Index: lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
===================================================================
--- lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
+++ lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
@@ -734,7 +734,6 @@
   addPass(&SIFoldOperandsID);
   addPass(&DeadMachineInstructionElimID);
   addPass(&SILoadStoreOptimizerID);
-  addPass(createSIShrinkInstructionsPass());
   if (EnableSDWAPeephole) {
     addPass(&SIPeepholeSDWAID);
     addPass(&MachineLICMID);
@@ -742,6 +741,7 @@
     addPass(&SIFoldOperandsID);
     addPass(&DeadMachineInstructionElimID);
   }
+  addPass(createSIShrinkInstructionsPass());
 }
 
 bool GCNPassConfig::addILPOpts() {
Index: lib/Target/AMDGPU/SIPeepholeSDWA.cpp
===================================================================
--- lib/Target/AMDGPU/SIPeepholeSDWA.cpp
+++ lib/Target/AMDGPU/SIPeepholeSDWA.cpp
@@ -129,7 +129,8 @@
   bool getNeg() const { return Neg; }
   bool getSext() const { return Sext; }
 
-  uint64_t getSrcMods() const;
+  uint64_t getSrcMods(const SIInstrInfo *TII,
+                      const MachineOperand *SrcOp) const;
 };
 
 class SDWADstOperand : public SDWAOperand {
@@ -240,13 +241,24 @@
   return SuperMask.all();
 }
 
-uint64_t SDWASrcOperand::getSrcMods() const {
+uint64_t SDWASrcOperand::getSrcMods(const SIInstrInfo *TII,
+                                    const MachineOperand *SrcOp) const {
   uint64_t Mods = 0;
+  const auto *MI = SrcOp->getParent();
+  if (TII->getNamedOperand(*MI, AMDGPU::OpName::src0) == SrcOp) {
+    if (auto *Mod = TII->getNamedOperand(*MI, AMDGPU::OpName::src0_modifiers)) {
+      Mods = Mod->getImm();
+    }
+  } else if (TII->getNamedOperand(*MI, AMDGPU::OpName::src1) == SrcOp) {
+    if (auto *Mod = TII->getNamedOperand(*MI, AMDGPU::OpName::src1_modifiers)) {
+      Mods = Mod->getImm();
+    }
+  }
   if (Abs || Neg) {
     assert(!Sext &&
            "Float and integer src modifiers can't be set simulteniously");
     Mods |= Abs ? SISrcMods::ABS : 0;
-    Mods |= Neg ? SISrcMods::NEG : 0;
+    Mods ^= Neg ? SISrcMods::NEG : 0;
   } else if (Sext) {
     Mods |= SISrcMods::SEXT;
   }
@@ -312,7 +324,7 @@
   }
   copyRegOperand(*Src, *getTargetOperand());
   SrcSel->setImm(getSrcSel());
-  SrcMods->setImm(getSrcMods());
+  SrcMods->setImm(getSrcMods(TII, Src));
   getTargetOperand()->setIsKill(false);
   return true;
 }
@@ -409,7 +421,10 @@
       switch (Opcode) {
       case AMDGPU::V_LSHRREV_B32_e32:
       case AMDGPU::V_ASHRREV_I32_e32:
-      case AMDGPU::V_LSHLREV_B32_e32: {
+      case AMDGPU::V_LSHLREV_B32_e32:
+      case AMDGPU::V_LSHRREV_B32_e64:
+      case AMDGPU::V_ASHRREV_I32_e64:
+      case AMDGPU::V_LSHLREV_B32_e64: {
         // from: v_lshrrev_b32_e32 v1, 16/24, v0
         // to SDWA src:v0 src_sel:WORD_1/BYTE_3
 
@@ -432,7 +447,8 @@
             TRI->isPhysicalRegister(Dst->getReg()))
           break;
 
-        if (Opcode == AMDGPU::V_LSHLREV_B32_e32) {
+        if (Opcode == AMDGPU::V_LSHLREV_B32_e32 ||
+            Opcode == AMDGPU::V_LSHLREV_B32_e64) {
           auto SDWADst = make_unique<SDWADstOperand>(
               Dst, Src1, *Imm == 16 ? WORD_1 : BYTE_3, UNUSED_PAD);
           DEBUG(dbgs() << "Match: " << MI << "To: " << *SDWADst << '\n');
@@ -441,7 +457,8 @@
         } else {
           auto SDWASrc = make_unique<SDWASrcOperand>(
               Src1, Dst, *Imm == 16 ? WORD_1 : BYTE_3, false, false,
-              Opcode == AMDGPU::V_LSHRREV_B32_e32 ? false : true);
+              Opcode != AMDGPU::V_LSHRREV_B32_e32 &&
+              Opcode != AMDGPU::V_LSHRREV_B32_e64);
           DEBUG(dbgs() << "Match: " << MI << "To: " << *SDWASrc << '\n');
           SDWAOperands[&MI] = std::move(SDWASrc);
           ++NumSDWAPatternsFound;
@@ -451,7 +468,10 @@
 
       case AMDGPU::V_LSHRREV_B16_e32:
       case AMDGPU::V_ASHRREV_I16_e32:
-      case AMDGPU::V_LSHLREV_B16_e32: {
+      case AMDGPU::V_LSHLREV_B16_e32:
+      case AMDGPU::V_LSHRREV_B16_e64:
+      case AMDGPU::V_ASHRREV_I16_e64:
+      case AMDGPU::V_LSHLREV_B16_e64: {
         // from: v_lshrrev_b16_e32 v1, 8, v0
         // to SDWA src:v0 src_sel:BYTE_1
 
@@ -472,7 +492,8 @@
             TRI->isPhysicalRegister(Dst->getReg()))
           break;
 
-        if (Opcode == AMDGPU::V_LSHLREV_B16_e32) {
+        if (Opcode == AMDGPU::V_LSHLREV_B16_e32 ||
+            Opcode == AMDGPU::V_LSHLREV_B16_e64) {
           auto SDWADst =
             make_unique<SDWADstOperand>(Dst, Src1, BYTE_1, UNUSED_PAD);
           DEBUG(dbgs() << "Match: " << MI << "To: " << *SDWADst << '\n');
@@ -481,7 +502,8 @@
         } else {
           auto SDWASrc = make_unique<SDWASrcOperand>(
               Src1, Dst, BYTE_1, false, false,
-              Opcode == AMDGPU::V_LSHRREV_B16_e32 ? false : true);
+              Opcode != AMDGPU::V_LSHRREV_B16_e32 &&
+              Opcode != AMDGPU::V_LSHRREV_B16_e64);
           DEBUG(dbgs() << "Match: " << MI << "To: " << *SDWASrc << '\n');
           SDWAOperands[&MI] = std::move(SDWASrc);
           ++NumSDWAPatternsFound;
@@ -549,20 +571,25 @@
         ++NumSDWAPatternsFound;
         break;
       }
-      case AMDGPU::V_AND_B32_e32: {
+      case AMDGPU::V_AND_B32_e32:
+      case AMDGPU::V_AND_B32_e64: {
         // e.g.:
         // from: v_and_b32_e32 v1, 0x0000ffff/0x000000ff, v0
         // to SDWA src:v0 src_sel:WORD_0/BYTE_0
 
         MachineOperand *Src0 = TII->getNamedOperand(MI, AMDGPU::OpName::src0);
+        MachineOperand *Src1 = TII->getNamedOperand(MI, AMDGPU::OpName::src1);
+        auto ValSrc = Src1;
         auto Imm = foldToImm(*Src0);
-        if (!Imm)
-          break;
 
-        if (*Imm != 0x0000ffff && *Imm != 0x000000ff)
+        if (!Imm) {
+          Imm = foldToImm(*Src1);
+          ValSrc = Src0;
+        }
+
+        if (!Imm || (*Imm != 0x0000ffff && *Imm != 0x000000ff))
           break;
 
