40010
-; GFX9-NEXT:    v_and_b32_e32 v5, 15, v5
-; GFX9-NEXT:    s_bfe_u32 s8, s2, 0x40014
-; GFX9-NEXT:    s_bfe_u32 s7, s0, 0x40010
-; GFX9-NEXT:    v_mov_b32_e32 v7, s6
-; GFX9-NEXT:    s_lshr_b32 s11, s2, 28
-; GFX9-NEXT:    s_bfe_u32 s2, s2, 0x40018
-; GFX9-NEXT:    s_bfe_u32 s9, s0, 0x40014
-; GFX9-NEXT:    v_mov_b32_e32 v8, s8
-; GFX9-NEXT:    s_bfe_u32 s10, s0, 0x40018
-; GFX9-NEXT:    v_mov_b32_e32 v9, s2
-; GFX9-NEXT:    s_lshr_b32 s0, s0, 28
+; GFX9-NEXT:    v_mov_b32_e32 v6, s14
+; GFX9-NEXT:    v_mul_u32_u24_e32 v3, s0, v3
+; GFX9-NEXT:    v_and_b32_e32 v3, 15, v3
+; GFX9-NEXT:    v_mov_b32_e32 v7, s13
+; GFX9-NEXT:    v_mov_b32_e32 v8, s12
+; GFX9-NEXT:    v_mov_b32_e32 v9, s11
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s4, v6, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s9, v4, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s8, v5, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s7, v6, v2
 ; GFX9-NEXT:    v_and_b32_e32 v2, 15, v2
-; GFX9-NEXT:    v_add_u32_e32 v2, v2, v5
-; GFX9-NEXT:    v_mad_u32_u24 v2, s7, v7, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s9, v8, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s10, v9, v2
-; GFX9-NEXT:    v_mov_b32_e32 v3, s11
-; GFX9-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX9-NEXT:    v_add_u32_e32 v2, v2, v3
+; GFX9-NEXT:    v_mad_u32_u24 v2, s6, v7, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s5, v8, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s4, v9, v2
+; GFX9-NEXT:    v_mov_b32_e32 v3, s10
+; GFX9-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX9-NEXT:    v_and_b32_e32 v2, 15, v2
 ; GFX9-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: udot8_acc4:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-DL-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_and_b32 s1, s2, 15
-; GFX9-DL-NEXT:    s_bfe_u32 s4, s2, 0x40004
+; GFX9-DL-NEXT:    s_and_b32 s9, s0, 15
+; GFX9-DL-NEXT:    s_and_b32 s16, s1, 15
+; GFX9-DL-NEXT:    s_bfe_u32 s15, s1, 0x40004
+; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s16
+; GFX9-DL-NEXT:    s_bfe_u32 s11, s1, 0x40018
+; GFX9-DL-NEXT:    s_bfe_u32 s12, s1, 0x40014
+; GFX9-DL-NEXT:    s_bfe_u32 s13, s1, 0x40010
+; GFX9-DL-NEXT:    s_bfe_u32 s14, s1, 0x40008
+; GFX9-DL-NEXT:    s_lshr_b32 s10, s1, 28
+; GFX9-DL-NEXT:    s_bfe_u32 s1, s1, 0x4000c
+; GFX9-DL-NEXT:    s_bfe_u32 s8, s0, 0x40004
+; GFX9-DL-NEXT:    v_mov_b32_e32 v5, s15
+; GFX9-DL-NEXT:    s_lshr_b32 s2, s0, 28
+; GFX9-DL-NEXT:    s_bfe_u32 s4, s0, 0x40018
+; GFX9-DL-NEXT:    s_bfe_u32 s5, s0, 0x40014
+; GFX9-DL-NEXT:    s_bfe_u32 s6, s0, 0x40010
+; GFX9-DL-NEXT:    s_bfe_u32 s7, s0, 0x40008
+; GFX9-DL-NEXT:    s_bfe_u32 s0, s0, 0x4000c
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-DL-NEXT:    s_bfe_u32 s6, s2, 0x40008
-; GFX9-DL-NEXT:    s_and_b32 s1, s0, 15
-; GFX9-DL-NEXT:    s_bfe_u32 s7, s2, 0x4000c
-; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s4
-; GFX9-DL-NEXT:    s_bfe_u32 s5, s0, 0x40004
-; GFX9-DL-NEXT:    v_mov_b32_e32 v5, s7
-; GFX9-DL-NEXT:    s_bfe_u32 s8, s0, 0x4000c
-; GFX9-DL-NEXT:    v_mov_b32_e32 v6, s6
-; GFX9-DL-NEXT:    s_bfe_u32 s4, s0, 0x40008
-; GFX9-DL-NEXT:    v_mul_u32_u24_e32 v5, s8, v5
-; GFX9-DL-NEXT:    s_bfe_u32 s6, s2, 0x40010
-; GFX9-DL-NEXT:    v_and_b32_e32 v5, 15, v5
-; GFX9-DL-NEXT:    s_bfe_u32 s8, s2, 0x40014
-; GFX9-DL-NEXT:    s_bfe_u32 s7, s0, 0x40010
-; GFX9-DL-NEXT:    v_mov_b32_e32 v7, s6
-; GFX9-DL-NEXT:    s_lshr_b32 s11, s2, 28
-; GFX9-DL-NEXT:    s_bfe_u32 s2, s2, 0x40018
-; GFX9-DL-NEXT:    s_bfe_u32 s9, s0, 0x40014
-; GFX9-DL-NEXT:    v_mov_b32_e32 v8, s8
-; GFX9-DL-NEXT:    s_bfe_u32 s10, s0, 0x40018
-; GFX9-DL-NEXT:    v_mov_b32_e32 v9, s2
-; GFX9-DL-NEXT:    s_lshr_b32 s0, s0, 28
+; GFX9-DL-NEXT:    v_mov_b32_e32 v6, s14
+; GFX9-DL-NEXT:    v_mul_u32_u24_e32 v3, s0, v3
+; GFX9-DL-NEXT:    v_and_b32_e32 v3, 15, v3
+; GFX9-DL-NEXT:    v_mov_b32_e32 v7, s13
+; GFX9-DL-NEXT:    v_mov_b32_e32 v8, s12
+; GFX9-DL-NEXT:    v_mov_b32_e32 v9, s11
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s4, v6, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s9, v4, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s8, v5, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s7, v6, v2
 ; GFX9-DL-NEXT:    v_and_b32_e32 v2, 15, v2
-; GFX9-DL-NEXT:    v_add_u32_e32 v2, v2, v5
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s7, v7, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s9, v8, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s10, v9, v2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s11
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX9-DL-NEXT:    v_add_u32_e32 v2, v2, v3
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s6, v7, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s5, v8, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s4, v9, v2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s10
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX9-DL-NEXT:    v_and_b32_e32 v2, 15, v2
 ; GFX9-DL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot8_acc4:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
+; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX10-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_and_b32 s2, s0, 15
 ; GFX10-DL-NEXT:    s_and_b32 s4, s1, 15
 ; GFX10-DL-NEXT:    s_bfe_u32 s5, s0, 0x40004
 ; GFX10-DL-NEXT:    s_bfe_u32 s6, s1, 0x40004
 ; GFX10-DL-NEXT:    s_bfe_u32 s7, s1, 0x40008
 ; GFX10-DL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX10-DL-NEXT:    v_mad_u32_u24 v2, s2, s4, v2
entry:
   ret void
 }
 
 ; TODO: Currently, permutation of udot8 is turned off due to a huge increase
 ; in the compile time.
 define amdgpu_kernel void @udot8_CommutationInsideMAD(<8 x i4> addrspace(1)* %src1,
 ; GFX7-LABEL: udot8_CommutationInsideMAD:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
 ; GFX7-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
 ; GFX7-NEXT:    buffer_load_ubyte v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
 ; GFX7-NEXT:    s_load_dword s1, s[10:11], 0x0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    s_lshr_b32 s2, s0, 28
-; GFX7-NEXT:    s_bfe_u32 s8, s0, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s15, s1, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s16, s1, 0x40014
 ; GFX7-NEXT:    s_bfe_u32 s17, s1, 0x40010
 ; GFX7-NEXT:    s_bfe_u32 s18, s1, 0x4000c
 ; GFX7-NEXT:    s_bfe_u32 s19, s1, 0x40008
 ; GFX7-NEXT:    s_bfe_u32 s20, s1, 0x40004
 ; GFX7-NEXT:    s_lshr_b32 s14, s1, 28
 ; GFX7-NEXT:    s_and_b32 s1, s1, 15
+; GFX7-NEXT:    s_bfe_u32 s8, s0, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s9, s0, 0x40014
 ; GFX7-NEXT:    s_bfe_u32 s10, s0, 0x40010
 ; GFX7-NEXT:    s_bfe_u32 s11, s0, 0x4000c
 ; GFX7-NEXT:    s_bfe_u32 s12, s0, 0x40008
 ; GFX7-NEXT:    s_bfe_u32 s13, s0, 0x40004
 ; GFX7-NEXT:    s_and_b32 s0, s0, 15
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s20

 ; GFX7-NEXT:    v_mov_b32_e32 v1, s14
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s2, v1, v0
 ; GFX7-NEXT:    v_and_b32_e32 v0, 15, v0
 ; GFX7-NEXT:    buffer_store_byte v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot8_CommutationInsideMAD:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    v_mov_b32_e32 v0, s4
+; GFX8-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX8-NEXT:    flat_load_ubyte v2, v[0:1]
+; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX8-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_and_b32 s1, s2, 15
-; GFX8-NEXT:    s_bfe_u32 s4, s2, 0x40004
+; GFX8-NEXT:    s_and_b32 s9, s0, 15
+; GFX8-NEXT:    s_and_b32 s16, s1, 15
+; GFX8-NEXT:    s_bfe_u32 s15, s1, 0x40004
+; GFX8-NEXT:    v_mov_b32_e32 v4, s16
+; GFX8-NEXT:    s_bfe_u32 s11, s1, 0x40018
+; GFX8-NEXT:    s_bfe_u32 s12, s1, 0x40014
+; GFX8-NEXT:    s_bfe_u32 s13, s1, 0x40010
+; GFX8-NEXT:    s_bfe_u32 s14, s1, 0x40008
+; GFX8-NEXT:    s_lshr_b32 s10, s1, 28
+; GFX8-NEXT:    s_bfe_u32 s1, s1, 0x4000c
+; GFX8-NEXT:    s_bfe_u32 s8, s0, 0x40004
+; GFX8-NEXT:    v_mov_b32_e32 v5, s15
+; GFX8-NEXT:    s_lshr_b32 s2, s0, 28
+; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x40018
+; GFX8-NEXT:    s_bfe_u32 s5, s0, 0x40014
+; GFX8-NEXT:    s_bfe_u32 s6, s0, 0x40010
+; GFX8-NEXT:    s_bfe_u32 s7, s0, 0x40008
+; GFX8-NEXT:    s_bfe_u32 s0, s0, 0x4000c
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s1
-; GFX8-NEXT:    s_bfe_u32 s6, s2, 0x40008
-; GFX8-NEXT:    s_and_b32 s1, s0, 15
-; GFX8-NEXT:    s_bfe_u32 s7, s2, 0x4000c
-; GFX8-NEXT:    v_mov_b32_e32 v4, s4
-; GFX8-NEXT:    s_bfe_u32 s5, s0, 0x40004
-; GFX8-NEXT:    v_mov_b32_e32 v5, s7
-; GFX8-NEXT:    s_bfe_u32 s8, s0, 0x4000c
-; GFX8-NEXT:    v_mov_b32_e32 v6, s6
-; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x40008
-; GFX8-NEXT:    v_mul_u32_u24_e32 v5, s8, v5
-; GFX8-NEXT:    s_bfe_u32 s6, s2, 0x40010
-; GFX8-NEXT:    v_and_b32_e32 v5, 15, v5
-; GFX8-NEXT:    s_bfe_u32 s8, s2, 0x40014
-; GFX8-NEXT:    s_bfe_u32 s7, s0, 0x40010
-; GFX8-NEXT:    v_mov_b32_e32 v7, s6
-; GFX8-NEXT:    s_lshr_b32 s11, s2, 28
-; GFX8-NEXT:    s_bfe_u32 s2, s2, 0x40018
-; GFX8-NEXT:    s_bfe_u32 s9, s0, 0x40014
-; GFX8-NEXT:    v_mov_b32_e32 v8, s8
-; GFX8-NEXT:    s_bfe_u32 s10, s0, 0x40018
-; GFX8-NEXT:    v_mov_b32_e32 v9, s2
-; GFX8-NEXT:    s_lshr_b32 s0, s0, 28
+; GFX8-NEXT:    v_mov_b32_e32 v6, s14
+; GFX8-NEXT:    v_mul_u32_u24_e32 v3, s0, v3
+; GFX8-NEXT:    v_and_b32_e32 v3, 15, v3
+; GFX8-NEXT:    v_mov_b32_e32 v7, s13
+; GFX8-NEXT:    v_mov_b32_e32 v8, s12
+; GFX8-NEXT:    v_mov_b32_e32 v9, s11
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
-; GFX8-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s4, v6, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s9, v4, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s8, v5, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s7, v6, v2
 ; GFX8-NEXT:    v_and_b32_e32 v2, 15, v2
-; GFX8-NEXT:    v_add_u32_e32 v2, vcc, v2, v5
-; GFX8-NEXT:    v_mad_u32_u24 v2, s7, v7, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s9, v8, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s10, v9, v2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s11
-; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX8-NEXT:    v_add_u32_e32 v2, vcc, v2, v3
+; GFX8-NEXT:    v_mad_u32_u24 v2, s6, v7, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v8, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s4, v9, v2
+; GFX8-NEXT:    v_mov_b32_e32 v3, s10
+; GFX8-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX8-NEXT:    v_and_b32_e32 v2, 15, v2
 ; GFX8-NEXT:    flat_store_byte v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: udot8_CommutationInsideMAD:
 ; GFX9:       ; %bb.0: ; %entry
-; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_and_b32 s1, s2, 15
-; GFX9-NEXT:    s_bfe_u32 s4, s2, 0x40004
+; GFX9-NEXT:    s_and_b32 s9, s0, 15
+; GFX9-NEXT:    s_and_b32 s16, s1, 15
+; GFX9-NEXT:    s_bfe_u32 s15, s1, 0x40004
+; GFX9-NEXT:    v_mov_b32_e32 v4, s16
+; GFX9-NEXT:    s_bfe_u32 s11, s1, 0x40018
+; GFX9-NEXT:    s_bfe_u32 s12, s1, 0x40014
+; GFX9-NEXT:    s_bfe_u32 s13, s1, 0x40010
+; GFX9-NEXT:    s_bfe_u32 s14, s1, 0x40008
+; GFX9-NEXT:    s_lshr_b32 s10, s1, 28
+; GFX9-NEXT:    s_bfe_u32 s1, s1, 0x4000c
+; GFX9-NEXT:    s_bfe_u32 s8, s0, 0x40004
+; GFX9-NEXT:    v_mov_b32_e32 v5, s15
+; GFX9-NEXT:    s_lshr_b32 s2, s0, 28
+; GFX9-NEXT:    s_bfe_u32 s4, s0, 0x40018
+; GFX9-NEXT:    s_bfe_u32 s5, s0, 0x40014
+; GFX9-NEXT:    s_bfe_u32 s6, s0, 0x40010
+; GFX9-NEXT:    s_bfe_u32 s7, s0, 0x40008
+; GFX9-NEXT:    s_bfe_u32 s0, s0, 0x4000c
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-NEXT:    s_bfe_u32 s6, s2, 0x40008
-; GFX9-NEXT:    s_and_b32 s1, s0, 15
-; GFX9-NEXT:    s_bfe_u32 s7, s2, 0x4000c
-; GFX9-NEXT:    v_mov_b32_e32 v4, s4
-; GFX9-NEXT:    s_bfe_u32 s5, s0, 0x40004
-; GFX9-NEXT:    v_mov_b32_e32 v5, s7
-; GFX9-NEXT:    s_bfe_u32 s8, s0, 0x4000c
-; GFX9-NEXT:    v_mov_b32_e32 v6, s6
-; GFX9-NEXT:    s_bfe_u32 s4, s0, 0x40008
-; GFX9-NEXT:    v_mul_u32_u24_e32 v5, s8, v5
-; GFX9-NEXT:    s_bfe_u32 s6, s2, 0x40010
-; GFX9-NEXT:    v_and_b32_e32 v5, 15, v5
-; GFX9-NEXT:    s_bfe_u32 s8, s2, 0x40014
-; GFX9-NEXT:    s_bfe_u32 s7, s0, 0x40010
-; GFX9-NEXT:    v_mov_b32_e32 v7, s6
-; GFX9-NEXT:    s_lshr_b32 s11, s2, 28
-; GFX9-NEXT:    s_bfe_u32 s2, s2, 0x40018
-; GFX9-NEXT:    s_bfe_u32 s9, s0, 0x40014
-; GFX9-NEXT:    v_mov_b32_e32 v8, s8
-; GFX9-NEXT:    s_bfe_u32 s10, s0, 0x40018
-; GFX9-NEXT:    v_mov_b32_e32 v9, s2
-; GFX9-NEXT:    s_lshr_b32 s0, s0, 28
+; GFX9-NEXT:    v_mov_b32_e32 v6, s14
+; GFX9-NEXT:    v_mul_u32_u24_e32 v3, s0, v3
+; GFX9-NEXT:    v_and_b32_e32 v3, 15, v3
+; GFX9-NEXT:    v_mov_b32_e32 v7, s13
+; GFX9-NEXT:    v_mov_b32_e32 v8, s12
+; GFX9-NEXT:    v_mov_b32_e32 v9, s11
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s4, v6, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s9, v4, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s8, v5, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s7, v6, v2
 ; GFX9-NEXT:    v_and_b32_e32 v2, 15, v2
-; GFX9-NEXT:    v_add_u32_e32 v2, v5, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s7, v7, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s9, v8, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s10, v9, v2
-; GFX9-NEXT:    v_mov_b32_e32 v3, s11
-; GFX9-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX9-NEXT:    v_add_u32_e32 v2, v3, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s6, v7, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s5, v8, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s4, v9, v2
+; GFX9-NEXT:    v_mov_b32_e32 v3, s10
+; GFX9-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX9-NEXT:    v_and_b32_e32 v2, 15, v2
 ; GFX9-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: udot8_CommutationInsideMAD:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-DL-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_and_b32 s1, s2, 15
-; GFX9-DL-NEXT:    s_bfe_u32 s4, s2, 0x40004
+; GFX9-DL-NEXT:    s_and_b32 s9, s0, 15
+; GFX9-DL-NEXT:    s_and_b32 s16, s1, 15
+; GFX9-DL-NEXT:    s_bfe_u32 s15, s1, 0x40004
+; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s16
+; GFX9-DL-NEXT:    s_bfe_u32 s11, s1, 0x40018
+; GFX9-DL-NEXT:    s_bfe_u32 s12, s1, 0x40014
+; GFX9-DL-NEXT:    s_bfe_u32 s13, s1, 0x40010
+; GFX9-DL-NEXT:    s_bfe_u32 s14, s1, 0x40008
+; GFX9-DL-NEXT:    s_lshr_b32 s10, s1, 28
+; GFX9-DL-NEXT:    s_bfe_u32 s1, s1, 0x4000c
+; GFX9-DL-NEXT:    s_bfe_u32 s8, s0, 0x40004
+; GFX9-DL-NEXT:    v_mov_b32_e32 v5, s15
+; GFX9-DL-NEXT:    s_lshr_b32 s2, s0, 28
+; GFX9-DL-NEXT:    s_bfe_u32 s4, s0, 0x40018
+; GFX9-DL-NEXT:    s_bfe_u32 s5, s0, 0x40014
+; GFX9-DL-NEXT:    s_bfe_u32 s6, s0, 0x40010
+; GFX9-DL-NEXT:    s_bfe_u32 s7, s0, 0x40008
+; GFX9-DL-NEXT:    s_bfe_u32 s0, s0, 0x4000c
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-DL-NEXT:    s_bfe_u32 s6, s2, 0x40008
-; GFX9-DL-NEXT:    s_and_b32 s1, s0, 15
-; GFX9-DL-NEXT:    s_bfe_u32 s7, s2, 0x4000c
-; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s4
-; GFX9-DL-NEXT:    s_bfe_u32 s5, s0, 0x40004
-; GFX9-DL-NEXT:    v_mov_b32_e32 v5, s7
-; GFX9-DL-NEXT:    s_bfe_u32 s8, s0, 0x4000c
-; GFX9-DL-NEXT:    v_mov_b32_e32 v6, s6
-; GFX9-DL-NEXT:    s_bfe_u32 s4, s0, 0x40008
-; GFX9-DL-NEXT:    v_mul_u32_u24_e32 v5, s8, v5
-; GFX9-DL-NEXT:    s_bfe_u32 s6, s2, 0x40010
-; GFX9-DL-NEXT:    v_and_b32_e32 v5, 15, v5
-; GFX9-DL-NEXT:    s_bfe_u32 s8, s2, 0x40014
-; GFX9-DL-NEXT:    s_bfe_u32 s7, s0, 0x40010
-; GFX9-DL-NEXT:    v_mov_b32_e32 v7, s6
-; GFX9-DL-NEXT:    s_lshr_b32 s11, s2, 28
-; GFX9-DL-NEXT:    s_bfe_u32 s2, s2, 0x40018
-; GFX9-DL-NEXT:    s_bfe_u32 s9, s0, 0x40014
-; GFX9-DL-NEXT:    v_mov_b32_e32 v8, s8
-; GFX9-DL-NEXT:    s_bfe_u32 s10, s0, 0x40018
-; GFX9-DL-NEXT:    v_mov_b32_e32 v9, s2
-; GFX9-DL-NEXT:    s_lshr_b32 s0, s0, 28
+; GFX9-DL-NEXT:    v_mov_b32_e32 v6, s14
+; GFX9-DL-NEXT:    v_mul_u32_u24_e32 v3, s0, v3
+; GFX9-DL-NEXT:    v_and_b32_e32 v3, 15, v3
+; GFX9-DL-NEXT:    v_mov_b32_e32 v7, s13
+; GFX9-DL-NEXT:    v_mov_b32_e32 v8, s12
+; GFX9-DL-NEXT:    v_mov_b32_e32 v9, s11
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s4, v6, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s9, v4, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s8, v5, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s7, v6, v2
 ; GFX9-DL-NEXT:    v_and_b32_e32 v2, 15, v2
-; GFX9-DL-NEXT:    v_add_u32_e32 v2, v5, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s7, v7, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s9, v8, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s10, v9, v2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s11
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX9-DL-NEXT:    v_add_u32_e32 v2, v3, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s6, v7, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s5, v8, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s4, v9, v2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s10
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX9-DL-NEXT:    v_and_b32_e32 v2, 15, v2
 ; GFX9-DL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot8_CommutationInsideMAD:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
+; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX10-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_and_b32 s2, s0, 15
 ; GFX10-DL-NEXT:    s_and_b32 s4, s1, 15
 ; GFX10-DL-NEXT:    s_bfe_u32 s5, s0, 0x40004
 ; GFX10-DL-NEXT:    s_bfe_u32 s6, s1, 0x40004
 ; GFX10-DL-NEXT:    s_bfe_u32 s7, s1, 0x40008
 ; GFX10-DL-NEXT:    s_bfe_u32 s8, s1, 0x4000c
 ; GFX10-DL-NEXT:    s_waitcnt vmcnt(0)

