diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index fa3dce256046f..80d4f67e18934 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -2101,6 +2101,11 @@ X86TargetLowering::X86TargetLowering(const X86TargetMachine &TM,
     // These operations are handled on non-VLX by artificially widening in
     // isel patterns.
 
+    for (MVT VT : {MVT::f16, MVT::f32, MVT::f64, MVT::v8f16, MVT::v4f32,
+                   MVT::v2f64, MVT::v16f16, MVT::v8f32, MVT::v4f64, MVT::v32f16,
+                   MVT::v16f32, MVT::v8f64})
+      setOperationAction(ISD::FLDEXP, VT, Custom);
+
     setOperationAction(ISD::STRICT_FP_TO_UINT,  MVT::v8i32, Custom);
     setOperationAction(ISD::STRICT_FP_TO_UINT,  MVT::v4i32, Custom);
     setOperationAction(ISD::STRICT_FP_TO_UINT,  MVT::v2i32, Custom);
@@ -19149,6 +19154,72 @@ SDValue X86TargetLowering::LowerINSERT_VECTOR_ELT(SDValue Op,
   return SDValue();
 }
 
+static SDValue LowerFLDEXP(SDValue Op, const X86Subtarget &Subtarget,
+                           SelectionDAG &DAG) {
+  SDLoc DL(Op);
+  SDValue X = Op.getOperand(0);
+  MVT XTy = X.getSimpleValueType();
+  SDValue Exp = Op.getOperand(1);
+
+  switch (XTy.SimpleTy) {
+  default:
+    return SDValue();
+  case MVT::f16:
+    if (!Subtarget.hasFP16())
+      X = DAG.getFPExtendOrRound(X, DL, MVT::f32);
+    [[fallthrough]];
+  case MVT::f32:
+  case MVT::f64: {
+    MVT VT = MVT::getVectorVT(X.getSimpleValueType(),
+                              128 / X.getSimpleValueType().getSizeInBits());
+    Exp = DAG.getNode(ISD::SINT_TO_FP, DL, X.getValueType(), Exp);
+    SDValue VX = DAG.getNode(ISD::SCALAR_TO_VECTOR, DL, VT, X);
+    SDValue VExp = DAG.getNode(ISD::SCALAR_TO_VECTOR, DL, VT, Exp);
+    SDValue Scalefs = DAG.getNode(X86ISD::SCALEFS, DL, VT, VX, VExp);
+    SDValue Final = DAG.getExtractVectorElt(DL, X.getValueType(), Scalefs, 0);
+    return DAG.getFPExtendOrRound(Final, DL, XTy);
+  }
+  case MVT::v4f32:
+  case MVT::v2f64:
+  case MVT::v8f32:
+  case MVT::v4f64:
+  case MVT::v16f32:
+  case MVT::v8f64:
+    if (XTy.getSizeInBits() == 512 || Subtarget.hasVLX()) {
+      Exp = DAG.getNode(ISD::SINT_TO_FP, DL, XTy, Exp);
+      return DAG.getNode(X86ISD::SCALEF, DL, XTy, X, Exp);
+    }
+    break;
+  case MVT::v8f16:
+  case MVT::v16f16:
+    if (Subtarget.hasFP16()) {
+      if (Subtarget.hasVLX()) {
+        Exp = DAG.getNode(ISD::SINT_TO_FP, DL, XTy, Exp);
+        return DAG.getNode(X86ISD::SCALEF, DL, XTy, X, Exp);
+      }
+      break;
+    }
+    X = DAG.getFPExtendOrRound(X, DL, XTy.changeVectorElementType(MVT::f32));
+    Exp = DAG.getSExtOrTrunc(Exp, DL,
+                             X.getSimpleValueType().changeTypeToInteger());
+    break;
+  case MVT::v32f16:
+    if (Subtarget.hasFP16()) {
+      Exp = DAG.getNode(ISD::SINT_TO_FP, DL, XTy, Exp);
+      return DAG.getNode(X86ISD::SCALEF, DL, XTy, X, Exp);
+    }
+    return splitVectorOp(Op, DAG, DL);
+  }
+  SDValue WideX = widenSubVector(X, true, Subtarget, DAG, DL, 512);
+  SDValue WideExp = widenSubVector(Exp, true, Subtarget, DAG, DL, 512);
+  Exp = DAG.getNode(ISD::SINT_TO_FP, DL, WideExp.getSimpleValueType(), Exp);
+  SDValue Scalef =
+      DAG.getNode(X86ISD::SCALEF, DL, WideX.getValueType(), WideX, WideExp);
+  SDValue Final =
+      DAG.getExtractSubvector(DL, X.getSimpleValueType(), Scalef, 0);
+  return DAG.getFPExtendOrRound(Final, DL, XTy);
+}
+
 static SDValue LowerSCALAR_TO_VECTOR(SDValue Op, const X86Subtarget &Subtarget,
                                      SelectionDAG &DAG) {
   SDLoc dl(Op);
@@ -33681,7 +33752,8 @@ SDValue X86TargetLowering::LowerOperation(SDValue Op, SelectionDAG &DAG) const {
   case ISD::ADDRSPACECAST:      return LowerADDRSPACECAST(Op, DAG);
   case X86ISD::CVTPS2PH:        return LowerCVTPS2PH(Op, DAG);
   case ISD::PREFETCH:           return LowerPREFETCH(Op, Subtarget, DAG);
-  // clang-format on
+  case ISD::FLDEXP:             return LowerFLDEXP(Op, Subtarget, DAG);
+    // clang-format on
   }
 }
 
diff --git a/llvm/test/CodeGen/X86/fold-int-pow2-with-fmul-or-fdiv.ll b/llvm/test/CodeGen/X86/fold-int-pow2-with-fmul-or-fdiv.ll
index 81529aff39ff1..3f63d2a483ae0 100644
--- a/llvm/test/CodeGen/X86/fold-int-pow2-with-fmul-or-fdiv.ll
+++ b/llvm/test/CodeGen/X86/fold-int-pow2-with-fmul-or-fdiv.ll
@@ -79,38 +79,54 @@ define <4 x float> @fmul_pow2_ldexp_4xfloat(<4 x i32> %i) {
 ; CHECK-SSE-NEXT:    .cfi_def_cfa_offset 8
 ; CHECK-SSE-NEXT:    retq
 ;
-; CHECK-AVX-LABEL: fmul_pow2_ldexp_4xfloat:
-; CHECK-AVX:       # %bb.0:
-; CHECK-AVX-NEXT:    subq $40, %rsp
-; CHECK-AVX-NEXT:    .cfi_def_cfa_offset 48
-; CHECK-AVX-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-AVX-NEXT:    vextractps $1, %xmm0, %edi
-; CHECK-AVX-NEXT:    vmovss {{.*#+}} xmm0 = [9.0E+0,0.0E+0,0.0E+0,0.0E+0]
-; CHECK-AVX-NEXT:    callq ldexpf@PLT
-; CHECK-AVX-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-AVX-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; CHECK-AVX-NEXT:    vmovd %xmm0, %edi
-; CHECK-AVX-NEXT:    vmovss {{.*#+}} xmm0 = [9.0E+0,0.0E+0,0.0E+0,0.0E+0]
-; CHECK-AVX-NEXT:    callq ldexpf@PLT
-; CHECK-AVX-NEXT:    vinsertps $16, (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; CHECK-AVX-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[2,3]
-; CHECK-AVX-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-AVX-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; CHECK-AVX-NEXT:    vextractps $2, %xmm0, %edi
-; CHECK-AVX-NEXT:    vmovss {{.*#+}} xmm0 = [9.0E+0,0.0E+0,0.0E+0,0.0E+0]
-; CHECK-AVX-NEXT:    callq ldexpf@PLT
-; CHECK-AVX-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-AVX-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-AVX-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-AVX-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; CHECK-AVX-NEXT:    vextractps $3, %xmm0, %edi
-; CHECK-AVX-NEXT:    vmovss {{.*#+}} xmm0 = [9.0E+0,0.0E+0,0.0E+0,0.0E+0]
-; CHECK-AVX-NEXT:    callq ldexpf@PLT
-; CHECK-AVX-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-AVX-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
-; CHECK-AVX-NEXT:    addq $40, %rsp
-; CHECK-AVX-NEXT:    .cfi_def_cfa_offset 8
-; CHECK-AVX-NEXT:    retq
+; CHECK-AVX2-LABEL: fmul_pow2_ldexp_4xfloat:
+; CHECK-AVX2:       # %bb.0:
+; CHECK-AVX2-NEXT:    subq $40, %rsp
+; CHECK-AVX2-NEXT:    .cfi_def_cfa_offset 48
+; CHECK-AVX2-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
+; CHECK-AVX2-NEXT:    vextractps $1, %xmm0, %edi
+; CHECK-AVX2-NEXT:    vmovss {{.*#+}} xmm0 = [9.0E+0,0.0E+0,0.0E+0,0.0E+0]
+; CHECK-AVX2-NEXT:    callq ldexpf@PLT
+; CHECK-AVX2-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