-        MachineOperand *Src1 = TII->getNamedOperand(MI, AMDGPU::OpName::src1);
         MachineOperand *Dst = TII->getNamedOperand(MI, AMDGPU::OpName::vdst);
       
         if (TRI->isPhysicalRegister(Src1->getReg()) ||
@@ -570,7 +597,7 @@
           break;
 
         auto SDWASrc = make_unique<SDWASrcOperand>(
-            Src1, Dst, *Imm == 0x0000ffff ? WORD_0 : BYTE_0);
+            ValSrc, Dst, *Imm == 0x0000ffff ? WORD_0 : BYTE_0);
         DEBUG(dbgs() << "Match: " << MI << "To: " << *SDWASrc << '\n');
         SDWAOperands[&MI] = std::move(SDWASrc);
         ++NumSDWAPatternsFound;
@@ -583,28 +610,37 @@
 
 bool SIPeepholeSDWA::isConvertibleToSDWA(const MachineInstr &MI) const {
   // Check if this instruction has opcode that supports SDWA
-  return AMDGPU::getSDWAOp(MI.getOpcode()) != -1;
+  unsigned Opc = MI.getOpcode();
+  if (AMDGPU::getSDWAOp(Opc) != -1)
+    return true;
+  int Opc32 = AMDGPU::getVOPe32(Opc);
+  if (Opc32 != -1 && AMDGPU::getSDWAOp(Opc32) != -1)
+    return true;
+  return false;
 }
 
 bool SIPeepholeSDWA::convertToSDWA(MachineInstr &MI,
                                    const SDWAOperandsVector &SDWAOperands) {
   // Convert to sdwa
   int SDWAOpcode = AMDGPU::getSDWAOp(MI.getOpcode());
+  if (SDWAOpcode == -1)
+    SDWAOpcode = AMDGPU::getSDWAOp(AMDGPU::getVOPe32(MI.getOpcode()));
   assert(SDWAOpcode != -1);
 
+  // Copy dst, if it is present in original then should also be present in SDWA
+  MachineOperand *Dst = TII->getNamedOperand(MI, AMDGPU::OpName::vdst);
+  if (!Dst && !TII->isVOPC(MI))
+    return false;
+
   const MCInstrDesc &SDWADesc = TII->get(SDWAOpcode);
 
   // Create SDWA version of instruction MI and initialize its operands
   MachineInstrBuilder SDWAInst =
     BuildMI(*MI.getParent(), MI, MI.getDebugLoc(), SDWADesc);
 
-  // Copy dst, if it is present in original then should also be present in SDWA
-  MachineOperand *Dst = TII->getNamedOperand(MI, AMDGPU::OpName::vdst);
   if (Dst) {
     assert(AMDGPU::getNamedOperandIdx(SDWAOpcode, AMDGPU::OpName::vdst) != -1);
     SDWAInst.add(*Dst);
-  } else {
-    assert(TII->isVOPC(MI));
   }
 
   // Copy src0, initialize src0_modifiers. All sdwa instructions has src0 and
@@ -614,7 +650,10 @@
     Src0 &&
     AMDGPU::getNamedOperandIdx(SDWAOpcode, AMDGPU::OpName::src0) != -1 &&
     AMDGPU::getNamedOperandIdx(SDWAOpcode, AMDGPU::OpName::src0_modifiers) != -1);
-  SDWAInst.addImm(0);
+  if (auto *Mod = TII->getNamedOperand(MI, AMDGPU::OpName::src0_modifiers))
+    SDWAInst.addImm(Mod->getImm());
+  else
+    SDWAInst.addImm(0);
   SDWAInst.add(*Src0);
 
   // Copy src1 if present, initialize src1_modifiers.
@@ -623,10 +662,11 @@
     assert(
       AMDGPU::getNamedOperandIdx(SDWAOpcode, AMDGPU::OpName::src1) != -1 &&
       AMDGPU::getNamedOperandIdx(SDWAOpcode, AMDGPU::OpName::src1_modifiers) != -1);
-    SDWAInst.addImm(0);
+    if (auto *Mod = TII->getNamedOperand(MI, AMDGPU::OpName::src1_modifiers))
+      SDWAInst.addImm(Mod->getImm());
+    else
+      SDWAInst.addImm(0);
     SDWAInst.add(*Src1);
-  } else {
-    assert(TII->isVOP1(MI));
   }
 
   if (SDWAOpcode == AMDGPU::V_MAC_F16_sdwa ||
Index: test/CodeGen/AMDGPU/add.v2i16.ll
===================================================================
--- test/CodeGen/AMDGPU/add.v2i16.ll
+++ test/CodeGen/AMDGPU/add.v2i16.ll
@@ -66,7 +66,7 @@
 
 ; VI-DAG: v_add_u16_e32 v{{[0-9]+}}, 0x7b, v{{[0-9]+}}
 ; VI-DAG: v_mov_b32_e32 v[[SCONST:[0-9]+]], 0x1c8
-; VI-DAG: v_add_u16_sdwa v{{[0-9]+}}, v[[SCONST]], v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_u16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v[[SCONST]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 define amdgpu_kernel void @v_test_add_v2i16_constant(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in0) #1 {
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep.out = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %out, i32 %tid
@@ -84,7 +84,7 @@
 
 ; VI-DAG: v_add_u16_e32 v{{[0-9]+}}, 0xfffffcb3, v{{[0-9]+}}
 ; VI-DAG: v_mov_b32_e32 v[[SCONST:[0-9]+]], 0xfffffc21
-; VI-DAG: v_add_u16_sdwa v{{[0-9]+}}, v[[SCONST]], v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_u16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v[[SCONST]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 define amdgpu_kernel void @v_test_add_v2i16_neg_constant(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in0) #1 {
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %gep.out = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %out, i32 %tid
@@ -101,7 +101,7 @@
 ; VI: v_mov_b32_e32 v[[SCONST:[0-9]+]], -1
 ; VI: flat_load_ushort [[LOAD0:v[0-9]+]]
 ; VI: flat_load_ushort [[LOAD1:v[0-9]+]]
-; VI-DAG: v_add_u16_sdwa v{{[0-9]+}}, v[[SCONST]], [[LOAD0]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_u16_sdwa v{{[0-9]+}}, [[LOAD0]], v[[SCONST]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI-DAG: v_add_u16_e32 v{{[0-9]+}}, -1, [[LOAD1]]
 ; VI: v_or_b32_e32
 define amdgpu_kernel void @v_test_add_v2i16_inline_neg1(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in0) #1 {
@@ -140,7 +140,7 @@
 
 ; VI-NOT: v_add_u16
 ; VI: v_mov_b32_e32 v[[K:[0-9]+]], 0x3f80
-; VI: v_add_u16_sdwa v{{[0-9]+}}, v[[K]], v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI: v_add_u16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v[[K]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI-NOT: v_add_u16
 ; VI: v_or_b32_e32
 define amdgpu_kernel void @v_test_add_v2i16_inline_fp_split(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in0) #1 {
Index: test/CodeGen/AMDGPU/ashr.v2i16.ll
===================================================================
--- test/CodeGen/AMDGPU/ashr.v2i16.ll
+++ test/CodeGen/AMDGPU/ashr.v2i16.ll
@@ -9,7 +9,7 @@
 ; GFX9: v_pk_ashrrev_i16 [[RESULT:v[0-9]+]], [[RHS]], [[VLHS]]
 
 ; VI: v_ashrrev_i32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
-; VI: v_or_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
+; VI: v_or_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
 