 ; GFX9-DL-LABEL: udot8_acc32_vecMul:
 ; GFX9-DL:       ; %bb.0: ; %entry
 ; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_load_dword s6, s[0:1], 0x0
 ; GFX9-DL-NEXT:    s_load_dword s4, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s6
+; GFX9-DL-NEXT:    v_dot8_u32_u4 v2, s4, v0, v1
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
-; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mov_b32_e32 v2, s2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s6
-; GFX9-DL-NEXT:    v_dot8_u32_u4 v2, s4, v2, v3
 ; GFX9-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot8_acc32_vecMul:
 ; GFX10-DL:       ; %bb.0: ; %entry
+; GFX10-DL-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0x34
 ; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    s_load_dword s2, s[0:1], 0x0
-; GFX10-DL-NEXT:    s_load_dword s4, s[4:5], 0x0
-; GFX10-DL-NEXT:    s_load_dword s5, s[6:7], 0x0
+; GFX10-DL-NEXT:    s_load_dword s0, s[8:9], 0x0
+; GFX10-DL-NEXT:    s_load_dword s1, s[4:5], 0x0
+; GFX10-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
-; GFX10-DL-NEXT:    v_dot8_u32_u4 v2, s4, s5, v0
 ; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_dot8_u32_u4 v2, s1, s2, v0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s8
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s9
 ; GFX10-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX10-DL-NEXT:    s_endpgm
                                               <8 x i4> addrspace(1)* %src2,
                                               i32 addrspace(1)* nocapture %dst) {
 entry:
   %vec1 = load <8 x i4>, <8 x i4> addrspace(1)* %src1
   %vec2 = load <8 x i4>, <8 x i4> addrspace(1)* %src2
 
entry:
   ret void
 }
 
 ; TODO: Clean up the code(by default pk_mad_I16 should be generated), then
 ; support the pattern.
 define amdgpu_kernel void @udot8_acc16_vecMul(<8 x i4> addrspace(1)* %src1,
 ; GFX7-LABEL: udot8_acc16_vecMul:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
 ; GFX7-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
 ; GFX7-NEXT:    buffer_load_ushort v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
 ; GFX7-NEXT:    s_load_dword s1, s[10:11], 0x0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    s_bfe_u32 s11, s0, 0x40004
-; GFX7-NEXT:    s_bfe_u32 s13, s0, 0x4000c
 ; GFX7-NEXT:    s_bfe_u32 s18, s1, 0x40004
 ; GFX7-NEXT:    s_bfe_u32 s20, s1, 0x4000c
-; GFX7-NEXT:    v_mov_b32_e32 v2, s20
 ; GFX7-NEXT:    v_mov_b32_e32 v4, s18
 ; GFX7-NEXT:    s_bfe_u32 s15, s1, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s16, s1, 0x40014
 ; GFX7-NEXT:    s_bfe_u32 s17, s1, 0x40010
 ; GFX7-NEXT:    s_and_b32 s19, s1, 15
 ; GFX7-NEXT:    s_lshr_b32 s14, s1, 28
 ; GFX7-NEXT:    s_bfe_u32 s1, s1, 0x40008
+; GFX7-NEXT:    s_bfe_u32 s13, s0, 0x4000c
+; GFX7-NEXT:    v_mov_b32_e32 v2, s20
 ; GFX7-NEXT:    v_mul_u32_u24_e32 v2, s13, v2
 ; GFX7-NEXT:    v_mul_u32_u24_e32 v4, s11, v4
 ; GFX7-NEXT:    s_lshr_b32 s2, s0, 28
 ; GFX7-NEXT:    s_bfe_u32 s8, s0, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s9, s0, 0x40014
 ; GFX7-NEXT:    s_bfe_u32 s10, s0, 0x40010
 ; GFX7-NEXT:    s_and_b32 s12, s0, 15
 ; GFX7-NEXT:    v_mov_b32_e32 v3, s19

 ; GFX7-NEXT:    v_mad_u32_u24 v0, s8, v7, v0
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s14
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s2, v1, v0
 ; GFX7-NEXT:    buffer_store_short v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot8_acc16_vecMul:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    v_mov_b32_e32 v0, s4
+; GFX8-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX8-NEXT:    flat_load_ushort v2, v[0:1]
+; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX8-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_and_b32 s1, s2, 15
-; GFX8-NEXT:    s_bfe_u32 s4, s2, 0x40004
+; GFX8-NEXT:    s_lshr_b32 s2, s0, 28
+; GFX8-NEXT:    s_bfe_u32 s11, s1, 0x40018
+; GFX8-NEXT:    s_bfe_u32 s12, s1, 0x40014
+; GFX8-NEXT:    s_bfe_u32 s13, s1, 0x40010
+; GFX8-NEXT:    s_bfe_u32 s14, s1, 0x4000c
+; GFX8-NEXT:    s_bfe_u32 s15, s1, 0x40008
+; GFX8-NEXT:    s_bfe_u32 s16, s1, 0x40004
+; GFX8-NEXT:    s_lshr_b32 s10, s1, 28
+; GFX8-NEXT:    s_and_b32 s1, s1, 15
+; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x40018
+; GFX8-NEXT:    s_bfe_u32 s5, s0, 0x40014
+; GFX8-NEXT:    s_bfe_u32 s6, s0, 0x40010
+; GFX8-NEXT:    s_bfe_u32 s7, s0, 0x4000c
+; GFX8-NEXT:    s_bfe_u32 s8, s0, 0x40008
+; GFX8-NEXT:    s_bfe_u32 s9, s0, 0x40004
+; GFX8-NEXT:    s_and_b32 s0, s0, 15
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s1
-; GFX8-NEXT:    v_mov_b32_e32 v4, s4
-; GFX8-NEXT:    s_and_b32 s1, s0, 15
-; GFX8-NEXT:    s_bfe_u32 s5, s0, 0x40004
-; GFX8-NEXT:    s_bfe_u32 s4, s2, 0x40008
-; GFX8-NEXT:    s_bfe_u32 s7, s2, 0x4000c
-; GFX8-NEXT:    v_mov_b32_e32 v5, s4
-; GFX8-NEXT:    s_bfe_u32 s6, s0, 0x40008
-; GFX8-NEXT:    s_bfe_u32 s8, s2, 0x40010
-; GFX8-NEXT:    v_mov_b32_e32 v6, s7
-; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x4000c
-; GFX8-NEXT:    s_bfe_u32 s9, s2, 0x40014
-; GFX8-NEXT:    s_bfe_u32 s7, s0, 0x40010
-; GFX8-NEXT:    v_mov_b32_e32 v7, s8
-; GFX8-NEXT:    s_bfe_u32 s11, s2, 0x40018
-; GFX8-NEXT:    s_bfe_u32 s10, s0, 0x40014
-; GFX8-NEXT:    v_mov_b32_e32 v8, s9
-; GFX8-NEXT:    s_bfe_u32 s12, s0, 0x40018
-; GFX8-NEXT:    s_lshr_b32 s2, s2, 28
+; GFX8-NEXT:    v_mov_b32_e32 v4, s16
+; GFX8-NEXT:    v_mov_b32_e32 v5, s15
+; GFX8-NEXT:    v_mov_b32_e32 v6, s14
+; GFX8-NEXT:    v_mov_b32_e32 v7, s13
+; GFX8-NEXT:    v_mov_b32_e32 v8, s12
 ; GFX8-NEXT:    v_mov_b32_e32 v9, s11
-; GFX8-NEXT:    s_lshr_b32 s0, s0, 28
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
-; GFX8-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
-; GFX8-NEXT:    v_and_b32_e32 v2, 0xffff, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s6, v5, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s4, v6, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s7, v7, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s10, v8, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s12, v9, v2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s9, v4, v2
+; GFX8-NEXT:    v_and_b32_e32 v2, 0xffff, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s8, v5, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s7, v6, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s6, v7, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v8, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s4, v9, v2
+; GFX8-NEXT:    v_mov_b32_e32 v3, s10
+; GFX8-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX8-NEXT:    flat_store_short v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: udot8_acc16_vecMul:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NEXT:    v_mov_b32_e32 v1, s1
-; GFX9-NEXT:    global_load_ushort v2, v[0:1], off
-; GFX9-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-NEXT:    s_load_dword s2, s[4:5], 0x0
+; GFX9-NEXT:    s_load_dword s6, s[6:7], 0x0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_and_b32 s1, s2, 15
-; GFX9-NEXT:    s_bfe_u32 s4, s2, 0x40004
-; GFX9-NEXT:    s_pack_ll_b32_b16 s1, s1, s4
-; GFX9-NEXT:    s_bfe_u32 s11, s2, 0x40018
-; GFX9-NEXT:    s_lshr_b32 s13, s2, 28
-; GFX9-NEXT:    s_and_b32 s4, s0, 15
-; GFX9-NEXT:    s_bfe_u32 s5, s0, 0x40004
+; GFX9-NEXT:    s_bfe_u32 s4, s2, 0x40018
+; GFX9-NEXT:    s_bfe_u32 s7, s6, 0x40018
+; GFX9-NEXT:    s_lshr_b32 s13, s6, 28
+; GFX9-NEXT:    s_lshr_b32 s5, s2, 28
+; GFX9-NEXT:    s_pack_ll_b32_b16 s7, s7, s13
+; GFX9-NEXT:    s_bfe_u32 s14, s6, 0x40010
+; GFX9-NEXT:    s_bfe_u32 s15, s6, 0x40014
 ; GFX9-NEXT:    s_pack_ll_b32_b16 s4, s4, s5
-; GFX9-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-NEXT:    s_pack_ll_b32_b16 s11, s11, s13
-; GFX9-NEXT:    s_bfe_u32 s1, s0, 0x40008
-; GFX9-NEXT:    s_bfe_u32 s5, s0, 0x4000c
-; GFX9-NEXT:    s_bfe_u32 s9, s0, 0x40010
-; GFX9-NEXT:    s_bfe_u32 s10, s0, 0x40014
-; GFX9-NEXT:    s_bfe_u32 s12, s0, 0x40018
-; GFX9-NEXT:    s_lshr_b32 s0, s0, 28
-; GFX9-NEXT:    s_bfe_u32 s6, s2, 0x40008
-; GFX9-NEXT:    s_bfe_u32 s7, s2, 0x4000c
+; GFX9-NEXT:    v_mov_b32_e32 v0, s7
+; GFX9-NEXT:    v_pk_mul_lo_u16 v2, s4, v0
+; GFX9-NEXT:    s_pack_ll_b32_b16 s4, s14, s15
 ; GFX9-NEXT:    s_bfe_u32 s8, s2, 0x40010
-; GFX9-NEXT:    s_bfe_u32 s2, s2, 0x40014
-; GFX9-NEXT:    s_pack_ll_b32_b16 s0, s12, s0
-; GFX9-NEXT:    v_mov_b32_e32 v4, s11
-; GFX9-NEXT:    v_pk_mul_lo_u16 v4, s0, v4
-; GFX9-NEXT:    s_pack_ll_b32_b16 s0, s8, s2
-; GFX9-NEXT:    v_mov_b32_e32 v5, s0
-; GFX9-NEXT:    s_pack_ll_b32_b16 s0, s6, s7
-; GFX9-NEXT:    v_pk_mul_lo_u16 v3, s4, v3
-; GFX9-NEXT:    s_pack_ll_b32_b16 s1, s1, s5
-; GFX9-NEXT:    v_mov_b32_e32 v6, s0
-; GFX9-NEXT:    v_pk_mul_lo_u16 v6, s1, v6
-; GFX9-NEXT:    s_pack_ll_b32_b16 s2, s9, s10
-; GFX9-NEXT:    v_pk_mul_lo_u16 v5, s2, v5
+; GFX9-NEXT:    s_bfe_u32 s9, s2, 0x40014
+; GFX9-NEXT:    s_bfe_u32 s16, s6, 0x40008
+; GFX9-NEXT:    s_bfe_u32 s17, s6, 0x4000c
+; GFX9-NEXT:    s_and_b32 s18, s6, 15
+; GFX9-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NEXT:    s_pack_ll_b32_b16 s5, s8, s9
+; GFX9-NEXT:    s_bfe_u32 s10, s2, 0x40008
+; GFX9-NEXT:    s_bfe_u32 s11, s2, 0x4000c
+; GFX9-NEXT:    s_bfe_u32 s6, s6, 0x40004
+; GFX9-NEXT:    s_pack_ll_b32_b16 s4, s16, s17
+; GFX9-NEXT:    v_pk_mul_lo_u16 v3, s5, v0
+; GFX9-NEXT:    s_and_b32 s12, s2, 15
+; GFX9-NEXT:    s_bfe_u32 s2, s2, 0x40004
+; GFX9-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NEXT:    s_pack_ll_b32_b16 s5, s10, s11
+; GFX9-NEXT:    s_pack_ll_b32_b16 s4, s18, s6
+; GFX9-NEXT:    v_pk_mul_lo_u16 v4, s5, v0
+; GFX9-NEXT:    s_pack_ll_b32_b16 s2, s12, s2
+; GFX9-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NEXT:    v_pk_mul_lo_u16 v5, s2, v0
+; GFX9-NEXT:    v_mov_b32_e32 v0, s0
+; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    global_load_ushort v6, v[0:1], off
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_add_u32_e32 v2, v3, v2
-; GFX9-NEXT:    v_add_u32_sdwa v2, v2, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
-; GFX9-NEXT:    v_add_u32_sdwa v2, v2, v6 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:BYTE_0
-; GFX9-NEXT:    v_add_u32_sdwa v2, v2, v6 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
-; GFX9-NEXT:    v_add_u32_e32 v2, v2, v5
-; GFX9-NEXT:    v_add_u32_sdwa v2, v2, v5 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
-; GFX9-NEXT:    v_add_u32_e32 v2, v2, v4
-; GFX9-NEXT:    v_add_u32_sdwa v2, v2, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; GFX9-NEXT:    v_add_u32_e32 v6, v5, v6
+; GFX9-NEXT:    v_add_u32_sdwa v5, v6, v5 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; GFX9-NEXT:    v_add_u32_sdwa v5, v5, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:BYTE_0
+; GFX9-NEXT:    v_add_u32_sdwa v4, v5, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; GFX9-NEXT:    v_add_u32_e32 v4, v4, v3
+; GFX9-NEXT:    v_add_u32_sdwa v3, v4, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; GFX9-NEXT:    v_add_u32_e32 v3, v3, v2
+; GFX9-NEXT:    v_add_u32_sdwa v2, v3, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
 ; GFX9-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: udot8_acc16_vecMul:
 ; GFX9-DL:       ; %bb.0: ; %entry
 ; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
-; GFX9-DL-NEXT:    global_load_ushort v2, v[0:1], off
-; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_load_dword s2, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_load_dword s6, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_and_b32 s1, s2, 15
-; GFX9-DL-NEXT:    s_bfe_u32 s4, s2, 0x40004
-; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s1, s1, s4
-; GFX9-DL-NEXT:    s_bfe_u32 s11, s2, 0x40018
-; GFX9-DL-NEXT:    s_lshr_b32 s13, s2, 28
-; GFX9-DL-NEXT:    s_and_b32 s4, s0, 15
-; GFX9-DL-NEXT:    s_bfe_u32 s5, s0, 0x40004
+; GFX9-DL-NEXT:    s_bfe_u32 s4, s2, 0x40018
+; GFX9-DL-NEXT:    s_bfe_u32 s7, s6, 0x40018
+; GFX9-DL-NEXT:    s_lshr_b32 s13, s6, 28
+; GFX9-DL-NEXT:    s_lshr_b32 s5, s2, 28
+; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s7, s7, s13
+; GFX9-DL-NEXT:    s_bfe_u32 s14, s6, 0x40010
+; GFX9-DL-NEXT:    s_bfe_u32 s15, s6, 0x40014
 ; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s4, s4, s5
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s11, s11, s13
-; GFX9-DL-NEXT:    s_bfe_u32 s1, s0, 0x40008
-; GFX9-DL-NEXT:    s_bfe_u32 s5, s0, 0x4000c
-; GFX9-DL-NEXT:    s_bfe_u32 s9, s0, 0x40010
-; GFX9-DL-NEXT:    s_bfe_u32 s10, s0, 0x40014
-; GFX9-DL-NEXT:    s_bfe_u32 s12, s0, 0x40018
-; GFX9-DL-NEXT:    s_lshr_b32 s0, s0, 28
-; GFX9-DL-NEXT:    s_bfe_u32 s6, s2, 0x40008
-; GFX9-DL-NEXT:    s_bfe_u32 s7, s2, 0x4000c
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s7
+; GFX9-DL-NEXT:    v_pk_mul_lo_u16 v2, s4, v0
+; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s4, s14, s15
 ; GFX9-DL-NEXT:    s_bfe_u32 s8, s2, 0x40010
-; GFX9-DL-NEXT:    s_bfe_u32 s2, s2, 0x40014
-; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s0, s12, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s11
-; GFX9-DL-NEXT:    v_pk_mul_lo_u16 v4, s0, v4
-; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s0, s8, s2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v5, s0
-; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s0, s6, s7
-; GFX9-DL-NEXT:    v_pk_mul_lo_u16 v3, s4, v3
-; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s1, s1, s5
-; GFX9-DL-NEXT:    v_mov_b32_e32 v6, s0
-; GFX9-DL-NEXT:    v_pk_mul_lo_u16 v6, s1, v6
-; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s2, s9, s10
-; GFX9-DL-NEXT:    v_pk_mul_lo_u16 v5, s2, v5
+; GFX9-DL-NEXT:    s_bfe_u32 s9, s2, 0x40014
+; GFX9-DL-NEXT:    s_bfe_u32 s16, s6, 0x40008
+; GFX9-DL-NEXT:    s_bfe_u32 s17, s6, 0x4000c
+; GFX9-DL-NEXT:    s_and_b32 s18, s6, 15
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s5, s8, s9
+; GFX9-DL-NEXT:    s_bfe_u32 s10, s2, 0x40008
+; GFX9-DL-NEXT:    s_bfe_u32 s11, s2, 0x4000c
+; GFX9-DL-NEXT:    s_bfe_u32 s6, s6, 0x40004
+; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s4, s16, s17
+; GFX9-DL-NEXT:    v_pk_mul_lo_u16 v3, s5, v0
+; GFX9-DL-NEXT:    s_and_b32 s12, s2, 15
+; GFX9-DL-NEXT:    s_bfe_u32 s2, s2, 0x40004
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s5, s10, s11
+; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s4, s18, s6
+; GFX9-DL-NEXT:    v_pk_mul_lo_u16 v4, s5, v0
+; GFX9-DL-NEXT:    s_pack_ll_b32_b16 s2, s12, s2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    v_pk_mul_lo_u16 v5, s2, v0
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    global_load_ushort v6, v[0:1], off
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-DL-NEXT:    v_add_u32_e32 v2, v3, v2
-; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
-; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v6 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:BYTE_0
-; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v6 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
-; GFX9-DL-NEXT:    v_add_u32_e32 v2, v2, v5
-; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v5 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
-; GFX9-DL-NEXT:    v_add_u32_e32 v2, v2, v4
-; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; GFX9-DL-NEXT:    v_add_u32_e32 v6, v5, v6
+; GFX9-DL-NEXT:    v_add_u32_sdwa v5, v6, v5 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; GFX9-DL-NEXT:    v_add_u32_sdwa v5, v5, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:BYTE_0
+; GFX9-DL-NEXT:    v_add_u32_sdwa v4, v5, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; GFX9-DL-NEXT:    v_add_u32_e32 v4, v4, v3
+; GFX9-DL-NEXT:    v_add_u32_sdwa v3, v4, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; GFX9-DL-NEXT:    v_add_u32_e32 v3, v3, v2
+; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v3, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
 ; GFX9-DL-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot8_acc16_vecMul:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
+; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX10-DL-NEXT:    global_load_ushort v2, v[0:1], off
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX10-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX10-DL-NEXT:    global_load_ushort v2, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_and_b32 s2, s0, 15
 ; GFX10-DL-NEXT:    s_bfe_u32 s6, s0, 0x40004
 ; GFX10-DL-NEXT:    s_and_b32 s4, s1, 15
 ; GFX10-DL-NEXT:    s_bfe_u32 s5, s1, 0x40004
 ; GFX10-DL-NEXT:    s_bfe_u32 s7, s1, 0x4000c
 ; GFX10-DL-NEXT:    s_bfe_u32 s8, s0, 0x4000c
 ; GFX10-DL-NEXT:    s_pack_ll_b32_b16 s2, s2, s6
entry:
   store i16 %add8, i16 addrspace(1)* %dst, align 4
   ret void
 }
 
 ; TODO: Cleanup the code to generate MAD; pattern should be recognized then.
 define amdgpu_kernel void @udot8_acc8_vecMul(<8 x i4> addrspace(1)* %src1,
 ; GFX7-LABEL: udot8_acc8_vecMul:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
 ; GFX7-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
 ; GFX7-NEXT:    buffer_load_ubyte v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
 ; GFX7-NEXT:    s_load_dword s1, s[10:11], 0x0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    s_bfe_u32 s2, s0, 0x4000c
-; GFX7-NEXT:    s_bfe_u32 s9, s0, 0x40004
 ; GFX7-NEXT:    s_bfe_u32 s14, s1, 0x4000c
 ; GFX7-NEXT:    s_bfe_u32 s16, s1, 0x40004
 ; GFX7-NEXT:    s_lshr_b32 s18, s1, 28
-; GFX7-NEXT:    v_mov_b32_e32 v6, s16
 ; GFX7-NEXT:    v_mov_b32_e32 v8, s14
 ; GFX7-NEXT:    s_bfe_u32 s15, s1, 0x40008
 ; GFX7-NEXT:    s_and_b32 s17, s1, 15
 ; GFX7-NEXT:    s_bfe_u32 s19, s1, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s20, s1, 0x40014
+; GFX7-NEXT:    s_bfe_u32 s9, s0, 0x40004
+; GFX7-NEXT:    v_mov_b32_e32 v6, s16
 ; GFX7-NEXT:    s_lshr_b32 s11, s0, 28
 ; GFX7-NEXT:    v_mov_b32_e32 v4, s18
 ; GFX7-NEXT:    v_mul_u32_u24_e32 v4, s11, v4
 ; GFX7-NEXT:    v_mul_u32_u24_e32 v6, s9, v6
 ; GFX7-NEXT:    v_mul_u32_u24_e32 v8, s2, v8
 ; GFX7-NEXT:    s_bfe_u32 s1, s1, 0x40010
 ; GFX7-NEXT:    s_bfe_u32 s8, s0, 0x40008
 ; GFX7-NEXT:    v_mov_b32_e32 v7, s15

