[AVX512] add PSHUFD Intrinsic

[oota-llvm.git] / test / CodeGen / X86 / avx512bwvl-intrinsics.ll

diff --git a/test/CodeGen/X86/avx512bwvl-intrinsics.ll b/test/CodeGen/X86/avx512bwvl-intrinsics.ll
index 17fbeee0bbe790bea3a50d69fda090dbf58eb0c4..cb907b3c42b0920377a706b3f3641129499f4c75 100644 (file)
--- a/test/CodeGen/X86/avx512bwvl-intrinsics.ll
+++ b/test/CodeGen/X86/avx512bwvl-intrinsics.ll
@@ -4509,3 +4509,252 @@ define <16 x i16>@test_int_x86_avx512_cvtmask2w_256(i16 %x0) {
   %res = call <16 x i16> @llvm.x86.avx512.cvtmask2w.256(i16 %x0)
   ret <16 x i16> %res
 }
+
+declare <8 x i16> @llvm.x86.avx512.mask.psrl.w.128(<8 x i16>, <8 x i16>, <8 x i16>, i8)
+
+define <8 x i16>@test_int_x86_avx512_mask_psrl_w_128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psrl_w_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vpsrlw %xmm1, %xmm0, %xmm2 {%k1}
+; CHECK-NEXT:    vpsrlw %xmm1, %xmm0, %xmm3 {%k1} {z}
+; CHECK-NEXT:    vpsrlw %xmm1, %xmm0, %xmm0
+; CHECK-NEXT:    vpaddw %xmm0, %xmm2, %xmm0
+; CHECK-NEXT:    vpaddw %xmm0, %xmm3, %xmm0
+; CHECK-NEXT:    retq
+  %res = call <8 x i16> @llvm.x86.avx512.mask.psrl.w.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3)
+  %res1 = call <8 x i16> @llvm.x86.avx512.mask.psrl.w.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 -1)
+  %res2 = call <8 x i16> @llvm.x86.avx512.mask.psrl.w.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> zeroinitializer, i8 %x3)
+  %res3 = add <8 x i16> %res, %res1
+  %res4 = add <8 x i16> %res2, %res3
+  ret <8 x i16> %res4
+}
+
+declare <16 x i16> @llvm.x86.avx512.mask.psrl.w.256(<16 x i16>, <8 x i16>, <16 x i16>, i16)
+
+define <16 x i16>@test_int_x86_avx512_mask_psrl_w_256(<16 x i16> %x0, <8 x i16> %x1, <16 x i16> %x2, i16 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psrl_w_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    kmovw %edi, %k1
+; CHECK-NEXT:    vpsrlw %xmm1, %ymm0, %ymm2 {%k1}
+; CHECK-NEXT:    vpsrlw %xmm1, %ymm0, %ymm3 {%k1} {z}
+; CHECK-NEXT:    vpsrlw %xmm1, %ymm0, %ymm0
+; CHECK-NEXT:    vpaddw %ymm0, %ymm2, %ymm0
+; CHECK-NEXT:    vpaddw %ymm3, %ymm0, %ymm0
+; CHECK-NEXT:    retq
+  %res = call <16 x i16> @llvm.x86.avx512.mask.psrl.w.256(<16 x i16> %x0, <8 x i16> %x1, <16 x i16> %x2, i16 %x3)
+  %res1 = call <16 x i16> @llvm.x86.avx512.mask.psrl.w.256(<16 x i16> %x0, <8 x i16> %x1, <16 x i16> %x2, i16 -1)
+  %res2 = call <16 x i16> @llvm.x86.avx512.mask.psrl.w.256(<16 x i16> %x0, <8 x i16> %x1, <16 x i16> zeroinitializer, i16 %x3)
+  %res3 = add <16 x i16> %res, %res1
+  %res4 = add <16 x i16> %res3, %res2
+  ret <16 x i16> %res4
