diff --git a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
--- a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -18581,6 +18581,13 @@
                                              V.getOperand(0), NewIndex);
             return DAG.getBitcast(NVT, NewExtract);
           }
+          if (NewExtNumElts == 1 &&
+              TLI.isOperationLegalOrCustom(ISD::EXTRACT_VECTOR_ELT, ScalarVT)) {
+            SDValue NewIndex = DAG.getVectorIdxConstant(IndexValScaled, DL);
+            SDValue NewExtract = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, DL, ScalarVT,
+                                             V.getOperand(0), NewIndex);
+            return DAG.getBitcast(NVT, NewExtract);
+          }
         }
       }
     }
diff --git a/llvm/test/CodeGen/AMDGPU/vector_shuffle.packed.ll b/llvm/test/CodeGen/AMDGPU/vector_shuffle.packed.ll
--- a/llvm/test/CodeGen/AMDGPU/vector_shuffle.packed.ll
+++ b/llvm/test/CodeGen/AMDGPU/vector_shuffle.packed.ll
@@ -5,9 +5,8 @@
 ; GFX9-LABEL: shuffle_v4f16_23uu:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[0:1], off
+; GFX9-NEXT:    global_load_dword v0, v[0:1], off offset:4
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, v1
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x half>, <4 x half> addrspace(1)* %arg0
   %val1 = load <4 x half>, <4 x half> addrspace(1)* %arg1
@@ -19,10 +18,10 @@
 ; GFX9-LABEL: shuffle_v4f16_234u:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[4:5], v[0:1], off
-; GFX9-NEXT:    global_load_dwordx2 v[1:2], v[2:3], off
+; GFX9-NEXT:    global_load_dwordx2 v[2:3], v[2:3], off
+; GFX9-NEXT:    global_load_dword v0, v[0:1], off offset:4
 ; GFX9-NEXT:    s_waitcnt vmcnt(1)
-; GFX9-NEXT:    v_mov_b32_e32 v0, v5
+; GFX9-NEXT:    v_mov_b32_e32 v1, v2
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x half>, <4 x half> addrspace(1)* %arg0
@@ -154,7 +153,7 @@
 ; GFX9-LABEL: shuffle_v4f16_0101:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[0:1], off
+; GFX9-NEXT:    global_load_dword v0, v[0:1], off
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    v_mov_b32_e32 v1, v0
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
@@ -181,9 +180,8 @@
 ; GFX9-LABEL: shuffle_v4f16_0145:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[0:1], off
-; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[1:2], v[2:3], off
+; GFX9-NEXT:    global_load_dword v0, v[0:1], off
+; GFX9-NEXT:    global_load_dword v1, v[2:3], off
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x half>, <4 x half> addrspace(1)* %arg0
@@ -196,11 +194,9 @@
 ; GFX9-LABEL: shuffle_v4f16_0167:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[0:1], off
-; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[1:2], v[2:3], off
+; GFX9-NEXT:    global_load_dword v0, v[0:1], off
+; GFX9-NEXT:    global_load_dword v1, v[2:3], off offset:4
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v1, v2
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x half>, <4 x half> addrspace(1)* %arg0
   %val1 = load <4 x half>, <4 x half> addrspace(1)* %arg1
@@ -226,9 +222,9 @@
 ; GFX9-LABEL: shuffle_v4f16_2323:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[0:1], off
+; GFX9-NEXT:    global_load_dword v0, v[0:1], off offset:4
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-NEXT:    v_mov_b32_e32 v1, v0
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x half>, <4 x half> addrspace(1)* %arg0
   %val1 = load <4 x half>, <4 x half> addrspace(1)* %arg1
