AVX-512: store <4 x i1> and <2 x i1> values in memory

[oota-llvm.git] / test / CodeGen / X86 / avx-arith.ll

diff --git a/test/CodeGen/X86/avx-arith.ll b/test/CodeGen/X86/avx-arith.ll
index 553e8acda9735490f4f6559231d4929e8f0b6488..792a9987e107eedd401b63353b1700b9b8eda0f7 100644 (file)
--- a/test/CodeGen/X86/avx-arith.ll
+++ b/test/CodeGen/X86/avx-arith.ll
@@ -38,7 +38,7 @@ entry:
 ; CHECK: vsubpd (%
 define <4 x double> @subpd256fold(<4 x double> %y, <4 x double>* nocapture %x) nounwind uwtable readonly ssp {
 entry:
-  %tmp2 = load <4 x double>* %x, align 32
+  %tmp2 = load <4 x double>, <4 x double>* %x, align 32
   %sub.i = fsub <4 x double> %y, %tmp2
   ret <4 x double> %sub.i
 }
@@ -53,7 +53,7 @@ entry:
 ; CHECK: vsubps (%
 define <8 x float> @subps256fold(<8 x float> %y, <8 x float>* nocapture %x) nounwind uwtable readonly ssp {
 entry:
-  %tmp2 = load <8 x float>* %x, align 32
+  %tmp2 = load <8 x float>, <8 x float>* %x, align 32
   %sub.i = fsub <8 x float> %y, %tmp2
   ret <8 x float> %sub.i
 }
@@ -131,3 +131,141 @@ entry:
 }
 