+; CHECK-AVX2-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; CHECK-AVX2-NEXT:    vmovd %xmm0, %edi
+; CHECK-AVX2-NEXT:    vmovss {{.*#+}} xmm0 = [9.0E+0,0.0E+0,0.0E+0,0.0E+0]
+; CHECK-AVX2-NEXT:    callq ldexpf@PLT
+; CHECK-AVX2-NEXT:    vinsertps $16, (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
+; CHECK-AVX2-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[2,3]
+; CHECK-AVX2-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
+; CHECK-AVX2-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; CHECK-AVX2-NEXT:    vextractps $2, %xmm0, %edi
+; CHECK-AVX2-NEXT:    vmovss {{.*#+}} xmm0 = [9.0E+0,0.0E+0,0.0E+0,0.0E+0]
+; CHECK-AVX2-NEXT:    callq ldexpf@PLT
+; CHECK-AVX2-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
+; CHECK-AVX2-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
+; CHECK-AVX2-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
+; CHECK-AVX2-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
+; CHECK-AVX2-NEXT:    vextractps $3, %xmm0, %edi
+; CHECK-AVX2-NEXT:    vmovss {{.*#+}} xmm0 = [9.0E+0,0.0E+0,0.0E+0,0.0E+0]
+; CHECK-AVX2-NEXT:    callq ldexpf@PLT
+; CHECK-AVX2-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
+; CHECK-AVX2-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
+; CHECK-AVX2-NEXT:    addq $40, %rsp
+; CHECK-AVX2-NEXT:    .cfi_def_cfa_offset 8
+; CHECK-AVX2-NEXT:    retq
+;
+; CHECK-ONLY-AVX512F-LABEL: fmul_pow2_ldexp_4xfloat:
+; CHECK-ONLY-AVX512F:       # %bb.0:
+; CHECK-ONLY-AVX512F-NEXT:    vbroadcastss {{.*#+}} xmm1 = [9.0E+0,9.0E+0,9.0E+0,9.0E+0]
+; CHECK-ONLY-AVX512F-NEXT:    vmovaps %xmm0, %xmm0
+; CHECK-ONLY-AVX512F-NEXT:    vscalefps %zmm0, %zmm1, %zmm0
+; CHECK-ONLY-AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
+; CHECK-ONLY-AVX512F-NEXT:    vzeroupper
+; CHECK-ONLY-AVX512F-NEXT:    retq
+;
+; CHECK-SKX-LABEL: fmul_pow2_ldexp_4xfloat:
+; CHECK-SKX:       # %bb.0:
+; CHECK-SKX-NEXT:    vcvtdq2ps %xmm0, %xmm0
+; CHECK-SKX-NEXT:    vbroadcastss {{.*#+}} xmm1 = [9.0E+0,9.0E+0,9.0E+0,9.0E+0]
+; CHECK-SKX-NEXT:    vscalefps %xmm0, %xmm1, %xmm0
+; CHECK-SKX-NEXT:    retq
   %r = call <4 x float> @llvm.ldexp.v4f32.v4i32(<4 x float> <float 9.000000e+00, float 9.000000e+00, float 9.000000e+00, float 9.000000e+00>, <4 x i32> %i)
   ret <4 x float> %r
 }
@@ -562,79 +578,11 @@ define <8 x half> @fmul_pow2_ldexp_8xhalf(<8 x i16> %i) {
 ;
 ; CHECK-AVX512F-LABEL: fmul_pow2_ldexp_8xhalf:
 ; CHECK-AVX512F:       # %bb.0:
-; CHECK-AVX512F-NEXT:    subq $72, %rsp
-; CHECK-AVX512F-NEXT:    .cfi_def_cfa_offset 80
-; CHECK-AVX512F-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
-; CHECK-AVX512F-NEXT:    vpextrw $7, %xmm0, %eax
-; CHECK-AVX512F-NEXT:    movswl %ax, %edi
-; CHECK-AVX512F-NEXT:    vmovss {{.*#+}} xmm0 = [8.192E+3,0.0E+0,0.0E+0,0.0E+0]
-; CHECK-AVX512F-NEXT:    callq ldexpf@PLT
-; CHECK-AVX512F-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; CHECK-AVX512F-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-AVX512F-NEXT:    vmovdqa (%rsp), %xmm0 # 16-byte Reload
-; CHECK-AVX512F-NEXT:    vpextrw $6, %xmm0, %eax
-; CHECK-AVX512F-NEXT:    movswl %ax, %edi
-; CHECK-AVX512F-NEXT:    vmovss {{.*#+}} xmm0 = [8.192E+3,0.0E+0,0.0E+0,0.0E+0]
-; CHECK-AVX512F-NEXT:    callq ldexpf@PLT
-; CHECK-AVX512F-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; CHECK-AVX512F-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; CHECK-AVX512F-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; CHECK-AVX512F-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-AVX512F-NEXT:    vmovdqa (%rsp), %xmm0 # 16-byte Reload
-; CHECK-AVX512F-NEXT:    vpextrw $5, %xmm0, %eax
-; CHECK-AVX512F-NEXT:    movswl %ax, %edi
-; CHECK-AVX512F-NEXT:    vmovss {{.*#+}} xmm0 = [8.192E+3,0.0E+0,0.0E+0,0.0E+0]
-; CHECK-AVX512F-NEXT:    callq ldexpf@PLT
-; CHECK-AVX512F-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; CHECK-AVX512F-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-AVX512F-NEXT:    vmovdqa (%rsp), %xmm0 # 16-byte Reload
-; CHECK-AVX512F-NEXT:    vpextrw $4, %xmm0, %eax
-; CHECK-AVX512F-NEXT:    movswl %ax, %edi
-; CHECK-AVX512F-NEXT:    vmovss {{.*#+}} xmm0 = [8.192E+3,0.0E+0,0.0E+0,0.0E+0]
-; CHECK-AVX512F-NEXT:    callq ldexpf@PLT
-; CHECK-AVX512F-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; CHECK-AVX512F-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; CHECK-AVX512F-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; CHECK-AVX512F-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; CHECK-AVX512F-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; CHECK-AVX512F-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-AVX512F-NEXT:    vmovdqa (%rsp), %xmm0 # 16-byte Reload
-; CHECK-AVX512F-NEXT:    vpextrw $3, %xmm0, %eax
-; CHECK-AVX512F-NEXT:    movswl %ax, %edi
-; CHECK-AVX512F-NEXT:    vmovss {{.*#+}} xmm0 = [8.192E+3,0.0E+0,0.0E+0,0.0E+0]
-; CHECK-AVX512F-NEXT:    callq ldexpf@PLT
-; CHECK-AVX512F-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; CHECK-AVX512F-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-AVX512F-NEXT:    vmovdqa (%rsp), %xmm0 # 16-byte Reload
-; CHECK-AVX512F-NEXT:    vpextrw $2, %xmm0, %eax
-; CHECK-AVX512F-NEXT:    movswl %ax, %edi
-; CHECK-AVX512F-NEXT:    vmovss {{.*#+}} xmm0 = [8.192E+3,0.0E+0,0.0E+0,0.0E+0]
-; CHECK-AVX512F-NEXT:    callq ldexpf@PLT
-; CHECK-AVX512F-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; CHECK-AVX512F-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; CHECK-AVX512F-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; CHECK-AVX512F-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-AVX512F-NEXT:    vmovdqa (%rsp), %xmm0 # 16-byte Reload
-; CHECK-AVX512F-NEXT:    vpextrw $1, %xmm0, %eax
-; CHECK-AVX512F-NEXT:    movswl %ax, %edi
-; CHECK-AVX512F-NEXT:    vmovss {{.*#+}} xmm0 = [8.192E+3,0.0E+0,0.0E+0,0.0E+0]
-; CHECK-AVX512F-NEXT:    callq ldexpf@PLT
-; CHECK-AVX512F-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; CHECK-AVX512F-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-AVX512F-NEXT:    vmovdqa (%rsp), %xmm0 # 16-byte Reload
-; CHECK-AVX512F-NEXT:    vmovd %xmm0, %eax
-; CHECK-AVX512F-NEXT:    movswl %ax, %edi
-; CHECK-AVX512F-NEXT:    vmovss {{.*#+}} xmm0 = [8.192E+3,0.0E+0,0.0E+0,0.0E+0]
-; CHECK-AVX512F-NEXT:    callq ldexpf@PLT
-; CHECK-AVX512F-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; CHECK-AVX512F-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; CHECK-AVX512F-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; CHECK-AVX512F-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; CHECK-AVX512F-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; CHECK-AVX512F-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; CHECK-AVX512F-NEXT:    # xmm0 = xmm0[0],mem[0]