 ; CI: v_ashrrev_i32_e32
 ; CI: v_and_b32_e32 v{{[0-9]+}}, 0xffff, v{{[0-9]+}}
Index: test/CodeGen/AMDGPU/fabs.f16.ll
===================================================================
--- test/CodeGen/AMDGPU/fabs.f16.ll
+++ test/CodeGen/AMDGPU/fabs.f16.ll
@@ -40,7 +40,7 @@
 ; VI: flat_load_ushort [[LO:v[0-9]+]]
 ; VI: v_mov_b32_e32 [[MASK:v[0-9]+]], 0x7fff{{$}}
 ; VI-DAG: v_and_b32_e32 [[FABS_LO:v[0-9]+]], [[MASK]], [[HI]]
-; VI-DAG: v_and_b32_sdwa [[FABS_HI:v[0-9]+]], [[MASK]], [[LO]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_and_b32_sdwa [[FABS_HI:v[0-9]+]], [[LO]], [[MASK]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI-DAG: v_or_b32_e32 v{{[0-9]+}}, [[FABS_HI]], [[FABS_LO]]
 ; VI: flat_store_dword
 
@@ -60,8 +60,8 @@
 ; CI: v_and_b32_e32 v{{[0-9]+}}, [[MASK]]
 
 ; VI: v_mov_b32_e32 [[MASK:v[0-9]+]], 0x7fff{{$}}
-; VI-DAG: v_and_b32_sdwa v{{[0-9]+}}, [[MASK]], v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
-; VI-DAG: v_and_b32_sdwa v{{[0-9]+}}, [[MASK]], v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_and_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, [[MASK]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_and_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, [[MASK]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI-DAG: v_and_b32_e32 v{{[0-9]+}}, [[MASK]], v{{[0-9]+}}
 ; VI-DAG: v_and_b32_e32 v{{[0-9]+}}, [[MASK]], v{{[0-9]+}}
 ; VI-DAG: v_or_b32_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
@@ -128,7 +128,7 @@
 ; CI: v_cvt_f16_f32
 
 ; VI: v_lshrrev_b32_e32 v{{[0-9]+}}, 16,
-; VI: v_mul_f16_e64 v{{[0-9]+}}, |v{{[0-9]+}}|, v{{[0-9]+}}
+; VI: v_mul_f16_sdwa v{{[0-9]+}}, |v{{[0-9]+}}|, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; VI: v_mul_f16_e64 v{{[0-9]+}}, |v{{[0-9]+}}|, v{{[0-9]+}}
 
 ; GFX9: v_and_b32_e32 [[FABS:v[0-9]+]], 0x7fff7fff, [[VAL]]
Index: test/CodeGen/AMDGPU/fadd.f16.ll
===================================================================
--- test/CodeGen/AMDGPU/fadd.f16.ll
+++ test/CodeGen/AMDGPU/fadd.f16.ll
@@ -78,7 +78,7 @@
 ; SI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_0]]
 
 ; VI-DAG: v_add_f16_e32 v[[R_F16_LO:[0-9]+]], v[[B_V2_F16]], v[[A_V2_F16]]
-; VI-DAG: v_add_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[B_V2_F16]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; VI-DAG: v_add_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[A_V2_F16]], v[[B_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
 ; VI:     v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_LO]]
 
 ; GCN: buffer_store_dword v[[R_V2_F16]]
@@ -108,7 +108,7 @@
 ; SI:  v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_0]]
 
 ; VI-DAG: v_mov_b32_e32 v[[CONST2:[0-9]+]], 0x4000
-; VI-DAG: v_add_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[CONST2]], v[[B_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; VI-DAG: v_add_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[B_V2_F16]], v[[CONST2]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; VI-DAG: v_add_f16_e32 v[[R_F16_0:[0-9]+]], 1.0, v[[B_V2_F16]]
 ; VI:     v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_0]]
 
@@ -137,7 +137,7 @@
 ; SI:  v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_0]]
 
 ; VI-DAG: v_mov_b32_e32 v[[CONST1:[0-9]+]], 0x3c00
-; VI-DAG: v_add_f16_sdwa v[[R_F16_0:[0-9]+]], v[[CONST1]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; VI-DAG: v_add_f16_sdwa v[[R_F16_0:[0-9]+]], v[[A_V2_F16]], v[[CONST1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; VI-DAG: v_add_f16_e32 v[[R_F16_1:[0-9]+]], 2.0, v[[A_V2_F16]]
 ; VI:     v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_0]], v[[R_F16_1]]
 
Index: test/CodeGen/AMDGPU/fcanonicalize.f16.ll
===================================================================
--- test/CodeGen/AMDGPU/fcanonicalize.f16.ll
+++ test/CodeGen/AMDGPU/fcanonicalize.f16.ll
@@ -278,9 +278,9 @@
 }
 
 ; GCN-LABEL: {{^}}s_test_canonicalize_var_v2f16:
-; VI: v_mul_f16_e64 [[REG0:v[0-9]+]], 1.0, {{s[0-9]+}}
-; VI-DAG: v_mul_f16_e64 [[REG1:v[0-9]+]], 1.0, {{s[0-9]+}}
-; VI-DAG: v_lshlrev_b32_e32 v{{[0-9]+}}, 16,
+; VI: v_mov_b32_e32 [[ONE:v[0-9]+]], 0x3c00
+; VI: v_mul_f16_sdwa [[REG0:v[0-9]+]], [[ONE]], {{v[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI: v_mul_f16_e64 [[REG1:v[0-9]+]], 1.0, {{s[0-9]+}}
 ; VI-NOT: v_and_b32
 
 ; GFX9: v_pk_mul_f16 [[REG:v[0-9]+]], 1.0, {{s[0-9]+$}}
Index: test/CodeGen/AMDGPU/fmul.f16.ll
===================================================================
--- test/CodeGen/AMDGPU/fmul.f16.ll
+++ test/CodeGen/AMDGPU/fmul.f16.ll
@@ -78,7 +78,7 @@
 ; SI:  v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_0]]
 
 ; VI-DAG: v_mul_f16_e32 v[[R_F16_LO:[0-9]+]], v[[B_V2_F16]], v[[A_V2_F16]]
-; VI-DAG: v_mul_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[B_V2_F16]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; VI-DAG: v_mul_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[A_V2_F16]], v[[B_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
 ; VI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_LO]]
 
 ; GCN: buffer_store_dword v[[R_V2_F16]]
@@ -105,7 +105,7 @@
 ; SI:  v_mul_f32_e32 v[[R_F32_1:[0-9]+]], 4.0, v[[B_F32_1]]
 ; SI:  v_cvt_f16_f32_e32 v[[R_F16_1:[0-9]+]], v[[R_F32_1]]
 ; VI-DAG:  v_mov_b32_e32 v[[CONST4:[0-9]+]], 0x4400
-; VI-DAG:  v_mul_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[CONST4]], v[[B_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; VI-DAG:  v_mul_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[B_V2_F16]], v[[CONST4]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; VI-DAG:  v_mul_f16_e32 v[[R_F16_0:[0-9]+]], 0x4200, v[[B_V2_F16]]
 ; SI-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[R_F16_1]]
 ; GCN: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_0]]
@@ -131,7 +131,7 @@
 ; SI:  v_mul_f32_e32 v[[R_F32_1:[0-9]+]], 0x40400000, v[[A_F32_1]]
 ; SI:  v_cvt_f16_f32_e32 v[[R_F16_1:[0-9]+]], v[[R_F32_1]]
 ; VI-DAG:  v_mov_b32_e32 v[[CONST3:[0-9]+]], 0x4200
-; VI-DAG:  v_mul_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[CONST3]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; VI-DAG:  v_mul_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[A_V2_F16]], v[[CONST3]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; VI-DAG:  v_mul_f16_e32 v[[R_F16_0:[0-9]+]], 4.0, v[[A_V2_F16]]
 ; SI-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[R_F16_1]]
 ; GCN: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_0]]
Index: test/CodeGen/AMDGPU/fneg-fabs.f16.ll
===================================================================
--- test/CodeGen/AMDGPU/fneg-fabs.f16.ll
+++ test/CodeGen/AMDGPU/fneg-fabs.f16.ll
@@ -73,7 +73,7 @@
 ; CIVI: s_mov_b32 [[MASK:s[0-9]+]], 0x8000{{$}}
 ; VI: v_mov_b32_e32 [[VMASK:v[0-9]+]], [[MASK]]
 ; CI: v_or_b32_e32 v{{[0-9]+}}, [[MASK]],
-; VI: v_or_b32_sdwa v{{[0-9]+}}, [[VMASK]], v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI: v_or_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, [[VMASK]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; CIVI: v_or_b32_e32 v{{[0-9]+}}, [[MASK]],
 ; CIVI: flat_store_dword
 
