[WebAssembly] Don't perform the returned-argument optimization on constants.

[oota-llvm.git] / test / CodeGen / ARM / vmul.ll

diff --git a/test/CodeGen/ARM/vmul.ll b/test/CodeGen/ARM/vmul.ll
index aa3cda0707355a5907602fab30c9dc4dc6a66e3f..0455190b4c9d849e2907b177d8fec144586d3a36 100644 (file)
--- a/test/CodeGen/ARM/vmul.ll
+++ b/test/CodeGen/ARM/vmul.ll
@@ -1,91 +1,91 @@
-; RUN: llc < %s -march=arm -mcpu=cortex-a8 | FileCheck %s
+; RUN: llc -mtriple=arm-eabi -mcpu=cortex-a8 %s -o - | FileCheck %s
 
 define <8 x i8> @vmuli8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
-;CHECK: vmuli8:
+;CHECK-LABEL: vmuli8:
 ;CHECK: vmul.i8
-       %tmp1 = load <8 x i8>* %A
-       %tmp2 = load <8 x i8>* %B
+       %tmp1 = load <8 x i8>, <8 x i8>* %A
+       %tmp2 = load <8 x i8>, <8 x i8>* %B
        %tmp3 = mul <8 x i8> %tmp1, %tmp2
        ret <8 x i8> %tmp3
 }
 
 define <4 x i16> @vmuli16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
-;CHECK: vmuli16:
+;CHECK-LABEL: vmuli16:
 ;CHECK: vmul.i16
-       %tmp1 = load <4 x i16>* %A
-       %tmp2 = load <4 x i16>* %B
+       %tmp1 = load <4 x i16>, <4 x i16>* %A
+       %tmp2 = load <4 x i16>, <4 x i16>* %B
        %tmp3 = mul <4 x i16> %tmp1, %tmp2
        ret <4 x i16> %tmp3
 }
 
 define <2 x i32> @vmuli32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
-;CHECK: vmuli32:
+;CHECK-LABEL: vmuli32:
 ;CHECK: vmul.i32
-       %tmp1 = load <2 x i32>* %A
-       %tmp2 = load <2 x i32>* %B
+       %tmp1 = load <2 x i32>, <2 x i32>* %A
+       %tmp2 = load <2 x i32>, <2 x i32>* %B
        %tmp3 = mul <2 x i32> %tmp1, %tmp2
        ret <2 x i32> %tmp3
 }
 
 define <2 x float> @vmulf32(<2 x float>* %A, <2 x float>* %B) nounwind {
-;CHECK: vmulf32:
+;CHECK-LABEL: vmulf32:
 ;CHECK: vmul.f32
-       %tmp1 = load <2 x float>* %A
-       %tmp2 = load <2 x float>* %B
+       %tmp1 = load <2 x float>, <2 x float>* %A
+       %tmp2 = load <2 x float>, <2 x float>* %B
        %tmp3 = fmul <2 x float> %tmp1, %tmp2
        ret <2 x float> %tmp3
 }
 
 define <8 x i8> @vmulp8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
-;CHECK: vmulp8:
+;CHECK-LABEL: vmulp8:
 ;CHECK: vmul.p8
-       %tmp1 = load <8 x i8>* %A
-       %tmp2 = load <8 x i8>* %B
+       %tmp1 = load <8 x i8>, <8 x i8>* %A
+       %tmp2 = load <8 x i8>, <8 x i8>* %B
        %tmp3 = call <8 x i8> @llvm.arm.neon.vmulp.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
        ret <8 x i8> %tmp3
 }
 
 define <16 x i8> @vmulQi8(<16 x i8>* %A, <16 x i8>* %B) nounwind {
-;CHECK: vmulQi8:
+;CHECK-LABEL: vmulQi8:
 ;CHECK: vmul.i8
-       %tmp1 = load <16 x i8>* %A
-       %tmp2 = load <16 x i8>* %B
+       %tmp1 = load <16 x i8>, <16 x i8>* %A
+       %tmp2 = load <16 x i8>, <16 x i8>* %B
        %tmp3 = mul <16 x i8> %tmp1, %tmp2
        ret <16 x i8> %tmp3
 }
 