-; CHECK-AVX512F-NEXT:    addq $72, %rsp
-; CHECK-AVX512F-NEXT:    .cfi_def_cfa_offset 8
+; CHECK-AVX512F-NEXT:    vbroadcastss {{.*#+}} ymm1 = [8.192E+3,8.192E+3,8.192E+3,8.192E+3,8.192E+3,8.192E+3,8.192E+3,8.192E+3]
+; CHECK-AVX512F-NEXT:    vpmovsxwd %xmm0, %ymm0
+; CHECK-AVX512F-NEXT:    vscalefps %zmm0, %zmm1, %zmm0
+; CHECK-AVX512F-NEXT:    vcvtps2ph $4, %ymm0, %xmm0
+; CHECK-AVX512F-NEXT:    vzeroupper
 ; CHECK-AVX512F-NEXT:    retq
   %r = call <8 x half> @llvm.ldexp.v8f16.v8i16(<8 x half> <half 0xH7000, half 0xH7000, half 0xH7000, half 0xH7000, half 0xH7000, half 0xH7000, half 0xH7000, half 0xH7000>, <8 x i16> %i)
   ret <8 x half> %r
diff --git a/llvm/test/CodeGen/X86/ldexp-avx512.ll b/llvm/test/CodeGen/X86/ldexp-avx512.ll
index 21491bc2cc8f5..bb6dc3162eb1f 100644
--- a/llvm/test/CodeGen/X86/ldexp-avx512.ll
+++ b/llvm/test/CodeGen/X86/ldexp-avx512.ll
@@ -1,16 +1,37 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 3
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,AVX512
-; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=CHECK,AVX512VL
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512F
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512f,+avx512fp16 | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512FP16
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl | FileCheck %s --check-prefixes=CHECK,AVX512VL,AVX512VLF
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+avx512fp16 | FileCheck %s --check-prefixes=CHECK,AVX512VLFP16
 
 define half @test_half(half %x, i32 %exp) nounwind {
-; CHECK-LABEL: test_half:
-; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    pushq %rax
-; CHECK-NEXT:    vcvtph2ps %xmm0, %xmm0
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; CHECK-NEXT:    popq %rax
-; CHECK-NEXT:    retq
+; AVX512F-LABEL: test_half:
+; AVX512F:       # %bb.0: # %entry
+; AVX512F-NEXT:    vcvtsi2ss %edi, %xmm15, %xmm1
+; AVX512F-NEXT:    vcvtph2ps %xmm0, %xmm0
+; AVX512F-NEXT:    vscalefss %xmm1, %xmm0, %xmm0
+; AVX512F-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
+; AVX512F-NEXT:    retq
+;
+; AVX512FP16-LABEL: test_half:
+; AVX512FP16:       # %bb.0: # %entry
+; AVX512FP16-NEXT:    vcvtsi2sh %edi, %xmm31, %xmm1
+; AVX512FP16-NEXT:    vscalefsh %xmm1, %xmm0, %xmm0
+; AVX512FP16-NEXT:    retq
+;
+; AVX512VL-LABEL: test_half:
+; AVX512VL:       # %bb.0: # %entry
+; AVX512VL-NEXT:    vcvtsi2ss %edi, %xmm15, %xmm1
+; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
+; AVX512VL-NEXT:    vscalefss %xmm1, %xmm0, %xmm0
+; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
+; AVX512VL-NEXT:    retq
+;
+; AVX512VLFP16-LABEL: test_half:
+; AVX512VLFP16:       # %bb.0: # %entry
+; AVX512VLFP16-NEXT:    vcvtsi2sh %edi, %xmm31, %xmm1
+; AVX512VLFP16-NEXT:    vscalefsh %xmm1, %xmm0, %xmm0
+; AVX512VLFP16-NEXT:    retq
 entry:
   %r = tail call fast half @llvm.ldexp.f16.i32(half %x, i32 %exp)
   ret half %r
@@ -20,7 +41,9 @@ declare half @llvm.ldexp.f16.i32(half, i32) memory(none)
 define float @test_float(float %x, i32 %exp) nounwind {
 ; CHECK-LABEL: test_float:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    jmp ldexpf@PLT # TAILCALL
+; CHECK-NEXT:    vcvtsi2ss %edi, %xmm15, %xmm1
+; CHECK-NEXT:    vscalefss %xmm1, %xmm0, %xmm0
+; CHECK-NEXT:    retq
 entry:
   %r = tail call fast float @ldexpf(float %x, i32 %exp)
   ret float %r
@@ -30,7 +53,9 @@ declare float @ldexpf(float, i32) memory(none)
 define double @test_double(double %x, i32 %exp) nounwind {
 ; CHECK-LABEL: test_double:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    jmp ldexp@PLT # TAILCALL
+; CHECK-NEXT:    vcvtsi2sd %edi, %xmm15, %xmm1
+; CHECK-NEXT:    vscalefsd %xmm1, %xmm0, %xmm0
+; CHECK-NEXT:    retq
 entry:
   %r = tail call fast double @ldexp(double %x, i32 %exp)
   ret double %r
@@ -48,220 +73,65 @@ entry:
 declare fp128 @ldexpl(fp128, i32) memory(none)
 
 define <8 x half> @test_ldexp_8xhalf(<8 x half> %x, <8 x i16> %exp) nounwind {
-; AVX512-LABEL: test_ldexp_8xhalf:
-; AVX512:       # %bb.0:
-; AVX512-NEXT:    subq $88, %rsp
-; AVX512-NEXT:    vmovdqa %xmm1, (%rsp) # 16-byte Spill
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpsrldq {{.*#+}} xmm2 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vcvtph2ps %xmm2, %xmm0
-; AVX512-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[3,3,3,3]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa (%rsp), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; AVX512-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa (%rsp), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[1,0]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa (%rsp), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; AVX512-NEXT:    vpsrlq $48, %xmm0, %xmm0
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa (%rsp), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[1,1,3,3]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa (%rsp), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm0
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa (%rsp), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vcvtph2ps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    vmovdqa (%rsp), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vmovd %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0]
-; AVX512-NEXT:    addq $88, %rsp
-; AVX512-NEXT:    retq
+; AVX512F-LABEL: test_ldexp_8xhalf:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    vpmovsxwd %xmm1, %ymm1
+; AVX512F-NEXT:    vcvtph2ps %xmm0, %ymm0
+; AVX512F-NEXT:    vscalefps %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vcvtps2ph $4, %ymm0, %xmm0
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512FP16-LABEL: test_ldexp_8xhalf:
+; AVX512FP16:       # %bb.0:
+; AVX512FP16-NEXT:    vxorps %xmm2, %xmm2, %xmm2
+; AVX512FP16-NEXT:    vinsertf32x4 $0, %xmm0, %zmm2, %zmm0
+; AVX512FP16-NEXT:    vmovaps %xmm1, %xmm1
+; AVX512FP16-NEXT:    vscalefph %zmm1, %zmm0, %zmm0
+; AVX512FP16-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
+; AVX512FP16-NEXT:    vzeroupper
+; AVX512FP16-NEXT:    retq
 ;
 ; AVX512VL-LABEL: test_ldexp_8xhalf:
 ; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    subq $88, %rsp
-; AVX512VL-NEXT:    vmovdqa %xmm1, (%rsp) # 16-byte Spill
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm2 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512VL-NEXT:    vcvtph2ps %xmm2, %xmm0