+}
+
+declare <8 x i16> @llvm.x86.avx512.mask.psrl.wi.128(<8 x i16>, i8, <8 x i16>, i8)
+
+define <8 x i16>@test_int_x86_avx512_mask_psrl_wi_128(<8 x i16> %x0, i8 %x1, <8 x i16> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psrl_wi_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %sil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vpsrlw $3, %xmm0, %xmm1 {%k1}
+; CHECK-NEXT:    vpsrlw $3, %xmm0, %xmm2 {%k1} {z}
+; CHECK-NEXT:    vpsrlw $3, %xmm0, %xmm0
+; CHECK-NEXT:    vpaddw %xmm0, %xmm1, %xmm0
+; CHECK-NEXT:    vpaddw %xmm0, %xmm2, %xmm0
+; CHECK-NEXT:    retq
+  %res = call <8 x i16> @llvm.x86.avx512.mask.psrl.wi.128(<8 x i16> %x0, i8 3, <8 x i16> %x2, i8 %x3)
+  %res1 = call <8 x i16> @llvm.x86.avx512.mask.psrl.wi.128(<8 x i16> %x0, i8 3, <8 x i16> %x2, i8 -1)
+  %res2 = call <8 x i16> @llvm.x86.avx512.mask.psrl.wi.128(<8 x i16> %x0, i8 3, <8 x i16> zeroinitializer, i8 %x3)
+  %res3 = add <8 x i16> %res, %res1
+  %res4 = add <8 x i16> %res2, %res3
+  ret <8 x i16> %res4
+}
+
+declare <16 x i16> @llvm.x86.avx512.mask.psrl.wi.256(<16 x i16>, i8, <16 x i16>, i16)
+
+define <16 x i16>@test_int_x86_avx512_mask_psrl_wi_256(<16 x i16> %x0, i8 %x1, <16 x i16> %x2, i16 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psrl_wi_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    kmovw %esi, %k1
+; CHECK-NEXT:    vpsrlw $3, %ymm0, %ymm1 {%k1}
+; CHECK-NEXT:    vpsrlw $3, %ymm0, %ymm2 {%k1} {z}
+; CHECK-NEXT:    vpsrlw $3, %ymm0, %ymm0
+; CHECK-NEXT:    vpaddw %ymm0, %ymm1, %ymm0
+; CHECK-NEXT:    vpaddw %ymm2, %ymm0, %ymm0
+; CHECK-NEXT:    retq
+  %res = call <16 x i16> @llvm.x86.avx512.mask.psrl.wi.256(<16 x i16> %x0, i8 3, <16 x i16> %x2, i16 %x3)
+  %res1 = call <16 x i16> @llvm.x86.avx512.mask.psrl.wi.256(<16 x i16> %x0, i8 3, <16 x i16> %x2, i16 -1)
+  %res2 = call <16 x i16> @llvm.x86.avx512.mask.psrl.wi.256(<16 x i16> %x0, i8 3, <16 x i16> zeroinitializer, i16 %x3)
+  %res3 = add <16 x i16> %res, %res1
+  %res4 = add <16 x i16> %res3, %res2
+  ret <16 x i16> %res4
+}
+
+declare <16 x i16> @llvm.x86.avx512.mask.psrlv16.hi(<16 x i16>, <16 x i16>, <16 x i16>, i16)
+
+define <16 x i16>@test_int_x86_avx512_mask_psrlv16_hi(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psrlv16_hi:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    kmovw %edi, %k1
+; CHECK-NEXT:    vpsrlvw %ymm1, %ymm0, %ymm2 {%k1}
+; CHECK-NEXT:    vpsrlvw %ymm1, %ymm0, %ymm3 {%k1} {z}
+; CHECK-NEXT:    vpsrlvw %ymm1, %ymm0, %ymm0
+; CHECK-NEXT:    vpaddw %ymm3, %ymm2, %ymm1
+; CHECK-NEXT:    vpaddw %ymm0, %ymm1, %ymm0
+; CHECK-NEXT:    retq