 define <8 x i16> @vmulQi16(<8 x i16>* %A, <8 x i16>* %B) nounwind {
-;CHECK: vmulQi16:
+;CHECK-LABEL: vmulQi16:
 ;CHECK: vmul.i16
-       %tmp1 = load <8 x i16>* %A
-       %tmp2 = load <8 x i16>* %B
+       %tmp1 = load <8 x i16>, <8 x i16>* %A
+       %tmp2 = load <8 x i16>, <8 x i16>* %B
        %tmp3 = mul <8 x i16> %tmp1, %tmp2
        ret <8 x i16> %tmp3
 }
 
 define <4 x i32> @vmulQi32(<4 x i32>* %A, <4 x i32>* %B) nounwind {
-;CHECK: vmulQi32:
+;CHECK-LABEL: vmulQi32:
 ;CHECK: vmul.i32
-       %tmp1 = load <4 x i32>* %A
-       %tmp2 = load <4 x i32>* %B
+       %tmp1 = load <4 x i32>, <4 x i32>* %A
+       %tmp2 = load <4 x i32>, <4 x i32>* %B
        %tmp3 = mul <4 x i32> %tmp1, %tmp2
        ret <4 x i32> %tmp3
 }
 
 define <4 x float> @vmulQf32(<4 x float>* %A, <4 x float>* %B) nounwind {
-;CHECK: vmulQf32:
+;CHECK-LABEL: vmulQf32:
 ;CHECK: vmul.f32
-       %tmp1 = load <4 x float>* %A
-       %tmp2 = load <4 x float>* %B
+       %tmp1 = load <4 x float>, <4 x float>* %A
+       %tmp2 = load <4 x float>, <4 x float>* %B
        %tmp3 = fmul <4 x float> %tmp1, %tmp2
        ret <4 x float> %tmp3
 }
 
 define <16 x i8> @vmulQp8(<16 x i8>* %A, <16 x i8>* %B) nounwind {
-;CHECK: vmulQp8:
+;CHECK-LABEL: vmulQp8:
 ;CHECK: vmul.p8
-       %tmp1 = load <16 x i8>* %A
-       %tmp2 = load <16 x i8>* %B
+       %tmp1 = load <16 x i8>, <16 x i8>* %A
+       %tmp2 = load <16 x i8>, <16 x i8>* %B
        %tmp3 = call <16 x i8> @llvm.arm.neon.vmulp.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
        ret <16 x i8> %tmp3
 }
@@ -148,10 +148,10 @@ entry:
 }
 
 define <8 x i16> @vmulls8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
-;CHECK: vmulls8:
+;CHECK-LABEL: vmulls8:
 ;CHECK: vmull.s8
-       %tmp1 = load <8 x i8>* %A
-       %tmp2 = load <8 x i8>* %B
+       %tmp1 = load <8 x i8>, <8 x i8>* %A
+       %tmp2 = load <8 x i8>, <8 x i8>* %B
        %tmp3 = sext <8 x i8> %tmp1 to <8 x i16>
        %tmp4 = sext <8 x i8> %tmp2 to <8 x i16>
        %tmp5 = mul <8 x i16> %tmp3, %tmp4
@@ -159,19 +159,19 @@ define <8 x i16> @vmulls8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
 }
 
 define <8 x i16> @vmulls8_int(<8 x i8>* %A, <8 x i8>* %B) nounwind {
-;CHECK: vmulls8_int:
+;CHECK-LABEL: vmulls8_int:
 ;CHECK: vmull.s8
-       %tmp1 = load <8 x i8>* %A
-       %tmp2 = load <8 x i8>* %B
+       %tmp1 = load <8 x i8>, <8 x i8>* %A
+       %tmp2 = load <8 x i8>, <8 x i8>* %B
        %tmp3 = call <8 x i16> @llvm.arm.neon.vmulls.v8i16(<8 x i8> %tmp1, <8 x i8> %tmp2)
        ret <8 x i16> %tmp3
 }
 