-; AVX512VL-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[3,3,3,3]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa (%rsp), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa (%rsp), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[1,0]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa (%rsp), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa (%rsp), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[1,1,3,3]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa (%rsp), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa (%rsp), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vcvtph2ps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vmovdqa (%rsp), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vmovd %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512VL-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0]
-; AVX512VL-NEXT:    addq $88, %rsp
+; AVX512VL-NEXT:    vpmovsxwd %xmm1, %ymm1
+; AVX512VL-NEXT:    vcvtph2ps %xmm0, %ymm0
+; AVX512VL-NEXT:    vscalefps %zmm1, %zmm0, %zmm0
+; AVX512VL-NEXT:    vcvtps2ph $4, %ymm0, %xmm0
+; AVX512VL-NEXT:    vzeroupper
 ; AVX512VL-NEXT:    retq
+;
+; AVX512VLFP16-LABEL: test_ldexp_8xhalf:
+; AVX512VLFP16:       # %bb.0:
+; AVX512VLFP16-NEXT:    vcvtw2ph %xmm1, %xmm1
+; AVX512VLFP16-NEXT:    vscalefph %xmm1, %xmm0, %xmm0
+; AVX512VLFP16-NEXT:    retq
   %r = call <8 x half> @llvm.ldexp.v8f16.v8i16(<8 x half> %x, <8 x i16> %exp)
   ret <8 x half> %r
 }
 declare <8 x half> @llvm.ldexp.v8f16.v8i16(<8 x half>, <8 x i16>)
 
 define <4 x float> @test_ldexp_4xfloat(<4 x float> %x, <4 x i32> %exp) nounwind {
-; CHECK-LABEL: test_ldexp_4xfloat:
-; CHECK:       # %bb.0:
-; CHECK-NEXT:    subq $56, %rsp
-; CHECK-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovd %xmm1, %edi
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $1, %xmm0, %edi
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $2, %xmm0, %edi
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $3, %xmm0, %edi
-; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
-; CHECK-NEXT:    addq $56, %rsp
-; CHECK-NEXT:    retq
+; AVX512-LABEL: test_ldexp_4xfloat:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vmovaps %xmm1, %xmm1
+; AVX512-NEXT:    vmovaps %xmm0, %xmm0
+; AVX512-NEXT:    vscalefps %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
+;
+; AVX512VL-LABEL: test_ldexp_4xfloat:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vcvtdq2ps %xmm1, %xmm1
+; AVX512VL-NEXT:    vscalefps %xmm1, %xmm0, %xmm0
+; AVX512VL-NEXT:    retq
+;
+; AVX512VLFP16-LABEL: test_ldexp_4xfloat:
+; AVX512VLFP16:       # %bb.0:
+; AVX512VLFP16-NEXT:    vcvtdq2ps %xmm1, %xmm1
+; AVX512VLFP16-NEXT:    vscalefps %xmm1, %xmm0, %xmm0
+; AVX512VLFP16-NEXT:    retq
   %r = call <4 x float> @llvm.ldexp.v4f32.v4i32(<4 x float> %x, <4 x i32> %exp)
   ret <4 x float> %r
 }
@@ -270,20 +140,13 @@ declare <4 x float> @llvm.ldexp.v4f32.v4i32(<4 x float>, <4 x i32>)
 define <2 x double> @test_ldexp_2xdouble(<2 x double> %x, <2 x i32> %exp) nounwind {
 ; CHECK-LABEL: test_ldexp_2xdouble:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    subq $56, %rsp
-; CHECK-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovd %xmm1, %edi
-; CHECK-NEXT:    callq ldexp@PLT
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $1, %xmm0, %edi
-; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq ldexp@PLT
-; CHECK-NEXT:    vmovapd {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; CHECK-NEXT:    addq $56, %rsp
+; CHECK-NEXT:    vcvtdq2pd %xmm1, %xmm2
+; CHECK-NEXT:    vscalefsd %xmm2, %xmm0, %xmm2
+; CHECK-NEXT:    vshufpd {{.*#+}} xmm0 = xmm0[1,0]
+; CHECK-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[1,1,1,1]
+; CHECK-NEXT:    vcvtdq2pd %xmm1, %xmm1
+; CHECK-NEXT:    vscalefsd %xmm1, %xmm0, %xmm0
+; CHECK-NEXT:    vunpcklpd {{.*#+}} xmm0 = xmm2[0],xmm0[0]
 ; CHECK-NEXT:    retq
   %r = call <2 x double> @llvm.ldexp.v2f64.v2i32(<2 x double> %x, <2 x i32> %exp)
   ret <2 x double> %r
@@ -291,1225 +154,133 @@ define <2 x double> @test_ldexp_2xdouble(<2 x double> %x, <2 x i32> %exp) nounwi
 declare <2 x double> @llvm.ldexp.v2f64.v2i32(<2 x double>, <2 x i32>)
 
 define <16 x half> @test_ldexp_16xhalf(<16 x half> %x, <16 x i16> %exp) nounwind {
-; AVX512-LABEL: test_ldexp_16xhalf:
-; AVX512:       # %bb.0:
-; AVX512-NEXT:    subq $168, %rsp
-; AVX512-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; AVX512-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm0
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vextracti128 $1, %ymm1, %xmm1
-; AVX512-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[3,3,3,3]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; AVX512-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; AVX512-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[1,0]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; AVX512-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
-; AVX512-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; AVX512-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[3,3,3,3]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
-; AVX512-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
-; AVX512-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; AVX512-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
-; AVX512-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[1,0]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
-; AVX512-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vpunpckldq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; AVX512-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; AVX512-NEXT:    vpsrlq $48, %xmm0, %xmm0
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; AVX512-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[1,1,3,3]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm0
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vcvtph2ps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vmovd %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vpunpckldq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; AVX512-NEXT:    vpsrlq $48, %xmm0, %xmm0
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
-; AVX512-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[1,1,3,3]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
-; AVX512-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm0
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
-; AVX512-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; AVX512-NEXT:    vcvtph2ps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
-; AVX512-NEXT:    vmovd %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; AVX512-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 32-byte Folded Reload