@@ -92,9 +92,9 @@
 ; CI: v_or_b32_e32 v{{[0-9]+}}, [[MASK]],
 ; CI: v_or_b32_e32 v{{[0-9]+}}, [[MASK]],
 ; VI: v_mov_b32_e32 [[VMASK:v[0-9]+]], [[MASK]]
-; VI: v_or_b32_sdwa v{{[0-9]+}}, [[VMASK]], v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI: v_or_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, [[VMASK]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32_e32 v{{[0-9]+}}, [[MASK]],
-; VI: v_or_b32_sdwa v{{[0-9]+}}, [[VMASK]], v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI: v_or_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, [[VMASK]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32_e32 v{{[0-9]+}}, [[MASK]],
 
 ; GFX9: s_mov_b32 [[MASK:s[0-9]+]], 0x80008000
@@ -116,7 +116,7 @@
 ; CI: v_mul_f32_e32 v{{[0-9]+}}, 4.0, v{{[0-9]+}}
 
 ; VI: v_mul_f16_e64 v{{[0-9]+}}, -|v{{[0-9]+}}|, 4.0
-; VI: v_mul_f16_e64 v{{[0-9]+}}, -|v{{[0-9]+}}|, 4.0
+; VI: v_mul_f16_sdwa v{{[0-9]+}}, -|v{{[0-9]+}}|, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 
 ; GFX9: s_and_b32 [[ABS:s[0-9]+]], s{{[0-9]+}}, 0x7fff7fff
 ; GFX9: v_pk_mul_f16 v{{[0-9]+}}, [[ABS]], 4.0 neg_lo:[1,0] neg_hi:[1,0]
Index: test/CodeGen/AMDGPU/fneg.f16.ll
===================================================================
--- test/CodeGen/AMDGPU/fneg.f16.ll
+++ test/CodeGen/AMDGPU/fneg.f16.ll
@@ -117,7 +117,7 @@
 ; CI: v_cvt_f16_f32
 
 ; VI: v_lshrrev_b32_e32 v{{[0-9]+}}, 16,
-; VI: v_mul_f16_e64 v{{[0-9]+}}, -v{{[0-9]+}}, v{{[0-9]+}}
+; VI: v_mul_f16_sdwa v{{[0-9]+}}, -v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; VI: v_mul_f16_e64 v{{[0-9]+}}, -v{{[0-9]+}}, v{{[0-9]+}}
 
 ; GFX9: v_pk_mul_f16 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} neg_lo:[1,0] neg_hi:[1,0]{{$}}
Index: test/CodeGen/AMDGPU/fptosi.f16.ll
===================================================================
--- test/CodeGen/AMDGPU/fptosi.f16.ll
+++ test/CodeGen/AMDGPU/fptosi.f16.ll
@@ -66,7 +66,7 @@
 ; VI: v_cvt_f32_f16_sdwa v[[A_F32_1:[0-9]+]], v[[A_V2_F16]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1
 ; VI: v_cvt_i32_f32_e32 v[[R_I16_0:[0-9]+]], v[[A_F32_0]]
 ; VI: v_cvt_i32_f32_sdwa v[[R_I16_1:[0-9]+]], v[[A_F32_1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD
-; VI: v_or_b32_sdwa v[[R_V2_I16:[0-9]+]], v[[R_I16_1]], v[[R_I16_0]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
+; VI: v_or_b32_sdwa v[[R_V2_I16:[0-9]+]], v[[R_I16_0]], v[[R_I16_1]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
 
 ; GCN: buffer_store_dword v[[R_V2_I16]]
 ; GCN: s_endpgm
Index: test/CodeGen/AMDGPU/fptoui.f16.ll
===================================================================
--- test/CodeGen/AMDGPU/fptoui.f16.ll
+++ test/CodeGen/AMDGPU/fptoui.f16.ll
@@ -66,7 +66,7 @@
 ; VI-DAG: v_cvt_f32_f16_sdwa v[[A_F32_0:[0-9]+]], v[[A_V2_F16]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1
 ; VI:      v_cvt_i32_f32_e32 v[[R_I16_1:[0-9]+]], v[[A_F32_1]]
 ; VI:      v_cvt_i32_f32_sdwa v[[R_I16_0:[0-9]+]], v[[A_F32_0]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD
-; VI:     v_or_b32_sdwa v[[R_V2_I16:[0-9]+]], v[[R_I16_0]], v[[R_I16_1]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
+; VI:     v_or_b32_sdwa v[[R_V2_I16:[0-9]+]], v[[R_I16_1]], v[[R_I16_0]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
 
 ; GCN:     buffer_store_dword v[[R_V2_I16]]
 ; GCN:     s_endpgm
Index: test/CodeGen/AMDGPU/fsub.f16.ll
===================================================================
--- test/CodeGen/AMDGPU/fsub.f16.ll
+++ test/CodeGen/AMDGPU/fsub.f16.ll
@@ -78,7 +78,7 @@
 ; SI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_0]]
 
 ; VI-DAG: v_subrev_f16_e32 v[[R_F16_0:[0-9]+]], v[[B_V2_F16]], v[[A_V2_F16]]
-; VI-DAG: v_subrev_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[B_V2_F16]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; VI-DAG: v_sub_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[A_V2_F16]], v[[B_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
 ; VI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_0]]
 
 ; GFX9: v_pk_add_f16 v[[R_V2_F16:[0-9]+]], v[[A_V2_F16]], v[[B_V2_F16]] neg_lo:[0,1] neg_hi:[0,1]
@@ -146,7 +146,7 @@
 ; SI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_0]]
 
 ; VI-DAG: v_mov_b32_e32 [[CONSTM1:v[0-9]+]], 0xbc00
-; VI-DAG: v_add_f16_sdwa v[[R_F16_HI:[0-9]+]], [[CONSTM1]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; VI-DAG: v_add_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[A_V2_F16]], [[CONSTM1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; VI-DAG: v_add_f16_e32 v[[R_F16_0:[0-9]+]], -2.0, v[[A_V2_F16]]
 ; VI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_0]]
 