 define <4 x i32> @vmulls16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
-;CHECK: vmulls16:
+;CHECK-LABEL: vmulls16:
 ;CHECK: vmull.s16
-       %tmp1 = load <4 x i16>* %A
-       %tmp2 = load <4 x i16>* %B
+       %tmp1 = load <4 x i16>, <4 x i16>* %A
+       %tmp2 = load <4 x i16>, <4 x i16>* %B
        %tmp3 = sext <4 x i16> %tmp1 to <4 x i32>
        %tmp4 = sext <4 x i16> %tmp2 to <4 x i32>
        %tmp5 = mul <4 x i32> %tmp3, %tmp4
@@ -179,19 +179,19 @@ define <4 x i32> @vmulls16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
 }
 
 define <4 x i32> @vmulls16_int(<4 x i16>* %A, <4 x i16>* %B) nounwind {
-;CHECK: vmulls16_int:
+;CHECK-LABEL: vmulls16_int:
 ;CHECK: vmull.s16
-       %tmp1 = load <4 x i16>* %A
-       %tmp2 = load <4 x i16>* %B
+       %tmp1 = load <4 x i16>, <4 x i16>* %A
+       %tmp2 = load <4 x i16>, <4 x i16>* %B
        %tmp3 = call <4 x i32> @llvm.arm.neon.vmulls.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
        ret <4 x i32> %tmp3
 }
 
 define <2 x i64> @vmulls32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
-;CHECK: vmulls32:
+;CHECK-LABEL: vmulls32:
 ;CHECK: vmull.s32
-       %tmp1 = load <2 x i32>* %A
-       %tmp2 = load <2 x i32>* %B
+       %tmp1 = load <2 x i32>, <2 x i32>* %A
+       %tmp2 = load <2 x i32>, <2 x i32>* %B
        %tmp3 = sext <2 x i32> %tmp1 to <2 x i64>
        %tmp4 = sext <2 x i32> %tmp2 to <2 x i64>
        %tmp5 = mul <2 x i64> %tmp3, %tmp4
@@ -199,19 +199,19 @@ define <2 x i64> @vmulls32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
 }
 
 define <2 x i64> @vmulls32_int(<2 x i32>* %A, <2 x i32>* %B) nounwind {
-;CHECK: vmulls32_int:
+;CHECK-LABEL: vmulls32_int:
 ;CHECK: vmull.s32
-       %tmp1 = load <2 x i32>* %A
-       %tmp2 = load <2 x i32>* %B
+       %tmp1 = load <2 x i32>, <2 x i32>* %A
+       %tmp2 = load <2 x i32>, <2 x i32>* %B
        %tmp3 = call <2 x i64> @llvm.arm.neon.vmulls.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
        ret <2 x i64> %tmp3
 }
 
 define <8 x i16> @vmullu8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
-;CHECK: vmullu8:
+;CHECK-LABEL: vmullu8:
 ;CHECK: vmull.u8
-       %tmp1 = load <8 x i8>* %A
-       %tmp2 = load <8 x i8>* %B
+       %tmp1 = load <8 x i8>, <8 x i8>* %A
+       %tmp2 = load <8 x i8>, <8 x i8>* %B
        %tmp3 = zext <8 x i8> %tmp1 to <8 x i16>
        %tmp4 = zext <8 x i8> %tmp2 to <8 x i16>
        %tmp5 = mul <8 x i16> %tmp3, %tmp4
@@ -219,19 +219,19 @@ define <8 x i16> @vmullu8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
 }
 
 define <8 x i16> @vmullu8_int(<8 x i8>* %A, <8 x i8>* %B) nounwind {
-;CHECK: vmullu8_int:
+;CHECK-LABEL: vmullu8_int:
 ;CHECK: vmull.u8
-       %tmp1 = load <8 x i8>* %A
-       %tmp2 = load <8 x i8>* %B
+       %tmp1 = load <8 x i8>, <8 x i8>* %A
+       %tmp2 = load <8 x i8>, <8 x i8>* %B
        %tmp3 = call <8 x i16> @llvm.arm.neon.vmullu.v8i16(<8 x i8> %tmp1, <8 x i8> %tmp2)
        ret <8 x i16> %tmp3
 }
 