 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v0, v7
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v0, v8
 ; GFX7-NEXT:    v_add_i32_e32 v0, vcc, v0, v2
 ; GFX7-NEXT:    buffer_store_byte v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot8_acc8_vecMul:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX8-NEXT:    s_mov_b32 s2, 0xffff
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    v_mov_b32_e32 v0, s4
+; GFX8-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX8-NEXT:    flat_load_ubyte v2, v[0:1]
-; GFX8-NEXT:    s_load_dword s1, s[4:5], 0x0
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    s_mov_b32 s0, 0xffff
-; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_bfe_u32 s8, s1, 0x40004
-; GFX8-NEXT:    s_bfe_u32 s10, s1, 0x4000c
-; GFX8-NEXT:    s_bfe_u32 s15, s2, 0x40004
-; GFX8-NEXT:    s_and_b32 s16, s2, 15
-; GFX8-NEXT:    s_bfe_u32 s17, s2, 0x4000c
-; GFX8-NEXT:    s_bfe_u32 s4, s1, 0x40014
-; GFX8-NEXT:    s_lshr_b32 s6, s1, 28
-; GFX8-NEXT:    s_bfe_u32 s11, s2, 0x40014
-; GFX8-NEXT:    s_bfe_u32 s12, s2, 0x40010
-; GFX8-NEXT:    s_lshr_b32 s13, s2, 28
-; GFX8-NEXT:    s_bfe_u32 s14, s2, 0x40018
-; GFX8-NEXT:    s_bfe_u32 s2, s2, 0x40008
-; GFX8-NEXT:    s_and_b32 s9, s1, 15
+; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX8-NEXT:    s_load_dword s1, s[6:7], 0x0
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX8-NEXT:    s_bfe_u32 s8, s0, 0x40004
+; GFX8-NEXT:    s_bfe_u32 s10, s0, 0x4000c
+; GFX8-NEXT:    s_bfe_u32 s15, s1, 0x40004
+; GFX8-NEXT:    s_and_b32 s16, s1, 15
+; GFX8-NEXT:    s_bfe_u32 s17, s1, 0x4000c
+; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x40014
+; GFX8-NEXT:    s_lshr_b32 s6, s0, 28
+; GFX8-NEXT:    s_bfe_u32 s11, s1, 0x40014
+; GFX8-NEXT:    s_bfe_u32 s12, s1, 0x40010
+; GFX8-NEXT:    s_lshr_b32 s13, s1, 28
+; GFX8-NEXT:    s_bfe_u32 s14, s1, 0x40018
+; GFX8-NEXT:    s_bfe_u32 s1, s1, 0x40008
+; GFX8-NEXT:    s_and_b32 s9, s0, 15
 ; GFX8-NEXT:    v_mov_b32_e32 v4, s17
 ; GFX8-NEXT:    v_mov_b32_e32 v5, s10
 ; GFX8-NEXT:    v_mov_b32_e32 v6, s16
 ; GFX8-NEXT:    v_mov_b32_e32 v7, s15
 ; GFX8-NEXT:    v_mov_b32_e32 v8, s8
 ; GFX8-NEXT:    v_mul_u32_u24_sdwa v4, v5, v4 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; GFX8-NEXT:    v_mul_u32_u24_e32 v5, s9, v6
 ; GFX8-NEXT:    v_mul_u32_u24_sdwa v6, v8, v7 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
-; GFX8-NEXT:    s_bfe_u32 s5, s1, 0x40010
-; GFX8-NEXT:    s_bfe_u32 s7, s1, 0x40018
+; GFX8-NEXT:    s_bfe_u32 s5, s0, 0x40010
+; GFX8-NEXT:    s_bfe_u32 s7, s0, 0x40018
 ; GFX8-NEXT:    v_mov_b32_e32 v9, s14
-; GFX8-NEXT:    s_bfe_u32 s1, s1, 0x40008
-; GFX8-NEXT:    v_mov_b32_e32 v3, s2
+; GFX8-NEXT:    s_bfe_u32 s0, s0, 0x40008
+; GFX8-NEXT:    v_mov_b32_e32 v3, s1
 ; GFX8-NEXT:    v_mov_b32_e32 v10, s13
 ; GFX8-NEXT:    v_mov_b32_e32 v11, s6
 ; GFX8-NEXT:    v_mov_b32_e32 v12, s12
 ; GFX8-NEXT:    v_mov_b32_e32 v13, s11
 ; GFX8-NEXT:    v_mov_b32_e32 v14, s4
-; GFX8-NEXT:    v_mul_u32_u24_e32 v3, s1, v3
+; GFX8-NEXT:    v_mul_u32_u24_e32 v3, s0, v3
 ; GFX8-NEXT:    v_or_b32_e32 v5, v5, v6
 ; GFX8-NEXT:    v_mul_u32_u24_e32 v7, s7, v9
 ; GFX8-NEXT:    v_mul_u32_u24_sdwa v8, v11, v10 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; GFX8-NEXT:    v_mul_u32_u24_e32 v9, s5, v12
 ; GFX8-NEXT:    v_mul_u32_u24_sdwa v10, v14, v13 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
-; GFX8-NEXT:    v_and_b32_e32 v5, s0, v5
+; GFX8-NEXT:    v_and_b32_e32 v5, s2, v5
 ; GFX8-NEXT:    v_or_b32_sdwa v3, v3, v4 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; GFX8-NEXT:    v_or_b32_e32 v9, v9, v10
 ; GFX8-NEXT:    v_or_b32_sdwa v7, v7, v8 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
-; GFX8-NEXT:    v_and_b32_e32 v4, s0, v9
+; GFX8-NEXT:    v_and_b32_e32 v4, s2, v9
 ; GFX8-NEXT:    v_or_b32_e32 v3, v5, v3
 ; GFX8-NEXT:    v_or_b32_e32 v6, v4, v7
 ; GFX8-NEXT:    v_lshrrev_b32_e32 v7, 8, v3
 ; GFX8-NEXT:    v_lshrrev_b32_e32 v8, 8, v6
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
 ; GFX8-NEXT:    v_add_u32_e32 v2, vcc, v2, v5
 ; GFX8-NEXT:    v_add_u32_e32 v2, vcc, v7, v2
 ; GFX8-NEXT:    v_add_u32_sdwa v2, vcc, v3, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_2 src1_sel:BYTE_0
 ; GFX8-NEXT:    v_add_u32_sdwa v2, vcc, v3, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:DWORD
 ; GFX8-NEXT:    v_add_u32_e32 v2, vcc, v2, v4
 ; GFX8-NEXT:    v_add_u32_e32 v2, vcc, v8, v2
 ; GFX8-NEXT:    v_add_u32_sdwa v2, vcc, v6, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; GFX8-NEXT:    v_add_u32_sdwa v2, vcc, v6, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_3 src1_sel:DWORD
 ; GFX8-NEXT:    flat_store_byte v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: udot8_acc8_vecMul:
 ; GFX9:       ; %bb.0: ; %entry
-; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX9-NEXT:    s_mov_b32 s2, 0xffff
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX9-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_bfe_u32 s4, s0, 0x40010
 ; GFX9-NEXT:    s_bfe_u32 s11, s1, 0x40010
 ; GFX9-NEXT:    s_bfe_u32 s12, s1, 0x40014
 ; GFX9-NEXT:    s_bfe_u32 s13, s1, 0x40018
 ; GFX9-NEXT:    s_lshr_b32 s14, s1, 28

 ; GFX9-NEXT:    v_add_u32_e32 v2, v2, v3
 ; GFX9-NEXT:    v_add_u32_sdwa v2, v2, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
 ; GFX9-NEXT:    v_add_u32_sdwa v2, v2, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
 ; GFX9-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: udot8_acc8_vecMul:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_mov_b32 s2, 0xffff
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-DL-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX9-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_bfe_u32 s4, s0, 0x40010
 ; GFX9-DL-NEXT:    s_bfe_u32 s11, s1, 0x40010
 ; GFX9-DL-NEXT:    s_bfe_u32 s12, s1, 0x40014
 ; GFX9-DL-NEXT:    s_bfe_u32 s13, s1, 0x40018
 ; GFX9-DL-NEXT:    s_lshr_b32 s14, s1, 28

 ; GFX9-DL-NEXT:    v_add_u32_e32 v2, v2, v3
 ; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
 ; GFX9-DL-NEXT:    v_add_u32_sdwa v2, v2, v4 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:BYTE_3
 ; GFX9-DL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot8_acc8_vecMul:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX10-DL-NEXT:    s_mov_b32 s2, 0xffff
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
+; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX10-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_bfe_u32 s4, s0, 0x40004
 ; GFX10-DL-NEXT:    s_bfe_u32 s5, s1, 0x40004
 ; GFX10-DL-NEXT:    s_and_b32 s6, s0, 15
 ; GFX10-DL-NEXT:    s_and_b32 s8, s1, 15
 ; GFX10-DL-NEXT:    s_bfe_u32 s7, s0, 0x4000c
 ; GFX10-DL-NEXT:    s_bfe_u32 s9, s1, 0x4000c
 ; GFX10-DL-NEXT:    v_mul_lo_u16_e64 v3, s4, s5
entry:
   store i8 %add8, i8 addrspace(1)* %dst, align 4
   ret void
 }
 
 ; TODO: Once the adictional "and+add" are removed, the pattern will be recognized.
 define amdgpu_kernel void @udot8_acc4_vecMul(<8 x i4> addrspace(1)* %src1,
 ; GFX7-LABEL: udot8_acc4_vecMul:
 ; GFX7:       ; %bb.0: ; %entry
-; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
 ; GFX7-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX7-NEXT:    s_mov_b32 s6, -1
+; GFX7-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
 ; GFX7-NEXT:    buffer_load_ubyte v0, off, s[4:7], 0
+; GFX7-NEXT:    s_load_dword s0, s[8:9], 0x0
 ; GFX7-NEXT:    s_load_dword s1, s[10:11], 0x0
 ; GFX7-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7-NEXT:    s_lshr_b32 s2, s0, 28
-; GFX7-NEXT:    s_bfe_u32 s8, s0, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s15, s1, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s16, s1, 0x40014
 ; GFX7-NEXT:    s_bfe_u32 s17, s1, 0x40010
 ; GFX7-NEXT:    s_bfe_u32 s18, s1, 0x4000c
 ; GFX7-NEXT:    s_bfe_u32 s19, s1, 0x40008
 ; GFX7-NEXT:    s_bfe_u32 s20, s1, 0x40004
 ; GFX7-NEXT:    s_lshr_b32 s14, s1, 28
 ; GFX7-NEXT:    s_and_b32 s1, s1, 15
+; GFX7-NEXT:    s_bfe_u32 s8, s0, 0x40018
 ; GFX7-NEXT:    s_bfe_u32 s9, s0, 0x40014
 ; GFX7-NEXT:    s_bfe_u32 s10, s0, 0x40010
 ; GFX7-NEXT:    s_bfe_u32 s11, s0, 0x4000c
 ; GFX7-NEXT:    s_bfe_u32 s12, s0, 0x40008
 ; GFX7-NEXT:    s_bfe_u32 s13, s0, 0x40004
 ; GFX7-NEXT:    s_and_b32 s0, s0, 15
 ; GFX7-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX7-NEXT:    v_mov_b32_e32 v2, s20

 ; GFX7-NEXT:    v_mov_b32_e32 v1, s14
 ; GFX7-NEXT:    v_mad_u32_u24 v0, s2, v1, v0
 ; GFX7-NEXT:    v_and_b32_e32 v0, 15, v0
 ; GFX7-NEXT:    buffer_store_byte v0, off, s[4:7], 0
 ; GFX7-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: udot8_acc4_vecMul:
 ; GFX8:       ; %bb.0: ; %entry
-; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX8-NEXT:    v_mov_b32_e32 v0, s0
-; GFX8-NEXT:    v_mov_b32_e32 v1, s1
+; GFX8-NEXT:    v_mov_b32_e32 v0, s4
+; GFX8-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX8-NEXT:    flat_load_ubyte v2, v[0:1]
+; GFX8-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX8-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX8-NEXT:    s_and_b32 s1, s2, 15
-; GFX8-NEXT:    s_bfe_u32 s4, s2, 0x40004
+; GFX8-NEXT:    s_and_b32 s9, s0, 15
+; GFX8-NEXT:    s_and_b32 s16, s1, 15
+; GFX8-NEXT:    s_bfe_u32 s15, s1, 0x40004
+; GFX8-NEXT:    v_mov_b32_e32 v4, s16
+; GFX8-NEXT:    s_bfe_u32 s11, s1, 0x40018
+; GFX8-NEXT:    s_bfe_u32 s12, s1, 0x40014
+; GFX8-NEXT:    s_bfe_u32 s13, s1, 0x40010
+; GFX8-NEXT:    s_bfe_u32 s14, s1, 0x40008
+; GFX8-NEXT:    s_lshr_b32 s10, s1, 28
+; GFX8-NEXT:    s_bfe_u32 s1, s1, 0x4000c
+; GFX8-NEXT:    s_bfe_u32 s8, s0, 0x40004
+; GFX8-NEXT:    v_mov_b32_e32 v5, s15
+; GFX8-NEXT:    s_lshr_b32 s2, s0, 28
+; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x40018
+; GFX8-NEXT:    s_bfe_u32 s5, s0, 0x40014
+; GFX8-NEXT:    s_bfe_u32 s6, s0, 0x40010
+; GFX8-NEXT:    s_bfe_u32 s7, s0, 0x40008
+; GFX8-NEXT:    s_bfe_u32 s0, s0, 0x4000c
 ; GFX8-NEXT:    v_mov_b32_e32 v3, s1
-; GFX8-NEXT:    s_bfe_u32 s6, s2, 0x40008
-; GFX8-NEXT:    s_and_b32 s1, s0, 15
-; GFX8-NEXT:    s_bfe_u32 s7, s2, 0x4000c
-; GFX8-NEXT:    v_mov_b32_e32 v4, s4
-; GFX8-NEXT:    s_bfe_u32 s5, s0, 0x40004
-; GFX8-NEXT:    v_mov_b32_e32 v5, s7
-; GFX8-NEXT:    s_bfe_u32 s8, s0, 0x4000c
-; GFX8-NEXT:    v_mov_b32_e32 v6, s6
-; GFX8-NEXT:    s_bfe_u32 s4, s0, 0x40008
-; GFX8-NEXT:    v_mul_u32_u24_e32 v5, s8, v5
-; GFX8-NEXT:    s_bfe_u32 s6, s2, 0x40010
-; GFX8-NEXT:    v_and_b32_e32 v5, 15, v5
-; GFX8-NEXT:    s_bfe_u32 s8, s2, 0x40014
-; GFX8-NEXT:    s_bfe_u32 s7, s0, 0x40010
-; GFX8-NEXT:    v_mov_b32_e32 v7, s6
-; GFX8-NEXT:    s_lshr_b32 s11, s2, 28
-; GFX8-NEXT:    s_bfe_u32 s2, s2, 0x40018
-; GFX8-NEXT:    s_bfe_u32 s9, s0, 0x40014
-; GFX8-NEXT:    v_mov_b32_e32 v8, s8
-; GFX8-NEXT:    s_bfe_u32 s10, s0, 0x40018
-; GFX8-NEXT:    v_mov_b32_e32 v9, s2
-; GFX8-NEXT:    s_lshr_b32 s0, s0, 28
+; GFX8-NEXT:    v_mov_b32_e32 v6, s14
+; GFX8-NEXT:    v_mul_u32_u24_e32 v3, s0, v3
+; GFX8-NEXT:    v_and_b32_e32 v3, 15, v3
+; GFX8-NEXT:    v_mov_b32_e32 v7, s13
+; GFX8-NEXT:    v_mov_b32_e32 v8, s12
+; GFX8-NEXT:    v_mov_b32_e32 v9, s11
 ; GFX8-NEXT:    s_waitcnt vmcnt(0)
-; GFX8-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s4, v6, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s9, v4, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s8, v5, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s7, v6, v2
 ; GFX8-NEXT:    v_and_b32_e32 v2, 15, v2
-; GFX8-NEXT:    v_add_u32_e32 v2, vcc, v5, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s7, v7, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s9, v8, v2
-; GFX8-NEXT:    v_mad_u32_u24 v2, s10, v9, v2
-; GFX8-NEXT:    v_mov_b32_e32 v3, s11
-; GFX8-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX8-NEXT:    v_add_u32_e32 v2, vcc, v3, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s6, v7, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s5, v8, v2
+; GFX8-NEXT:    v_mad_u32_u24 v2, s4, v9, v2
+; GFX8-NEXT:    v_mov_b32_e32 v3, s10
+; GFX8-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX8-NEXT:    v_and_b32_e32 v2, 15, v2
 ; GFX8-NEXT:    flat_store_byte v[0:1], v2
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: udot8_acc4_vecMul:
 ; GFX9:       ; %bb.0: ; %entry
-; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_and_b32 s1, s2, 15
-; GFX9-NEXT:    s_bfe_u32 s4, s2, 0x40004
+; GFX9-NEXT:    s_and_b32 s9, s0, 15
+; GFX9-NEXT:    s_and_b32 s16, s1, 15
+; GFX9-NEXT:    s_bfe_u32 s15, s1, 0x40004
+; GFX9-NEXT:    v_mov_b32_e32 v4, s16
+; GFX9-NEXT:    s_bfe_u32 s11, s1, 0x40018
+; GFX9-NEXT:    s_bfe_u32 s12, s1, 0x40014
+; GFX9-NEXT:    s_bfe_u32 s13, s1, 0x40010
+; GFX9-NEXT:    s_bfe_u32 s14, s1, 0x40008
+; GFX9-NEXT:    s_lshr_b32 s10, s1, 28
+; GFX9-NEXT:    s_bfe_u32 s1, s1, 0x4000c
+; GFX9-NEXT:    s_bfe_u32 s8, s0, 0x40004
+; GFX9-NEXT:    v_mov_b32_e32 v5, s15
+; GFX9-NEXT:    s_lshr_b32 s2, s0, 28
+; GFX9-NEXT:    s_bfe_u32 s4, s0, 0x40018
+; GFX9-NEXT:    s_bfe_u32 s5, s0, 0x40014
+; GFX9-NEXT:    s_bfe_u32 s6, s0, 0x40010
+; GFX9-NEXT:    s_bfe_u32 s7, s0, 0x40008
+; GFX9-NEXT:    s_bfe_u32 s0, s0, 0x4000c
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-NEXT:    s_bfe_u32 s6, s2, 0x40008
-; GFX9-NEXT:    s_and_b32 s1, s0, 15
-; GFX9-NEXT:    s_bfe_u32 s7, s2, 0x4000c
-; GFX9-NEXT:    v_mov_b32_e32 v4, s4
-; GFX9-NEXT:    s_bfe_u32 s5, s0, 0x40004
-; GFX9-NEXT:    v_mov_b32_e32 v5, s7
-; GFX9-NEXT:    s_bfe_u32 s8, s0, 0x4000c
-; GFX9-NEXT:    v_mov_b32_e32 v6, s6
-; GFX9-NEXT:    s_bfe_u32 s4, s0, 0x40008
-; GFX9-NEXT:    v_mul_u32_u24_e32 v5, s8, v5
-; GFX9-NEXT:    s_bfe_u32 s6, s2, 0x40010
-; GFX9-NEXT:    v_and_b32_e32 v5, 15, v5
-; GFX9-NEXT:    s_bfe_u32 s8, s2, 0x40014
-; GFX9-NEXT:    s_bfe_u32 s7, s0, 0x40010
-; GFX9-NEXT:    v_mov_b32_e32 v7, s6
-; GFX9-NEXT:    s_lshr_b32 s11, s2, 28
-; GFX9-NEXT:    s_bfe_u32 s2, s2, 0x40018
-; GFX9-NEXT:    s_bfe_u32 s9, s0, 0x40014
-; GFX9-NEXT:    v_mov_b32_e32 v8, s8
-; GFX9-NEXT:    s_bfe_u32 s10, s0, 0x40018
-; GFX9-NEXT:    v_mov_b32_e32 v9, s2
-; GFX9-NEXT:    s_lshr_b32 s0, s0, 28
+; GFX9-NEXT:    v_mov_b32_e32 v6, s14
+; GFX9-NEXT:    v_mul_u32_u24_e32 v3, s0, v3
+; GFX9-NEXT:    v_and_b32_e32 v3, 15, v3
+; GFX9-NEXT:    v_mov_b32_e32 v7, s13
+; GFX9-NEXT:    v_mov_b32_e32 v8, s12
+; GFX9-NEXT:    v_mov_b32_e32 v9, s11
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s4, v6, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s9, v4, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s8, v5, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s7, v6, v2
 ; GFX9-NEXT:    v_and_b32_e32 v2, 15, v2
-; GFX9-NEXT:    v_add_u32_e32 v2, v2, v5
-; GFX9-NEXT:    v_mad_u32_u24 v2, s7, v7, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s9, v8, v2
-; GFX9-NEXT:    v_mad_u32_u24 v2, s10, v9, v2
-; GFX9-NEXT:    v_mov_b32_e32 v3, s11
-; GFX9-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX9-NEXT:    v_add_u32_e32 v2, v2, v3
+; GFX9-NEXT:    v_mad_u32_u24 v2, s6, v7, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s5, v8, v2
+; GFX9-NEXT:    v_mad_u32_u24 v2, s4, v9, v2
+; GFX9-NEXT:    v_mov_b32_e32 v3, s10
+; GFX9-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX9-NEXT:    v_and_b32_e32 v2, 15, v2
 ; GFX9-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX9-DL-LABEL: udot8_acc4_vecMul:
 ; GFX9-DL:       ; %bb.0: ; %entry
-; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX9-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX9-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_load_dword s2, s[6:7], 0x0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-DL-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX9-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
+; GFX9-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
 ; GFX9-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-DL-NEXT:    s_and_b32 s1, s2, 15
-; GFX9-DL-NEXT:    s_bfe_u32 s4, s2, 0x40004
+; GFX9-DL-NEXT:    s_and_b32 s9, s0, 15
+; GFX9-DL-NEXT:    s_and_b32 s16, s1, 15
+; GFX9-DL-NEXT:    s_bfe_u32 s15, s1, 0x40004
+; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s16
+; GFX9-DL-NEXT:    s_bfe_u32 s11, s1, 0x40018
+; GFX9-DL-NEXT:    s_bfe_u32 s12, s1, 0x40014
+; GFX9-DL-NEXT:    s_bfe_u32 s13, s1, 0x40010
+; GFX9-DL-NEXT:    s_bfe_u32 s14, s1, 0x40008
+; GFX9-DL-NEXT:    s_lshr_b32 s10, s1, 28
+; GFX9-DL-NEXT:    s_bfe_u32 s1, s1, 0x4000c
+; GFX9-DL-NEXT:    s_bfe_u32 s8, s0, 0x40004
+; GFX9-DL-NEXT:    v_mov_b32_e32 v5, s15
+; GFX9-DL-NEXT:    s_lshr_b32 s2, s0, 28
+; GFX9-DL-NEXT:    s_bfe_u32 s4, s0, 0x40018
+; GFX9-DL-NEXT:    s_bfe_u32 s5, s0, 0x40014
+; GFX9-DL-NEXT:    s_bfe_u32 s6, s0, 0x40010
+; GFX9-DL-NEXT:    s_bfe_u32 s7, s0, 0x40008
+; GFX9-DL-NEXT:    s_bfe_u32 s0, s0, 0x4000c
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-DL-NEXT:    s_bfe_u32 s6, s2, 0x40008
-; GFX9-DL-NEXT:    s_and_b32 s1, s0, 15
-; GFX9-DL-NEXT:    s_bfe_u32 s7, s2, 0x4000c
-; GFX9-DL-NEXT:    v_mov_b32_e32 v4, s4
-; GFX9-DL-NEXT:    s_bfe_u32 s5, s0, 0x40004
-; GFX9-DL-NEXT:    v_mov_b32_e32 v5, s7
-; GFX9-DL-NEXT:    s_bfe_u32 s8, s0, 0x4000c
-; GFX9-DL-NEXT:    v_mov_b32_e32 v6, s6
-; GFX9-DL-NEXT:    s_bfe_u32 s4, s0, 0x40008
-; GFX9-DL-NEXT:    v_mul_u32_u24_e32 v5, s8, v5
-; GFX9-DL-NEXT:    s_bfe_u32 s6, s2, 0x40010
-; GFX9-DL-NEXT:    v_and_b32_e32 v5, 15, v5
-; GFX9-DL-NEXT:    s_bfe_u32 s8, s2, 0x40014
-; GFX9-DL-NEXT:    s_bfe_u32 s7, s0, 0x40010
-; GFX9-DL-NEXT:    v_mov_b32_e32 v7, s6
-; GFX9-DL-NEXT:    s_lshr_b32 s11, s2, 28
-; GFX9-DL-NEXT:    s_bfe_u32 s2, s2, 0x40018
-; GFX9-DL-NEXT:    s_bfe_u32 s9, s0, 0x40014
-; GFX9-DL-NEXT:    v_mov_b32_e32 v8, s8
-; GFX9-DL-NEXT:    s_bfe_u32 s10, s0, 0x40018
-; GFX9-DL-NEXT:    v_mov_b32_e32 v9, s2
-; GFX9-DL-NEXT:    s_lshr_b32 s0, s0, 28
+; GFX9-DL-NEXT:    v_mov_b32_e32 v6, s14
+; GFX9-DL-NEXT:    v_mul_u32_u24_e32 v3, s0, v3
+; GFX9-DL-NEXT:    v_and_b32_e32 v3, 15, v3
+; GFX9-DL-NEXT:    v_mov_b32_e32 v7, s13
+; GFX9-DL-NEXT:    v_mov_b32_e32 v8, s12
+; GFX9-DL-NEXT:    v_mov_b32_e32 v9, s11
 ; GFX9-DL-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s1, v3, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s5, v4, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s4, v6, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s9, v4, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s8, v5, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s7, v6, v2
 ; GFX9-DL-NEXT:    v_and_b32_e32 v2, 15, v2
-; GFX9-DL-NEXT:    v_add_u32_e32 v2, v2, v5
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s7, v7, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s9, v8, v2
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s10, v9, v2
-; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s11
-; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s0, v3, v2
+; GFX9-DL-NEXT:    v_add_u32_e32 v2, v2, v3
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s6, v7, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s5, v8, v2
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s4, v9, v2
+; GFX9-DL-NEXT:    v_mov_b32_e32 v3, s10
+; GFX9-DL-NEXT:    v_mad_u32_u24 v2, s2, v3, v2
 ; GFX9-DL-NEXT:    v_and_b32_e32 v2, 15, v2
 ; GFX9-DL-NEXT:    global_store_byte v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot8_acc4_vecMul:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
+; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
+; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_load_dword s0, s[4:5], 0x0
 ; GFX10-DL-NEXT:    s_load_dword s1, s[6:7], 0x0
-; GFX10-DL-NEXT:    global_load_ubyte v2, v[0:1], off
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-DL-NEXT:    s_and_b32 s2, s0, 15
 ; GFX10-DL-NEXT:    s_and_b32 s4, s1, 15
 ; GFX10-DL-NEXT:    s_bfe_u32 s5, s0, 0x40004
 ; GFX10-DL-NEXT:    s_bfe_u32 s6, s1, 0x40004
 ; GFX10-DL-NEXT:    s_bfe_u32 s7, s1, 0x40008
 ; GFX10-DL-NEXT:    s_waitcnt vmcnt(0)
 ; GFX10-DL-NEXT:    v_mad_u32_u24 v2, s2, s4, v2