Index: test/CodeGen/AMDGPU/immv216.ll
===================================================================
--- test/CodeGen/AMDGPU/immv216.ll
+++ test/CodeGen/AMDGPU/immv216.ll
@@ -124,7 +124,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, 0, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONST0:v[0-9]+]], 0
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONST0]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONST0]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_0.0_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
@@ -142,7 +142,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, 0.5, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONST05:v[0-9]+]], 0x3800
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONST05]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONST05]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_0.5_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
@@ -160,7 +160,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, -0.5, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONSTM05:v[0-9]+]], 0xb800
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONSTM05]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONSTM05]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_neg_0.5_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
@@ -178,7 +178,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, 1.0, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONST1:v[0-9]+]], 0x3c00
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONST1]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONST1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_1.0_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
@@ -196,7 +196,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, -1.0, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONSTM1:v[0-9]+]], 0xbc00
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONSTM1]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONSTM1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_neg_1.0_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
@@ -214,7 +214,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, 2.0, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONST2:v[0-9]+]], 0x4000
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONST2]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONST2]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_2.0_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
@@ -232,7 +232,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, -2.0, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONSTM2:v[0-9]+]], 0xc000
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONSTM2]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONSTM2]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_neg_2.0_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
@@ -250,7 +250,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, 4.0, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONST4:v[0-9]+]], 0x4400
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONST4]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONST4]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_4.0_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
@@ -268,7 +268,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, -4.0, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONSTM4:v[0-9]+]], 0xc400
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONSTM4]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONSTM4]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_neg_4.0_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
@@ -285,7 +285,7 @@
 ; VI: v_mov_b32_e32 [[CONST05:v[0-9]+]], 0x3800
 ; VI: buffer_load_dword
 ; VI-NOT: and
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONST05]], v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, [[CONST05]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, 0.5, v{{[0-9]+}}
 ; VI: v_or_b32
 ; VI: buffer_store_dword
@@ -306,7 +306,7 @@
 ; VI-DAG: buffer_load_dword
 ; VI-NOT: and
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, [[K]], v{{[0-9]+}}
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[K]], v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, [[K]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; VI: v_or_b32_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
 ; VI: buffer_store_dword
 define amdgpu_kernel void @commute_add_literal_v2f16(<2 x half> addrspace(1)* %out, <2 x half> addrspace(1)* %in) #0 {
@@ -325,7 +325,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, 1, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONST1:v[0-9]+]], 1
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONST1]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONST1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_1_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
@@ -343,7 +343,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, 2, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONST2:v[0-9]+]], 2
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONST2]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONST2]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_2_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
@@ -361,7 +361,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, 16, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONST16:v[0-9]+]], 16
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONST16]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONST16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_16_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
@@ -379,7 +379,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, -1, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONSTM1:v[0-9]+]], 0xffff
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONSTM1]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONSTM1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_neg_1_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
@@ -397,7 +397,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, -2, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONSTM2:v[0-9]+]], 0xfffe
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONSTM2]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONSTM2]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_neg_2_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
@@ -415,7 +415,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, -16, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONSTM16:v[0-9]+]], 0xfff0
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONSTM16]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONSTM16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_neg_16_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
@@ -433,7 +433,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, 63, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONST63:v[0-9]+]], 63
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONST63]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONST63]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_63_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
@@ -451,7 +451,7 @@
 ; VI: buffer_load_ushort [[VAL1:v[0-9]+]]
 ; VI-DAG: v_add_f16_e32 v{{[0-9]+}}, 64, [[VAL0]]
 ; VI-DAG: v_mov_b32_e32 [[CONST64:v[0-9]+]], 64
-; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[CONST64]], [[VAL1]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_f16_sdwa v{{[0-9]+}}, [[VAL1]], [[CONST64]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_or_b32
 ; VI: buffer_store_dword
 define amdgpu_kernel void @add_inline_imm_64_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %x) #0 {
Index: test/CodeGen/AMDGPU/insert_vector_elt.v2i16.ll
===================================================================
--- test/CodeGen/AMDGPU/insert_vector_elt.v2i16.ll
+++ test/CodeGen/AMDGPU/insert_vector_elt.v2i16.ll
@@ -261,7 +261,7 @@
 ; VI: v_mov_b32_e32 [[K:v[0-9]+]], 0x3e70000
 ; GCN-DAG: flat_load_dword [[VEC:v[0-9]+]]
 ; CI: v_or_b32_e32 [[RES:v[0-9]+]], 0x3e70000, [[VEC]]
-; VI: v_or_b32_sdwa [[RES:v[0-9]+]], [[K]], [[VEC]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
+; VI: v_or_b32_sdwa [[RES:v[0-9]+]], [[VEC]], [[K]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
 
 ; GFX9-DAG: s_movk_i32 [[K:s[0-9]+]], 0x3e7
 ; GFX9-DAG: v_and_b32_e32 [[ELT0:v[0-9]+]], 0xffff, [[VEC]]
@@ -285,7 +285,7 @@
 ; CI:   v_and_b32_e32 [[ELT0:v[0-9]+]], 0xffff, [[VEC]]
 ; GFX9: v_and_b32_e32 [[ELT0:v[0-9]+]], 0xffff, [[VEC]]
 ; CI: v_or_b32_e32 [[RES:v[0-9]+]], 0xfff10000, [[ELT0]]
-; VI: v_or_b32_sdwa [[RES:v[0-9]+]], [[K]], [[VEC]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
+; VI: v_or_b32_sdwa [[RES:v[0-9]+]], [[VEC]], [[K]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
 ; GFX9: v_lshl_or_b32 [[RES:v[0-9]+]], -15, 16, [[ELT0]]
 ; GCN: flat_store_dword v{{\[[0-9]+:[0-9]+\]}}, [[RES]]
 define amdgpu_kernel void @v_insertelement_v2i16_1_inlineimm(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in) #0 {
@@ -345,7 +345,7 @@
 ; VI: v_mov_b32_e32 [[K:v[0-9]+]], 0x45000000
 ; GCN-DAG: flat_load_dword [[VEC:v[0-9]+]]
 ; CI: v_or_b32_e32 [[RES:v[0-9]+]], 0x45000000, [[VEC]]
-; VI: v_or_b32_sdwa [[RES:v[0-9]+]], [[K]], [[VEC]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
+; VI: v_or_b32_sdwa [[RES:v[0-9]+]], [[VEC]], [[K]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
 
 ; GFX9-DAG: s_movk_i32 [[K:s[0-9]+]], 0x4500
 ; GFX9-DAG: v_and_b32_e32 [[ELT0:v[0-9]+]], 0xffff, [[VEC]]
@@ -369,7 +369,7 @@
 ; CI: v_and_b32_e32 [[ELT0:v[0-9]+]], 0xffff, [[VEC]]
 ; GFX9: v_and_b32_e32 [[ELT0:v[0-9]+]], 0xffff, [[VEC]]
 ; CI: v_or_b32_e32 [[RES:v[0-9]+]], 0x230000, [[ELT0]]
-; VI: v_or_b32_sdwa [[RES:v[0-9]+]], [[K]], [[VEC]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
+; VI: v_or_b32_sdwa [[RES:v[0-9]+]], [[VEC]], [[K]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
 ; GFX9: v_lshl_or_b32 [[RES:v[0-9]+]], 35, 16, [[ELT0]]
 ; GCN: flat_store_dword v{{\[[0-9]+:[0-9]+\]}}, [[RES]]
 define amdgpu_kernel void @v_insertelement_v2f16_1_inlineimm(<2 x half> addrspace(1)* %out, <2 x half> addrspace(1)* %in) #0 {
Index: test/CodeGen/AMDGPU/llvm.fmuladd.f16.ll
===================================================================
--- test/CodeGen/AMDGPU/llvm.fmuladd.f16.ll
+++ test/CodeGen/AMDGPU/llvm.fmuladd.f16.ll
@@ -118,7 +118,7 @@
 ; SI:  v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_LO]]
 
 ; VI-FLUSH:     v_lshrrev_b32_e32 v[[A_F16_1:[0-9]+]], 16, v[[A_V2_F16]]
-; VI-FLUSH-DAG: v_mac_f16_sdwa v[[A_F16_1]], v[[C_V2_F16]], v[[B_V2_F16]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; VI-FLUSH-DAG: v_mac_f16_sdwa v[[A_F16_1]], v[[B_V2_F16]], v[[C_V2_F16]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
 ; VI-FLUSH-DAG: v_mac_f16_e32 v[[A_V2_F16]], v[[C_V2_F16]], v[[B_V2_F16]]
 ; VI-FLUSH-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[A_F16_1]]
 ; VI-FLUSH-NOT: v_and_b32
Index: test/CodeGen/AMDGPU/llvm.maxnum.f16.ll
===================================================================
--- test/CodeGen/AMDGPU/llvm.maxnum.f16.ll
+++ test/CodeGen/AMDGPU/llvm.maxnum.f16.ll
@@ -82,7 +82,7 @@
 ; SI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_0]]
 