-; AVX512-NEXT:    # ymm0 = ymm0[0],mem[0],ymm0[2],mem[2]
-; AVX512-NEXT:    addq $168, %rsp
-; AVX512-NEXT:    retq
+; AVX512F-LABEL: test_ldexp_16xhalf:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    vpmovsxwd %ymm1, %zmm1
+; AVX512F-NEXT:    vcvtph2ps %ymm0, %zmm0
+; AVX512F-NEXT:    vscalefps %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vcvtps2ph $4, %zmm0, %ymm0
+; AVX512F-NEXT:    retq
+;
+; AVX512FP16-LABEL: test_ldexp_16xhalf:
+; AVX512FP16:       # %bb.0:
+; AVX512FP16-NEXT:    vxorps %xmm2, %xmm2, %xmm2
+; AVX512FP16-NEXT:    vinsertf64x4 $0, %ymm0, %zmm2, %zmm0
+; AVX512FP16-NEXT:    vmovaps %ymm1, %ymm1
+; AVX512FP16-NEXT:    vscalefph %zmm1, %zmm0, %zmm0
+; AVX512FP16-NEXT:    # kill: def $ymm0 killed $ymm0 killed $zmm0
+; AVX512FP16-NEXT:    retq
 ;
 ; AVX512VL-LABEL: test_ldexp_16xhalf:
 ; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    subq $168, %rsp
-; AVX512VL-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; AVX512VL-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; AVX512VL-NEXT:    vextracti128 $1, %ymm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vextracti128 $1, %ymm1, %xmm1
-; AVX512VL-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[3,3,3,3]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; AVX512VL-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[1,0]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
-; AVX512VL-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; AVX512VL-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[3,3,3,3]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
-; AVX512VL-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
-; AVX512VL-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
-; AVX512VL-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[1,0]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
-; AVX512VL-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vpunpckldq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512VL-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; AVX512VL-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; AVX512VL-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[1,1,3,3]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
-; AVX512VL-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vcvtph2ps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vmovd %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vpunpckldq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
-; AVX512VL-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[1,1,3,3]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
-; AVX512VL-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
-; AVX512VL-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; AVX512VL-NEXT:    vcvtph2ps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm1 # 32-byte Reload
-; AVX512VL-NEXT:    vmovd %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512VL-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 32-byte Folded Reload
-; AVX512VL-NEXT:    # ymm0 = ymm0[0],mem[0],ymm0[2],mem[2]
-; AVX512VL-NEXT:    addq $168, %rsp
+; AVX512VL-NEXT:    vpmovsxwd %ymm1, %zmm1
+; AVX512VL-NEXT:    vcvtph2ps %ymm0, %zmm0
+; AVX512VL-NEXT:    vscalefps %zmm1, %zmm0, %zmm0
+; AVX512VL-NEXT:    vcvtps2ph $4, %zmm0, %ymm0
 ; AVX512VL-NEXT:    retq
+;
+; AVX512VLFP16-LABEL: test_ldexp_16xhalf:
+; AVX512VLFP16:       # %bb.0:
+; AVX512VLFP16-NEXT:    vcvtw2ph %ymm1, %ymm1
+; AVX512VLFP16-NEXT:    vscalefph %ymm1, %ymm0, %ymm0
+; AVX512VLFP16-NEXT:    retq
   %r = call <16 x half> @llvm.ldexp.v16f16.v16i16(<16 x half> %x, <16 x i16> %exp)
   ret <16 x half> %r
 }
 declare <16 x half> @llvm.ldexp.v16f16.v16i16(<16 x half>, <16 x i16>)
 
 define <8 x float> @test_ldexp_8xfloat(<8 x float> %x, <8 x i32> %exp) nounwind {
-; CHECK-LABEL: test_ldexp_8xfloat:
-; CHECK:       # %bb.0:
-; CHECK-NEXT:    subq $120, %rsp
-; CHECK-NEXT:    vmovdqu %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; CHECK-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; CHECK-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vextracti128 $1, %ymm1, %xmm1
-; CHECK-NEXT:    vmovdqa %xmm1, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovd %xmm1, %edi
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $1, %xmm0, %edi
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $2, %xmm0, %edi
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $3, %xmm0, %edi
-; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; CHECK-NEXT:    vmovd %xmm0, %edi
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; CHECK-NEXT:    # kill: def $xmm0 killed $xmm0 killed $ymm0
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; CHECK-NEXT:    vextractps $1, %xmm0, %edi
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; CHECK-NEXT:    vextractps $2, %xmm0, %edi
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; CHECK-NEXT:    vextractps $3, %xmm0, %edi
-; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
-; CHECK-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; CHECK-NEXT:    addq $120, %rsp
-; CHECK-NEXT:    retq
+; AVX512-LABEL: test_ldexp_8xfloat:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vmovaps %ymm1, %ymm1
+; AVX512-NEXT:    vmovaps %ymm0, %ymm0
+; AVX512-NEXT:    vscalefps %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    # kill: def $ymm0 killed $ymm0 killed $zmm0
+; AVX512-NEXT:    retq
+;
+; AVX512VL-LABEL: test_ldexp_8xfloat:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vcvtdq2ps %ymm1, %ymm1
+; AVX512VL-NEXT:    vscalefps %ymm1, %ymm0, %ymm0
+; AVX512VL-NEXT:    retq
+;
+; AVX512VLFP16-LABEL: test_ldexp_8xfloat:
+; AVX512VLFP16:       # %bb.0:
+; AVX512VLFP16-NEXT:    vcvtdq2ps %ymm1, %ymm1
+; AVX512VLFP16-NEXT:    vscalefps %ymm1, %ymm0, %ymm0
+; AVX512VLFP16-NEXT:    retq
   %r = call <8 x float> @llvm.ldexp.v8f32.v8i32(<8 x float> %x, <8 x i32> %exp)
   ret <8 x float> %r
 }
 declare <8 x float> @llvm.ldexp.v8f32.v8i32(<8 x float>, <8 x i32>)
 
 define <4 x double> @test_ldexp_4xdouble(<4 x double> %x, <4 x i32> %exp) nounwind {
-; CHECK-LABEL: test_ldexp_4xdouble:
-; CHECK:       # %bb.0:
-; CHECK-NEXT:    subq $88, %rsp
-; CHECK-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; CHECK-NEXT:    vextractps $2, %xmm1, %edi
-; CHECK-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexp@PLT
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $3, %xmm0, %edi
-; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq ldexp@PLT
-; CHECK-NEXT:    vmovapd {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; CHECK-NEXT:    vmovapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vmovd %xmm0, %edi