@@ -240,10 +236,8 @@
 ; GFX9-LABEL: shuffle_v4f16_2345:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[4:5], v[0:1], off
-; GFX9-NEXT:    global_load_dwordx2 v[1:2], v[2:3], off
-; GFX9-NEXT:    s_waitcnt vmcnt(1)
-; GFX9-NEXT:    v_mov_b32_e32 v0, v5
+; GFX9-NEXT:    global_load_dword v0, v[0:1], off offset:4
+; GFX9-NEXT:    global_load_dword v1, v[2:3], off
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x half>, <4 x half> addrspace(1)* %arg0
@@ -256,10 +250,9 @@
 ; GFX9-LABEL: shuffle_v4f16_2367:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[4:5], v[0:1], off
-; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[2:3], off
+; GFX9-NEXT:    global_load_dword v0, v[0:1], off offset:4
+; GFX9-NEXT:    global_load_dword v1, v[2:3], off offset:4
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, v5
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x half>, <4 x half> addrspace(1)* %arg0
   %val1 = load <4 x half>, <4 x half> addrspace(1)* %arg1
@@ -271,10 +264,11 @@
 ; GFX9-LABEL: shuffle_v4f16_4501:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[4:5], v[0:1], off
-; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[2:3], off
+; GFX9-NEXT:    global_load_dword v2, v[2:3], off
+; GFX9-NEXT:    global_load_dword v1, v[0:1], off
+; GFX9-NEXT:    s_waitcnt vmcnt(1)
+; GFX9-NEXT:    v_mov_b32_e32 v0, v2
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v1, v4
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x half>, <4 x half> addrspace(1)* %arg0
   %val1 = load <4 x half>, <4 x half> addrspace(1)* %arg1
@@ -286,10 +280,11 @@
 ; GFX9-LABEL: shuffle_v4f16_4523:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[0:1], off
-; GFX9-NEXT:    global_load_dwordx2 v[2:3], v[2:3], off
-; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    global_load_dword v2, v[2:3], off
+; GFX9-NEXT:    global_load_dword v1, v[0:1], off offset:4
+; GFX9-NEXT:    s_waitcnt vmcnt(1)
 ; GFX9-NEXT:    v_mov_b32_e32 v0, v2
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x half>, <4 x half> addrspace(1)* %arg0
   %val1 = load <4 x half>, <4 x half> addrspace(1)* %arg1
@@ -301,7 +296,7 @@
 ; GFX9-LABEL: shuffle_v4f16_4545:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[2:3], off
+; GFX9-NEXT:    global_load_dword v0, v[2:3], off
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    v_mov_b32_e32 v1, v0
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
@@ -328,11 +323,11 @@
 ; GFX9-LABEL: shuffle_v4f16_6701:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[4:5], v[0:1], off
-; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[2:3], off
+; GFX9-NEXT:    global_load_dword v2, v[2:3], off offset:4
+; GFX9-NEXT:    global_load_dword v1, v[0:1], off
+; GFX9-NEXT:    s_waitcnt vmcnt(1)
+; GFX9-NEXT:    v_mov_b32_e32 v0, v2
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, v1
-; GFX9-NEXT:    v_mov_b32_e32 v1, v4
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x half>, <4 x half> addrspace(1)* %arg0
   %val1 = load <4 x half>, <4 x half> addrspace(1)* %arg1
@@ -344,10 +339,11 @@
 ; GFX9-LABEL: shuffle_v4f16_6723:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[0:1], off
-; GFX9-NEXT:    global_load_dwordx2 v[2:3], v[2:3], off
+; GFX9-NEXT:    global_load_dword v2, v[2:3], off offset:4
+; GFX9-NEXT:    global_load_dword v1, v[0:1], off offset:4
+; GFX9-NEXT:    s_waitcnt vmcnt(1)
+; GFX9-NEXT:    v_mov_b32_e32 v0, v2