 ; VI-DAG: v_max_f16_e32 v[[R_F16_0:[0-9]+]], v[[B_V2_F16]], v[[A_V2_F16]]
-; VI-DAG: v_max_f16_sdwa v[[R_F16_1:[0-9]+]], v[[B_V2_F16]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; VI-DAG: v_max_f16_sdwa v[[R_F16_1:[0-9]+]], v[[A_V2_F16]], v[[B_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
 ; VI-NOT: and
 ; VI:     v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_1]], v[[R_F16_0]]
 
@@ -110,7 +110,7 @@
 ; SI:  v_max_f32_e32 v[[R_F32_1:[0-9]+]], 4.0, v[[B_F32_1]]
 ; SI:  v_cvt_f16_f32_e32 v[[R_F16_1:[0-9]+]], v[[R_F32_1]]
 ; VI-DAG:  v_mov_b32_e32 [[CONST4:v[0-9]+]], 0x4400
-; VI-DAG:  v_max_f16_sdwa v[[R_F16_HI:[0-9]+]], [[CONST4]], v[[B_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; VI-DAG:  v_max_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[B_V2_F16]], [[CONST4]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; VI-DAG:  v_max_f16_e32 v[[R_F16_0:[0-9]+]], 0x4200, v[[B_V2_F16]]
 
 ; SI-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[R_F16_1]]
@@ -138,7 +138,7 @@
 ; SI:  v_max_f32_e32 v[[R_F32_1:[0-9]+]], 0x40400000, v[[A_F32_1]]
 ; SI:  v_cvt_f16_f32_e32 v[[R_F16_1:[0-9]+]], v[[R_F32_1]]
 ; VI-DAG:  v_mov_b32_e32 [[CONST3:v[0-9]+]], 0x4200
-; VI-DAG:  v_max_f16_sdwa v[[R_F16_HI:[0-9]+]], [[CONST3]], v[[B_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; VI-DAG:  v_max_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[B_V2_F16]], [[CONST3]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; VI-DAG:  v_max_f16_e32 v[[R_F16_0:[0-9]+]], 4.0, v[[A_V2_F16]]
 
 ; SI-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[R_F16_1]]
Index: test/CodeGen/AMDGPU/llvm.minnum.f16.ll
===================================================================
--- test/CodeGen/AMDGPU/llvm.minnum.f16.ll
+++ test/CodeGen/AMDGPU/llvm.minnum.f16.ll
@@ -81,7 +81,7 @@
 ; SI:     v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_0]]
 
 ; VI-DAG: v_min_f16_e32 v[[R_F16_0:[0-9]+]], v[[B_V2_F16]], v[[A_V2_F16]]
-; VI-DAG: v_min_f16_sdwa v[[R_F16_1:[0-9]+]], v[[B_V2_F16]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; VI-DAG: v_min_f16_sdwa v[[R_F16_1:[0-9]+]], v[[A_V2_F16]], v[[B_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
 ; VI-NOT: and
 ; VI:     v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_1]], v[[R_F16_0]]
 
@@ -111,7 +111,7 @@
 ; SI-DAG:  v_cvt_f16_f32_e32 v[[R_F16_1:[0-9]+]], v[[R_F32_1]]
 
 ; VI-DAG:  v_mov_b32_e32 [[CONST4:v[0-9]+]], 0x4400
-; VI-DAG:  v_min_f16_sdwa v[[R_F16_HI:[0-9]+]], [[CONST4]], v[[B_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; VI-DAG:  v_min_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[B_V2_F16]], [[CONST4]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; VI-DAG:  v_min_f16_e32 v[[R_F16_0:[0-9]+]], 0x4200, v[[B_V2_F16]]
 
 ; SI-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[R_F16_1]]
@@ -139,7 +139,7 @@
 ; SI:  v_min_f32_e32 v[[R_F32_1:[0-9]+]], 0x40400000, v[[A_F32_1]]
 ; SI:  v_cvt_f16_f32_e32 v[[R_F16_1:[0-9]+]], v[[R_F32_1]]
 ; VI-DAG:  v_mov_b32_e32 [[CONST3:v[0-9]+]], 0x4200
-; VI-DAG:  v_min_f16_sdwa v[[R_F16_HI:[0-9]+]], [[CONST3]], v[[B_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; VI-DAG:  v_min_f16_sdwa v[[R_F16_HI:[0-9]+]], v[[B_V2_F16]], [[CONST3]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; VI-DAG:  v_min_f16_e32 v[[R_F16_0:[0-9]+]], 4.0, v[[A_V2_F16]]
 
 ; SI-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[R_F16_1]]
Index: test/CodeGen/AMDGPU/scratch-simple.ll
===================================================================
--- test/CodeGen/AMDGPU/scratch-simple.ll
+++ test/CodeGen/AMDGPU/scratch-simple.ll
@@ -12,8 +12,10 @@
 ; 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-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-DAG: v_mov_b32_e32 [[C200:v[0-9]+]], 0x200
+; GCN-DAG: v_mov_b32_e32 [[C400:v[0-9]+]], 0x400
+; GCN-DAG: v_or_b32_e32 [[LO_OFF:v[0-9]+]], [[C200]], [[CLAMP_IDX]]
+; GCN-DAG: v_or_b32_e32 [[HI_OFF:v[0-9]+]], [[C400]], [[CLAMP_IDX]]
 
 ; GCN: buffer_load_dword {{v[0-9]+}}, [[LO_OFF]], {{s\[[0-9]+:[0-9]+\]}}, [[SWO]] offen
 ; GCN: buffer_load_dword {{v[0-9]+}}, [[HI_OFF]], {{s\[[0-9]+:[0-9]+\]}}, [[SWO]] offen
Index: test/CodeGen/AMDGPU/sdwa-peephole.ll
===================================================================
--- test/CodeGen/AMDGPU/sdwa-peephole.ll
+++ test/CodeGen/AMDGPU/sdwa-peephole.ll
@@ -74,7 +74,7 @@
 
 ; SDWA-DAG: v_mul_u32_u24_sdwa v[[DST_MUL_LO:[0-9]+]], v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:WORD_0
 ; SDWA-DAG: v_mul_u32_u24_sdwa v[[DST_MUL_HI:[0-9]+]], v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
-; SDWA: v_or_b32_sdwa v{{[0-9]+}}, v[[DST_MUL_HI]], v[[DST_MUL_LO]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
+; SDWA: v_or_b32_sdwa v{{[0-9]+}}, v[[DST_MUL_LO]], v[[DST_MUL_HI]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
 
 define amdgpu_kernel void @mul_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %ina, <2 x i16> addrspace(1)* %inb) {
 entry:
@@ -97,8 +97,8 @@
 ; SDWA-DAG: v_mul_u32_u24_sdwa v[[DST_MUL1:[0-9]+]], v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
 ; SDWA-DAG: v_mul_u32_u24_sdwa v[[DST_MUL2:[0-9]+]], v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:WORD_0
 ; SDWA-DAG: v_mul_u32_u24_sdwa v[[DST_MUL3:[0-9]+]], v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
-; SDWA-DAG: v_or_b32_sdwa v{{[0-9]+}}, v[[DST_MUL3]], v[[DST_MUL2]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
-; SDWA-DAG: v_or_b32_sdwa v{{[0-9]+}}, v[[DST_MUL1]], v[[DST_MUL0]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
+; SDWA-DAG: v_or_b32_sdwa v{{[0-9]+}}, v[[DST_MUL2]], v[[DST_MUL3]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
+; SDWA-DAG: v_or_b32_sdwa v{{[0-9]+}}, v[[DST_MUL0]], v[[DST_MUL1]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
 