 declare float @sqrtf(float) readnone
+
+
+; CHECK: vextractf128 $1
+; CHECK-NEXT: vextractf128 $1
+; CHECK-NEXT: vpaddq %xmm
+; CHECK-NEXT: vpaddq %xmm
+; CHECK-NEXT: vinsertf128 $1
+define <4 x i64> @vpaddq(<4 x i64> %i, <4 x i64> %j) nounwind readnone {
+  %x = add <4 x i64> %i, %j
+  ret <4 x i64> %x
+}
+
+; CHECK: vextractf128 $1
+; CHECK-NEXT: vextractf128 $1
+; CHECK-NEXT: vpaddd %xmm
+; CHECK-NEXT: vpaddd %xmm
+; CHECK-NEXT: vinsertf128 $1
+define <8 x i32> @vpaddd(<8 x i32> %i, <8 x i32> %j) nounwind readnone {
+  %x = add <8 x i32> %i, %j
+  ret <8 x i32> %x
+}
+
+; CHECK: vextractf128 $1
+; CHECK-NEXT: vextractf128 $1
+; CHECK-NEXT: vpaddw %xmm
+; CHECK-NEXT: vpaddw %xmm
+; CHECK-NEXT: vinsertf128 $1
+define <16 x i16> @vpaddw(<16 x i16> %i, <16 x i16> %j) nounwind readnone {
+  %x = add <16 x i16> %i, %j
+  ret <16 x i16> %x
+}
+
+; CHECK: vextractf128 $1
+; CHECK-NEXT: vextractf128 $1
+; CHECK-NEXT: vpaddb %xmm
+; CHECK-NEXT: vpaddb %xmm
+; CHECK-NEXT: vinsertf128 $1
+define <32 x i8> @vpaddb(<32 x i8> %i, <32 x i8> %j) nounwind readnone {
+  %x = add <32 x i8> %i, %j
+  ret <32 x i8> %x
+}
+
+; CHECK: vextractf128 $1
+; CHECK-NEXT: vextractf128 $1
+; CHECK-NEXT: vpsubq %xmm
+; CHECK-NEXT: vpsubq %xmm
+; CHECK-NEXT: vinsertf128 $1
+define <4 x i64> @vpsubq(<4 x i64> %i, <4 x i64> %j) nounwind readnone {
+  %x = sub <4 x i64> %i, %j
+  ret <4 x i64> %x
+}
+
+; CHECK: vextractf128 $1
+; CHECK-NEXT: vextractf128 $1
+; CHECK-NEXT: vpsubd %xmm
+; CHECK-NEXT: vpsubd %xmm
+; CHECK-NEXT: vinsertf128 $1
+define <8 x i32> @vpsubd(<8 x i32> %i, <8 x i32> %j) nounwind readnone {
+  %x = sub <8 x i32> %i, %j
+  ret <8 x i32> %x
+}
+
+; CHECK: vextractf128 $1
+; CHECK-NEXT: vextractf128 $1
+; CHECK-NEXT: vpsubw %xmm
+; CHECK-NEXT: vpsubw %xmm
+; CHECK-NEXT: vinsertf128 $1
+define <16 x i16> @vpsubw(<16 x i16> %i, <16 x i16> %j) nounwind readnone {
+  %x = sub <16 x i16> %i, %j
+  ret <16 x i16> %x
+}
+
+; CHECK: vextractf128 $1
+; CHECK-NEXT: vextractf128 $1
+; CHECK-NEXT: vpsubb %xmm
+; CHECK-NEXT: vpsubb %xmm
+; CHECK-NEXT: vinsertf128 $1
+define <32 x i8> @vpsubb(<32 x i8> %i, <32 x i8> %j) nounwind readnone {
+  %x = sub <32 x i8> %i, %j
+  ret <32 x i8> %x
+}
+
+; CHECK: vextractf128 $1
+; CHECK-NEXT: vextractf128 $1
+; CHECK-NEXT: vpmulld %xmm
+; CHECK-NEXT: vpmulld %xmm
+; CHECK-NEXT: vinsertf128 $1
+define <8 x i32> @vpmulld(<8 x i32> %i, <8 x i32> %j) nounwind readnone {
+  %x = mul <8 x i32> %i, %j
+  ret <8 x i32> %x
+}
+
+; CHECK: vextractf128 $1
+; CHECK-NEXT: vextractf128 $1
+; CHECK-NEXT: vpmullw %xmm
+; CHECK-NEXT: vpmullw %xmm
+; CHECK-NEXT: vinsertf128 $1
+define <16 x i16> @vpmullw(<16 x i16> %i, <16 x i16> %j) nounwind readnone {
+  %x = mul <16 x i16> %i, %j
+  ret <16 x i16> %x
+}
+
+; CHECK: vextractf128 $1
+; CHECK-NEXT: vextractf128 $1
+; CHECK-NEXT: vpmuludq %xmm
+; CHECK-NEXT: vpsrlq $32, %xmm
+; CHECK-NEXT: vpmuludq %xmm
+; CHECK-NEXT: vpsllq $32, %xmm
+; CHECK-NEXT: vpaddq %xmm
+; CHECK-NEXT: vpsrlq $32, %xmm
+; CHECK-NEXT: vpmuludq %xmm
+; CHECK-NEXT: vpsllq $32, %xmm
+; CHECK-NEXT: vpaddq %xmm
+; CHECK-NEXT: vpmuludq %xmm
+; CHECK-NEXT: vpsrlq $32, %xmm
+; CHECK-NEXT: vpmuludq %xmm
+; CHECK-NEXT: vpsllq $32, %xmm
+; CHECK-NEXT: vpaddq %xmm
+; CHECK-NEXT: vpsrlq $32, %xmm
+; CHECK-NEXT: vpmuludq %xmm
+; CHECK-NEXT: vpsllq $32, %xmm
+; CHECK-NEXT: vpaddq %xmm
+; CHECK-NEXT: vinsertf128 $1
+define <4 x i64> @mul-v4i64(<4 x i64> %i, <4 x i64> %j) nounwind readnone {
+  %x = mul <4 x i64> %i, %j
+  ret <4 x i64> %x
+}
+
+declare <4 x float> @llvm.x86.sse.sqrt.ss(<4 x float>) nounwind readnone
+
+define <4 x float> @int_sqrt_ss() {
+; CHECK: int_sqrt_ss
+; CHECK: vsqrtss
+ %x0 = load float, float addrspace(1)* undef, align 8
+ %x1 = insertelement <4 x float> undef, float %x0, i32 0
+ %x2 = call <4 x float> @llvm.x86.sse.sqrt.ss(<4 x float> %x1) nounwind
+ ret <4 x float> %x2
+}