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; CHECK-NEXT:    # kill: def $xmm0 killed $xmm0 killed $ymm0
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexp@PLT
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $1, %xmm0, %edi
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq ldexp@PLT
-; CHECK-NEXT:    vmovapd (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; CHECK-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; CHECK-NEXT:    addq $88, %rsp
-; CHECK-NEXT:    retq
+; AVX512-LABEL: test_ldexp_4xdouble:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vmovapd %xmm1, %xmm1
+; AVX512-NEXT:    vmovapd %ymm0, %ymm0
+; AVX512-NEXT:    vscalefpd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    # kill: def $ymm0 killed $ymm0 killed $zmm0
+; AVX512-NEXT:    retq
+;
+; AVX512VL-LABEL: test_ldexp_4xdouble:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vcvtdq2pd %xmm1, %ymm1
+; AVX512VL-NEXT:    vscalefpd %ymm1, %ymm0, %ymm0
+; AVX512VL-NEXT:    retq
+;
+; AVX512VLFP16-LABEL: test_ldexp_4xdouble:
+; AVX512VLFP16:       # %bb.0:
+; AVX512VLFP16-NEXT:    vcvtdq2pd %xmm1, %ymm1
+; AVX512VLFP16-NEXT:    vscalefpd %ymm1, %ymm0, %ymm0
+; AVX512VLFP16-NEXT:    retq
   %r = call <4 x double> @llvm.ldexp.v4f64.v4i32(<4 x double> %x, <4 x i32> %exp)
   ret <4 x double> %r
 }
 declare <4 x double> @llvm.ldexp.v4f64.v4i32(<4 x double>, <4 x i32>)
 
 define <32 x half> @test_ldexp_32xhalf(<32 x half> %x, <32 x i16> %exp) nounwind {
-; AVX512-LABEL: test_ldexp_32xhalf:
-; AVX512:       # %bb.0:
-; AVX512-NEXT:    subq $360, %rsp # imm = 0x168
-; AVX512-NEXT:    vmovdqu64 %zmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
-; AVX512-NEXT:    vmovdqu64 %zmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
-; AVX512-NEXT:    vextracti32x4 $3, %zmm0, %xmm0
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vextracti32x4 $3, %zmm1, %xmm1
-; AVX512-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; AVX512-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[3,3,3,3]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; AVX512-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[1,0]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vpunpckldq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; AVX512-NEXT:    vextracti32x4 $2, %zmm0, %xmm0
-; AVX512-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
-; AVX512-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512-NEXT:    vextracti32x4 $2, %zmm1, %xmm1
-; AVX512-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpermilps $255, (%rsp), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[3,3,3,3]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqa (%rsp), %xmm0 # 16-byte Reload
-; AVX512-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpermilpd $1, (%rsp), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[1,0]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; AVX512-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; AVX512-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm0
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512-NEXT:    vextracti128 $1, %ymm1, %xmm1
-; AVX512-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[3,3,3,3]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; AVX512-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[1,0]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; AVX512-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[3,3,3,3]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; AVX512-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[1,0]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; AVX512-NEXT:    vinserti64x4 $1, {{[-0-9]+}}(%r{{[sb]}}p), %zmm0, %zmm0 # 32-byte Folded Reload
-; AVX512-NEXT:    vmovdqu64 %zmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; AVX512-NEXT:    vpsrlq $48, %xmm0, %xmm0
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[1,1,3,3]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm0
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vcvtph2ps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vmovd %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqa (%rsp), %xmm0 # 16-byte Reload
-; AVX512-NEXT:    vpsrlq $48, %xmm0, %xmm0
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[1,1,3,3]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqa (%rsp), %xmm0 # 16-byte Reload
-; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm0
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vcvtph2ps (%rsp), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vmovd %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; AVX512-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; AVX512-NEXT:    vpsrlq $48, %xmm0, %xmm0
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[1,1,3,3]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm0
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; AVX512-NEXT:    vcvtph2ps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512-NEXT:    vmovd %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; AVX512-NEXT:    vpsrlq $48, %xmm0, %xmm0
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; AVX512-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = mem[1,1,3,3]
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
-; AVX512-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm0
-; AVX512-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512-NEXT:    vcvtph2ps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512-NEXT:    vmovd %xmm1, %eax
-; AVX512-NEXT:    movswl %ax, %edi
-; AVX512-NEXT:    vzeroupper
-; AVX512-NEXT:    callq ldexpf@PLT
-; AVX512-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512-NEXT:    vpunpckldq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; AVX512-NEXT:    vinserti64x4 $1, {{[-0-9]+}}(%r{{[sb]}}p), %zmm0, %zmm0 # 32-byte Folded Reload
-; AVX512-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %zmm0, %zmm0 # 64-byte Folded Reload
-; AVX512-NEXT:    # zmm0 = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
-; AVX512-NEXT:    addq $360, %rsp # imm = 0x168
-; AVX512-NEXT:    retq
+; AVX512F-LABEL: test_ldexp_32xhalf:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    vpmovsxwd %ymm1, %zmm2
+; AVX512F-NEXT:    vcvtph2ps %ymm0, %zmm3
+; AVX512F-NEXT:    vscalefps %zmm2, %zmm3, %zmm2
+; AVX512F-NEXT:    vcvtps2ph $4, %zmm2, %ymm2
+; AVX512F-NEXT:    vextracti64x4 $1, %zmm1, %ymm1
+; AVX512F-NEXT:    vpmovsxwd %ymm1, %zmm1
+; AVX512F-NEXT:    vextractf64x4 $1, %zmm0, %ymm0
+; AVX512F-NEXT:    vcvtph2ps %ymm0, %zmm0