 define amdgpu_kernel void @mul_v4i16(<4 x i16> addrspace(1)* %out, <4 x i16> addrspace(1)* %ina, <4 x i16> addrspace(1)* %inb) {
 entry:
@@ -125,10 +125,10 @@
 ; SDWA-DAG: v_mul_u32_u24_sdwa v[[DST_MUL5:[0-9]+]], v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
 ; SDWA-DAG: v_mul_u32_u24_sdwa v[[DST_MUL6:[0-9]+]], v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:WORD_0
 ; SDWA-DAG: v_mul_u32_u24_sdwa v[[DST_MUL7:[0-9]+]], v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
-; SDWA-DAG: v_or_b32_sdwa v{{[0-9]+}}, v[[DST_MUL7]], v[[DST_MUL6]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
-; SDWA-DAG: v_or_b32_sdwa v{{[0-9]+}}, v[[DST_MUL5]], v[[DST_MUL4]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
-; SDWA-DAG: v_or_b32_sdwa v{{[0-9]+}}, v[[DST_MUL3]], v[[DST_MUL2]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
-; SDWA-DAG: v_or_b32_sdwa v{{[0-9]+}}, v[[DST_MUL1]], v[[DST_MUL0]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
+; SDWA-DAG: v_or_b32_sdwa v{{[0-9]+}}, v[[DST_MUL6]], v[[DST_MUL7]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
+; SDWA-DAG: v_or_b32_sdwa v{{[0-9]+}}, v[[DST_MUL4]], v[[DST_MUL5]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
+; SDWA-DAG: v_or_b32_sdwa v{{[0-9]+}}, v[[DST_MUL2]], v[[DST_MUL3]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
+; SDWA-DAG: v_or_b32_sdwa v{{[0-9]+}}, v[[DST_MUL0]], v[[DST_MUL1]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
 
 define amdgpu_kernel void @mul_v8i16(<8 x i16> addrspace(1)* %out, <8 x i16> addrspace(1)* %ina, <8 x i16> addrspace(1)* %inb) {
 entry:
@@ -347,8 +347,8 @@
 ; NOSDWA-NOT: v_mul_u32_u24_sdwa
 ; SDWA-DAG: v_mov_b32_e32 v[[M321:[0-9]+]], 0x141
 ; SDWA-DAG: v_mov_b32_e32 v[[M123:[0-9]+]], 0x7b
-; SDWA-DAG: v_mul_u32_u24_sdwa v{{[0-9]+}}, v[[M123]], v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
-; SDWA-DAG: v_mul_u32_u24_sdwa v{{[0-9]+}}, v[[M321]], v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; SDWA-DAG: v_mul_u32_u24_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v[[M123]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
+; SDWA-DAG: v_mul_u32_u24_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v[[M321]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 
 define amdgpu_kernel void @immediate_mul_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in) {
 entry:
@@ -367,7 +367,7 @@
 ; NOSDWA: v_or_b32_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
 ; NOSDWA-NOT: v_mul_u32_u24_sdwa
 
-; SDWA: v_mul_u32_u24_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; SDWA: v_mul_u32_u24_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 
 define amdgpu_kernel void @mulmul_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %ina, <2 x i16> addrspace(1)* %inb) {
 entry:
@@ -408,9 +408,9 @@
 ; NOSDWA-NOT: v_and_b32_sdwa
 ; NOSDWA-NOT: v_or_b32_sdwa
 
-; SDWA-DAG: v_and_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; SDWA-DAG: v_and_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; SDWA-DAG: v_lshlrev_b16_e32 v{{[0-9]+}}, 8, v{{[0-9]+}}
-; SDWA-DAG: v_and_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; SDWA-DAG: v_and_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; SDWA-DAG: v_lshlrev_b16_e32 v{{[0-9]+}}, 8, v{{[0-9]+}}
 ; SDWA: v_or_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 
Index: test/CodeGen/AMDGPU/shl.v2i16.ll
===================================================================
--- test/CodeGen/AMDGPU/shl.v2i16.ll
+++ test/CodeGen/AMDGPU/shl.v2i16.ll
@@ -10,7 +10,7 @@
 
 ; VI: v_lshlrev_b32_e32
 ; VI: v_lshlrev_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
-; VI: v_or_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
+; VI: v_or_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
 
 ; CI: v_lshlrev_b32_e32
 ; CI: v_and_b32_e32 v{{[0-9]+}}, 0xffff, v{{[0-9]+}}
Index: test/CodeGen/AMDGPU/sminmax.v2i16.ll
===================================================================
--- test/CodeGen/AMDGPU/sminmax.v2i16.ll
+++ test/CodeGen/AMDGPU/sminmax.v2i16.ll
@@ -10,11 +10,11 @@
 
 ; VI: v_sub_i32_e32
 ; VI-DAG: v_sub_i32_e32
-; VI: v_max_i32_sdwa v{{[0-9]+}}, sext(v{{[0-9]+}}), v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
-; VI: v_max_i32_sdwa v{{[0-9]+}}, sext(v{{[0-9]+}}), v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
+; VI: v_max_i32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, sext(v{{[0-9]+}}) dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
+; VI: v_max_i32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, sext(v{{[0-9]+}}) dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
 ; VI: v_add_i32_e32
 ; VI: v_add_i32_e32
-; VI: v_or_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_0
+; VI: v_or_b32_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_0 src1_sel:DWORD
 
 ; CI: v_sub_i32_e32
 ; CI-DAG: v_sub_i32_e32
@@ -47,7 +47,7 @@
 ; VI: v_max_i16_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
 ; VI: v_max_i16_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
 ; VI: v_add_u16_e32 v{{[0-9]+}}, 2, v{{[0-9]+}}
-; VI: v_add_u16_sdwa v{{[0-9]+}}, [[TWO]], v{{[0-9]+}}  dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI: v_add_u16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, [[TWO]]  dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI-NOT: v_and_b32
 ; VI: v_or_b32_e32
 define amdgpu_kernel void @v_abs_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %src) #0 {
Index: test/CodeGen/AMDGPU/sub.v2i16.ll
===================================================================
--- test/CodeGen/AMDGPU/sub.v2i16.ll
+++ test/CodeGen/AMDGPU/sub.v2i16.ll
@@ -5,7 +5,7 @@
 ; GCN-LABEL: {{^}}v_test_sub_v2i16:
 ; GFX9: v_pk_sub_i16 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
 
-; VI: v_subrev_u16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI: v_sub_u16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI: v_subrev_u16_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
 define amdgpu_kernel void @v_test_sub_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in0, <2 x i16> addrspace(1)* %in1) #1 {
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
@@ -62,7 +62,7 @@
 ; GFX9: v_pk_sub_i16 v{{[0-9]+}}, v{{[0-9]+}}, [[CONST]]
 
 ; VI-DAG: v_mov_b32_e32 [[K:v[0-9]+]], 0xfffffe38
-; VI-DAG: v_add_u16_sdwa v{{[0-9]+}}, [[K]], v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_u16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, [[K]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI-DAG: v_add_u16_e32 v{{[0-9]+}}, 0xffffff85, v{{[0-9]+}}
 define amdgpu_kernel void @v_test_sub_v2i16_constant(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in0) #1 {
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
@@ -80,7 +80,7 @@
 ; GFX9: v_pk_sub_i16 v{{[0-9]+}}, v{{[0-9]+}}, [[CONST]]
 
 ; VI-DAG: v_mov_b32_e32 [[K:v[0-9]+]], 0x3df
-; VI-DAG: v_add_u16_sdwa v{{[0-9]+}}, [[K]], v{{[0-9]+}}
+; VI-DAG: v_add_u16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, [[K]]
 ; VI-DAG: v_add_u16_e32 v{{[0-9]+}}, 0x34d, v{{[0-9]+}}
 define amdgpu_kernel void @v_test_sub_v2i16_neg_constant(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in0) #1 {
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
@@ -98,7 +98,7 @@
 ; VI: v_mov_b32_e32 [[ONE:v[0-9]+]], 1
 ; VI: flat_load_ushort [[LOAD0:v[0-9]+]]
 ; VI: flat_load_ushort [[LOAD1:v[0-9]+]]
-; VI-DAG: v_add_u16_sdwa v{{[0-9]+}}, [[ONE]], [[LOAD0]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI-DAG: v_add_u16_sdwa v{{[0-9]+}}, [[LOAD0]], [[ONE]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI-DAG: v_add_u16_e32 v{{[0-9]+}}, 1, [[LOAD1]]
 ; VI: v_or_b32_e32
 define amdgpu_kernel void @v_test_sub_v2i16_inline_neg1(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in0) #1 {
@@ -137,7 +137,7 @@
 