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, v3
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x half>, <4 x half> addrspace(1)* %arg0
   %val1 = load <4 x half>, <4 x half> addrspace(1)* %arg1
@@ -373,9 +369,9 @@
 ; GFX9-LABEL: shuffle_v4f16_6767:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[2:3], off
+; GFX9-NEXT:    global_load_dword v0, v[2:3], off offset:4
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, v1
+; GFX9-NEXT:    v_mov_b32_e32 v1, v0
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x half>, <4 x half> addrspace(1)* %arg0
   %val1 = load <4 x half>, <4 x half> addrspace(1)* %arg1
@@ -388,13 +384,12 @@
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX9-NEXT:    global_load_dwordx2 v[2:3], v[2:3], off
-; GFX9-NEXT:    global_load_dwordx2 v[4:5], v[0:1], off
-; GFX9-NEXT:    v_mov_b32_e32 v0, 0xffff
+; GFX9-NEXT:    global_load_dword v0, v[0:1], off offset:4
+; GFX9-NEXT:    v_mov_b32_e32 v1, 0xffff
 ; GFX9-NEXT:    s_waitcnt vmcnt(1)
-; GFX9-NEXT:    v_and_b32_sdwa v0, v0, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
-; GFX9-NEXT:    v_lshl_or_b32 v1, v3, 16, v0
+; GFX9-NEXT:    v_and_b32_sdwa v1, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; GFX9-NEXT:    v_lshl_or_b32 v1, v3, 16, v1
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, v5
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x half>, <4 x half> addrspace(1)* %arg0
   %val1 = load <4 x half>, <4 x half> addrspace(1)* %arg1
@@ -407,11 +402,12 @@
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX9-NEXT:    global_load_dwordx2 v[2:3], v[2:3], off
-; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[0:1], off
-; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    global_load_dword v1, v[0:1], off offset:4
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 0xffff
+; GFX9-NEXT:    s_waitcnt vmcnt(1)
 ; GFX9-NEXT:    v_and_b32_sdwa v0, v0, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
 ; GFX9-NEXT:    v_lshl_or_b32 v0, v3, 16, v0
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x half>, <4 x half> addrspace(1)* %arg0
   %val1 = load <4 x half>, <4 x half> addrspace(1)* %arg1
@@ -485,13 +481,12 @@
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX9-NEXT:    global_load_dwordx2 v[2:3], v[2:3], off
-; GFX9-NEXT:    global_load_dwordx2 v[4:5], v[0:1], off
-; GFX9-NEXT:    v_mov_b32_e32 v0, 0xffff
+; GFX9-NEXT:    global_load_dword v0, v[0:1], off offset:4
+; GFX9-NEXT:    v_mov_b32_e32 v1, 0xffff
 ; GFX9-NEXT:    s_waitcnt vmcnt(1)
-; GFX9-NEXT:    v_and_b32_sdwa v0, v0, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
-; GFX9-NEXT:    v_lshl_or_b32 v1, v3, 16, v0
+; GFX9-NEXT:    v_and_b32_sdwa v1, v1, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:WORD_1
+; GFX9-NEXT:    v_lshl_or_b32 v1, v3, 16, v1
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v0, v5
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x i16>, <4 x i16> addrspace(1)* %arg0
   %val1 = load <4 x i16>, <4 x i16> addrspace(1)* %arg1
@@ -503,11 +498,9 @@
 ; GFX9-LABEL: shuffle_v4i16_0167:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[0:1], v[0:1], off
-; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[1:2], v[2:3], off
+; GFX9-NEXT:    global_load_dword v0, v[0:1], off
+; GFX9-NEXT:    global_load_dword v1, v[2:3], off offset:4
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_mov_b32_e32 v1, v2
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x i16>, <4 x i16> addrspace(1)* %arg0
