[X86] combine-sse41-intrinsics.ll - add AVX2 test coverage

Author: RKSimon

llvm/test/CodeGen/X86/combine-sse41-intrinsics.ll

@@ -1,6 +1,7 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=sse4.1 | FileCheck %s --check-prefixes=CHECK,SSE
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx    | FileCheck %s --check-prefixes=CHECK,AVX
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx    | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx2   | FileCheck %s --check-prefixes=CHECK,AVX,AVX2
 
 
 define <2 x double> @test_x86_sse41_blend_pd(<2 x double> %a0, <2 x double> %a1) {
@@ -146,12 +147,18 @@ define <16 x i8> @demandedelts_pblendvb(<16 x i8> %a0, <16 x i8> %a1, <16 x i8>
 ; SSE-NEXT:    movdqa %xmm3, %xmm0
 ; SSE-NEXT:    retq
 ;
-; AVX-LABEL: demandedelts_pblendvb:
-; AVX:       # %bb.0:
-; AVX-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    retq
+; AVX1-LABEL: demandedelts_pblendvb:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: demandedelts_pblendvb:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpbroadcastb %xmm0, %xmm0
+; AVX2-NEXT:    retq
   %1 = shufflevector <16 x i8> %a0, <16 x i8> undef, <16 x i32> zeroinitializer
   %2 = shufflevector <16 x i8> %a1, <16 x i8> undef, <16 x i32> zeroinitializer
   %3 = shufflevector <16 x i8> %a2, <16 x i8> undef, <16 x i32> zeroinitializer
@@ -193,15 +200,28 @@ define <4 x float> @demandedbits_uitofp_blendvps(<4 x float> %a0, <4 x float> %a
 ; SSE-NEXT:    movaps %xmm3, %xmm0
 ; SSE-NEXT:    retq
 ;
-; AVX-LABEL: demandedbits_uitofp_blendvps:
-; AVX:       # %bb.0:
-; AVX-NEXT:    vpblendw {{.*#+}} xmm3 = xmm2[0],mem[1],xmm2[2],mem[3],xmm2[4],mem[5],xmm2[6],mem[7]
-; AVX-NEXT:    vpsrld $16, %xmm2, %xmm2
-; AVX-NEXT:    vpblendw {{.*#+}} xmm2 = xmm2[0],mem[1],xmm2[2],mem[3],xmm2[4],mem[5],xmm2[6],mem[7]
-; AVX-NEXT:    vsubps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
-; AVX-NEXT:    vaddps %xmm2, %xmm3, %xmm2
-; AVX-NEXT:    vblendvps %xmm2, %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    retq
+; AVX1-LABEL: demandedbits_uitofp_blendvps:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm3 = xmm2[0],mem[1],xmm2[2],mem[3],xmm2[4],mem[5],xmm2[6],mem[7]
+; AVX1-NEXT:    vpsrld $16, %xmm2, %xmm2
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm2[0],mem[1],xmm2[2],mem[3],xmm2[4],mem[5],xmm2[6],mem[7]
+; AVX1-NEXT:    vsubps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
+; AVX1-NEXT:    vaddps %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vblendvps %xmm2, %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: demandedbits_uitofp_blendvps:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm3 = [1258291200,1258291200,1258291200,1258291200]
+; AVX2-NEXT:    vpblendw {{.*#+}} xmm3 = xmm2[0],xmm3[1],xmm2[2],xmm3[3],xmm2[4],xmm3[5],xmm2[6],xmm3[7]
+; AVX2-NEXT:    vpsrld $16, %xmm2, %xmm2
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm4 = [1392508928,1392508928,1392508928,1392508928]
+; AVX2-NEXT:    vpblendw {{.*#+}} xmm2 = xmm2[0],xmm4[1],xmm2[2],xmm4[3],xmm2[4],xmm4[5],xmm2[6],xmm4[7]
+; AVX2-NEXT:    vbroadcastss {{.*#+}} xmm4 = [5.49764202E+11,5.49764202E+11,5.49764202E+11,5.49764202E+11]
+; AVX2-NEXT:    vsubps %xmm4, %xmm2, %xmm2
+; AVX2-NEXT:    vaddps %xmm2, %xmm3, %xmm2
+; AVX2-NEXT:    vblendvps %xmm2, %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    retq
   %cvt = uitofp <4 x i32> %a2 to <4 x float>
   %sel = tail call noundef <4 x float> @llvm.x86.sse41.blendvps(<4 x float> %a0, <4 x float> %a1, <4 x float> %cvt)
   ret <4 x float> %sel