+; AVX512F-NEXT:    vscalefps %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vcvtps2ph $4, %zmm0, %ymm0
+; AVX512F-NEXT:    vinsertf64x4 $1, %ymm0, %zmm2, %zmm0
+; AVX512F-NEXT:    retq
+;
+; AVX512FP16-LABEL: test_ldexp_32xhalf:
+; AVX512FP16:       # %bb.0:
+; AVX512FP16-NEXT:    vcvtw2ph %zmm1, %zmm1
+; AVX512FP16-NEXT:    vscalefph %zmm1, %zmm0, %zmm0
+; AVX512FP16-NEXT:    retq
 ;
 ; AVX512VL-LABEL: test_ldexp_32xhalf:
 ; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    subq $360, %rsp # imm = 0x168
-; AVX512VL-NEXT:    vmovdqu64 %zmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
-; AVX512VL-NEXT:    vmovdqu64 %zmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
-; AVX512VL-NEXT:    vextracti32x4 $3, %zmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vextracti32x4 $3, %zmm1, %xmm1
-; AVX512VL-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; AVX512VL-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[3,3,3,3]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[1,0]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vpunpckldq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; AVX512VL-NEXT:    vextracti32x4 $2, %zmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
-; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512VL-NEXT:    vextracti32x4 $2, %zmm1, %xmm1
-; AVX512VL-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpermilps $255, (%rsp), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[3,3,3,3]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vmovdqa (%rsp), %xmm0 # 16-byte Reload
-; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpermilpd $1, (%rsp), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[1,0]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512VL-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; AVX512VL-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; AVX512VL-NEXT:    vextracti128 $1, %ymm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512VL-NEXT:    vextracti128 $1, %ymm1, %xmm1
-; AVX512VL-NEXT:    vmovdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[3,3,3,3]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[1,0]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512VL-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[3,3,3,3]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512VL-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512VL-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[1,0]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512VL-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512VL-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vinserti64x4 $1, {{[-0-9]+}}(%r{{[sb]}}p), %zmm0, %zmm0 # 32-byte Folded Reload
-; AVX512VL-NEXT:    vmovdqu64 %zmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
-; AVX512VL-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[1,1,3,3]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vcvtph2ps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vmovd %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpsrlq $48, (%rsp), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vmovshdup (%rsp), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[1,1,3,3]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpsrld $16, (%rsp), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vcvtph2ps (%rsp), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vmovd %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512VL-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vmovdqu %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; AVX512VL-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[1,1,3,3]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; AVX512VL-NEXT:    vcvtph2ps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; AVX512VL-NEXT:    vmovd %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512VL-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vpsrlq $48, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512VL-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; AVX512VL-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = mem[1,1,3,3]
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512VL-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
-; AVX512VL-NEXT:    vpsrld $16, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512VL-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; AVX512VL-NEXT:    vcvtph2ps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm1 # 64-byte Reload
-; AVX512VL-NEXT:    vmovd %xmm1, %eax
-; AVX512VL-NEXT:    movswl %ax, %edi
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    callq ldexpf@PLT
-; AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpunpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; AVX512VL-NEXT:    vpunpckldq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; AVX512VL-NEXT:    vinserti128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; AVX512VL-NEXT:    vinserti64x4 $1, {{[-0-9]+}}(%r{{[sb]}}p), %zmm0, %zmm0 # 32-byte Folded Reload
-; AVX512VL-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %zmm0, %zmm0 # 64-byte Folded Reload
-; AVX512VL-NEXT:    # zmm0 = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
-; AVX512VL-NEXT:    addq $360, %rsp # imm = 0x168
+; AVX512VL-NEXT:    vpmovsxwd %ymm1, %zmm2
+; AVX512VL-NEXT:    vcvtph2ps %ymm0, %zmm3
+; AVX512VL-NEXT:    vscalefps %zmm2, %zmm3, %zmm2
+; AVX512VL-NEXT:    vcvtps2ph $4, %zmm2, %ymm2
+; AVX512VL-NEXT:    vextracti64x4 $1, %zmm1, %ymm1
+; AVX512VL-NEXT:    vpmovsxwd %ymm1, %zmm1
+; AVX512VL-NEXT:    vextractf64x4 $1, %zmm0, %ymm0
+; AVX512VL-NEXT:    vcvtph2ps %ymm0, %zmm0
+; AVX512VL-NEXT:    vscalefps %zmm1, %zmm0, %zmm0
+; AVX512VL-NEXT:    vcvtps2ph $4, %zmm0, %ymm0
+; AVX512VL-NEXT:    vinsertf64x4 $1, %ymm0, %zmm2, %zmm0
 ; AVX512VL-NEXT:    retq
+;
+; AVX512VLFP16-LABEL: test_ldexp_32xhalf:
+; AVX512VLFP16:       # %bb.0:
+; AVX512VLFP16-NEXT:    vcvtw2ph %zmm1, %zmm1
+; AVX512VLFP16-NEXT:    vscalefph %zmm1, %zmm0, %zmm0
+; AVX512VLFP16-NEXT:    retq
   %r = call <32 x half> @llvm.ldexp.v32f16.v32i16(<32 x half> %x, <32 x i16> %exp)
   ret <32 x half> %r
 }
@@ -1518,148 +289,8 @@ declare <32 x half> @llvm.ldexp.v32f16.v32i16(<32 x half>, <32 x i16>)
 define <16 x float> @test_ldexp_16xfloat(<16 x float> %x, <16 x i32> %exp) nounwind {
 ; CHECK-LABEL: test_ldexp_16xfloat:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    subq $216, %rsp
-; CHECK-NEXT:    vmovdqu64 %zmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
-; CHECK-NEXT:    vmovups %zmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
-; CHECK-NEXT:    vextractf32x4 $3, %zmm0, %xmm0