   %val1 = load <4 x i16>, <4 x i16> addrspace(1)* %arg1
@@ -590,12 +583,11 @@
 ; GFX9-LABEL: shuffle_v4f16_2333:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[1:2], v[0:1], off
+; GFX9-NEXT:    global_load_dword v0, v[0:1], off offset:4
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_lshrrev_b32_e32 v0, 16, v2
-; GFX9-NEXT:    v_and_b32_e32 v1, 0xffff, v0
-; GFX9-NEXT:    v_lshl_or_b32 v1, v0, 16, v1
-; GFX9-NEXT:    v_mov_b32_e32 v0, v2
+; GFX9-NEXT:    v_lshrrev_b32_e32 v1, 16, v0
+; GFX9-NEXT:    v_and_b32_e32 v2, 0xffff, v1
+; GFX9-NEXT:    v_lshl_or_b32 v1, v1, 16, v2
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x half>, <4 x half> addrspace(1)* %arg0
   %val1 = load <4 x half>, <4 x half> addrspace(1)* %arg1
@@ -607,12 +599,11 @@
 ; GFX9-LABEL: shuffle_v4f16_6667:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX9-NEXT:    global_load_dwordx2 v[1:2], v[0:1], off
+; GFX9-NEXT:    global_load_dword v0, v[0:1], off offset:4
 ; GFX9-NEXT:    s_waitcnt vmcnt(0)
-; GFX9-NEXT:    v_lshrrev_b32_e32 v0, 16, v2
-; GFX9-NEXT:    v_and_b32_e32 v1, 0xffff, v0
-; GFX9-NEXT:    v_lshl_or_b32 v1, v0, 16, v1
-; GFX9-NEXT:    v_mov_b32_e32 v0, v2
+; GFX9-NEXT:    v_lshrrev_b32_e32 v1, 16, v0
+; GFX9-NEXT:    v_and_b32_e32 v2, 0xffff, v1
+; GFX9-NEXT:    v_lshl_or_b32 v1, v1, 16, v2
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
   %val0 = load <4 x half>, <4 x half> addrspace(1)* %arg0
   %val1 = load <4 x half>, <4 x half> addrspace(1)* %arg1
diff --git a/llvm/test/CodeGen/ARM/vdup.ll b/llvm/test/CodeGen/ARM/vdup.ll
--- a/llvm/test/CodeGen/ARM/vdup.ll
+++ b/llvm/test/CodeGen/ARM/vdup.ll
@@ -429,8 +429,8 @@
 define <2 x float> @check_f32(<4 x float> %v) nounwind {
 ; CHECK-LABEL: check_f32:
 ; CHECK:       @ %bb.0:
-; CHECK-NEXT:    vmov d17, r2, r3
-; CHECK-NEXT:    vdup.32 d16, d17[1]
+; CHECK-NEXT:    vmov d16, r2, r3
+; CHECK-NEXT:    vdup.32 d16, d16[1]
 ; CHECK-NEXT:    vmov r0, r1, d16
 ; CHECK-NEXT:    mov pc, lr
   %x = extractelement <4 x float> %v, i32 3
@@ -442,8 +442,8 @@
 define <2 x i32> @check_i32(<4 x i32> %v) nounwind {
 ; CHECK-LABEL: check_i32:
 ; CHECK:       @ %bb.0:
-; CHECK-NEXT:    vmov d17, r2, r3
-; CHECK-NEXT:    vdup.32 d16, d17[1]
+; CHECK-NEXT:    vmov d16, r2, r3
+; CHECK-NEXT:    vdup.32 d16, d16[1]
 ; CHECK-NEXT:    vmov r0, r1, d16
 ; CHECK-NEXT:    mov pc, lr
   %x = extractelement <4 x i32> %v, i32 3
diff --git a/llvm/test/CodeGen/ARM/vext.ll b/llvm/test/CodeGen/ARM/vext.ll
--- a/llvm/test/CodeGen/ARM/vext.ll
+++ b/llvm/test/CodeGen/ARM/vext.ll
@@ -183,10 +183,10 @@
 ; CHECK:       @ %bb.0:
 ; CHECK-NEXT:    vld1.64 {d16, d17}, [r0]
 ; CHECK-NEXT:    vext.16 d16, d16, d17, #3
-; CHECK-NEXT:    vorr d17, d16, d16
-; CHECK-NEXT:    vld1.64 {d18, d19}, [r1]
-; CHECK-NEXT:    vuzp.16 d16, d17
-; CHECK-NEXT:    vzip.16 d16, d18
+; CHECK-NEXT:    vorr d18, d16, d16
+; CHECK-NEXT:    vldr d17, [r1]
+; CHECK-NEXT:    vuzp.16 d16, d18
+; CHECK-NEXT:    vzip.16 d16, d17
 ; CHECK-NEXT:    vmov r0, r1, d16
 ; CHECK-NEXT:    mov pc, lr
         %tmp1 = load <8 x i16>, <8 x i16>* %A
@@ -216,17 +216,15 @@
 define <4 x i16> @test_multisource(<32 x i16>* %B) nounwind {
 ; CHECK-LABEL: test_multisource:
 ; CHECK:       @ %bb.0:
+; CHECK-NEXT:    vldr d18, [r0, #32]
 ; CHECK-NEXT:    mov r1, r0
-; CHECK-NEXT:    add r2, r0, #48
-; CHECK-NEXT:    add r0, r0, #32
+; CHECK-NEXT:    vorr d22, d18, d18
 ; CHECK-NEXT:    vld1.16 {d16, d17}, [r1:128]!