 ; GFX9-DL-NEXT:    v_dot8_u32_u4 v2, s4, v0, v1
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX9-DL-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX9-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-DL-NEXT:    s_endpgm
 ;
 ; GFX10-DL-LABEL: udot8_variant1:
 ; GFX10-DL:       ; %bb.0: ; %entry
-; GFX10-DL-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
-; GFX10-DL-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
+; GFX10-DL-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX10-DL-NEXT:    ; implicit-def: $vcc_hi
 ; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    s_load_dword s2, s[0:1], 0x0
-; GFX10-DL-NEXT:    s_load_dword s3, s[4:5], 0x0
-; GFX10-DL-NEXT:    s_load_dword s4, s[6:7], 0x0
-; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s2
-; GFX10-DL-NEXT:    v_dot8_u32_u4 v2, s4, s3, v0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s0
-; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s1
+; GFX10-DL-NEXT:    s_load_dword s6, s[4:5], 0x0
+; GFX10-DL-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX10-DL-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX10-DL-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s6
+; GFX10-DL-NEXT:    v_dot8_u32_u4 v2, s1, s0, v0
+; GFX10-DL-NEXT:    v_mov_b32_e32 v0, s4
+; GFX10-DL-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX10-DL-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX10-DL-NEXT:    s_endpgm
                                           i32 addrspace(1)* %v2addr,
                                           i32 addrspace(1)* %dst) {
 entry:
   %v1 = load i32, i32 addrspace(1)* %v1addr, align 4
   %v2 = load i32, i32 addrspace(1)* %v2addr, align 4
   %and = and i32 %v1, 15

llvm/test/CodeGen/AMDGPU/insert_vector_dynelt.ll

 }
 
 ; GCN-LABEL: {{^}}byte8_inselt:
 ; GCN-NOT: v_movrel
 ; GCN-NOT: buffer_
 ; GCN:     s_lshl_b32 [[SEL:s[0-9]+]], s{{[0-9]+}}, 3
 ; GCN:     s_lshl_b64 s[{{[0-9:]+}}], s[{{[0-9:]+}}], [[SEL]]
 ; GCN:     s_mov_b32 [[K:s[0-9]+]], 0x1010101
-; GCN:     s_and_b32 s3, s1, [[K]]
+; GCN:     s_and_b32 s{{[0-9]+}}, s{{[0-9]+}}, [[K]]
 ; GCN:     s_and_b32 s{{[0-9]+}}, s{{[0-9]+}}, [[K]]
 ; GCN:     s_andn2_b64 s[{{[0-9:]+}}], s[{{[0-9:]+}}], s[{{[0-9:]+}}]
 ; GCN:     s_or_b64 s[{{[0-9:]+}}], s[{{[0-9:]+}}], s[{{[0-9:]+}}]
 define amdgpu_kernel void @byte8_inselt(<8 x i8> addrspace(1)* %out, <8 x i8> %vec, i32 %sel) {
 entry:
   %v = insertelement <8 x i8> %vec, i8 1, i32 %sel
   store <8 x i8> %v, <8 x i8> addrspace(1)* %out
   ret void

llvm/test/CodeGen/AMDGPU/insert_vector_elt.ll

; VI-NEXT:    s_endpgm
   %vecins = insertelement <4 x double> %a, double 8.0, i32 %b
   store <4 x double> %vecins, <4 x double> addrspace(1)* %out, align 16
   ret void
 }
 
 define amdgpu_kernel void @dynamic_insertelement_v8f64(<8 x double> addrspace(1)* %out, <8 x double> %a, i32 %b) #0 {
 ; SI-LABEL: dynamic_insertelement_v8f64:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[8:9], s[4:5], 0x0
-; SI-NEXT:    s_load_dwordx16 s[12:27], s[4:5], 0x10
-; SI-NEXT:    s_load_dword s4, s[4:5], 0x20
+; SI-NEXT:    s_load_dwordx16 s[8:23], s[4:5], 0x10
+; SI-NEXT:    s_load_dword s6, s[4:5], 0x20
 ; SI-NEXT:    v_mov_b32_e32 v16, 64
-; SI-NEXT:    s_mov_b32 s11, 0x100f000
-; SI-NEXT:    s_mov_b32 s10, -1
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    v_mov_b32_e32 v0, s12
-; SI-NEXT:    s_and_b32 s4, s4, 7
-; SI-NEXT:    s_lshl_b32 s4, s4, 3
-; SI-NEXT:    v_mov_b32_e32 v1, s13
-; SI-NEXT:    v_mov_b32_e32 v2, s14
-; SI-NEXT:    v_mov_b32_e32 v3, s15
-; SI-NEXT:    v_mov_b32_e32 v4, s16
-; SI-NEXT:    v_mov_b32_e32 v5, s17
-; SI-NEXT:    v_mov_b32_e32 v6, s18
-; SI-NEXT:    v_mov_b32_e32 v7, s19
-; SI-NEXT:    v_mov_b32_e32 v8, s20
-; SI-NEXT:    v_mov_b32_e32 v9, s21
-; SI-NEXT:    v_mov_b32_e32 v10, s22
-; SI-NEXT:    v_mov_b32_e32 v11, s23
-; SI-NEXT:    v_mov_b32_e32 v12, s24
-; SI-NEXT:    v_mov_b32_e32 v13, s25
-; SI-NEXT:    v_mov_b32_e32 v14, s26
-; SI-NEXT:    v_mov_b32_e32 v15, s27
+; SI-NEXT:    v_mov_b32_e32 v2, s10
+; SI-NEXT:    s_and_b32 s6, s6, 7
+; SI-NEXT:    v_mov_b32_e32 v3, s11
+; SI-NEXT:    s_lshl_b32 s6, s6, 3
+; SI-NEXT:    v_mov_b32_e32 v0, s8
+; SI-NEXT:    v_mov_b32_e32 v1, s9
+; SI-NEXT:    v_mov_b32_e32 v4, s12
+; SI-NEXT:    v_mov_b32_e32 v5, s13
+; SI-NEXT:    v_mov_b32_e32 v6, s14
+; SI-NEXT:    v_mov_b32_e32 v7, s15
+; SI-NEXT:    v_mov_b32_e32 v8, s16
+; SI-NEXT:    v_mov_b32_e32 v9, s17
+; SI-NEXT:    v_mov_b32_e32 v10, s18
+; SI-NEXT:    v_mov_b32_e32 v11, s19
+; SI-NEXT:    v_mov_b32_e32 v12, s20
+; SI-NEXT:    v_mov_b32_e32 v13, s21
+; SI-NEXT:    v_mov_b32_e32 v14, s22
+; SI-NEXT:    v_mov_b32_e32 v15, s23
 ; SI-NEXT:    buffer_store_dwordx4 v[12:15], off, s[0:3], s7 offset:112
 ; SI-NEXT:    buffer_store_dwordx4 v[8:11], off, s[0:3], s7 offset:96
 ; SI-NEXT:    buffer_store_dwordx4 v[4:7], off, s[0:3], s7 offset:80
 ; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], s7 offset:64
-; SI-NEXT:    v_or_b32_e32 v16, s4, v16
+; SI-NEXT:    v_or_b32_e32 v16, s6, v16
 ; SI-NEXT:    v_mov_b32_e32 v0, 0
 ; SI-NEXT:    v_mov_b32_e32 v1, 0x40200000
 ; SI-NEXT:    buffer_store_dwordx2 v[0:1], v16, s[0:3], s7 offen
 ; SI-NEXT:    buffer_load_dwordx4 v[0:3], off, s[0:3], s7 offset:64
 ; SI-NEXT:    buffer_load_dwordx4 v[4:7], off, s[0:3], s7 offset:80
 ; SI-NEXT:    buffer_load_dwordx4 v[8:11], off, s[0:3], s7 offset:96
 ; SI-NEXT:    buffer_load_dwordx4 v[12:15], off, s[0:3], s7 offset:112
-; SI-NEXT:    s_waitcnt vmcnt(0)
+; SI-NEXT:    s_load_dwordx2 s[8:9], s[4:5], 0x0
+; SI-NEXT:    s_mov_b32 s11, 0x100f000
+; SI-NEXT:    s_mov_b32 s10, -1
+; SI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; SI-NEXT:    buffer_store_dwordx4 v[12:15], off, s[8:11], 0 offset:48
 ; SI-NEXT:    buffer_store_dwordx4 v[8:11], off, s[8:11], 0 offset:32
 ; SI-NEXT:    buffer_store_dwordx4 v[4:7], off, s[8:11], 0 offset:16
 ; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[8:11], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: dynamic_insertelement_v8f64:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx2 s[8:9], s[4:5], 0x0
-; VI-NEXT:    s_load_dwordx16 s[12:27], s[4:5], 0x40
-; VI-NEXT:    s_load_dword s4, s[4:5], 0x80
+; VI-NEXT:    s_load_dword s6, s[4:5], 0x80
+; VI-NEXT:    s_load_dwordx16 s[8:23], s[4:5], 0x40
 ; VI-NEXT:    v_mov_b32_e32 v16, 64
-; VI-NEXT:    s_mov_b32 s11, 0x1100f000
-; VI-NEXT:    s_mov_b32 s10, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v0, s12
-; VI-NEXT:    s_and_b32 s4, s4, 7
-; VI-NEXT:    s_lshl_b32 s4, s4, 3
-; VI-NEXT:    v_mov_b32_e32 v1, s13
-; VI-NEXT:    v_mov_b32_e32 v2, s14
-; VI-NEXT:    v_mov_b32_e32 v3, s15
-; VI-NEXT:    v_mov_b32_e32 v4, s16
-; VI-NEXT:    v_mov_b32_e32 v5, s17
-; VI-NEXT:    v_mov_b32_e32 v6, s18
-; VI-NEXT:    v_mov_b32_e32 v7, s19
-; VI-NEXT:    v_mov_b32_e32 v8, s20
-; VI-NEXT:    v_mov_b32_e32 v9, s21
-; VI-NEXT:    v_mov_b32_e32 v10, s22
-; VI-NEXT:    v_mov_b32_e32 v11, s23
-; VI-NEXT:    v_mov_b32_e32 v12, s24
-; VI-NEXT:    v_mov_b32_e32 v13, s25
-; VI-NEXT:    v_mov_b32_e32 v14, s26
-; VI-NEXT:    v_mov_b32_e32 v15, s27
+; VI-NEXT:    s_and_b32 s6, s6, 7
+; VI-NEXT:    v_mov_b32_e32 v2, s10
+; VI-NEXT:    v_mov_b32_e32 v3, s11
+; VI-NEXT:    s_lshl_b32 s6, s6, 3
+; VI-NEXT:    v_mov_b32_e32 v0, s8
+; VI-NEXT:    v_mov_b32_e32 v1, s9
+; VI-NEXT:    v_mov_b32_e32 v4, s12
+; VI-NEXT:    v_mov_b32_e32 v5, s13
+; VI-NEXT:    v_mov_b32_e32 v6, s14
+; VI-NEXT:    v_mov_b32_e32 v7, s15
+; VI-NEXT:    v_mov_b32_e32 v8, s16
+; VI-NEXT:    v_mov_b32_e32 v9, s17
+; VI-NEXT:    v_mov_b32_e32 v10, s18
+; VI-NEXT:    v_mov_b32_e32 v11, s19
+; VI-NEXT:    v_mov_b32_e32 v12, s20
+; VI-NEXT:    v_mov_b32_e32 v13, s21
+; VI-NEXT:    v_mov_b32_e32 v14, s22
+; VI-NEXT:    v_mov_b32_e32 v15, s23
 ; VI-NEXT:    buffer_store_dwordx4 v[12:15], off, s[0:3], s7 offset:112
 ; VI-NEXT:    buffer_store_dwordx4 v[8:11], off, s[0:3], s7 offset:96
 ; VI-NEXT:    buffer_store_dwordx4 v[4:7], off, s[0:3], s7 offset:80
 ; VI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], s7 offset:64
-; VI-NEXT:    v_or_b32_e32 v16, s4, v16
+; VI-NEXT:    v_or_b32_e32 v16, s6, v16
 ; VI-NEXT:    v_mov_b32_e32 v0, 0
 ; VI-NEXT:    v_mov_b32_e32 v1, 0x40200000
 ; VI-NEXT:    buffer_store_dwordx2 v[0:1], v16, s[0:3], s7 offen
 ; VI-NEXT:    buffer_load_dwordx4 v[0:3], off, s[0:3], s7 offset:64
 ; VI-NEXT:    buffer_load_dwordx4 v[4:7], off, s[0:3], s7 offset:80
 ; VI-NEXT:    buffer_load_dwordx4 v[8:11], off, s[0:3], s7 offset:96
 ; VI-NEXT:    buffer_load_dwordx4 v[12:15], off, s[0:3], s7 offset:112
-; VI-NEXT:    s_waitcnt vmcnt(0)
+; VI-NEXT:    s_load_dwordx2 s[8:9], s[4:5], 0x0
+; VI-NEXT:    s_mov_b32 s11, 0x1100f000
+; VI-NEXT:    s_mov_b32 s10, -1
+; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; VI-NEXT:    buffer_store_dwordx4 v[12:15], off, s[8:11], 0 offset:48
 ; VI-NEXT:    buffer_store_dwordx4 v[8:11], off, s[8:11], 0 offset:32
 ; VI-NEXT:    buffer_store_dwordx4 v[4:7], off, s[8:11], 0 offset:16
 ; VI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[8:11], 0
 ; VI-NEXT:    s_endpgm
   %vecins = insertelement <8 x double> %a, double 8.0, i32 %b
   store <8 x double> %vecins, <8 x double> addrspace(1)* %out, align 16
   ret void
 }
 
 declare <4 x float> @llvm.amdgcn.image.gather4.lz.2d.v4f32.f32(i32, float, float, <8 x i32>, <4 x i32>, i1, i32, i32) #1
 
 attributes #0 = { nounwind }
 attributes #1 = { nounwind readnone }

llvm/test/CodeGen/AMDGPU/insert_vector_elt.v2i16.ll

 
 
 define amdgpu_kernel void @s_insertelement_v2i16_0_reg(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(4)* %vec.ptr, [8 x i32], i16 %elt) #0 {
 ; GFX9-LABEL: s_insertelement_v2i16_0_reg:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; GFX9-NEXT:    s_load_dword s4, s[4:5], 0x30
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_load_dword s2, s[2:3], 0x0
 ; GFX9-NEXT:    v_mov_b32_e32 v0, s0
+; GFX9-NEXT:    s_load_dword s0, s[2:3], 0x0
 ; GFX9-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_pack_lh_b32_b16 s0, s4, s2
+; GFX9-NEXT:    s_pack_lh_b32_b16 s0, s4, s0
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s0
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
 ;
 ; VI-LABEL: s_insertelement_v2i16_0_reg:
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; VI-NEXT:    s_load_dword s4, s[4:5], 0x30
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_load_dword s2, s[2:3], 0x0
 ; VI-NEXT:    v_mov_b32_e32 v0, s0
+; VI-NEXT:    s_load_dword s0, s[2:3], 0x0
 ; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    s_and_b32 s0, s4, 0xffff
+; VI-NEXT:    s_and_b32 s1, s4, 0xffff
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_and_b32 s1, s2, 0xffff0000
-; VI-NEXT:    s_or_b32 s0, s0, s1
+; VI-NEXT:    s_and_b32 s0, s0, 0xffff0000
+; VI-NEXT:    s_or_b32 s0, s1, s0
 ; VI-NEXT:    v_mov_b32_e32 v2, s0
 ; VI-NEXT:    flat_store_dword v[0:1], v2
 ; VI-NEXT:    s_endpgm
 ;
 ; CI-LABEL: s_insertelement_v2i16_0_reg:
 ; CI:       ; %bb.0:
 ; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; CI-NEXT:    s_load_dword s4, s[4:5], 0xc
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
-; CI-NEXT:    s_load_dword s2, s[2:3], 0x0
 ; CI-NEXT:    v_mov_b32_e32 v0, s0
+; CI-NEXT:    s_load_dword s0, s[2:3], 0x0
 ; CI-NEXT:    v_mov_b32_e32 v1, s1
 ; CI-NEXT:    s_and_b32 s1, s4, 0xffff
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
-; CI-NEXT:    s_and_b32 s0, s2, 0xffff0000
+; CI-NEXT:    s_and_b32 s0, s0, 0xffff0000
 ; CI-NEXT:    s_or_b32 s0, s1, s0
 ; CI-NEXT:    v_mov_b32_e32 v2, s0
 ; CI-NEXT:    flat_store_dword v[0:1], v2
 ; CI-NEXT:    s_endpgm
   %vec = load <2 x i16>, <2 x i16> addrspace(4)* %vec.ptr
   %vecins = insertelement <2 x i16> %vec, i16 %elt, i32 0
   store <2 x i16> %vecins, <2 x i16> addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @s_insertelement_v2i16_0_multi_use_hi_reg(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(4)* %vec.ptr, [8 x i32], i16 %elt) #0 {
 ; GFX9-LABEL: s_insertelement_v2i16_0_multi_use_hi_reg:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; GFX9-NEXT:    s_load_dword s4, s[4:5], 0x30
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_load_dword s2, s[2:3], 0x0
 ; GFX9-NEXT:    v_mov_b32_e32 v0, s0
+; GFX9-NEXT:    s_load_dword s0, s[2:3], 0x0
 ; GFX9-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_lshr_b32 s0, s2, 16
+; GFX9-NEXT:    s_lshr_b32 s0, s0, 16
 ; GFX9-NEXT:    s_pack_ll_b32_b16 s1, s4, s0
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    ;;#ASMSTART
 ; GFX9-NEXT:    ; use s0
 ; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; VI-LABEL: s_insertelement_v2i16_0_multi_use_hi_reg:
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; VI-NEXT:    s_load_dword s4, s[4:5], 0x30
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_load_dword s2, s[2:3], 0x0
 ; VI-NEXT:    v_mov_b32_e32 v0, s0
+; VI-NEXT:    s_load_dword s0, s[2:3], 0x0
 ; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    s_and_b32 s0, s4, 0xffff
+; VI-NEXT:    s_and_b32 s1, s4, 0xffff
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_lshr_b32 s1, s2, 16
-; VI-NEXT:    s_and_b32 s2, s2, 0xffff0000
-; VI-NEXT:    s_or_b32 s0, s0, s2
+; VI-NEXT:    s_lshr_b32 s2, s0, 16
+; VI-NEXT:    s_and_b32 s0, s0, 0xffff0000
+; VI-NEXT:    s_or_b32 s0, s1, s0
 ; VI-NEXT:    v_mov_b32_e32 v2, s0
 ; VI-NEXT:    flat_store_dword v[0:1], v2
 ; VI-NEXT:    ;;#ASMSTART
-; VI-NEXT:    ; use s1
+; VI-NEXT:    ; use s2
 ; VI-NEXT:    ;;#ASMEND
 ; VI-NEXT:    s_endpgm
 ;
 ; CI-LABEL: s_insertelement_v2i16_0_multi_use_hi_reg:
 ; CI:       ; %bb.0:
 ; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
-; CI-NEXT:    s_load_dword s4, s[4:5], 0xc
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
 ; CI-NEXT:    s_load_dword s2, s[2:3], 0x0
-; CI-NEXT:    v_mov_b32_e32 v1, s1
 ; CI-NEXT:    v_mov_b32_e32 v0, s0
-; CI-NEXT:    s_and_b32 s0, s4, 0xffff
+; CI-NEXT:    s_load_dword s0, s[4:5], 0xc
+; CI-NEXT:    v_mov_b32_e32 v1, s1
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
 ; CI-NEXT:    s_lshr_b32 s1, s2, 16
 ; CI-NEXT:    s_lshl_b32 s2, s1, 16
+; CI-NEXT:    s_and_b32 s0, s0, 0xffff
 ; CI-NEXT:    s_or_b32 s0, s0, s2
 ; CI-NEXT:    v_mov_b32_e32 v2, s0
 ; CI-NEXT:    flat_store_dword v[0:1], v2
 ; CI-NEXT:    ;;#ASMSTART
 ; CI-NEXT:    ; use s1
 ; CI-NEXT:    ;;#ASMEND
 ; CI-NEXT:    s_endpgm
   %vec = load <2 x i16>, <2 x i16> addrspace(4)* %vec.ptr
   %elt1 = extractelement <2 x i16> %vec, i32 1
   %vecins = insertelement <2 x i16> %vec, i16 %elt, i32 0
   store <2 x i16> %vecins, <2 x i16> addrspace(1)* %out
   %use1 = zext i16 %elt1 to i32
   call void asm sideeffect "; use $0", "s"(i32 %use1) #0
   ret void
 }
 