+  %res = call <16 x i16> @llvm.x86.avx512.mask.psrlv16.hi(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 %x3)
+  %res1 = call <16 x i16> @llvm.x86.avx512.mask.psrlv16.hi(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> zeroinitializer, i16 %x3)
+  %res2 = call <16 x i16> @llvm.x86.avx512.mask.psrlv16.hi(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 -1)
+  %res3 = add <16 x i16> %res, %res1
+  %res4 = add <16 x i16> %res3, %res2
+  ret <16 x i16> %res4
+}
+
+declare <8 x i16> @llvm.x86.avx512.mask.psrlv8.hi(<8 x i16>, <8 x i16>, <8 x i16>, i8)
+
+define <8 x i16>@test_int_x86_avx512_mask_psrlv8_hi(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psrlv8_hi:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vpsrlvw %xmm1, %xmm0, %xmm2 {%k1}
+; CHECK-NEXT:    vpsrlvw %xmm1, %xmm0, %xmm3 {%k1} {z}
+; CHECK-NEXT:    vpsrlvw %xmm1, %xmm0, %xmm0
+; CHECK-NEXT:    vpaddw %xmm3, %xmm2, %xmm1
+; CHECK-NEXT:    vpaddw %xmm0, %xmm1, %xmm0
+; CHECK-NEXT:    retq
+  %res = call <8 x i16> @llvm.x86.avx512.mask.psrlv8.hi(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3)
+  %res1 = call <8 x i16> @llvm.x86.avx512.mask.psrlv8.hi(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> zeroinitializer, i8 %x3)
+  %res2 = call <8 x i16> @llvm.x86.avx512.mask.psrlv8.hi(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 -1)
+  %res3 = add <8 x i16> %res, %res1
+  %res4 = add <8 x i16> %res3, %res2
+  ret <8 x i16> %res4
+}
+
+declare <8 x i16> @llvm.x86.avx512.mask.psra.w.128(<8 x i16>, <8 x i16>, <8 x i16>, i8)
+
+define <8 x i16>@test_int_x86_avx512_mask_psra_w_128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psra_w_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %dil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vpsraw %xmm1, %xmm0, %xmm2 {%k1}
+; CHECK-NEXT:    vpsraw %xmm1, %xmm0, %xmm3 {%k1} {z}
+; CHECK-NEXT:    vpsraw %xmm1, %xmm0, %xmm0
+; CHECK-NEXT:    vpaddw %xmm3, %xmm2, %xmm1
+; CHECK-NEXT:    vpaddw %xmm0, %xmm1, %xmm0
+; CHECK-NEXT:    retq
+  %res = call <8 x i16> @llvm.x86.avx512.mask.psra.w.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3)
+  %res1 = call <8 x i16> @llvm.x86.avx512.mask.psra.w.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> zeroinitializer, i8 %x3)
+  %res2 = call <8 x i16> @llvm.x86.avx512.mask.psra.w.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 -1)
+  %res3 = add <8 x i16> %res, %res1
+  %res4 = add <8 x i16> %res3, %res2
+  ret <8 x i16> %res4
+}
+
+declare <8 x i16> @llvm.x86.avx512.mask.psra.wi.128(<8 x i16>, i8, <8 x i16>, i8)
+