-; CHECK-NEXT:    vld1.64 {d20, d21}, [r0:128]
-; CHECK-NEXT:    vorr d24, d20, d20
-; CHECK-NEXT:    vld1.64 {d18, d19}, [r2:128]
-; CHECK-NEXT:    vld1.64 {d22, d23}, [r1:128]
-; CHECK-NEXT:    vzip.16 d24, d18
-; CHECK-NEXT:    vtrn.16 q8, q11
-; CHECK-NEXT:    vext.16 d18, d20, d24, #2
+; CHECK-NEXT:    vldr d19, [r0, #48]
+; CHECK-NEXT:    vld1.64 {d20, d21}, [r1:128]
+; CHECK-NEXT:    vzip.16 d22, d19
+; CHECK-NEXT:    vtrn.16 q8, q10
+; CHECK-NEXT:    vext.16 d18, d18, d22, #2
 ; CHECK-NEXT:    vext.16 d16, d18, d16, #2
 ; CHECK-NEXT:    vext.16 d16, d16, d16, #2
 ; CHECK-NEXT:    vmov r0, r1, d16
diff --git a/llvm/test/CodeGen/ARM/vpadd.ll b/llvm/test/CodeGen/ARM/vpadd.ll
--- a/llvm/test/CodeGen/ARM/vpadd.ll
+++ b/llvm/test/CodeGen/ARM/vpadd.ll
@@ -285,11 +285,11 @@
 define void @addCombineToVPADDL_s8(<16 x i8> *%cbcr, <4 x i16> *%X) nounwind ssp {
 ; CHECK-LABEL: addCombineToVPADDL_s8:
 ; CHECK:       @ %bb.0:
-; CHECK-NEXT:    vld1.64 {d16, d17}, [r0]
-; CHECK-NEXT:    vext.8 d18, d16, d16, #1
+; CHECK-NEXT:    vldr d16, [r0]
+; CHECK-NEXT:    vext.8 d17, d16, d16, #1
 ; CHECK-NEXT:    vshl.i16 d16, d16, #8
-; CHECK-NEXT:    vshl.i16 d18, d18, #8
-; CHECK-NEXT:    vshr.s16 d17, d18, #8
+; CHECK-NEXT:    vshl.i16 d17, d17, #8
+; CHECK-NEXT:    vshr.s16 d17, d17, #8
 ; CHECK-NEXT:    vsra.s16 d17, d16, #8
 ; CHECK-NEXT:    vstr d17, [r1]
 ; CHECK-NEXT:    mov pc, lr
@@ -347,11 +347,11 @@
 define void @addCombineToVPADDL_u8(<16 x i8> *%cbcr, <4 x i16> *%X) nounwind ssp {
 ; CHECK-LABEL: addCombineToVPADDL_u8:
 ; CHECK:       @ %bb.0:
-; CHECK-NEXT:    vld1.64 {d16, d17}, [r0]
-; CHECK-NEXT:    vext.8 d18, d16, d16, #1
+; CHECK-NEXT:    vldr d16, [r0]
+; CHECK-NEXT:    vext.8 d17, d16, d16, #1
 ; CHECK-NEXT:    vbic.i16 d16, #0xff00
-; CHECK-NEXT:    vbic.i16 d18, #0xff00
-; CHECK-NEXT:    vadd.i16 d16, d18, d16
+; CHECK-NEXT:    vbic.i16 d17, #0xff00
+; CHECK-NEXT:    vadd.i16 d16, d17, d16
 ; CHECK-NEXT:    vstr d16, [r1]
 ; CHECK-NEXT:    mov pc, lr
   %tmp = load <16 x i8>, <16 x i8>* %cbcr
@@ -368,7 +368,7 @@
 define void @addCombineToVPADDL_u8_early_zext(<16 x i8> *%cbcr, <4 x i16> *%X) nounwind ssp {
 ; CHECK-LABEL: addCombineToVPADDL_u8_early_zext:
 ; CHECK:       @ %bb.0:
-; CHECK-NEXT:    vld1.64 {d16, d17}, [r0]
+; CHECK-NEXT:    vldr d16, [r0]
 ; CHECK-NEXT:    vmovl.u8 q8, d16
 ; CHECK-NEXT:    vpadd.i16 d16, d16, d17
 ; CHECK-NEXT:    vstr d16, [r1]
diff --git a/llvm/test/CodeGen/ARM/vuzp.ll b/llvm/test/CodeGen/ARM/vuzp.ll
--- a/llvm/test/CodeGen/ARM/vuzp.ll
