diff --git a/llvm/test/CodeGen/X86/combine-gfni.ll b/llvm/test/CodeGen/X86/combine-gfni.ll new file mode 100644 index 0000000000000..21ea17a20e0fd --- /dev/null +++ b/llvm/test/CodeGen/X86/combine-gfni.ll @@ -0,0 +1,104 @@ +; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py +; RUN: llc < %s -mtriple=x86_64-- -mattr=+gfni | FileCheck %s --check-prefixes=SSE +; RUN: llc < %s -mtriple=x86_64-- -mattr=+gfni,+avx | FileCheck %s --check-prefixes=AVX +; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+gfni,+avx512bw | FileCheck %s --check-prefixes=AVX512 + +define <16 x i8> @gf2p8affineqb_freeze(<16 x i8> %a0, <16 x i8> %a1, <16 x i8> %a2) { +; SSE-LABEL: gf2p8affineqb_freeze: +; SSE: # %bb.0: +; SSE-NEXT: pxor %xmm3, %xmm3 +; SSE-NEXT: pcmpgtb %xmm2, %xmm3 +; SSE-NEXT: gf2p8affineqb $11, %xmm1, %xmm1 +; SSE-NEXT: pand %xmm3, %xmm1 +; SSE-NEXT: pandn %xmm0, %xmm3 +; SSE-NEXT: por %xmm1, %xmm3 +; SSE-NEXT: movdqa %xmm3, %xmm0 +; SSE-NEXT: retq +; +; AVX-LABEL: gf2p8affineqb_freeze: +; AVX: # %bb.0: +; AVX-NEXT: vgf2p8affineqb $11, %xmm1, %xmm1, %xmm1 +; AVX-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 +; AVX-NEXT: retq +; +; AVX512-LABEL: gf2p8affineqb_freeze: +; AVX512: # %bb.0: +; AVX512-NEXT: vpmovb2m %xmm2, %k1 +; AVX512-NEXT: vgf2p8affineqb $11, %xmm1, %xmm1, %xmm1 +; AVX512-NEXT: vmovdqu8 %xmm1, %xmm0 {%k1} +; AVX512-NEXT: retq + %i = icmp slt <16 x i8> %a2, zeroinitializer + %g = call <16 x i8> @llvm.x86.vgf2p8affineqb.128(<16 x i8> %a1, <16 x i8> %a1, i8 11) + %f = freeze <16 x i8> %g + %r = select <16 x i1> %i, <16 x i8> %f, <16 x i8> %a0 + ret <16 x i8> %r +} + +define <16 x i8> @gf2p8affineinvqb_freeze(<16 x i8> %a0, <16 x i8> %a1, <16 x i8> %a2) { +; SSE-LABEL: gf2p8affineinvqb_freeze: +; SSE: # %bb.0: +; SSE-NEXT: pxor %xmm3, %xmm3 +; SSE-NEXT: pcmpgtb %xmm2, %xmm3 +; SSE-NEXT: gf2p8affineinvqb $11, %xmm1, %xmm1 +; SSE-NEXT: pand %xmm3, %xmm1 +; SSE-NEXT: pandn %xmm0, %xmm3 +; SSE-NEXT: por %xmm1, %xmm3 +; SSE-NEXT: movdqa %xmm3, %xmm0 +; SSE-NEXT: retq +; +; AVX-LABEL: gf2p8affineinvqb_freeze: +; AVX: # %bb.0: +; AVX-NEXT: vgf2p8affineinvqb $11, %xmm1, %xmm1, %xmm1 +; AVX-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 +; AVX-NEXT: retq +; +; AVX512-LABEL: gf2p8affineinvqb_freeze: +; AVX512: # %bb.0: +; AVX512-NEXT: vpmovb2m %xmm2, %k1 +; AVX512-NEXT: vgf2p8affineinvqb $11, %xmm1, %xmm1, %xmm1 +; AVX512-NEXT: vmovdqu8 %xmm1, %xmm0 {%k1} +; AVX512-NEXT: retq + %i = icmp slt <16 x i8> %a2, zeroinitializer + %g = call <16 x i8> @llvm.x86.vgf2p8affineinvqb.128(<16 x i8> %a1, <16 x i8> %a1, i8 11) + %f = freeze <16 x i8> %g + %r = select <16 x i1> %i, <16 x i8> %f, <16 x i8> %a0 + ret <16 x i8> %r +} + +define <16 x i8> @gf2p8mulb_freeze(<16 x i8> %a0, <16 x i8> %a1, <16 x i8> %a2) { +; SSE-LABEL: gf2p8mulb_freeze: +; SSE: # %bb.0: +; SSE-NEXT: pxor %xmm3, %xmm3 +; SSE-NEXT: pcmpgtb %xmm2, %xmm3 +; SSE-NEXT: gf2p8mulb %xmm1, %xmm1 +; SSE-NEXT: pand %xmm3, %xmm1 +; SSE-NEXT: pandn %xmm0, %xmm3 +; SSE-NEXT: por %xmm1, %xmm3 +; SSE-NEXT: movdqa %xmm3, %xmm0 +; SSE-NEXT: retq +; +; AVX-LABEL: gf2p8mulb_freeze: +; AVX: # %bb.0: +; AVX-NEXT: vgf2p8mulb %xmm1, %xmm1, %xmm1 +; AVX-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 +; AVX-NEXT: retq +; +; AVX512-LABEL: gf2p8mulb_freeze: +; AVX512: # %bb.0: +; AVX512-NEXT: vpmovb2m %xmm2, %k1 +; AVX512-NEXT: vgf2p8mulb %xmm1, %xmm1, %xmm1 +; AVX512-NEXT: vmovdqu8 %xmm1, %xmm0 {%k1} +; AVX512-NEXT: retq + %i = icmp slt <16 x i8> %a2, zeroinitializer + %g = call <16 x i8> @llvm.x86.vgf2p8mulb.128(<16 x i8> %a1, <16 x i8> %a1) + %f = freeze <16 x i8> %g + %r = select <16 x i1> %i, <16 x i8> %f, <16 x i8> %a0 + ret <16 x i8> %r +} + +declare <16 x i8> @llvm.x86.vgf2p8affineqb.128(<16 x i8>, <16 x i8>, i8) +declare <32 x i8> @llvm.x86.vgf2p8affineqb.256(<32 x i8>, <32 x i8>, i8) +declare <16 x i8> @llvm.x86.vgf2p8affineinvqb.128(<16 x i8>, <16 x i8>, i8) +declare <32 x i8> @llvm.x86.vgf2p8affineinvqb.256(<32 x i8>, <32 x i8>, i8) +declare <16 x i8> @llvm.x86.vgf2p8mulb.128(<16 x i8>, <16 x i8>) +declare <32 x i8> @llvm.x86.vgf2p8mulb.256(<32 x i8>, <32 x i8>)