 define amdgpu_kernel void @s_insertelement_v2i16_0_reghi(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(4)* %vec.ptr, [8 x i32], i32 %elt.arg) #0 {
 ; GFX9-LABEL: s_insertelement_v2i16_0_reghi:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; GFX9-NEXT:    s_load_dword s4, s[4:5], 0x30
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_load_dword s2, s[2:3], 0x0
 ; GFX9-NEXT:    v_mov_b32_e32 v0, s0
+; GFX9-NEXT:    s_load_dword s0, s[2:3], 0x0
 ; GFX9-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_pack_hh_b32_b16 s0, s4, s2
+; GFX9-NEXT:    s_pack_hh_b32_b16 s0, s4, s0
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s0
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
 ;
 ; VI-LABEL: s_insertelement_v2i16_0_reghi:
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; VI-NEXT:    s_load_dword s4, s[4:5], 0x30
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_load_dword s2, s[2:3], 0x0
 ; VI-NEXT:    v_mov_b32_e32 v0, s0
+; VI-NEXT:    s_load_dword s0, s[2:3], 0x0
 ; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    s_lshr_b32 s0, s4, 16
+; VI-NEXT:    s_lshr_b32 s1, s4, 16
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_and_b32 s1, s2, 0xffff0000
-; VI-NEXT:    s_or_b32 s0, s0, s1
+; VI-NEXT:    s_and_b32 s0, s0, 0xffff0000
+; VI-NEXT:    s_or_b32 s0, s1, s0
 ; VI-NEXT:    v_mov_b32_e32 v2, s0
 ; VI-NEXT:    flat_store_dword v[0:1], v2
 ; VI-NEXT:    s_endpgm
 ;
 ; CI-LABEL: s_insertelement_v2i16_0_reghi:
 ; CI:       ; %bb.0:
 ; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; CI-NEXT:    s_load_dword s4, s[4:5], 0xc
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
-; CI-NEXT:    s_load_dword s2, s[2:3], 0x0
 ; CI-NEXT:    v_mov_b32_e32 v0, s0
+; CI-NEXT:    s_load_dword s0, s[2:3], 0x0
 ; CI-NEXT:    v_mov_b32_e32 v1, s1
 ; CI-NEXT:    s_lshr_b32 s1, s4, 16
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
-; CI-NEXT:    s_and_b32 s0, s2, 0xffff0000
+; CI-NEXT:    s_and_b32 s0, s0, 0xffff0000
 ; CI-NEXT:    s_or_b32 s0, s1, s0
 ; CI-NEXT:    v_mov_b32_e32 v2, s0
 ; CI-NEXT:    flat_store_dword v[0:1], v2
 ; CI-NEXT:    s_endpgm
   %vec = load <2 x i16>, <2 x i16> addrspace(4)* %vec.ptr
   %elt.hi = lshr i32 %elt.arg, 16
   %elt = trunc i32 %elt.hi to i16
   %vecins = insertelement <2 x i16> %vec, i16 %elt, i32 0
   store <2 x i16> %vecins, <2 x i16> addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @s_insertelement_v2i16_0_reghi_multi_use_1(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(4)* %vec.ptr, i32 %elt.arg) #0 {
 ; GFX9-LABEL: s_insertelement_v2i16_0_reghi_multi_use_1:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; GFX9-NEXT:    s_load_dword s4, s[4:5], 0x10
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_load_dword s2, s[2:3], 0x0
 ; GFX9-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NEXT:    s_lshr_b32 s0, s4, 16
+; GFX9-NEXT:    s_load_dword s0, s[2:3], 0x0
 ; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    s_lshr_b32 s1, s4, 16
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_pack_lh_b32_b16 s1, s0, s2
-; GFX9-NEXT:    v_mov_b32_e32 v2, s1
+; GFX9-NEXT:    s_pack_lh_b32_b16 s0, s1, s0
+; GFX9-NEXT:    v_mov_b32_e32 v2, s0
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use s0
+; GFX9-NEXT:    ; use s1
 ; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; VI-LABEL: s_insertelement_v2i16_0_reghi_multi_use_1:
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; VI-NEXT:    s_load_dword s4, s[4:5], 0x10
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_load_dword s2, s[2:3], 0x0
 ; VI-NEXT:    v_mov_b32_e32 v0, s0
+; VI-NEXT:    s_load_dword s0, s[2:3], 0x0
 ; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    s_lshr_b32 s0, s4, 16
+; VI-NEXT:    s_lshr_b32 s1, s4, 16
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_and_b32 s1, s2, 0xffff0000
-; VI-NEXT:    s_or_b32 s1, s0, s1
-; VI-NEXT:    v_mov_b32_e32 v2, s1
+; VI-NEXT:    s_and_b32 s0, s0, 0xffff0000
+; VI-NEXT:    s_or_b32 s0, s1, s0
+; VI-NEXT:    v_mov_b32_e32 v2, s0
 ; VI-NEXT:    flat_store_dword v[0:1], v2
 ; VI-NEXT:    ;;#ASMSTART
-; VI-NEXT:    ; use s0
+; VI-NEXT:    ; use s1
 ; VI-NEXT:    ;;#ASMEND
 ; VI-NEXT:    s_endpgm
 ;
 ; CI-LABEL: s_insertelement_v2i16_0_reghi_multi_use_1:
 ; CI:       ; %bb.0:
 ; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; CI-NEXT:    s_load_dword s4, s[4:5], 0x4
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
-; CI-NEXT:    s_load_dword s2, s[2:3], 0x0
 ; CI-NEXT:    v_mov_b32_e32 v0, s0
+; CI-NEXT:    s_load_dword s0, s[2:3], 0x0
 ; CI-NEXT:    v_mov_b32_e32 v1, s1
-; CI-NEXT:    s_lshr_b32 s0, s4, 16
+; CI-NEXT:    s_lshr_b32 s1, s4, 16
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
-; CI-NEXT:    s_and_b32 s1, s2, 0xffff0000
-; CI-NEXT:    s_or_b32 s1, s0, s1
-; CI-NEXT:    v_mov_b32_e32 v2, s1
+; CI-NEXT:    s_and_b32 s0, s0, 0xffff0000
+; CI-NEXT:    s_or_b32 s0, s1, s0
+; CI-NEXT:    v_mov_b32_e32 v2, s0
 ; CI-NEXT:    flat_store_dword v[0:1], v2
 ; CI-NEXT:    ;;#ASMSTART
-; CI-NEXT:    ; use s0
+; CI-NEXT:    ; use s1
 ; CI-NEXT:    ;;#ASMEND
 ; CI-NEXT:    s_endpgm
   %vec = load <2 x i16>, <2 x i16> addrspace(4)* %vec.ptr
   %elt.hi = lshr i32 %elt.arg, 16
   %elt = trunc i32 %elt.hi to i16
   %vecins = insertelement <2 x i16> %vec, i16 %elt, i32 0
   store <2 x i16> %vecins, <2 x i16> addrspace(1)* %out
   %use1 = zext i16 %elt to i32
   call void asm sideeffect "; use $0", "s"(i32 %use1) #0
   ret void
 }
 
 define amdgpu_kernel void @s_insertelement_v2i16_0_reghi_both_multi_use_1(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(4)* %vec.ptr, i32 %elt.arg) #0 {
 ; GFX9-LABEL: s_insertelement_v2i16_0_reghi_both_multi_use_1:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; GFX9-NEXT:    s_load_dword s4, s[4:5], 0x10
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_load_dword s2, s[2:3], 0x0
 ; GFX9-NEXT:    v_mov_b32_e32 v0, s0
+; GFX9-NEXT:    s_load_dword s0, s[2:3], 0x0
 ; GFX9-NEXT:    v_mov_b32_e32 v1, s1
-; GFX9-NEXT:    s_lshr_b32 s0, s4, 16
+; GFX9-NEXT:    s_lshr_b32 s1, s4, 16
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_lshr_b32 s1, s2, 16
-; GFX9-NEXT:    s_pack_ll_b32_b16 s2, s0, s1
+; GFX9-NEXT:    s_lshr_b32 s0, s0, 16
+; GFX9-NEXT:    s_pack_ll_b32_b16 s2, s1, s0
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use s0
+; GFX9-NEXT:    ; use s1
 ; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    ;;#ASMSTART
-; GFX9-NEXT:    ; use s1
+; GFX9-NEXT:    ; use s0
 ; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
 ;
 ; VI-LABEL: s_insertelement_v2i16_0_reghi_both_multi_use_1:
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; VI-NEXT:    s_load_dword s4, s[4:5], 0x10
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_load_dword s2, s[2:3], 0x0
 ; VI-NEXT:    v_mov_b32_e32 v0, s0
+; VI-NEXT:    s_load_dword s0, s[2:3], 0x0
 ; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    s_lshr_b32 s0, s4, 16
+; VI-NEXT:    s_lshr_b32 s1, s4, 16
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_lshr_b32 s1, s2, 16
-; VI-NEXT:    s_and_b32 s2, s2, 0xffff0000
-; VI-NEXT:    s_or_b32 s2, s0, s2
-; VI-NEXT:    v_mov_b32_e32 v2, s2
+; VI-NEXT:    s_lshr_b32 s2, s0, 16
+; VI-NEXT:    s_and_b32 s0, s0, 0xffff0000
+; VI-NEXT:    s_or_b32 s0, s1, s0
+; VI-NEXT:    v_mov_b32_e32 v2, s0
 ; VI-NEXT:    flat_store_dword v[0:1], v2
 ; VI-NEXT:    ;;#ASMSTART
-; VI-NEXT:    ; use s0
+; VI-NEXT:    ; use s1
 ; VI-NEXT:    ;;#ASMEND
 ; VI-NEXT:    ;;#ASMSTART
-; VI-NEXT:    ; use s1
+; VI-NEXT:    ; use s2
 ; VI-NEXT:    ;;#ASMEND
 ; VI-NEXT:    s_endpgm
 ;
 ; CI-LABEL: s_insertelement_v2i16_0_reghi_both_multi_use_1:
 ; CI:       ; %bb.0:
 ; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
-; CI-NEXT:    s_load_dword s4, s[4:5], 0x4
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
 ; CI-NEXT:    s_load_dword s2, s[2:3], 0x0
-; CI-NEXT:    v_mov_b32_e32 v1, s1
 ; CI-NEXT:    v_mov_b32_e32 v0, s0
-; CI-NEXT:    s_lshr_b32 s0, s4, 16
+; CI-NEXT:    s_load_dword s0, s[4:5], 0x4
+; CI-NEXT:    v_mov_b32_e32 v1, s1
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
 ; CI-NEXT:    s_lshr_b32 s1, s2, 16
 ; CI-NEXT:    s_lshl_b32 s2, s1, 16
+; CI-NEXT:    s_lshr_b32 s0, s0, 16
 ; CI-NEXT:    s_or_b32 s2, s0, s2
 ; CI-NEXT:    v_mov_b32_e32 v2, s2
 ; CI-NEXT:    flat_store_dword v[0:1], v2
 ; CI-NEXT:    ;;#ASMSTART
 ; CI-NEXT:    ; use s0
 ; CI-NEXT:    ;;#ASMEND
 ; CI-NEXT:    ;;#ASMSTART
 ; CI-NEXT:    ; use s1

 }
 
 define amdgpu_kernel void @s_insertelement_v2i16_1_reg(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(4)* %vec.ptr, [8 x i32], i16 %elt) #0 {
 ; GFX9-LABEL: s_insertelement_v2i16_1_reg:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; GFX9-NEXT:    s_load_dword s4, s[4:5], 0x30
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_load_dword s2, s[2:3], 0x0
 ; GFX9-NEXT:    v_mov_b32_e32 v0, s0
+; GFX9-NEXT:    s_load_dword s0, s[2:3], 0x0
 ; GFX9-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_pack_ll_b32_b16 s0, s2, s4
+; GFX9-NEXT:    s_pack_ll_b32_b16 s0, s0, s4
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s0
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
 ;
 ; VI-LABEL: s_insertelement_v2i16_1_reg:
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; VI-NEXT:    s_load_dword s4, s[4:5], 0x30
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_load_dword s2, s[2:3], 0x0
 ; VI-NEXT:    v_mov_b32_e32 v0, s0
+; VI-NEXT:    s_load_dword s0, s[2:3], 0x0
 ; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    s_lshl_b32 s0, s4, 16
+; VI-NEXT:    s_lshl_b32 s1, s4, 16
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_and_b32 s1, s2, 0xffff
-; VI-NEXT:    s_or_b32 s0, s1, s0
+; VI-NEXT:    s_and_b32 s0, s0, 0xffff
+; VI-NEXT:    s_or_b32 s0, s0, s1
 ; VI-NEXT:    v_mov_b32_e32 v2, s0
 ; VI-NEXT:    flat_store_dword v[0:1], v2
 ; VI-NEXT:    s_endpgm
 ;
 ; CI-LABEL: s_insertelement_v2i16_1_reg:
 ; CI:       ; %bb.0:
 ; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; CI-NEXT:    s_load_dword s4, s[4:5], 0xc
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
-; CI-NEXT:    s_load_dword s2, s[2:3], 0x0
 ; CI-NEXT:    v_mov_b32_e32 v0, s0
+; CI-NEXT:    s_load_dword s0, s[2:3], 0x0
 ; CI-NEXT:    v_mov_b32_e32 v1, s1
 ; CI-NEXT:    s_lshl_b32 s1, s4, 16
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
-; CI-NEXT:    s_and_b32 s0, s2, 0xffff
+; CI-NEXT:    s_and_b32 s0, s0, 0xffff
 ; CI-NEXT:    s_or_b32 s0, s0, s1
 ; CI-NEXT:    v_mov_b32_e32 v2, s0
 ; CI-NEXT:    flat_store_dword v[0:1], v2
 ; CI-NEXT:    s_endpgm
   %vec = load <2 x i16>, <2 x i16> addrspace(4)* %vec.ptr
   %vecins = insertelement <2 x i16> %vec, i16 %elt, i32 1
   store <2 x i16> %vecins, <2 x i16> addrspace(1)* %out
   ret void
; CI-NEXT:    s_endpgm
   store <2 x half> %vecins, <2 x half> addrspace(1)* %out.gep
   ret void
 }
 
 ; FIXME: Enable for others when argument load not split
 define amdgpu_kernel void @s_insertelement_v2i16_dynamic(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(4)* %vec.ptr, i32 addrspace(4)* %idx.ptr) #0 {
 ; GFX9-LABEL: s_insertelement_v2i16_dynamic:
 ; GFX9:       ; %bb.0:
-; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
-; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[4:5], 0x10
+; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x10
+; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[4:5], 0x0
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0x3e703e7
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, s0
-; GFX9-NEXT:    v_mov_b32_e32 v1, s1
-; GFX9-NEXT:    s_load_dword s0, s[4:5], 0x0
-; GFX9-NEXT:    s_load_dword s1, s[2:3], 0x0
+; GFX9-NEXT:    s_load_dword s0, s[0:1], 0x0
+; GFX9-NEXT:    s_load_dword s1, s[6:7], 0x0
+; GFX9-NEXT:    v_mov_b32_e32 v0, s4
+; GFX9-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_lshl_b32 s0, s0, 4
 ; GFX9-NEXT:    s_lshl_b32 s0, 0xffff, s0
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s1
 ; GFX9-NEXT:    v_bfi_b32 v2, s0, v2, v3
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
 ;
 ; VI-LABEL: s_insertelement_v2i16_dynamic:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
-; VI-NEXT:    s_load_dwordx4 s[4:7], s[4:5], 0x10
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x10
+; VI-NEXT:    s_load_dwordx4 s[4:7], s[4:5], 0x0
 ; VI-NEXT:    v_mov_b32_e32 v2, 0x3e703e7
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v0, s0
-; VI-NEXT:    v_mov_b32_e32 v1, s1
-; VI-NEXT:    s_load_dword s0, s[4:5], 0x0
-; VI-NEXT:    s_load_dword s1, s[2:3], 0x0
+; VI-NEXT:    s_load_dword s0, s[0:1], 0x0
+; VI-NEXT:    s_load_dword s1, s[6:7], 0x0
+; VI-NEXT:    v_mov_b32_e32 v0, s4
+; VI-NEXT:    v_mov_b32_e32 v1, s5
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
 ; VI-NEXT:    s_lshl_b32 s0, s0, 4
 ; VI-NEXT:    s_lshl_b32 s0, 0xffff, s0
 ; VI-NEXT:    v_mov_b32_e32 v3, s1
 ; VI-NEXT:    v_bfi_b32 v2, s0, v2, v3
 ; VI-NEXT:    flat_store_dword v[0:1], v2
 ; VI-NEXT:    s_endpgm
 ;
 ; CI-LABEL: s_insertelement_v2i16_dynamic:
 ; CI:       ; %bb.0:
-; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
-; CI-NEXT:    s_load_dwordx4 s[4:7], s[4:5], 0x4
+; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x4
+; CI-NEXT:    s_load_dwordx4 s[4:7], s[4:5], 0x0
 ; CI-NEXT:    v_mov_b32_e32 v2, 0x3e703e7
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
-; CI-NEXT:    v_mov_b32_e32 v0, s0
-; CI-NEXT:    v_mov_b32_e32 v1, s1
-; CI-NEXT:    s_load_dword s0, s[4:5], 0x0
-; CI-NEXT:    s_load_dword s1, s[2:3], 0x0
+; CI-NEXT:    s_load_dword s0, s[0:1], 0x0
+; CI-NEXT:    s_load_dword s1, s[6:7], 0x0
+; CI-NEXT:    v_mov_b32_e32 v0, s4
+; CI-NEXT:    v_mov_b32_e32 v1, s5
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
 ; CI-NEXT:    s_lshl_b32 s0, s0, 4
 ; CI-NEXT:    s_lshl_b32 s0, 0xffff, s0
 ; CI-NEXT:    v_mov_b32_e32 v3, s1
 ; CI-NEXT:    v_bfi_b32 v2, s0, v2, v3
 ; CI-NEXT:    flat_store_dword v[0:1], v2
 ; CI-NEXT:    s_endpgm
   %idx = load volatile i32, i32 addrspace(4)* %idx.ptr
; CI-NEXT:    s_endpgm
   %vecins = insertelement <2 x i16> %vec, i16 999, i32 %idx
   store <2 x i16> %vecins, <2 x i16> addrspace(1)* %out.gep
   ret void
 }
 
 define amdgpu_kernel void @v_insertelement_v2f16_dynamic_vgpr(<2 x half> addrspace(1)* %out, <2 x half> addrspace(1)* %in, i32 addrspace(1)* %idx.ptr) #0 {
 ; GFX9-LABEL: v_insertelement_v2f16_dynamic_vgpr:
 ; GFX9:       ; %bb.0:
-; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
-; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[4:5], 0x10
+; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x10
+; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[4:5], 0x0
 ; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX9-NEXT:    s_mov_b32 s6, 0xffff
-; GFX9-NEXT:    s_mov_b32 s7, 0x12341234
-; GFX9-NEXT:    v_mov_b32_e32 v3, s3
-; GFX9-NEXT:    v_mov_b32_e32 v1, s5
-; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s4, v2
+; GFX9-NEXT:    s_mov_b32 s2, 0xffff
+; GFX9-NEXT:    s_mov_b32 s3, 0x12341234
+; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s0, v2
 ; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
 ; GFX9-NEXT:    global_load_dword v4, v[0:1], off
-; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s2, v2
+; GFX9-NEXT:    v_mov_b32_e32 v3, s7
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s6, v2
 ; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v3, vcc
 ; GFX9-NEXT:    global_load_dword v3, v[0:1], off
-; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s0, v2
-; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s4, v2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s5
 ; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
 ; GFX9-NEXT:    s_waitcnt vmcnt(1)
 ; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 4, v4
-; GFX9-NEXT:    v_lshlrev_b32_e64 v2, v2, s6
+; GFX9-NEXT:    v_lshlrev_b32_e64 v2, v2, s2
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_bfi_b32 v2, v2, s7, v3
+; GFX9-NEXT:    v_bfi_b32 v2, v2, s3, v3
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
 ;
 ; VI-LABEL: v_insertelement_v2f16_dynamic_vgpr:
 ; VI:       ; %bb.0:
-; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
-; VI-NEXT:    s_load_dwordx4 s[4:7], s[4:5], 0x10
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x10
+; VI-NEXT:    s_load_dwordx4 s[4:7], s[4:5], 0x0
 ; VI-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    s_mov_b32 s6, 0xffff
-; VI-NEXT:    s_mov_b32 s7, 0x12341234
-; VI-NEXT:    v_mov_b32_e32 v3, s3
-; VI-NEXT:    v_mov_b32_e32 v1, s5
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v2
+; VI-NEXT:    s_mov_b32 s2, 0xffff
+; VI-NEXT:    s_mov_b32 s3, 0x12341234
+; VI-NEXT:    v_mov_b32_e32 v1, s1
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v2
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v4, v[0:1]
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s2, v2
+; VI-NEXT:    v_mov_b32_e32 v3, s7
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s6, v2
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v3, vcc
 ; VI-NEXT:    flat_load_dword v3, v[0:1]
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v2
-; VI-NEXT:    v_mov_b32_e32 v1, s1
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s4, v2
+; VI-NEXT:    v_mov_b32_e32 v1, s5
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    s_waitcnt vmcnt(1) lgkmcnt(1)
 ; VI-NEXT:    v_lshlrev_b32_e32 v2, 4, v4
-; VI-NEXT:    v_lshlrev_b32_e64 v2, v2, s6
+; VI-NEXT:    v_lshlrev_b32_e64 v2, v2, s2
 ; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
-; VI-NEXT:    v_bfi_b32 v2, v2, s7, v3
+; VI-NEXT:    v_bfi_b32 v2, v2, s3, v3
 ; VI-NEXT:    flat_store_dword v[0:1], v2
 ; VI-NEXT:    s_endpgm
 ;
 ; CI-LABEL: v_insertelement_v2f16_dynamic_vgpr:
 ; CI:       ; %bb.0:
-; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
-; CI-NEXT:    s_load_dwordx4 s[4:7], s[4:5], 0x4
+; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x4
+; CI-NEXT:    s_load_dwordx4 s[4:7], s[4:5], 0x0
 ; CI-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
-; CI-NEXT:    s_mov_b32 s6, 0x12341234
-; CI-NEXT:    v_mov_b32_e32 v3, s3
-; CI-NEXT:    v_mov_b32_e32 v1, s5
-; CI-NEXT:    v_add_i32_e32 v0, vcc, s4, v2
+; CI-NEXT:    s_mov_b32 s2, 0x12341234
+; CI-NEXT:    v_mov_b32_e32 v1, s1
+; CI-NEXT:    v_add_i32_e32 v0, vcc, s0, v2
 ; CI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; CI-NEXT:    flat_load_dword v4, v[0:1]
-; CI-NEXT:    v_add_i32_e32 v0, vcc, s2, v2
+; CI-NEXT:    v_mov_b32_e32 v3, s7
+; CI-NEXT:    v_add_i32_e32 v0, vcc, s6, v2
 ; CI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v3, vcc
 ; CI-NEXT:    flat_load_dword v3, v[0:1]
-; CI-NEXT:    v_add_i32_e32 v0, vcc, s0, v2
-; CI-NEXT:    v_mov_b32_e32 v1, s1
+; CI-NEXT:    v_add_i32_e32 v0, vcc, s4, v2
+; CI-NEXT:    v_mov_b32_e32 v1, s5
 ; CI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; CI-NEXT:    s_waitcnt vmcnt(1) lgkmcnt(1)
 ; CI-NEXT:    v_lshlrev_b32_e32 v2, 4, v4
 ; CI-NEXT:    v_lshl_b32_e32 v2, 0xffff, v2
 ; CI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
-; CI-NEXT:    v_bfi_b32 v2, v2, s6, v3
+; CI-NEXT:    v_bfi_b32 v2, v2, s2, v3
 ; CI-NEXT:    flat_store_dword v[0:1], v2
 ; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x() #1
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <2 x half>, <2 x half> addrspace(1)* %in, i64 %tid.ext
   %idx.gep = getelementptr inbounds i32, i32 addrspace(1)* %idx.ptr, i64 %tid.ext
   %out.gep = getelementptr inbounds <2 x half>, <2 x half> addrspace(1)* %out, i64 %tid.ext
   %idx = load i32, i32 addrspace(1)* %idx.gep
; CI-NEXT:    s_endpgm
   ret void
 }
 