 define <4 x i32> @vmullu16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
-;CHECK: vmullu16:
+;CHECK-LABEL: vmullu16:
 ;CHECK: vmull.u16
-       %tmp1 = load <4 x i16>* %A
-       %tmp2 = load <4 x i16>* %B
+       %tmp1 = load <4 x i16>, <4 x i16>* %A
+       %tmp2 = load <4 x i16>, <4 x i16>* %B
        %tmp3 = zext <4 x i16> %tmp1 to <4 x i32>
        %tmp4 = zext <4 x i16> %tmp2 to <4 x i32>
        %tmp5 = mul <4 x i32> %tmp3, %tmp4
@@ -239,19 +239,19 @@ define <4 x i32> @vmullu16(<4 x i16>* %A, <4 x i16>* %B) nounwind {
 }
 
 define <4 x i32> @vmullu16_int(<4 x i16>* %A, <4 x i16>* %B) nounwind {
-;CHECK: vmullu16_int:
+;CHECK-LABEL: vmullu16_int:
 ;CHECK: vmull.u16
-       %tmp1 = load <4 x i16>* %A
-       %tmp2 = load <4 x i16>* %B
+       %tmp1 = load <4 x i16>, <4 x i16>* %A
+       %tmp2 = load <4 x i16>, <4 x i16>* %B
        %tmp3 = call <4 x i32> @llvm.arm.neon.vmullu.v4i32(<4 x i16> %tmp1, <4 x i16> %tmp2)
        ret <4 x i32> %tmp3
 }
 
 define <2 x i64> @vmullu32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
-;CHECK: vmullu32:
+;CHECK-LABEL: vmullu32:
 ;CHECK: vmull.u32
-       %tmp1 = load <2 x i32>* %A
-       %tmp2 = load <2 x i32>* %B
+       %tmp1 = load <2 x i32>, <2 x i32>* %A
+       %tmp2 = load <2 x i32>, <2 x i32>* %B
        %tmp3 = zext <2 x i32> %tmp1 to <2 x i64>
        %tmp4 = zext <2 x i32> %tmp2 to <2 x i64>
        %tmp5 = mul <2 x i64> %tmp3, %tmp4
@@ -259,19 +259,19 @@ define <2 x i64> @vmullu32(<2 x i32>* %A, <2 x i32>* %B) nounwind {
 }
 
 define <2 x i64> @vmullu32_int(<2 x i32>* %A, <2 x i32>* %B) nounwind {
-;CHECK: vmullu32_int:
+;CHECK-LABEL: vmullu32_int:
 ;CHECK: vmull.u32
-       %tmp1 = load <2 x i32>* %A
-       %tmp2 = load <2 x i32>* %B
+       %tmp1 = load <2 x i32>, <2 x i32>* %A
+       %tmp2 = load <2 x i32>, <2 x i32>* %B
        %tmp3 = call <2 x i64> @llvm.arm.neon.vmullu.v2i64(<2 x i32> %tmp1, <2 x i32> %tmp2)
        ret <2 x i64> %tmp3
 }
 