 ; VI-NOT: v_subrev_i16
 ; VI: v_mov_b32_e32 [[K:v[0-9]+]], 0xffffc080
-; VI: v_add_u16_sdwa v{{[0-9]+}}, [[K]], v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
+; VI: v_add_u16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, [[K]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
 ; VI-NOT: v_subrev_i16
 ; VI: v_or_b32_e32
 define amdgpu_kernel void @v_test_sub_v2i16_inline_fp_split(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %in0) #1 {
@@ -252,7 +252,7 @@
 ; GFX9: v_pk_sub_i16
 ; GFX9: v_lshrrev_b32_e32 v{{[0-9]+}}, 16, v{{[0-9]+}}
 
-; VI: v_subrev_u16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; VI: v_sub_u16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
 ; VI: v_subrev_u16_e32
 
 ; GCN: v_bfe_i32 v{{[0-9]+}}, v{{[0-9]+}}, 0, 16
Index: test/CodeGen/AMDGPU/v_mac_f16.ll
===================================================================
--- test/CodeGen/AMDGPU/v_mac_f16.ll
+++ test/CodeGen/AMDGPU/v_mac_f16.ll
@@ -304,14 +304,14 @@
 ; GCN: {{buffer|flat}}_load_dword v[[C_V2_F16:[0-9]+]]
 
 ; SI:  v_cvt_f32_f16_e32 v[[A_F32_0:[0-9]+]], v[[A_V2_F16]]
-; SI: v_lshrrev_b32_e32 v[[A_F16_1:[0-9]+]], 16, v[[A_V2_F16]]
-; SI:  v_cvt_f32_f16_e32 v[[B_F32_0:[0-9]+]], v[[B_V2_F16]]
-; SI-DAG: v_lshrrev_b32_e32 v[[B_F16_1:[0-9]+]], 16, v[[B_V2_F16]]
-; SI-DAG: v_lshrrev_b32_e32 v[[C_F16_1:[0-9]+]], 16, v[[C_V2_F16]]
-; SI-DAG:  v_cvt_f32_f16_e32 v[[C_F32_0:[0-9]+]], v[[C_V2_F16]]
-; SI-DAG:  v_cvt_f32_f16_e32 v[[A_F32_1:[0-9]+]], v[[A_F16_1]]
+; SI:  v_lshrrev_b32_e32 v[[A_F16_1:[0-9]+]], 16, v[[A_V2_F16]]
+; SI:  v_cvt_f32_f16_e32 v[[A_F32_1:[0-9]+]], v[[A_F16_1]]
+; SI:  v_lshrrev_b32_e32 v[[B_F16_1:[0-9]+]], 16, v[[B_V2_F16]]
 ; SI:  v_cvt_f32_f16_e32 v[[B_F32_1:[0-9]+]], v[[B_F16_1]]
+; SI:  v_cvt_f32_f16_e32 v[[B_F32_0:[0-9]+]], v[[B_V2_F16]]
+; SI:  v_lshrrev_b32_e32 v[[C_F16_1:[0-9]+]], 16, v[[C_V2_F16]]
 ; SI:  v_cvt_f32_f16_e32 v[[C_F32_1:[0-9]+]], v[[C_F16_1]]
+; SI-DAG:  v_cvt_f32_f16_e32 v[[C_F32_0:[0-9]+]], v[[C_V2_F16]]
 ; SI-DAG:  v_mac_f32_e32 v[[C_F32_0]], v[[B_F32_0]], v[[A_F32_0]]
 ; SI-DAG:  v_cvt_f16_f32_e32 v[[R_F16_LO:[0-9]+]], v[[C_F32_0]]
 ; SI-DAG:  v_mac_f32_e32 v[[C_F32_1]], v[[B_F32_1]], v[[A_F32_1]]
@@ -320,12 +320,12 @@
 ; VI-NOT: and
 ; SI:  v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_LO]]
 
-; VI-DAG: v_lshrrev_b32_e32 v[[A_F16_1:[0-9]+]], 16, v[[A_V2_F16]]
-; VI-DAG: v_mac_f16_sdwa v[[A_F16_1]], v[[C_V2_F16]], v[[B_V2_F16]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
-; VI-DAG: v_mac_f16_e32 v[[A_V2_F16]], v[[C_V2_F16]], v[[B_V2_F16]]
-; VI-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[A_F16_1]]
+; VI-DAG: v_lshrrev_b32_e32 v[[C_F16_1:[0-9]+]], 16, v[[C_V2_F16]]
+; VI-DAG: v_mac_f16_sdwa v[[C_F16_1]], v[[A_V2_F16]], v[[B_V2_F16]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; VI-DAG: v_mac_f16_e32 v[[C_V2_F16]], v[[B_V2_F16]], v[[A_V2_F16]]
+; VI-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[C_F16_1]]
 ; VI-NOT: and
-; VI:  v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[A_V2_F16]]
+; VI:  v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[C_V2_F16]]
 
 ; GCN: {{buffer|flat}}_store_dword v[[R_V2_F16]]
 ; GCN: s_endpgm
@@ -336,7 +336,9 @@
     <2 x half> addrspace(1)* %c) #0 {
 entry:
   %a.val = load <2 x half>, <2 x half> addrspace(1)* %a
+  call void @llvm.amdgcn.s.barrier() #2
   %b.val = load <2 x half>, <2 x half> addrspace(1)* %b
+  call void @llvm.amdgcn.s.barrier() #2
   %c.val = load <2 x half>, <2 x half> addrspace(1)* %c
 
   %t.val = fmul <2 x half> %a.val, %b.val
@@ -485,7 +487,7 @@
 ; VI-DAG:  v_mov_b32_e32 [[ZERO:v[0-9]+]], 0
 ; VI-DAG:  v_sub_f16_e32 v[[NEG_A1:[0-9]+]], 0, v{{[0-9]+}}
 ; VI-DAG:  v_sub_f16_sdwa v[[NEG_A0:[0-9]+]], [[ZERO]], v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
-; VI-DAG:  v_mac_f16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v[[NEG_A0]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
+; VI-DAG:  v_mac_f16_sdwa v{{[0-9]+}}, v[[NEG_A0]], v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
 ; VI-DAG:  v_mac_f16_e32 v{{[0-9]+}}, v{{[0-9]+}}, v[[NEG_A1]]
 
 ; GCN: s_endpgm
@@ -517,7 +519,7 @@
 ; VI:  v_mov_b32_e32 [[ZERO:v[0-9]+]], 0
 ; VI:  v_sub_f16_e32 v[[NEG_A1:[0-9]+]], 0, v{{[0-9]+}}
 ; VI:  v_sub_f16_sdwa v[[NEG_A0:[0-9]+]], [[ZERO]], v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
-; VI-DAG:  v_mac_f16_sdwa v{{[0-9]+}}, v[[NEG_A0]], v{{[0-9]+}} dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; VI-DAG:  v_mac_f16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v[[NEG_A0]] dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
 ; VI-DAG:  v_mac_f16_e32 v{{[0-9]+}}, v[[NEG_A1]], v{{[0-9]+}}
 
 ; GCN: s_endpgm
@@ -670,5 +672,8 @@
   ret void
 }
 
+declare void @llvm.amdgcn.s.barrier() #2
+
 attributes #0 = { nounwind "no-signed-zeros-fp-math"="false" }
 attributes #1 = { nounwind "no-signed-zeros-fp-math"="true" }
+attributes #2 = { nounwind convergent }