 ; FIXME: Better code on CI?
 define amdgpu_kernel void @v_insertelement_v4i16_dynamic_vgpr(<4 x i16> addrspace(1)* %out, <4 x i16> addrspace(1)* %in, i32 %val) #0 {
 ; GFX9-LABEL: v_insertelement_v4i16_dynamic_vgpr:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
-; GFX9-NEXT:    global_load_dword v4, v[0:1], off
-; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 3, v0
+; GFX9-NEXT:    global_load_dword v2, v[0:1], off
+; GFX9-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
 ; GFX9-NEXT:    s_load_dword s6, s[4:5], 0x10
 ; GFX9-NEXT:    s_mov_b32 s5, 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_mov_b32_e32 v1, s3
-; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s2, v2
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s2, v4
 ; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
 ; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[0:1], off
 ; GFX9-NEXT:    s_mov_b32 s4, 0xffff
-; GFX9-NEXT:    v_mov_b32_e32 v3, s1
-; GFX9-NEXT:    v_add_co_u32_e32 v2, vcc, s0, v2
+; GFX9-NEXT:    v_mov_b32_e32 v5, s1
+; GFX9-NEXT:    v_add_co_u32_e32 v4, vcc, s0, v4
 ; GFX9-NEXT:    s_pack_ll_b32_b16 s1, s6, s6
-; GFX9-NEXT:    v_addc_co_u32_e32 v3, vcc, 0, v3, vcc
+; GFX9-NEXT:    v_addc_co_u32_e32 v5, vcc, 0, v5, vcc
 ; GFX9-NEXT:    s_waitcnt vmcnt(1)
-; GFX9-NEXT:    v_lshlrev_b32_e32 v4, 4, v4
-; GFX9-NEXT:    v_lshlrev_b64 v[4:5], v4, s[4:5]
+; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 4, v2
+; GFX9-NEXT:    v_lshlrev_b64 v[2:3], v2, s[4:5]
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_bfi_b32 v1, v5, s1, v1
-; GFX9-NEXT:    v_bfi_b32 v0, v4, s1, v0
-; GFX9-NEXT:    global_store_dwordx2 v[2:3], v[0:1], off
+; GFX9-NEXT:    v_bfi_b32 v1, v3, s1, v1
+; GFX9-NEXT:    v_bfi_b32 v0, v2, s1, v0
+; GFX9-NEXT:    global_store_dwordx2 v[4:5], v[0:1], off
 ; GFX9-NEXT:    s_endpgm
 ;
 ; VI-LABEL: v_insertelement_v4i16_dynamic_vgpr:
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
 ; VI-NEXT:    flat_load_dword v4, v[0:1]
 ; VI-NEXT:    v_lshlrev_b32_e32 v2, 3, v0

 ; CI-NEXT:    v_mov_b32_e32 v1, s3
 ; CI-NEXT:    v_add_i32_e32 v0, vcc, s2, v2
 ; CI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; CI-NEXT:    flat_load_dwordx2 v[0:1], v[0:1]
 ; CI-NEXT:    s_mov_b32 s5, 0
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
 ; CI-NEXT:    s_lshl_b32 s2, s6, 16
 ; CI-NEXT:    s_and_b32 s3, s6, s4
-; CI-NEXT:    v_mov_b32_e32 v3, s1
 ; CI-NEXT:    v_add_i32_e32 v2, vcc, s0, v2
-; CI-NEXT:    s_or_b32 s1, s3, s2
+; CI-NEXT:    v_mov_b32_e32 v3, s1
+; CI-NEXT:    s_or_b32 s0, s3, s2
 ; CI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v3, vcc
 ; CI-NEXT:    s_waitcnt vmcnt(1)
 ; CI-NEXT:    v_lshlrev_b32_e32 v4, 4, v4
 ; CI-NEXT:    v_lshl_b64 v[4:5], s[4:5], v4
 ; CI-NEXT:    s_waitcnt vmcnt(0)
-; CI-NEXT:    v_bfi_b32 v1, v5, s1, v1
-; CI-NEXT:    v_bfi_b32 v0, v4, s1, v0
+; CI-NEXT:    v_bfi_b32 v1, v5, s0, v1
+; CI-NEXT:    v_bfi_b32 v0, v4, s0, v0
 ; CI-NEXT:    flat_store_dwordx2 v[2:3], v[0:1]
 ; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x() #1
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <4 x i16>, <4 x i16> addrspace(1)* %in, i64 %tid.ext
   %out.gep = getelementptr inbounds <4 x i16>, <4 x i16> addrspace(1)* %out, i64 %tid.ext
   %idx.val = load volatile i32, i32 addrspace(1)* undef
   %vec = load <4 x i16>, <4 x i16> addrspace(1)* %in.gep

llvm/test/CodeGen/AMDGPU/llvm.amdgcn.s.barrier.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck --check-prefixes=GCN,GFX8,VARIANT0 %s
 ; RUN: llc -march=amdgcn -mattr=+auto-waitcnt-before-barrier -verify-machineinstrs < %s | FileCheck --check-prefixes=GCN,GFX8,VARIANT1 %s
 ; RUN: llc -march=amdgcn -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck --check-prefixes=GCN,GFX9,VARIANT2 %s
 ; RUN: llc -march=amdgcn -mcpu=gfx900 -mattr=+auto-waitcnt-before-barrier -verify-machineinstrs < %s | FileCheck --check-prefixes=GCN,GFX9,VARIANT3 %s
 
 define amdgpu_kernel void @test_barrier(i32 addrspace(1)* %out, i32 %size) #0 {
 ; VARIANT0-LABEL: test_barrier:
 ; VARIANT0:       ; %bb.0: ; %entry
 ; VARIANT0-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
 ; VARIANT0-NEXT:    s_load_dword s2, s[0:1], 0xb
-; VARIANT0-NEXT:    v_not_b32_e32 v3, v0
 ; VARIANT0-NEXT:    s_mov_b32 s7, 0xf000
 ; VARIANT0-NEXT:    s_mov_b32 s6, 0
 ; VARIANT0-NEXT:    v_lshlrev_b32_e32 v1, 2, v0
 ; VARIANT0-NEXT:    v_mov_b32_e32 v2, 0
+; VARIANT0-NEXT:    v_not_b32_e32 v3, v0
 ; VARIANT0-NEXT:    s_waitcnt lgkmcnt(0)
 ; VARIANT0-NEXT:    buffer_store_dword v0, v[1:2], s[4:7], 0 addr64
 ; VARIANT0-NEXT:    s_waitcnt vmcnt(0) expcnt(0)
 ; VARIANT0-NEXT:    s_barrier
 ; VARIANT0-NEXT:    v_add_i32_e32 v3, vcc, s2, v3
 ; VARIANT0-NEXT:    v_ashrrev_i32_e32 v4, 31, v3
 ; VARIANT0-NEXT:    v_lshl_b64 v[3:4], v[3:4], 2
 ; VARIANT0-NEXT:    buffer_load_dword v0, v[3:4], s[4:7], 0 addr64
 ; VARIANT0-NEXT:    s_waitcnt vmcnt(0)
 ; VARIANT0-NEXT:    buffer_store_dword v0, v[1:2], s[4:7], 0 addr64
 ; VARIANT0-NEXT:    s_endpgm
 ;
 ; VARIANT1-LABEL: test_barrier:
 ; VARIANT1:       ; %bb.0: ; %entry
 ; VARIANT1-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
 ; VARIANT1-NEXT:    s_load_dword s2, s[0:1], 0xb
-; VARIANT1-NEXT:    v_not_b32_e32 v3, v0
 ; VARIANT1-NEXT:    s_mov_b32 s7, 0xf000
 ; VARIANT1-NEXT:    s_mov_b32 s6, 0
 ; VARIANT1-NEXT:    v_lshlrev_b32_e32 v1, 2, v0
 ; VARIANT1-NEXT:    v_mov_b32_e32 v2, 0
+; VARIANT1-NEXT:    v_not_b32_e32 v3, v0
 ; VARIANT1-NEXT:    s_waitcnt lgkmcnt(0)
 ; VARIANT1-NEXT:    buffer_store_dword v0, v[1:2], s[4:7], 0 addr64
 ; VARIANT1-NEXT:    s_barrier
 ; VARIANT1-NEXT:    v_add_i32_e32 v3, vcc, s2, v3
 ; VARIANT1-NEXT:    v_ashrrev_i32_e32 v4, 31, v3
 ; VARIANT1-NEXT:    v_lshl_b64 v[3:4], v[3:4], 2
 ; VARIANT1-NEXT:    s_waitcnt expcnt(0)
 ; VARIANT1-NEXT:    buffer_load_dword v0, v[3:4], s[4:7], 0 addr64

llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ubfe.ll

 
 ; Make sure that SimplifyDemandedBits doesn't cause the and to be
 ; reduced to the bits demanded by the bfe.
 
 ; XXX: The operand to v_bfe_u32 could also just directly be the load register.
 define amdgpu_kernel void @simplify_bfe_u32_multi_use_arg(i32 addrspace(1)* %out0,
 ; SI-LABEL: simplify_bfe_u32_multi_use_arg:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_mov_b32 s3, 0xf000
 ; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0xd
+; SI-NEXT:    s_mov_b32 s3, 0xf000
 ; SI-NEXT:    s_mov_b32 s2, -1
 ; SI-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0x9
 ; SI-NEXT:    s_mov_b32 s6, s2
 ; SI-NEXT:    s_mov_b32 s7, s3
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
 ; SI-NEXT:    buffer_load_dword v0, off, s[4:7], 0
 ; SI-NEXT:    s_mov_b32 s0, s10
 ; SI-NEXT:    s_mov_b32 s1, s11

llvm/test/CodeGen/AMDGPU/llvm.round.f64.ll

; CI-NEXT:    s_endpgm
   %result = call double @llvm.round.f64(double %x) #1
   store double %result, double addrspace(1)* %out.gep
   ret void
 }
 
 define amdgpu_kernel void @round_v2f64(<2 x double> addrspace(1)* %out, <2 x double> %in) #0 {
 ; SI-LABEL: round_v2f64:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
 ; SI-NEXT:    s_load_dwordx4 s[8:11], s[0:1], 0xd
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
 ; SI-NEXT:    s_mov_b32 s6, -1
 ; SI-NEXT:    s_movk_i32 s18, 0xfc01
 ; SI-NEXT:    s_mov_b32 s3, 0xfffff
-; SI-NEXT:    s_mov_b32 s2, s6
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
 ; SI-NEXT:    s_bfe_u32 s0, s11, 0xb0014
 ; SI-NEXT:    s_add_i32 s19, s0, s18
+; SI-NEXT:    s_mov_b32 s2, s6
 ; SI-NEXT:    s_lshr_b64 s[0:1], s[2:3], s19
 ; SI-NEXT:    s_and_b64 s[12:13], s[10:11], s[0:1]
 ; SI-NEXT:    s_mov_b32 s15, 0x80000
 ; SI-NEXT:    s_mov_b32 s14, 0
 ; SI-NEXT:    s_lshr_b64 s[16:17], s[14:15], s19
 ; SI-NEXT:    v_cmp_ne_u64_e64 vcc, s[12:13], 0
 ; SI-NEXT:    v_mov_b32_e32 v1, s16
 ; SI-NEXT:    v_mov_b32_e32 v0, s17
; CI-NEXT:    s_endpgm
   %result = call <2 x double> @llvm.round.v2f64(<2 x double> %in) #1
   store <2 x double> %result, <2 x double> addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @round_v4f64(<4 x double> addrspace(1)* %out, <4 x double> %in) #0 {
 ; SI-LABEL: round_v4f64:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
 ; SI-NEXT:    s_load_dwordx8 s[8:15], s[0:1], 0x11
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
 ; SI-NEXT:    s_mov_b32 s6, -1
 ; SI-NEXT:    s_movk_i32 s22, 0xfc01
 ; SI-NEXT:    s_mov_b32 s3, 0xfffff
-; SI-NEXT:    s_mov_b32 s2, s6
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
 ; SI-NEXT:    s_bfe_u32 s0, s11, 0xb0014
 ; SI-NEXT:    s_add_i32 s23, s0, s22
+; SI-NEXT:    s_mov_b32 s2, s6
 ; SI-NEXT:    s_lshr_b64 s[0:1], s[2:3], s23
 ; SI-NEXT:    s_and_b64 s[16:17], s[10:11], s[0:1]
 ; SI-NEXT:    s_mov_b32 s19, 0x80000
 ; SI-NEXT:    s_mov_b32 s18, 0
 ; SI-NEXT:    s_lshr_b64 s[20:21], s[18:19], s23
 ; SI-NEXT:    v_cmp_ne_u64_e64 vcc, s[16:17], 0
 ; SI-NEXT:    v_mov_b32_e32 v1, s20
 ; SI-NEXT:    v_mov_b32_e32 v0, s21

 ; SI-NEXT:    s_mov_b32 s7, 0xf000
 ; SI-NEXT:    v_cndmask_b32_e64 v4, v4, v8, s[0:1]
 ; SI-NEXT:    buffer_store_dwordx4 v[4:7], off, s[4:7], 0 offset:16
 ; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; CI-LABEL: round_v4f64:
 ; CI:       ; %bb.0:
-; CI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
-; CI-NEXT:    s_load_dwordx8 s[8:15], s[0:1], 0x11
-; CI-NEXT:    s_brev_b32 s2, -2
+; CI-NEXT:    s_load_dwordx8 s[4:11], s[0:1], 0x11
+; CI-NEXT:    s_brev_b32 s12, -2
 ; CI-NEXT:    v_mov_b32_e32 v12, 0x3ff00000
-; CI-NEXT:    s_mov_b32 s7, 0xf000
-; CI-NEXT:    s_mov_b32 s6, -1
+; CI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
+; CI-NEXT:    s_mov_b32 s3, 0xf000
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
-; CI-NEXT:    v_trunc_f64_e32 v[0:1], s[10:11]
-; CI-NEXT:    v_mov_b32_e32 v4, s11
-; CI-NEXT:    v_add_f64 v[2:3], s[10:11], -v[0:1]
-; CI-NEXT:    v_bfi_b32 v4, s2, v12, v4
+; CI-NEXT:    v_trunc_f64_e32 v[0:1], s[6:7]
+; CI-NEXT:    v_mov_b32_e32 v4, s7
+; CI-NEXT:    v_add_f64 v[2:3], s[6:7], -v[0:1]
+; CI-NEXT:    v_bfi_b32 v4, s12, v12, v4
 ; CI-NEXT:    v_cmp_ge_f64_e64 vcc, |v[2:3]|, 0.5
-; CI-NEXT:    v_trunc_f64_e32 v[8:9], s[8:9]
+; CI-NEXT:    v_trunc_f64_e32 v[8:9], s[4:5]
 ; CI-NEXT:    v_cndmask_b32_e32 v3, 0, v4, vcc
 ; CI-NEXT:    v_mov_b32_e32 v2, 0
 ; CI-NEXT:    v_add_f64 v[2:3], v[0:1], v[2:3]
-; CI-NEXT:    v_add_f64 v[0:1], s[8:9], -v[8:9]
-; CI-NEXT:    v_mov_b32_e32 v4, s9
+; CI-NEXT:    v_add_f64 v[0:1], s[4:5], -v[8:9]
+; CI-NEXT:    v_mov_b32_e32 v4, s5
 ; CI-NEXT:    v_cmp_ge_f64_e64 vcc, |v[0:1]|, 0.5
-; CI-NEXT:    v_bfi_b32 v4, s2, v12, v4
+; CI-NEXT:    v_bfi_b32 v4, s12, v12, v4
 ; CI-NEXT:    v_cndmask_b32_e32 v1, 0, v4, vcc
-; CI-NEXT:    v_trunc_f64_e32 v[4:5], s[14:15]
-; CI-NEXT:    v_mov_b32_e32 v10, s15
-; CI-NEXT:    v_add_f64 v[6:7], s[14:15], -v[4:5]
-; CI-NEXT:    v_bfi_b32 v10, s2, v12, v10
+; CI-NEXT:    v_trunc_f64_e32 v[4:5], s[10:11]
+; CI-NEXT:    v_mov_b32_e32 v10, s11
+; CI-NEXT:    v_add_f64 v[6:7], s[10:11], -v[4:5]
+; CI-NEXT:    v_bfi_b32 v10, s12, v12, v10
 ; CI-NEXT:    v_cmp_ge_f64_e64 vcc, |v[6:7]|, 0.5
 ; CI-NEXT:    v_mov_b32_e32 v6, 0
 ; CI-NEXT:    v_cndmask_b32_e32 v7, 0, v10, vcc
-; CI-NEXT:    v_trunc_f64_e32 v[10:11], s[12:13]
+; CI-NEXT:    v_trunc_f64_e32 v[10:11], s[8:9]
 ; CI-NEXT:    v_add_f64 v[6:7], v[4:5], v[6:7]
-; CI-NEXT:    v_add_f64 v[4:5], s[12:13], -v[10:11]
-; CI-NEXT:    v_mov_b32_e32 v13, s13
+; CI-NEXT:    v_add_f64 v[4:5], s[8:9], -v[10:11]
+; CI-NEXT:    v_mov_b32_e32 v13, s9
 ; CI-NEXT:    v_cmp_ge_f64_e64 vcc, |v[4:5]|, 0.5
-; CI-NEXT:    v_bfi_b32 v12, s2, v12, v13
+; CI-NEXT:    v_bfi_b32 v12, s12, v12, v13
 ; CI-NEXT:    v_mov_b32_e32 v0, 0
 ; CI-NEXT:    v_cndmask_b32_e32 v5, 0, v12, vcc
 ; CI-NEXT:    v_mov_b32_e32 v4, 0
 ; CI-NEXT:    v_add_f64 v[4:5], v[10:11], v[4:5]
 ; CI-NEXT:    v_add_f64 v[0:1], v[8:9], v[0:1]
-; CI-NEXT:    buffer_store_dwordx4 v[4:7], off, s[4:7], 0 offset:16
-; CI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
+; CI-NEXT:    s_mov_b32 s2, -1
+; CI-NEXT:    buffer_store_dwordx4 v[4:7], off, s[0:3], 0 offset:16
+; CI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
 ; CI-NEXT:    s_endpgm
   %result = call <4 x double> @llvm.round.v4f64(<4 x double> %in) #1
   store <4 x double> %result, <4 x double> addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @round_v8f64(<8 x double> addrspace(1)* %out, <8 x double> %in) #0 {
 ; SI-LABEL: round_v8f64:
 ; SI:       ; %bb.0:
-; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
 ; SI-NEXT:    s_load_dwordx16 s[8:23], s[0:1], 0x19
+; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
 ; SI-NEXT:    s_mov_b32 s6, -1
 ; SI-NEXT:    s_movk_i32 s30, 0xfc01
 ; SI-NEXT:    s_mov_b32 s3, 0xfffff
-; SI-NEXT:    s_mov_b32 s2, s6
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
 ; SI-NEXT:    s_bfe_u32 s0, s11, 0xb0014
 ; SI-NEXT:    s_add_i32 s31, s0, s30
+; SI-NEXT:    s_mov_b32 s2, s6
 ; SI-NEXT:    s_lshr_b64 s[0:1], s[2:3], s31
 ; SI-NEXT:    s_and_b64 s[26:27], s[10:11], s[0:1]
 ; SI-NEXT:    s_mov_b32 s25, 0x80000
 ; SI-NEXT:    s_mov_b32 s24, 0
 ; SI-NEXT:    s_lshr_b64 s[28:29], s[24:25], s31
 ; SI-NEXT:    v_cmp_ne_u64_e64 vcc, s[26:27], 0
 ; SI-NEXT:    v_mov_b32_e32 v1, s28
 ; SI-NEXT:    v_mov_b32_e32 v0, s29

 ; SI-NEXT:    buffer_store_dwordx4 v[12:15], off, s[4:7], 0 offset:48
 ; SI-NEXT:    buffer_store_dwordx4 v[8:11], off, s[4:7], 0 offset:32
 ; SI-NEXT:    buffer_store_dwordx4 v[4:7], off, s[4:7], 0 offset:16
 ; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; CI-LABEL: round_v8f64:
 ; CI:       ; %bb.0:
-; CI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
 ; CI-NEXT:    s_load_dwordx16 s[8:23], s[0:1], 0x19
 ; CI-NEXT:    s_brev_b32 s2, -2
 ; CI-NEXT:    v_mov_b32_e32 v16, 0x3ff00000
+; CI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
 ; CI-NEXT:    s_mov_b32 s7, 0xf000
-; CI-NEXT:    s_mov_b32 s6, -1
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
 ; CI-NEXT:    v_trunc_f64_e32 v[0:1], s[10:11]
 ; CI-NEXT:    v_mov_b32_e32 v4, s11
 ; CI-NEXT:    v_add_f64 v[2:3], s[10:11], -v[0:1]
 ; CI-NEXT:    v_bfi_b32 v4, s2, v16, v4
 ; CI-NEXT:    v_cmp_ge_f64_e64 vcc, |v[2:3]|, 0.5
 ; CI-NEXT:    v_mov_b32_e32 v2, 0
 ; CI-NEXT:    v_cndmask_b32_e32 v3, 0, v4, vcc