+++ b/llvm/test/CodeGen/ARM/vuzp.ll
@@ -522,12 +522,11 @@
 define %struct.uint8x8x2_t @vuzp_extract_subvector(<16 x i8> %t) #0 {
 ; CHECK-LABEL: vuzp_extract_subvector:
 ; CHECK:       @ %bb.0:
-; CHECK-NEXT:    vmov d17, r2, r3
-; CHECK-NEXT:    vmov d16, r0, r1
-; CHECK-NEXT:    vorr d18, d17, d17
-; CHECK-NEXT:    vuzp.8 d16, d18
-; CHECK-NEXT:    vmov r0, r1, d16
-; CHECK-NEXT:    vmov r2, r3, d18
+; CHECK-NEXT:    vmov d16, r2, r3
+; CHECK-NEXT:    vmov d17, r0, r1
+; CHECK-NEXT:    vuzp.8 d17, d16
+; CHECK-NEXT:    vmov r0, r1, d17
+; CHECK-NEXT:    vmov r2, r3, d16
 ; CHECK-NEXT:    mov pc, lr
 
   %vuzp.i = shufflevector <16 x i8> %t, <16 x i8> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
diff --git a/llvm/test/CodeGen/X86/sse41.ll b/llvm/test/CodeGen/X86/sse41.ll
--- a/llvm/test/CodeGen/X86/sse41.ll
+++ b/llvm/test/CodeGen/X86/sse41.ll
@@ -97,9 +97,8 @@
 ; X86-AVX512:       ## %bb.0: ## %entry
 ; X86-AVX512-NEXT:    movl L_g16$non_lazy_ptr, %eax ## encoding: [0xa1,A,A,A,A]
 ; X86-AVX512-NEXT:    ## fixup A - offset: 1, value: L_g16$non_lazy_ptr, kind: FK_Data_4
-; X86-AVX512-NEXT:    vpbroadcastw (%eax), %xmm0 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x79,0x00]
-; X86-AVX512-NEXT:    vpmovzxbq %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x32,0xc0]
-; X86-AVX512-NEXT:    ## xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
+; X86-AVX512-NEXT:    vpmovzxbq (%eax), %xmm0 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x32,0x00]
+; X86-AVX512-NEXT:    ## xmm0 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
 ; X86-AVX512-NEXT:    retl ## encoding: [0xc3]
 ;
 ; X64-SSE-LABEL: pmovzxbq_1:
@@ -122,9 +121,8 @@
 ; X64-AVX512:       ## %bb.0: ## %entry
 ; X64-AVX512-NEXT:    movq _g16@{{.*}}(%rip), %rax ## encoding: [0x48,0x8b,0x05,A,A,A,A]
 ; X64-AVX512-NEXT:    ## fixup A - offset: 3, value: _g16@GOTPCREL-4, kind: reloc_riprel_4byte_movq_load
-; X64-AVX512-NEXT:    vpbroadcastw (%rax), %xmm0 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x79,0x00]
-; X64-AVX512-NEXT:    vpmovzxbq %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x32,0xc0]
-; X64-AVX512-NEXT:    ## xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
+; X64-AVX512-NEXT:    vpmovzxbq (%rax), %xmm0 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x32,0x00]
+; X64-AVX512-NEXT:    ## xmm0 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
 ; X64-AVX512-NEXT:    retq ## encoding: [0xc3]
 entry:
 	%0 = load i16, i16* @g16, align 2		; <i16> [#uses=1]