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vextracti32x4 $3, %zmm1, %xmm1
-; CHECK-NEXT:    vmovdqa %xmm1, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovd %xmm1, %edi
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $1, %xmm0, %edi
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $2, %xmm0, %edi
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $3, %xmm0, %edi
-; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; CHECK-NEXT:    vextractf32x4 $2, %zmm0, %xmm1
-; CHECK-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; CHECK-NEXT:    vextracti32x4 $2, %zmm0, %xmm0
-; CHECK-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovd %xmm0, %edi
-; CHECK-NEXT:    vmovaps %xmm1, %xmm0
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $1, %xmm0, %edi
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $2, %xmm0, %edi
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $3, %xmm0, %edi
-; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
-; CHECK-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; CHECK-NEXT:    vmovups %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; CHECK-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; CHECK-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; CHECK-NEXT:    vextracti128 $1, %ymm0, %xmm0
-; CHECK-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovd %xmm0, %edi
-; CHECK-NEXT:    vmovaps %xmm1, %xmm0
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $1, %xmm0, %edi
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $2, %xmm0, %edi
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps (%rsp), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $3, %xmm0, %edi
-; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovdqu64 {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; CHECK-NEXT:    vmovd %xmm0, %edi
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; CHECK-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; CHECK-NEXT:    vextractps $1, %xmm0, %edi
-; CHECK-NEXT:    vmovshdup {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,1,3,3]
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; CHECK-NEXT:    vextractps $2, %xmm0, %edi
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1],xmm0[0],xmm1[3]
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; CHECK-NEXT:    vextractps $3, %xmm0, %edi
-; CHECK-NEXT:    vpermilps $255, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[3,3,3,3]
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexpf@PLT
-; CHECK-NEXT:    vmovaps (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0,1,2],xmm0[0]
-; CHECK-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; CHECK-NEXT:    vinsertf64x4 $1, {{[-0-9]+}}(%r{{[sb]}}p), %zmm0, %zmm0 # 32-byte Folded Reload
-; CHECK-NEXT:    addq $216, %rsp
+; CHECK-NEXT:    vcvtdq2ps %zmm1, %zmm1
+; CHECK-NEXT:    vscalefps %zmm1, %zmm0, %zmm0
 ; CHECK-NEXT:    retq
   %r = call <16 x float> @llvm.ldexp.v16f32.v16i32(<16 x float> %x, <16 x i32> %exp)
   ret <16 x float> %r
@@ -1669,81 +300,13 @@ declare <16 x float> @llvm.ldexp.v16f32.v16i32(<16 x float>, <16 x i32>)
 define <8 x double> @test_ldexp_8xdouble(<8 x double> %x, <8 x i32> %exp) nounwind {
 ; CHECK-LABEL: test_ldexp_8xdouble:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    subq $184, %rsp
-; CHECK-NEXT:    vmovups %ymm1, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; CHECK-NEXT:    vmovups %zmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 64-byte Spill
-; CHECK-NEXT:    vextractf32x4 $3, %zmm0, %xmm0
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vextractf128 $1, %ymm1, %xmm1
-; CHECK-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vextractps $2, %xmm1, %edi
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexp@PLT
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $3, %xmm0, %edi
-; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq ldexp@PLT
-; CHECK-NEXT:    vmovapd {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; CHECK-NEXT:    vmovapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; CHECK-NEXT:    vextractf32x4 $2, %zmm0, %xmm1
-; CHECK-NEXT:    vmovaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vmovd %xmm0, %edi
-; CHECK-NEXT:    vmovaps %xmm1, %xmm0
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexp@PLT
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; CHECK-NEXT:    vextractps $1, %xmm0, %edi
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    callq ldexp@PLT
-; CHECK-NEXT:    vmovapd (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; CHECK-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; CHECK-NEXT:    vmovupd %ymm0, {{[-0-9]+}}(%r{{[sb]}}p) # 32-byte Spill
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; CHECK-NEXT:    vextractps $2, %xmm0, %edi
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; CHECK-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexp@PLT
-; CHECK-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; CHECK-NEXT:    vextractps $3, %xmm0, %edi
-; CHECK-NEXT:    vpermilpd $1, (%rsp), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexp@PLT
-; CHECK-NEXT:    vmovapd {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; CHECK-NEXT:    vmovapd %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; CHECK-NEXT:    vmovdqu {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; CHECK-NEXT:    vmovd %xmm0, %edi
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %zmm0 # 64-byte Reload
-; CHECK-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexp@PLT
-; CHECK-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; CHECK-NEXT:    vmovups {{[-0-9]+}}(%r{{[sb]}}p), %ymm0 # 32-byte Reload
-; CHECK-NEXT:    vextractps $1, %xmm0, %edi
-; CHECK-NEXT:    vpermilpd $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; CHECK-NEXT:    # xmm0 = mem[1,0]
-; CHECK-NEXT:    vzeroupper
-; CHECK-NEXT:    callq ldexp@PLT
-; CHECK-NEXT:    vmovapd (%rsp), %xmm1 # 16-byte Reload
-; CHECK-NEXT:    vunpcklpd {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; CHECK-NEXT:    vinsertf128 $1, {{[-0-9]+}}(%r{{[sb]}}p), %ymm0, %ymm0 # 16-byte Folded Reload
-; CHECK-NEXT:    vinsertf64x4 $1, {{[-0-9]+}}(%r{{[sb]}}p), %zmm0, %zmm0 # 32-byte Folded Reload
-; CHECK-NEXT:    addq $184, %rsp
+; CHECK-NEXT:    vcvtdq2pd %ymm1, %zmm1
+; CHECK-NEXT:    vscalefpd %zmm1, %zmm0, %zmm0
 ; CHECK-NEXT:    retq
   %r = call <8 x double> @llvm.ldexp.v8f64.v8i32(<8 x double> %x, <8 x i32> %exp)
   ret <8 x double> %r
 }
 declare <8 x double> @llvm.ldexp.v8f64.v8i32(<8 x double>, <8 x i32>)
 
+;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
+; AVX512VLF: {{.*}}