 define <8 x i16> @vmullp8(<8 x i8>* %A, <8 x i8>* %B) nounwind {
-;CHECK: vmullp8:
+;CHECK-LABEL: vmullp8:
 ;CHECK: vmull.p8
-       %tmp1 = load <8 x i8>* %A
-       %tmp2 = load <8 x i8>* %B
+       %tmp1 = load <8 x i8>, <8 x i8>* %A
+       %tmp2 = load <8 x i8>, <8 x i8>* %B
        %tmp3 = call <8 x i16> @llvm.arm.neon.vmullp.v8i16(<8 x i8> %tmp1, <8 x i8> %tmp2)
        ret <8 x i16> %tmp3
 }
@@ -441,13 +441,13 @@ define <2 x i64> @vmull_extvec_u32(<2 x i32> %arg) nounwind {
 ; rdar://9197392
 define void @distribute(i16* %dst, i8* %src, i32 %mul) nounwind {
 entry:
-; CHECK: distribute:
+; CHECK-LABEL: distribute:
 ; CHECK: vmull.u8 [[REG1:(q[0-9]+)]], d{{.*}}, [[REG2:(d[0-9]+)]]
 ; CHECK: vmlal.u8 [[REG1]], d{{.*}}, [[REG2]]
   %0 = trunc i32 %mul to i8
   %1 = insertelement <8 x i8> undef, i8 %0, i32 0
   %2 = shufflevector <8 x i8> %1, <8 x i8> undef, <8 x i32> zeroinitializer
-  %3 = tail call <16 x i8> @llvm.arm.neon.vld1.v16i8(i8* %src, i32 1)
+  %3 = tail call <16 x i8> @llvm.arm.neon.vld1.v16i8.p0i8(i8* %src, i32 1)
   %4 = bitcast <16 x i8> %3 to <2 x double>
   %5 = extractelement <2 x double> %4, i32 1
   %6 = bitcast double %5 to <8 x i8>
@@ -459,13 +459,13 @@ entry:
   %12 = add <8 x i16> %7, %11
   %13 = mul <8 x i16> %12, %8
   %14 = bitcast i16* %dst to i8*
-  tail call void @llvm.arm.neon.vst1.v8i16(i8* %14, <8 x i16> %13, i32 2)
+  tail call void @llvm.arm.neon.vst1.p0i8.v8i16(i8* %14, <8 x i16> %13, i32 2)
   ret void
 }
 
-declare <16 x i8> @llvm.arm.neon.vld1.v16i8(i8*, i32) nounwind readonly
+declare <16 x i8> @llvm.arm.neon.vld1.v16i8.p0i8(i8*, i32) nounwind readonly
 
-declare void @llvm.arm.neon.vst1.v8i16(i8*, <8 x i16>, i32) nounwind
+declare void @llvm.arm.neon.vst1.p0i8.v8i16(i8*, <8 x i16>, i32) nounwind
 
 ; Take advantage of the Cortex-A8 multiplier accumulator forward.
 
@@ -480,7 +480,7 @@ entry:
   %0 = trunc i32 %mul to i8
   %1 = insertelement <8 x i8> undef, i8 %0, i32 0
   %2 = shufflevector <8 x i8> %1, <8 x i8> undef, <8 x i32> zeroinitializer
-  %3 = tail call <16 x i8> @llvm.arm.neon.vld1.v16i8(i8* %src, i32 1)
+  %3 = tail call <16 x i8> @llvm.arm.neon.vld1.v16i8.p0i8(i8* %src, i32 1)
   %4 = bitcast <16 x i8> %3 to <2 x double>
   %5 = extractelement <2 x double> %4, i32 1
   %6 = bitcast double %5 to <8 x i8>
@@ -488,7 +488,7 @@ entry:
   %8 = bitcast double %7 to <8 x i8>
   %9 = add <8 x i8> %6, %8
   %10 = mul <8 x i8> %9, %2
-  %11 = getelementptr inbounds %struct.uint8x8_t* %dst, i32 0, i32 0
+  %11 = getelementptr inbounds %struct.uint8x8_t, %struct.uint8x8_t* %dst, i32 0, i32 0
   store <8 x i8> %10, <8 x i8>* %11, align 8
   ret void
 }
@@ -502,7 +502,7 @@ entry:
   %0 = trunc i32 %mul to i8
   %1 = insertelement <8 x i8> undef, i8 %0, i32 0
   %2 = shufflevector <8 x i8> %1, <8 x i8> undef, <8 x i32> zeroinitializer
-  %3 = tail call <16 x i8> @llvm.arm.neon.vld1.v16i8(i8* %src, i32 1)
+  %3 = tail call <16 x i8> @llvm.arm.neon.vld1.v16i8.p0i8(i8* %src, i32 1)
   %4 = bitcast <16 x i8> %3 to <2 x double>
   %5 = extractelement <2 x double> %4, i32 1
   %6 = bitcast double %5 to <8 x i8>
@@ -510,11 +510,22 @@ entry:
   %8 = bitcast double %7 to <8 x i8>
   %9 = add <8 x i8> %6, %8
   %10 = mul <8 x i8> %2, %9
-  %11 = getelementptr inbounds %struct.uint8x8_t* %dst, i32 0, i32 0
+  %11 = getelementptr inbounds %struct.uint8x8_t, %struct.uint8x8_t* %dst, i32 0, i32 0
   store <8 x i8> %10, <8 x i8>* %11, align 8
   ret void
 }
 