 ; CI-NEXT:    v_mov_b32_e32 v14, 0
 ; CI-NEXT:    v_mov_b32_e32 v17, s17
 ; CI-NEXT:    v_add_f64 v[8:9], v[8:9], v[14:15]
 ; CI-NEXT:    v_add_f64 v[14:15], s[16:17], -v[12:13]
 ; CI-NEXT:    v_bfi_b32 v19, s2, v16, v17
 ; CI-NEXT:    v_trunc_f64_e32 v[16:17], s[18:19]
 ; CI-NEXT:    v_cmp_ge_f64_e64 vcc, |v[14:15]|, 0.5
 ; CI-NEXT:    v_add_f64 v[14:15], s[18:19], -v[16:17]
+; CI-NEXT:    s_mov_b32 s6, -1
 ; CI-NEXT:    v_cmp_ge_f64_e64 s[0:1], |v[14:15]|, 0.5
 ; CI-NEXT:    v_mov_b32_e32 v14, 0
 ; CI-NEXT:    v_cndmask_b32_e64 v15, 0, v18, s[0:1]
 ; CI-NEXT:    v_add_f64 v[14:15], v[16:17], v[14:15]
 ; CI-NEXT:    v_cndmask_b32_e32 v17, 0, v19, vcc
 ; CI-NEXT:    v_mov_b32_e32 v16, 0
 ; CI-NEXT:    v_add_f64 v[12:13], v[12:13], v[16:17]
 ; CI-NEXT:    buffer_store_dwordx4 v[8:11], off, s[4:7], 0 offset:48

llvm/test/CodeGen/AMDGPU/lshr.v2i16.ll

 ; VI-NEXT:    v_lshrrev_b16_sdwa v3, v4, v3 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
 ; VI-NEXT:    v_or_b32_e32 v2, v2, v3
 ; VI-NEXT:    flat_store_dword v[0:1], v2
 ; VI-NEXT:    s_endpgm
 ;
 ; CI-LABEL: lshr_v_s_v2i16:
 ; CI:       ; %bb.0:
 ; CI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; CI-NEXT:    s_load_dword s8, s[0:1], 0xd
-; CI-NEXT:    s_mov_b32 s3, 0xf000
-; CI-NEXT:    s_mov_b32 s2, 0
+; CI-NEXT:    s_mov_b32 s11, 0xf000
+; CI-NEXT:    s_mov_b32 s10, 0
 ; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; CI-NEXT:    s_waitcnt lgkmcnt(0)
-; CI-NEXT:    s_mov_b64 s[0:1], s[6:7]
 ; CI-NEXT:    v_mov_b32_e32 v1, 0
-; CI-NEXT:    buffer_load_dword v2, v[0:1], s[0:3], 0 addr64
-; CI-NEXT:    s_lshr_b32 s9, s8, 16
-; CI-NEXT:    s_mov_b32 s10, 0xffff
-; CI-NEXT:    s_and_b32 s8, s8, s10
-; CI-NEXT:    s_mov_b64 s[6:7], s[2:3]
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[8:9], s[6:7]
+; CI-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; CI-NEXT:    s_load_dword s0, s[0:1], 0xd
+; CI-NEXT:    s_mov_b32 s1, 0xffff
+; CI-NEXT:    s_mov_b64 s[6:7], s[10:11]
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_lshr_b32 s2, s0, 16
+; CI-NEXT:    s_and_b32 s0, s0, s1
 ; CI-NEXT:    s_waitcnt vmcnt(0)
 ; CI-NEXT:    v_lshrrev_b32_e32 v3, 16, v2
-; CI-NEXT:    v_and_b32_e32 v2, s10, v2
-; CI-NEXT:    v_lshrrev_b32_e32 v3, s9, v3
-; CI-NEXT:    v_lshrrev_b32_e32 v2, s8, v2
+; CI-NEXT:    v_and_b32_e32 v2, s1, v2
+; CI-NEXT:    v_lshrrev_b32_e32 v3, s2, v3
+; CI-NEXT:    v_lshrrev_b32_e32 v2, s0, v2
 ; CI-NEXT:    v_lshlrev_b32_e32 v3, 16, v3
 ; CI-NEXT:    v_or_b32_e32 v2, v2, v3
 ; CI-NEXT:    buffer_store_dword v2, v[0:1], s[4:7], 0 addr64
 ; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
   %out.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %out, i64 %tid.ext

 ; VI-NEXT:    v_lshrrev_b16_sdwa v3, v3, v4 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; VI-NEXT:    v_or_b32_e32 v2, v2, v3
 ; VI-NEXT:    flat_store_dword v[0:1], v2
 ; VI-NEXT:    s_endpgm
 ;
 ; CI-LABEL: lshr_s_v_v2i16:
 ; CI:       ; %bb.0:
 ; CI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; CI-NEXT:    s_load_dword s8, s[0:1], 0xd
-; CI-NEXT:    s_mov_b32 s3, 0xf000
-; CI-NEXT:    s_mov_b32 s2, 0
+; CI-NEXT:    s_mov_b32 s11, 0xf000
+; CI-NEXT:    s_mov_b32 s10, 0
 ; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; CI-NEXT:    s_waitcnt lgkmcnt(0)
-; CI-NEXT:    s_mov_b64 s[0:1], s[6:7]
 ; CI-NEXT:    v_mov_b32_e32 v1, 0
-; CI-NEXT:    buffer_load_dword v2, v[0:1], s[0:3], 0 addr64
-; CI-NEXT:    s_lshr_b32 s9, s8, 16
-; CI-NEXT:    s_mov_b32 s10, 0xffff
-; CI-NEXT:    s_and_b32 s8, s8, s10
-; CI-NEXT:    s_mov_b64 s[6:7], s[2:3]
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[8:9], s[6:7]
+; CI-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; CI-NEXT:    s_load_dword s0, s[0:1], 0xd
+; CI-NEXT:    s_mov_b32 s1, 0xffff
+; CI-NEXT:    s_mov_b64 s[6:7], s[10:11]
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_lshr_b32 s2, s0, 16
+; CI-NEXT:    s_and_b32 s0, s0, s1
 ; CI-NEXT:    s_waitcnt vmcnt(0)
 ; CI-NEXT:    v_lshrrev_b32_e32 v3, 16, v2
-; CI-NEXT:    v_and_b32_e32 v2, s10, v2
-; CI-NEXT:    v_lshr_b32_e32 v3, s9, v3
-; CI-NEXT:    v_lshr_b32_e32 v2, s8, v2
+; CI-NEXT:    v_and_b32_e32 v2, s1, v2
+; CI-NEXT:    v_lshr_b32_e32 v3, s2, v3
+; CI-NEXT:    v_lshr_b32_e32 v2, s0, v2
 ; CI-NEXT:    v_lshlrev_b32_e32 v3, 16, v3
 ; CI-NEXT:    v_or_b32_e32 v2, v2, v3
 ; CI-NEXT:    buffer_store_dword v2, v[0:1], s[4:7], 0 addr64
 ; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
   %out.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %out, i64 %tid.ext

llvm/test/CodeGen/AMDGPU/madak.ll

 ; RUN: llc -march=amdgcn -mcpu=tahiti -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN,GFX6,GFX6_8_9,MAD %s
 ; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN,GFX8,GFX6_8_9,GFX8_9,GFX8_9_10,MAD %s
 ; RUN: llc -march=amdgcn -mcpu=gfx900 -verify-machineinstrs  -amdgpu-enable-global-sgpr-addr < %s | FileCheck -check-prefixes=GCN,GFX9,GFX6_8_9,GFX8_9,GFX8_9_10,MAD %s
 ; RUN: llc -march=amdgcn -mcpu=gfx1010 -verify-machineinstrs -amdgpu-enable-global-sgpr-addr < %s | FileCheck -check-prefixes=GCN,GFX10,GFX8_9_10,GFX10-MAD %s
 ; RUN: llc -march=amdgcn -mcpu=gfx1010 -verify-machineinstrs -fp-contract=fast -amdgpu-enable-global-sgpr-addr < %s | FileCheck -check-prefixes=GCN,GFX10,GFX8_9_10,FMA %s
 
 declare i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
 declare float @llvm.fabs.f32(float) nounwind readnone
 
 ; GCN-LABEL: {{^}}madak_f32:
 ; GFX6:   buffer_load_dword [[VA:v[0-9]+]]
 ; GFX6:   buffer_load_dword [[VB:v[0-9]+]]
 ; GFX8: {{flat|global}}_load_dword [[VA:v[0-9]+]]
 ; GFX8: {{flat|global}}_load_dword [[VB:v[0-9]+]]
 ; GFX9: {{flat|global}}_load_dword [[VA:v[0-9]+]]
 ; GFX9: {{flat|global}}_load_dword [[VB:v[0-9]+]]
-; GFX10: {{flat|global}}_load_dword [[VA:v[0-9]+]]
 ; GFX10: {{flat|global}}_load_dword [[VB:v[0-9]+]]
+; GFX10: {{flat|global}}_load_dword [[VA:v[0-9]+]]
 ; MAD:   v_madak_f32 {{v[0-9]+}}, [[VA]], [[VB]], 0x41200000
 ; GFX10-MAD:   v_madak_f32 {{v[0-9]+}}, [[VA]], [[VB]], 0x41200000
 ; FMA:   v_fmaak_f32 {{v[0-9]+}}, [[VA]], [[VB]], 0x41200000
 define amdgpu_kernel void @madak_f32(float addrspace(1)* noalias %out, float addrspace(1)* noalias %in.a, float addrspace(1)* noalias %in.b) nounwind {
   %tid = tail call i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
   %in.a.gep = getelementptr float, float addrspace(1)* %in.a, i32 %tid
   %in.b.gep = getelementptr float, float addrspace(1)* %in.b, i32 %tid
   %out.gep = getelementptr float, float addrspace(1)* %out, i32 %tid

 
 ; GCN-LABEL: {{^}}madak_inline_imm_f32:
 ; GFX6:   buffer_load_dword [[VA:v[0-9]+]]
 ; GFX6:   buffer_load_dword [[VB:v[0-9]+]]
 ; GFX8: {{flat|global}}_load_dword [[VA:v[0-9]+]]
 ; GFX8: {{flat|global}}_load_dword [[VB:v[0-9]+]]
 ; GFX9: {{flat|global}}_load_dword [[VA:v[0-9]+]]
 ; GFX9: {{flat|global}}_load_dword [[VB:v[0-9]+]]
-; GFX10: {{flat|global}}_load_dword [[VA:v[0-9]+]]
 ; GFX10: {{flat|global}}_load_dword [[VB:v[0-9]+]]
+; GFX10: {{flat|global}}_load_dword [[VA:v[0-9]+]]
 ; MAD:   v_mad_f32 {{v[0-9]+}}, [[VA]], [[VB]], 4.0
 ; GFX10-MAD:   v_mad_f32 {{v[0-9]+}}, [[VA]], [[VB]], 4.0
 ; FMA:   v_fma_f32 {{v[0-9]+}}, [[VA]], [[VB]], 4.0
 define amdgpu_kernel void @madak_inline_imm_f32(float addrspace(1)* noalias %out, float addrspace(1)* noalias %in.a, float addrspace(1)* noalias %in.b) nounwind {
   %tid = tail call i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
   %in.a.gep = getelementptr float, float addrspace(1)* %in.a, i32 %tid
   %in.b.gep = getelementptr float, float addrspace(1)* %in.b, i32 %tid
   %out.gep = getelementptr float, float addrspace(1)* %out, i32 %tid

llvm/test/CodeGen/AMDGPU/memory_clause.ll

 ; GCN-NEXT:    s_load_dwordx4 s[0:3], s[16:17], 0x0
 ; GCN-NEXT:    s_load_dwordx4 s[4:7], s[16:17], 0x10
 ; GCN-NEXT:    s_load_dwordx4 s[8:11], s[16:17], 0x20
 ; GCN-NEXT:    s_load_dwordx4 s[12:15], s[16:17], 0x30
 ; GCN-NEXT:    v_mov_b32_e32 v12, s18
 ; GCN-NEXT:    s_waitcnt lgkmcnt(0)
 ; GCN-NEXT:    v_mov_b32_e32 v0, s0
 ; GCN-NEXT:    v_mov_b32_e32 v4, s4
+; GCN-NEXT:    v_mov_b32_e32 v8, s8
 ; GCN-NEXT:    v_mov_b32_e32 v13, s19
 ; GCN-NEXT:    v_mov_b32_e32 v1, s1
 ; GCN-NEXT:    v_mov_b32_e32 v2, s2
 ; GCN-NEXT:    v_mov_b32_e32 v3, s3
 ; GCN-NEXT:    v_mov_b32_e32 v5, s5
 ; GCN-NEXT:    v_mov_b32_e32 v6, s6
 ; GCN-NEXT:    v_mov_b32_e32 v7, s7
-; GCN-NEXT:    v_mov_b32_e32 v8, s8
 ; GCN-NEXT:    v_mov_b32_e32 v9, s9
 ; GCN-NEXT:    v_mov_b32_e32 v10, s10
 ; GCN-NEXT:    v_mov_b32_e32 v11, s11
 ; GCN-NEXT:    s_nop 0
 ; GCN-NEXT:    s_nop 0
 ; GCN-NEXT:    global_store_dwordx4 v[12:13], v[0:3], off
 ; GCN-NEXT:    global_store_dwordx4 v[12:13], v[4:7], off offset:16
+; GCN-NEXT:    global_store_dwordx4 v[12:13], v[8:11], off offset:32
 ; GCN-NEXT:    v_mov_b32_e32 v0, s12
 ; GCN-NEXT:    v_mov_b32_e32 v1, s13
 ; GCN-NEXT:    v_mov_b32_e32 v2, s14
 ; GCN-NEXT:    v_mov_b32_e32 v3, s15
 ; GCN-NEXT:    s_nop 0
 ; GCN-NEXT:    s_nop 0
-; GCN-NEXT:    global_store_dwordx4 v[12:13], v[8:11], off offset:32
 ; GCN-NEXT:    global_store_dwordx4 v[12:13], v[0:3], off offset:48
 ; GCN-NEXT:    s_endpgm
 bb:
   %tmp = load <4 x i32>, <4 x i32> addrspace(1)* %arg, align 16
   %tmp2 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg, i64 1
   %tmp3 = load <4 x i32>, <4 x i32> addrspace(1)* %tmp2, align 16
   %tmp4 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg1, i64 1
   %tmp5 = getelementptr inbounds <4 x i32>, <4 x i32> addrspace(1)* %arg, i64 2
bb:
   store <4 x i32> %tmp15, <4 x i32> addrspace(5)* %tmp16, align 16
   ret void
 }
 
 define amdgpu_kernel void @vector_clause_indirect(i64 addrspace(1)* noalias nocapture readonly %arg, <4 x i32> addrspace(1)* noalias nocapture readnone %arg1, <4 x i32> addrspace(1)* noalias nocapture %arg2) {
 ; GCN-LABEL: vector_clause_indirect:
 ; GCN:       ; %bb.0: ; %bb
 ; GCN-NEXT:    s_load_dwordx2 s[2:3], s[0:1], 0x24
-; GCN-NEXT:    v_mov_b32_e32 v1, 0
 ; GCN-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
+; GCN-NEXT:    v_mov_b32_e32 v1, 0
 ; GCN-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x34
 ; GCN-NEXT:    s_waitcnt lgkmcnt(0)
 ; GCN-NEXT:    global_load_dwordx2 v[8:9], v[0:1], s[2:3]
+; GCN-NEXT:    v_mov_b32_e32 v11, s5
+; GCN-NEXT:    v_mov_b32_e32 v10, s4
 ; GCN-NEXT:    s_nop 0
 ; GCN-NEXT:    s_waitcnt vmcnt(0)
 ; GCN-NEXT:    s_nop 0
-; GCN-NEXT:    global_load_dwordx4 v[0:3], v[8:9], off
 ; GCN-NEXT:    global_load_dwordx4 v[4:7], v[8:9], off offset:16
-; GCN-NEXT:    v_mov_b32_e32 v9, s5
-; GCN-NEXT:    v_mov_b32_e32 v8, s4
+; GCN-NEXT:    global_load_dwordx4 v[0:3], v[8:9], off
 ; GCN-NEXT:    s_nop 0
-; GCN-NEXT:    s_waitcnt vmcnt(1)
+; GCN-NEXT:    s_waitcnt vmcnt(0)
 ; GCN-NEXT:    s_nop 0
-; GCN-NEXT:    global_store_dwordx4 v[8:9], v[0:3], off
-; GCN-NEXT:    s_waitcnt vmcnt(1)
-; GCN-NEXT:    global_store_dwordx4 v[8:9], v[4:7], off offset:16
+; GCN-NEXT:    global_store_dwordx4 v[10:11], v[0:3], off
+; GCN-NEXT:    global_store_dwordx4 v[10:11], v[4:7], off offset:16
 ; GCN-NEXT:    s_endpgm
 bb:
   %tmp = tail call i32 @llvm.amdgcn.workitem.id.x()
   %tmp3 = zext i32 %tmp to i64
   %tmp4 = getelementptr inbounds i64, i64 addrspace(1)* %arg, i64 %tmp3
   %tmp5 = bitcast i64 addrspace(1)* %tmp4 to <4 x i32> addrspace(1)* addrspace(1)*
   %tmp6 = load <4 x i32> addrspace(1)*, <4 x i32> addrspace(1)* addrspace(1)* %tmp5, align 8
   %tmp7 = load <4 x i32>, <4 x i32> addrspace(1)* %tmp6, align 16

llvm/test/CodeGen/AMDGPU/promote-constOffset-to-imm.ll

 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ;
-; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:2048
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-4096
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:2048
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:2048
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
+; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:2048
 ;
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048

 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ;
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-4096
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
+; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-4096
+; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:2048
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-4096
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:2048
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off
-; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-4096
-; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off
 ;
-; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
+; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
+; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
-; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off
 entry:
   %call = tail call i64 @_Z13get_global_idj(i32 0)
   %conv = and i64 %call, 255

 ; GFX8:    flat_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}]
 ;
 ; GFX9:    global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off offset:1024
 ; GFX9:    global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off offset:2048
 ; GFX9:    global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off offset:3072
 ; GFX9:    global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off{{$}}
+; GFX9:    global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX9:    global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off offset:1024
 ; GFX9:    global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX9:    global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off offset:1024
 ; GFX9:    global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off offset:2048
 ; GFX9:    global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off offset:3072
-; GFX9:    global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off{{$}}
 ;
-; GFX10:   global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off offset:1024
+; GFX10:   global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off offset:1024
 ; GFX10:   global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off offset:-2048
 ; GFX10:   global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off offset:1024
 ; GFX10:   global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off offset:1024
 ; GFX10:   global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off offset:-2048
 ; GFX10:   global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dword {{v[0-9]+}}, v[{{[0-9]+:[0-9]+}}], off offset:1024

 
 define amdgpu_kernel void @Offset64(i8 addrspace(1)*  %buffer) {
 ; GCN-LABEL: Offset64:
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ;
-; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-4096
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:2048
+; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-4096
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ;
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:-2048
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 entry:
   %call = tail call i64 @_Z13get_global_idj(i32 0)

 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ; GFX8:    flat_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
 ;
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:2048
-; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:2048
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:2048
+; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX9:    global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off offset:2048
 ;
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}
 ; GFX10:   global_load_dwordx2 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], off{{$}}

llvm/test/CodeGen/AMDGPU/scratch-simple.ll

 ; GFX10_W64-DAG: s_mov_b32 s7, 0x31e16000
 ; GCN-DAG: v_lshlrev_b32_e32 [[BYTES:v[0-9]+]], 2, v0
 ; GCN-DAG: v_and_b32_e32 [[CLAMP_IDX:v[0-9]+]], 0x1fc, [[BYTES]]
 ; GCN-NOT: s_mov_b32 s0
 
 ; GCN-DAG: v_or_b32_e32 [[LO_OFF:v[0-9]+]], 0x200, [[CLAMP_IDX]]
 ; GCN-DAG: v_or_b32_e32 [[HI_OFF:v[0-9]+]], 0x400, [[CLAMP_IDX]]
 
-; GCN: buffer_load_dword {{v[0-9]+}}, [[LO_OFF]], {{s\[[0-9]+:[0-9]+\]}}, s0 offen
-; GCN: buffer_load_dword {{v[0-9]+}}, [[HI_OFF]], {{s\[[0-9]+:[0-9]+\]}}, s0 offen
+; GCN-DAG: buffer_load_dword {{v[0-9]+}}, [[LO_OFF]], {{s\[[0-9]+:[0-9]+\]}}, s0 offen
+; GCN-DAG: buffer_load_dword {{v[0-9]+}}, [[HI_OFF]], {{s\[[0-9]+:[0-9]+\]}}, s0 offen
 define amdgpu_ps float @ps_main(i32 %idx) {
   %v1 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0xBFEA477C60000000, float 0xBFEBE5DC60000000, float 0xBFEC71C720000000, float 0xBFEBE5DC60000000, float 0xBFEA477C60000000, float 0xBFE7A693C0000000, float 0xBFE41CFEA0000000, float 0x3FDF9B13E0000000, float 0x3FDF9B1380000000, float 0x3FD5C53B80000000, float 0x3FD5C53B00000000, float 0x3FC6326AC0000000, float 0x3FC63269E0000000, float 0xBEE05CEB00000000, float 0xBEE086A320000000, float 0xBFC63269E0000000, float 0xBFC6326AC0000000, float 0xBFD5C53B80000000, float 0xBFD5C53B80000000, float 0xBFDF9B13E0000000, float 0xBFDF9B1460000000, float 0xBFE41CFE80000000, float 0x3FE7A693C0000000, float 0x3FEA477C20000000, float 0x3FEBE5DC40000000, float 0x3FEC71C6E0000000, float 0x3FEBE5DC40000000, float 0x3FEA477C20000000, float 0x3FE7A693C0000000, float 0xBFE41CFE80000000>, i32 %idx
   %v2 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0x3FE7A693C0000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFEBE5DC40000000, float 0x3FEBE5DC40000000, float 0xBFEC71C720000000, float 0x3FEC71C6E0000000, float 0xBFEBE5DC60000000, float 0x3FEBE5DC40000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFE7A693C0000000, float 0x3FE7A69380000000, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFE80000000>, i32 %idx
   %r = fadd float %v1, %v2
   ret float %r
 }
 
 ; GCN-LABEL: {{^}}vs_main:

 ; GCN-LABEL: {{^}}hs_ir_uses_scratch_offset:
 ; GCN: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
 
 ; SIVI-NOT: s_mov_b32 s6
 ; SIVI: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, s6 offen
 ; SIVI: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, s6 offen
 
 ; GFX9_10-NOT: s_mov_b32 s5
-; GFX9_10: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, s5 offen
-; GFX9_10: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, s5 offen
+; GFX9_10-DAG: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, s5 offen
+; GFX9_10-DAG: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, s5 offen
 
-; GCN: s_mov_b32 s2, s5
+; GCN-DAG: s_mov_b32 s2, s5
 define amdgpu_hs <{i32, i32, i32, float}> @hs_ir_uses_scratch_offset(i32 inreg, i32 inreg, i32 inreg, i32 inreg, i32 inreg, i32 inreg %swo, i32 %idx) {
   %v1 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0xBFEA477C60000000, float 0xBFEBE5DC60000000, float 0xBFEC71C720000000, float 0xBFEBE5DC60000000, float 0xBFEA477C60000000, float 0xBFE7A693C0000000, float 0xBFE41CFEA0000000, float 0x3FDF9B13E0000000, float 0x3FDF9B1380000000, float 0x3FD5C53B80000000, float 0x3FD5C53B00000000, float 0x3FC6326AC0000000, float 0x3FC63269E0000000, float 0xBEE05CEB00000000, float 0xBEE086A320000000, float 0xBFC63269E0000000, float 0xBFC6326AC0000000, float 0xBFD5C53B80000000, float 0xBFD5C53B80000000, float 0xBFDF9B13E0000000, float 0xBFDF9B1460000000, float 0xBFE41CFE80000000, float 0x3FE7A693C0000000, float 0x3FEA477C20000000, float 0x3FEBE5DC40000000, float 0x3FEC71C6E0000000, float 0x3FEBE5DC40000000, float 0x3FEA477C20000000, float 0x3FE7A693C0000000, float 0xBFE41CFE80000000>, i32 %idx
   %v2 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0x3FE7A693C0000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFEBE5DC40000000, float 0x3FEBE5DC40000000, float 0xBFEC71C720000000, float 0x3FEC71C6E0000000, float 0xBFEBE5DC60000000, float 0x3FEBE5DC40000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFE7A693C0000000, float 0x3FE7A69380000000, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFE80000000>, i32 %idx
   %f = fadd float %v1, %v2
   %r1 = insertvalue <{i32, i32, i32, float}> undef, i32 %swo, 2
   %r2 = insertvalue <{i32, i32, i32, float}> %r1, float %f, 3
   ret <{i32, i32, i32, float}> %r2
 }
 
 ; GCN-LABEL: {{^}}gs_ir_uses_scratch_offset:
 ; GCN: s_mov_b32 s8, SCRATCH_RSRC_DWORD0
 