+define <8 x i16>@test_int_x86_avx512_mask_psra_wi_128(<8 x i16> %x0, i8 %x1, <8 x i16> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psra_wi_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %sil, %eax
+; CHECK-NEXT:    kmovw %eax, %k1
+; CHECK-NEXT:    vpsraw $3, %xmm0, %xmm1 {%k1}
+; CHECK-NEXT:    vpsraw $3, %xmm0, %xmm2 {%k1} {z}
+; CHECK-NEXT:    vpsraw $3, %xmm0, %xmm0
+; CHECK-NEXT:    vpaddw %xmm2, %xmm1, %xmm1
+; CHECK-NEXT:    vpaddw %xmm0, %xmm1, %xmm0
+; CHECK-NEXT:    retq
+  %res = call <8 x i16> @llvm.x86.avx512.mask.psra.wi.128(<8 x i16> %x0, i8 3, <8 x i16> %x2, i8 %x3)
+  %res1 = call <8 x i16> @llvm.x86.avx512.mask.psra.wi.128(<8 x i16> %x0, i8 3, <8 x i16> zeroinitializer, i8 %x3)
+  %res2 = call <8 x i16> @llvm.x86.avx512.mask.psra.wi.128(<8 x i16> %x0, i8 3, <8 x i16> %x2, i8 -1)
+  %res3 = add <8 x i16> %res, %res1
+  %res4 = add <8 x i16> %res3, %res2
+  ret <8 x i16> %res4
+}
+
+declare <16 x i16> @llvm.x86.avx512.mask.psra.w.256(<16 x i16>, <8 x i16>, <16 x i16>, i16)
+
+define <16 x i16>@test_int_x86_avx512_mask_psra_w_256(<16 x i16> %x0, <8 x i16> %x1, <16 x i16> %x2, i16 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psra_w_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    kmovw %edi, %k1
+; CHECK-NEXT:    vpsraw %xmm1, %ymm0, %ymm2 {%k1}
+; CHECK-NEXT:    vpsraw %xmm1, %ymm0, %ymm3 {%k1} {z}
+; CHECK-NEXT:    vpsraw %xmm1, %ymm0, %ymm0
+; CHECK-NEXT:    vpaddw %ymm3, %ymm2, %ymm1
+; CHECK-NEXT:    vpaddw %ymm0, %ymm1, %ymm0
+; CHECK-NEXT:    retq
+  %res = call <16 x i16> @llvm.x86.avx512.mask.psra.w.256(<16 x i16> %x0, <8 x i16> %x1, <16 x i16> %x2, i16 %x3)
+  %res1 = call <16 x i16> @llvm.x86.avx512.mask.psra.w.256(<16 x i16> %x0, <8 x i16> %x1, <16 x i16> zeroinitializer, i16 %x3)
+  %res2 = call <16 x i16> @llvm.x86.avx512.mask.psra.w.256(<16 x i16> %x0, <8 x i16> %x1, <16 x i16> %x2, i16 -1)
+  %res3 = add <16 x i16> %res, %res1
+  %res4 = add <16 x i16> %res3, %res2
+  ret <16 x i16> %res4
+}
+
+declare <16 x i16> @llvm.x86.avx512.mask.psra.wi.256(<16 x i16>, i8, <16 x i16>, i16)
+
+define <16 x i16>@test_int_x86_avx512_mask_psra_wi_256(<16 x i16> %x0, i8 %x1, <16 x i16> %x2, i16 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_psra_wi_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    kmovw %esi, %k1
+; CHECK-NEXT:    vpsraw $3, %ymm0, %ymm1 {%k1}
+; CHECK-NEXT:    vpsraw $3, %ymm0, %ymm2 {%k1} {z}
+; CHECK-NEXT:    vpsraw $3, %ymm0, %ymm0
+; CHECK-NEXT:    vpaddw %ymm2, %ymm1, %ymm1
+; CHECK-NEXT:    vpaddw %ymm0, %ymm1, %ymm0
+; CHECK-NEXT:    retq
+  %res = call <16 x i16> @llvm.x86.avx512.mask.psra.wi.256(<16 x i16> %x0, i8 3, <16 x i16> %x2, i16 %x3)