+define <8 x i8> @no_distribute(<8 x i8> %a, <8 x i8> %b) nounwind {
+entry:
+; CHECK: no_distribute
+; CHECK: vadd.i8
+; CHECK: vmul.i8
+; CHECK-NOT: vmla.i8
+  %0 = add <8 x i8> %a, %b
+  %1 = mul <8x i8> %0, %0
+  ret <8 x i8> %1
+}
+
 ; If one operand has a zero-extend and the other a sign-extend, vmull
 ; cannot be used.
 define i16 @vmullWithInconsistentExtensions(<8 x i8> %vec) {
@@ -548,8 +559,8 @@ for.body33.lr.ph:                                 ; preds = %for.body
 
 for.body33:                                       ; preds = %for.body33, %for.body33.lr.ph
   %add45 = add i32 undef, undef
-  %vld155 = tail call <16 x i8> @llvm.arm.neon.vld1.v16i8(i8* undef, i32 1)
-  %0 = load i32** undef, align 4
+  %vld155 = tail call <16 x i8> @llvm.arm.neon.vld1.v16i8.p0i8(i8* undef, i32 1)
+  %0 = load i32*, i32** undef, align 4
   %shuffle.i250 = shufflevector <2 x i64> undef, <2 x i64> undef, <1 x i32> zeroinitializer
   %1 = bitcast <1 x i64> %shuffle.i250 to <8 x i8>
   %vmovl.i249 = zext <8 x i8> %1 to <8 x i16>
@@ -605,7 +616,7 @@ declare <8 x i8> @llvm.arm.neon.vqmovnu.v8i8(<8 x i16>) nounwind readnone
 ; PR15970
 define void @no_illegal_types_vmull_sext(<4 x i32> %a) {
 entry:
-  %wide.load283.i = load <4 x i8>* undef, align 1
+  %wide.load283.i = load <4 x i8>, <4 x i8>* undef, align 1
   %0 = sext <4 x i8> %wide.load283.i to <4 x i32>
   %1 = sub nsw <4 x i32> %0, %a
   %2 = mul nsw <4 x i32> %1, %1
@@ -615,7 +626,7 @@ entry:
 }
 define void @no_illegal_types_vmull_zext(<4 x i32> %a) {
 entry:
-  %wide.load283.i = load <4 x i8>* undef, align 1
+  %wide.load283.i = load <4 x i8>, <4 x i8>* undef, align 1
   %0 = zext <4 x i8> %wide.load283.i to <4 x i32>
   %1 = sub nsw <4 x i32> %0, %a
   %2 = mul nsw <4 x i32> %1, %1
@@ -623,3 +634,21 @@ entry:
   store <4 x i32> %predphi290.v.i, <4 x i32>* undef, align 4
   ret void
 }
+
+define void @foo(<4 x float> * %a, <4 x float>* nocapture %dst, float* nocapture readonly %src) nounwind {
+;   Look for doing a normal scalar FP load rather than an to-all-lanes load.
+;   e.g., "ldr s0, [r2]" rathern than "vld1.32  {d18[], d19[]}, [r2:32]"
+;   Then check that the vector multiply has folded the splat to all lanes
+;   and used a vector * scalar instruction.
+; CHECK: vldr  {{s[0-9]+}}, [r2]
+; CHECK: vmul.f32  q8, q8, d0[0]
+  %tmp = load float, float* %src, align 4
+  %tmp5 = load <4 x float>, <4 x float>* %a, align 4
+  %tmp6 = insertelement <4 x float> undef, float %tmp, i32 0
+  %tmp7 = insertelement <4 x float> %tmp6, float %tmp, i32 1
+  %tmp8 = insertelement <4 x float> %tmp7, float %tmp, i32 2
+  %tmp9 = insertelement <4 x float> %tmp8, float %tmp, i32 3
+  %tmp10 = fmul <4 x float> %tmp9, %tmp5
+  store <4 x float> %tmp10, <4 x float>* %dst, align 4
+  ret void
+}