 ; SIVI: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, s6 offen
 ; SIVI: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, s6 offen
 
-; GFX9_10: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, s5 offen
-; GFX9_10: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, s5 offen
+; GFX9_10-DAG: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, s5 offen
+; GFX9_10-DAG: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, s5 offen
 
-; GCN: s_mov_b32 s2, s5
+; GCN-DAG: s_mov_b32 s2, s5
 define amdgpu_gs <{i32, i32, i32, float}> @gs_ir_uses_scratch_offset(i32 inreg, i32 inreg, i32 inreg, i32 inreg, i32 inreg, i32 inreg %swo, i32 %idx) {
   %v1 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0xBFEA477C60000000, float 0xBFEBE5DC60000000, float 0xBFEC71C720000000, float 0xBFEBE5DC60000000, float 0xBFEA477C60000000, float 0xBFE7A693C0000000, float 0xBFE41CFEA0000000, float 0x3FDF9B13E0000000, float 0x3FDF9B1380000000, float 0x3FD5C53B80000000, float 0x3FD5C53B00000000, float 0x3FC6326AC0000000, float 0x3FC63269E0000000, float 0xBEE05CEB00000000, float 0xBEE086A320000000, float 0xBFC63269E0000000, float 0xBFC6326AC0000000, float 0xBFD5C53B80000000, float 0xBFD5C53B80000000, float 0xBFDF9B13E0000000, float 0xBFDF9B1460000000, float 0xBFE41CFE80000000, float 0x3FE7A693C0000000, float 0x3FEA477C20000000, float 0x3FEBE5DC40000000, float 0x3FEC71C6E0000000, float 0x3FEBE5DC40000000, float 0x3FEA477C20000000, float 0x3FE7A693C0000000, float 0xBFE41CFE80000000>, i32 %idx
   %v2 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0x3FE7A693C0000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFEBE5DC40000000, float 0x3FEBE5DC40000000, float 0xBFEC71C720000000, float 0x3FEC71C6E0000000, float 0xBFEBE5DC60000000, float 0x3FEBE5DC40000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFE7A693C0000000, float 0x3FE7A69380000000, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFE80000000>, i32 %idx
   %f = fadd float %v1, %v2
   %r1 = insertvalue <{i32, i32, i32, float}> undef, i32 %swo, 2
   %r2 = insertvalue <{i32, i32, i32, float}> %r1, float %f, 3
   ret <{i32, i32, i32, float}> %r2
 }

llvm/test/CodeGen/AMDGPU/setcc-limit-load-shrink.ll

 ; RUN: llc -march=amdgcn -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
 
 ; GCN-LABEL: {{^}}const_load_no_shrink_dword_to_unaligned_byte:
-; GCN: s_load_dword [[LD:s[0-9]+]],
+; GCN: s_load_dword [[LD:s[0-9]+]], s[{{[0-9:]+}}], 0x0
 ; GCN: s_bfe_i32 s{{[0-9]+}}, [[LD]], 0x10013
 define amdgpu_kernel void @const_load_no_shrink_dword_to_unaligned_byte(i32 addrspace(1)* %out, i32 addrspace(4)* %in, i32 %x) {
   %ptr = getelementptr i32, i32 addrspace(4)* %in, i32 %x
   %load = load i32, i32 addrspace(4)* %ptr, align 4
   %and = and i32 %load, 524288
   %cmp = icmp eq i32 %and, 0
   %sel = select i1 %cmp, i32 0, i32 -1
   store i32 %sel, i32 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: const_load_no_shrink_dword_to_aligned_byte:
-; GCN: s_load_dword [[LD:s[0-9]+]],
+; GCN: s_load_dword [[LD:s[0-9]+]], s[{{[0-9:]+}}], 0x0
 ; GCN: s_bfe_i32 s{{[0-9]+}}, [[LD]], 0x10003
 define amdgpu_kernel void @const_load_no_shrink_dword_to_aligned_byte(i32 addrspace(1)* %out, i32 addrspace(4)* %in, i32 %x) {
   %ptr = getelementptr i32, i32 addrspace(4)* %in, i32 %x
   %load = load i32, i32 addrspace(4)* %ptr, align 4
   %and = and i32 %load, 8
   %cmp = icmp eq i32 %and, 0
   %sel = select i1 %cmp, i32 0, i32 -1
   store i32 %sel, i32 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: global_load_no_shrink_dword_to_unaligned_byte:
-; GCN: s_load_dword [[LD:s[0-9]+]],
+; GCN: s_load_dword [[LD:s[0-9]+]], s[{{[0-9:]+}}], 0x0
 ; GCN: s_bfe_i32 s{{[0-9]+}}, [[LD]], 0x10013
 define amdgpu_kernel void @global_load_no_shrink_dword_to_unaligned_byte(i32 addrspace(1)* noalias %out, i32 addrspace(1)* noalias %in, i32 %x) {
   %ptr = getelementptr i32, i32 addrspace(1)* %in, i32 %x
   %load = load i32, i32 addrspace(1)* %ptr, align 4
   %and = and i32 %load, 524288
   %cmp = icmp eq i32 %and, 0
   %sel = select i1 %cmp, i32 0, i32 -1
   store i32 %sel, i32 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: global_load_no_shrink_dword_to_aligned_byte:
-; GCN: s_load_dword [[LD:s[0-9]+]],
+; GCN: s_load_dword [[LD:s[0-9]+]], s[{{[0-9:]+}}], 0x0
 ; GCN: s_bfe_i32 s{{[0-9]+}}, [[LD]], 0x10003
 define amdgpu_kernel void @global_load_no_shrink_dword_to_aligned_byte(i32 addrspace(1)* %out, i32 addrspace(1)* %in, i32 %x) {
   %ptr = getelementptr i32, i32 addrspace(1)* %in, i32 %x
   %load = load i32, i32 addrspace(1)* %ptr, align 4
   %and = and i32 %load, 8
   %cmp = icmp eq i32 %and, 0
   %sel = select i1 %cmp, i32 0, i32 -1
   store i32 %sel, i32 addrspace(1)* %out

llvm/test/CodeGen/AMDGPU/shl.v2i16.ll

 ; VI-NEXT:    v_lshlrev_b16_sdwa v3, v4, v3 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
 ; VI-NEXT:    v_or_b32_e32 v2, v2, v3
 ; VI-NEXT:    flat_store_dword v[0:1], v2
 ; VI-NEXT:    s_endpgm
 ;
 ; CI-LABEL: shl_v_s_v2i16:
 ; CI:       ; %bb.0:
 ; CI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; CI-NEXT:    s_load_dword s8, s[0:1], 0xd
-; CI-NEXT:    s_mov_b32 s3, 0xf000
-; CI-NEXT:    s_mov_b32 s2, 0
+; CI-NEXT:    s_mov_b32 s11, 0xf000
+; CI-NEXT:    s_mov_b32 s10, 0
 ; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; CI-NEXT:    s_waitcnt lgkmcnt(0)
-; CI-NEXT:    s_mov_b64 s[0:1], s[6:7]
 ; CI-NEXT:    v_mov_b32_e32 v1, 0
-; CI-NEXT:    buffer_load_dword v2, v[0:1], s[0:3], 0 addr64
-; CI-NEXT:    s_mov_b32 s9, 0xffff
-; CI-NEXT:    s_lshr_b32 s10, s8, 16
-; CI-NEXT:    s_and_b32 s8, s8, s9
-; CI-NEXT:    s_mov_b64 s[6:7], s[2:3]
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[8:9], s[6:7]
+; CI-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; CI-NEXT:    s_load_dword s0, s[0:1], 0xd
+; CI-NEXT:    s_mov_b32 s1, 0xffff
+; CI-NEXT:    s_mov_b64 s[6:7], s[10:11]
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_lshr_b32 s2, s0, 16
+; CI-NEXT:    s_and_b32 s0, s0, s1
 ; CI-NEXT:    s_waitcnt vmcnt(0)
 ; CI-NEXT:    v_lshrrev_b32_e32 v3, 16, v2
-; CI-NEXT:    v_lshlrev_b32_e32 v2, s8, v2
-; CI-NEXT:    v_lshlrev_b32_e32 v3, s10, v3
-; CI-NEXT:    v_and_b32_e32 v2, s9, v2
+; CI-NEXT:    v_lshlrev_b32_e32 v2, s0, v2
+; CI-NEXT:    v_lshlrev_b32_e32 v3, s2, v3
+; CI-NEXT:    v_and_b32_e32 v2, s1, v2
 ; CI-NEXT:    v_lshlrev_b32_e32 v3, 16, v3
 ; CI-NEXT:    v_or_b32_e32 v2, v2, v3
 ; CI-NEXT:    buffer_store_dword v2, v[0:1], s[4:7], 0 addr64
 ; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
   %out.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %out, i64 %tid.ext

 ; VI-NEXT:    v_lshlrev_b16_sdwa v3, v3, v4 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; VI-NEXT:    v_or_b32_e32 v2, v2, v3
 ; VI-NEXT:    flat_store_dword v[0:1], v2
 ; VI-NEXT:    s_endpgm
 ;
 ; CI-LABEL: shl_s_v_v2i16:
 ; CI:       ; %bb.0:
 ; CI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x9
-; CI-NEXT:    s_load_dword s8, s[0:1], 0xd
-; CI-NEXT:    s_mov_b32 s3, 0xf000
-; CI-NEXT:    s_mov_b32 s2, 0
+; CI-NEXT:    s_mov_b32 s11, 0xf000
+; CI-NEXT:    s_mov_b32 s10, 0
 ; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
-; CI-NEXT:    s_waitcnt lgkmcnt(0)
-; CI-NEXT:    s_mov_b64 s[0:1], s[6:7]
 ; CI-NEXT:    v_mov_b32_e32 v1, 0
-; CI-NEXT:    buffer_load_dword v2, v[0:1], s[0:3], 0 addr64
-; CI-NEXT:    s_mov_b32 s0, 0xffff
-; CI-NEXT:    s_lshr_b32 s1, s8, 16
-; CI-NEXT:    s_mov_b64 s[6:7], s[2:3]
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_mov_b64 s[8:9], s[6:7]
+; CI-NEXT:    buffer_load_dword v2, v[0:1], s[8:11], 0 addr64
+; CI-NEXT:    s_load_dword s0, s[0:1], 0xd
+; CI-NEXT:    s_mov_b32 s1, 0xffff
+; CI-NEXT:    s_mov_b64 s[6:7], s[10:11]
+; CI-NEXT:    s_waitcnt lgkmcnt(0)
+; CI-NEXT:    s_lshr_b32 s2, s0, 16
 ; CI-NEXT:    s_waitcnt vmcnt(0)
-; CI-NEXT:    v_and_b32_e32 v3, s0, v2
+; CI-NEXT:    v_and_b32_e32 v3, s1, v2
 ; CI-NEXT:    v_lshrrev_b32_e32 v2, 16, v2
-; CI-NEXT:    v_lshl_b32_e32 v2, s1, v2
-; CI-NEXT:    v_lshl_b32_e32 v3, s8, v3
+; CI-NEXT:    v_lshl_b32_e32 v2, s2, v2
+; CI-NEXT:    v_lshl_b32_e32 v3, s0, v3
 ; CI-NEXT:    v_lshlrev_b32_e32 v2, 16, v2
-; CI-NEXT:    v_and_b32_e32 v3, s0, v3
+; CI-NEXT:    v_and_b32_e32 v3, s1, v3
 ; CI-NEXT:    v_or_b32_e32 v2, v3, v2
 ; CI-NEXT:    buffer_store_dword v2, v[0:1], s[4:7], 0 addr64
 ; CI-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %in.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
   %out.gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %out, i64 %tid.ext
   %vgpr = load <2 x i16>, <2 x i16> addrspace(1)* %in.gep

llvm/test/CodeGen/AMDGPU/sign_extend.ll

 define amdgpu_kernel void @v_sext_i1_to_i16_with_and(i16 addrspace(1)* %out, i32 %a, i32 %b, i32 %c) nounwind {
 ; SI-LABEL: v_sext_i1_to_i16_with_and:
 ; SI:       ; %bb.0:
 ; SI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
 ; SI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0xb
 ; SI-NEXT:    s_mov_b32 s7, 0xf000
 ; SI-NEXT:    s_mov_b32 s6, -1
 ; SI-NEXT:    s_waitcnt lgkmcnt(0)
-; SI-NEXT:    v_mov_b32_e32 v1, s2
 ; SI-NEXT:    v_cmp_eq_u32_e32 vcc, s0, v0
-; SI-NEXT:    v_cmp_eq_u32_e64 s[0:1], s1, v1
+; SI-NEXT:    v_mov_b32_e32 v0, s2
+; SI-NEXT:    v_cmp_eq_u32_e64 s[0:1], s1, v0
 ; SI-NEXT:    s_and_b64 s[0:1], vcc, s[0:1]
 ; SI-NEXT:    v_cndmask_b32_e64 v0, 0, -1, s[0:1]
 ; SI-NEXT:    buffer_store_short v0, off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
 ; VI-LABEL: v_sext_i1_to_i16_with_and:
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
 ; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x2c
 ; VI-NEXT:    s_mov_b32 s7, 0xf000
 ; VI-NEXT:    s_mov_b32 s6, -1
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
-; VI-NEXT:    v_mov_b32_e32 v1, s2
 ; VI-NEXT:    v_cmp_eq_u32_e32 vcc, s0, v0
-; VI-NEXT:    v_cmp_eq_u32_e64 s[0:1], s1, v1
+; VI-NEXT:    v_mov_b32_e32 v0, s2
+; VI-NEXT:    v_cmp_eq_u32_e64 s[0:1], s1, v0
 ; VI-NEXT:    s_and_b64 s[0:1], vcc, s[0:1]
 ; VI-NEXT:    v_cndmask_b32_e64 v0, 0, -1, s[0:1]
 ; VI-NEXT:    buffer_store_short v0, off, s[4:7], 0
 ; VI-NEXT:    s_endpgm
   %tid = tail call i32 @llvm.amdgcn.workitem.id.x() #1
   %cmp0 = icmp eq i32 %a, %tid
   %cmp1 = icmp eq i32 %b, %c
   %cmp = and i1 %cmp0, %cmp1

llvm/test/CodeGen/AMDGPU/smrd-vccz-bug.ll

 ; RUN: llc -march=amdgcn -mcpu=verde -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=VCCZ-BUG %s
 ; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=VCCZ-BUG %s
 ; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
 
 ; GCN-FUNC: {{^}}vccz_workaround:
-; GCN: s_load_dword s{{[0-9]+}}, s[{{[0-9]+:[0-9]+}}], 0x0
-; GCN: v_cmp_neq_f32_e64 {{[^,]*}}, s{{[0-9]+}}, 0{{$}}
+; GCN-DAG: s_load_dword s{{[0-9]+}}, s[{{[0-9]+:[0-9]+}}], 0x0
+; GCN-DAG: v_cmp_neq_f32_e64 {{[^,]*}}, s{{[0-9]+}}, 0{{$}}
 ; VCCZ-BUG: s_waitcnt lgkmcnt(0)
 ; VCCZ-BUG: s_mov_b64 vcc, vcc
 ; GCN-NOT: s_mov_b64 vcc, vcc
 ; GCN: s_cbranch_vccnz [[EXIT:[0-9A-Za-z_]+]]
 ; GCN: buffer_store_dword
 ; GCN: [[EXIT]]:
 ; GCN: s_endpgm
 define amdgpu_kernel void @vccz_workaround(i32 addrspace(4)* %in, i32 addrspace(1)* %out, float %cond) {

llvm/test/CodeGen/AMDGPU/store-weird-sizes.ll

 ; HAWAII-LABEL: local_store_i65:
 ; HAWAII:       ; %bb.0:
 ; HAWAII-NEXT:    s_load_dword s2, s[4:5], 0x0
 ; HAWAII-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x2
 ; HAWAII-NEXT:    s_load_dword s3, s[4:5], 0x4
 ; HAWAII-NEXT:    s_mov_b32 m0, -1
 ; HAWAII-NEXT:    s_waitcnt lgkmcnt(0)
 ; HAWAII-NEXT:    v_mov_b32_e32 v2, s2
+; HAWAII-NEXT:    s_and_b32 s3, s3, 1
+; HAWAII-NEXT:    v_mov_b32_e32 v0, s3
+; HAWAII-NEXT:    ds_write_b8 v2, v0 offset:8
 ; HAWAII-NEXT:    v_mov_b32_e32 v0, s0
 ; HAWAII-NEXT:    v_mov_b32_e32 v1, s1
-; HAWAII-NEXT:    s_and_b32 s0, s3, 1
-; HAWAII-NEXT:    v_mov_b32_e32 v3, s0
-; HAWAII-NEXT:    ds_write_b8 v2, v3 offset:8
 ; HAWAII-NEXT:    ds_write_b64 v2, v[0:1]
 ; HAWAII-NEXT:    s_endpgm
 ;
 ; FIJI-LABEL: local_store_i65:
 ; FIJI:       ; %bb.0:
 ; FIJI-NEXT:    s_load_dword s2, s[4:5], 0x0
 ; FIJI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x8
 ; FIJI-NEXT:    s_load_dword s3, s[4:5], 0x10
 ; FIJI-NEXT:    s_mov_b32 m0, -1
 ; FIJI-NEXT:    s_waitcnt lgkmcnt(0)
 ; FIJI-NEXT:    v_mov_b32_e32 v2, s2
+; FIJI-NEXT:    s_and_b32 s3, s3, 1
+; FIJI-NEXT:    v_mov_b32_e32 v0, s3
+; FIJI-NEXT:    ds_write_b8 v2, v0 offset:8
 ; FIJI-NEXT:    v_mov_b32_e32 v0, s0
 ; FIJI-NEXT:    v_mov_b32_e32 v1, s1
-; FIJI-NEXT:    s_and_b32 s0, s3, 1
-; FIJI-NEXT:    v_mov_b32_e32 v3, s0
-; FIJI-NEXT:    ds_write_b8 v2, v3 offset:8
 ; FIJI-NEXT:    ds_write_b64 v2, v[0:1]
 ; FIJI-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: local_store_i65:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_load_dword s2, s[4:5], 0x0
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x8
 ; GFX9-NEXT:    s_load_dword s3, s[4:5], 0x10
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s2
 ; GFX9-NEXT:    v_mov_b32_e32 v0, s0
+; GFX9-NEXT:    s_and_b32 s3, s3, 1
 ; GFX9-NEXT:    v_mov_b32_e32 v1, s1
-; GFX9-NEXT:    s_and_b32 s0, s3, 1
-; GFX9-NEXT:    v_mov_b32_e32 v3, s0
+; GFX9-NEXT:    v_mov_b32_e32 v3, s3
 ; GFX9-NEXT:    ds_write_b8 v2, v3 offset:8
 ; GFX9-NEXT:    ds_write_b64 v2, v[0:1]
 ; GFX9-NEXT:    s_endpgm
   store i65 %arg, i65 addrspace(3)* %ptr, align 8
   ret void
 }
 
 define void @local_store_i13(i13 addrspace(3)* %ptr, i13 %arg) #0 {

llvm/test/CodeGen/AMDGPU/sub.v2i16.ll

; VI-NEXT:    s_endpgm
   %add = sub <2 x i16> %a, %b
   store <2 x i16> %add, <2 x i16> addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @s_test_sub_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(4)* %in0, <2 x i16> addrspace(4)* %in1) #1 {
 ; GFX9-LABEL: s_test_sub_v2i16:
 ; GFX9:       ; %bb.0:
-; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-NEXT:    s_load_dwordx2 s[8:9], s[0:1], 0x34
+; GFX9-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; GFX9-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX9-NEXT:    s_mov_b32 s2, -1
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_mov_b32 s0, s4
 ; GFX9-NEXT:    s_mov_b32 s1, s5
 ; GFX9-NEXT:    s_load_dword s4, s[6:7], 0x0
 ; GFX9-NEXT:    s_load_dword s5, s[8:9], 0x0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)

 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_mov_b32_e32 v1, s7
 ; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s6, v2
 ; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
 ; GFX9-NEXT:    v_mov_b32_e32 v3, s1
 ; GFX9-NEXT:    v_add_co_u32_e32 v2, vcc, s0, v2
 ; GFX9-NEXT:    v_addc_co_u32_e32 v3, vcc, 0, v3, vcc
 ; GFX9-NEXT:    global_load_dword v0, v[0:1], off
-; GFX9-NEXT:    global_load_dword v1, v[2:3], off
+; GFX9-NEXT:    global_load_dword v2, v[2:3], off
 ; GFX9-NEXT:    s_mov_b32 s0, s4
-; GFX9-NEXT:    v_mov_b32_e32 v3, 0
 ; GFX9-NEXT:    s_mov_b32 s1, s5
-; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_pk_sub_i16 v1, v0, v1
-; GFX9-NEXT:    v_and_b32_e32 v0, 0xffff, v1
-; GFX9-NEXT:    v_lshrrev_b32_e32 v2, 16, v1
+; GFX9-NEXT:    v_mov_b32_e32 v3, 0
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_pk_sub_i16 v2, v0, v2
+; GFX9-NEXT:    v_and_b32_e32 v0, 0xffff, v2
+; GFX9-NEXT:    v_lshrrev_b32_e32 v2, 16, v2
 ; GFX9-NEXT:    buffer_store_dwordx4 v[0:3], off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; VI-LABEL: v_test_sub_v2i16_zext_to_v2i64:
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dwordx4 s[4:7], s[0:1], 0x24
 ; VI-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x34
 ; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0

llvm/test/CodeGen/AMDGPU/wwm-reserved.ll

; GFX9-O0: buffer_load_dword v7
   %tmp137 = tail call i64 @llvm.amdgcn.wwm.i64(i64 %tmp136)
   %tmp138 = bitcast i64 %tmp137 to <2 x i32>
   call void @llvm.amdgcn.raw.buffer.store.v2i32(<2 x i32> %tmp138, <4 x i32> %tmp14, i32 4, i32 0, i32 0)
   ret void
 }
 
 define amdgpu_cs void @_amdgpu_cs_main(<4 x i32> inreg %desc, i32 %index) {
   %tmp17 = shl i32 %index, 5
-; GFX9: buffer_load_dwordx4
+; GFX9-DAG: buffer_load_dwordx4
   %tmp18 = tail call <4 x i32> @llvm.amdgcn.s.buffer.load.v4i32(<4 x i32> %desc, i32 %tmp17, i32 0)
   %.i0.upto1.bc = bitcast <4 x i32> %tmp18 to <2 x i64>
   %tmp19 = or i32 %tmp17, 16
-; GFX9: buffer_load_dwordx2
+; GFX9-DAG: buffer_load_dwordx2
   %tmp20 = tail call <2 x i32> @llvm.amdgcn.s.buffer.load.v2i32(<4 x i32> %desc, i32 %tmp19, i32 0)
   %.i0.upto1.extract = extractelement <2 x i64> %.i0.upto1.bc, i32 0
   %tmp22 = tail call i64 @llvm.amdgcn.set.inactive.i64(i64 %.i0.upto1.extract, i64 9223372036854775807)
   %tmp97 = tail call i64 @llvm.amdgcn.wwm.i64(i64 %tmp22)
   %.i1.upto1.extract = extractelement <2 x i64> %.i0.upto1.bc, i32 1
   %tmp99 = tail call i64 @llvm.amdgcn.set.inactive.i64(i64 %.i1.upto1.extract, i64 9223372036854775807)
   %tmp174 = tail call i64 @llvm.amdgcn.wwm.i64(i64 %tmp99)
   %.i25 = bitcast <2 x i32> %tmp20 to i64