+  %res1 = call <16 x i16> @llvm.x86.avx512.mask.psra.wi.256(<16 x i16> %x0, i8 3, <16 x i16> zeroinitializer, i16 %x3)
+  %res2 = call <16 x i16> @llvm.x86.avx512.mask.psra.wi.256(<16 x i16> %x0, i8 3, <16 x i16> %x2, i16 -1)
+  %res3 = add <16 x i16> %res, %res1
+  %res4 = add <16 x i16> %res3, %res2
+  ret <16 x i16> %res4
+}
+
+declare <4 x i32> @llvm.x86.avx512.mask.pshuf.d.128(<4 x i32>, i16, <4 x i32>, i8)
+
+define <4 x i32>@test_int_x86_avx512_mask_pshuf_d_128(<4 x i32> %x0, i16 %x1, <4 x i32> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pshuf_d_128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %sil, %eax 
+; CHECK-NEXT:    kmovw %eax, %k1 
+; CHECK-NEXT:    vpshufd $3, %xmm0, %xmm1 {%k1} 
+; CHECK-NEXT:    vpshufd $3, %xmm0, %xmm2 {%k1} {z} 
+; CHECK-NEXT:    vpshufd $3, %xmm0, %xmm0 
+; CHECK-NEXT:    ## xmm0 = xmm0[3,0,0,0]
+; CHECK-NEXT:    vpaddd %xmm2, %xmm1, %xmm1 
+; CHECK-NEXT:    vpaddd %xmm0, %xmm1, %xmm0 
+; CHECK-NEXT:    retq 
+       %res = call <4 x i32> @llvm.x86.avx512.mask.pshuf.d.128(<4 x i32> %x0, i16 3, <4 x i32> %x2, i8 %x3)
+       %res1 = call <4 x i32> @llvm.x86.avx512.mask.pshuf.d.128(<4 x i32> %x0, i16 3, <4 x i32> zeroinitializer, i8 %x3)
+       %res2 = call <4 x i32> @llvm.x86.avx512.mask.pshuf.d.128(<4 x i32> %x0, i16 3, <4 x i32> %x2, i8 -1)
+       %res3 = add <4 x i32> %res, %res1
+       %res4 = add <4 x i32> %res3, %res2
+       ret <4 x i32> %res4
+}
+
+declare <8 x i32> @llvm.x86.avx512.mask.pshuf.d.256(<8 x i32>, i16, <8 x i32>, i8)
+
+define <8 x i32>@test_int_x86_avx512_mask_pshuf_d_256(<8 x i32> %x0, i16 %x1, <8 x i32> %x2, i8 %x3) {
+; CHECK-LABEL: test_int_x86_avx512_mask_pshuf_d_256:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    movzbl %sil, %eax 
+; CHECK-NEXT:    kmovw %eax, %k1 
+; CHECK-NEXT:    vpshufd $3, %ymm0, %ymm1 {%k1} 
+; CHECK-NEXT:    vpshufd $3, %ymm0, %ymm2 {%k1} {z} 
+; CHECK-NEXT:    vpshufd $3, %ymm0, %ymm0 
+; CHECK-NEXT:    ## ymm0 = ymm0[3,0,0,0,7,4,4,4]
+; CHECK-NEXT:    vpaddd %ymm2, %ymm1, %ymm1 
+; CHECK-NEXT:    vpaddd %ymm0, %ymm1, %ymm0 
+; CHECK-NEXT:    retq 
+       %res = call <8 x i32> @llvm.x86.avx512.mask.pshuf.d.256(<8 x i32> %x0, i16 3, <8 x i32> %x2, i8 %x3)
+       %res1 = call <8 x i32> @llvm.x86.avx512.mask.pshuf.d.256(<8 x i32> %x0, i16 3, <8 x i32> zeroinitializer, i8 %x3)
+       %res2 = call <8 x i32> @llvm.x86.avx512.mask.pshuf.d.256(<8 x i32> %x0, i16 3, <8 x i32> %x2, i8 -1)
+       %res3 = add <8 x i32> %res, %res1
+       %res4 = add <8 x i32> %res3, %res2
+       ret <8 x i32> %res4
+}