1 files changed, 162 insertions, 0 deletions
diff --git a/llvm/test/CodeGen/X86/sad.ll b/llvm/test/CodeGen/X86/sad.ll
index f19fea31d7c..d89c46f2b7b 100644
--- a/llvm/test/CodeGen/X86/sad.ll
+++ b/llvm/test/CodeGen/X86/sad.ll
@@ -1505,3 +1505,165 @@ bb:
   ret i32 %tmp29
 }
 
+; This test contains two absolute difference patterns joined by an add. The result of that add is then reduced to a single element.
+; SelectionDAGBuilder should tag the joining add as a vector reduction. We neeed to recognize that both sides can use psadbw.
+define i32 @sad_double_reduction(<16 x i8>* %arg, <16 x i8>* %arg1, <16 x i8>* %arg2, <16 x i8>* %arg3) {
+; SSE2-LABEL: sad_double_reduction:
+; SSE2:       # %bb.0: # %bb
+; SSE2-NEXT:    movdqu (%rdi), %xmm0
+; SSE2-NEXT:    movdqu (%rsi), %xmm4
+; SSE2-NEXT:    pxor %xmm5, %xmm5
+; SSE2-NEXT:    movdqa %xmm0, %xmm3
+; SSE2-NEXT:    punpckhbw {{.*#+}} xmm3 = xmm3[8],xmm5[8],xmm3[9],xmm5[9],xmm3[10],xmm5[10],xmm3[11],xmm5[11],xmm3[12],xmm5[12],xmm3[13],xmm5[13],xmm3[14],xmm5[14],xmm3[15],xmm5[15]
+; SSE2-NEXT:    movdqa %xmm3, %xmm1
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1],xmm1[2],xmm5[2],xmm1[3],xmm5[3]
+; SSE2-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm5[4],xmm3[5],xmm5[5],xmm3[6],xmm5[6],xmm3[7],xmm5[7]
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1],xmm0[2],xmm5[2],xmm0[3],xmm5[3],xmm0[4],xmm5[4],xmm0[5],xmm5[5],xmm0[6],xmm5[6],xmm0[7],xmm5[7]
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm5[4],xmm2[5],xmm5[5],xmm2[6],xmm5[6],xmm2[7],xmm5[7]
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1],xmm0[2],xmm5[2],xmm0[3],xmm5[3]
+; SSE2-NEXT:    movdqa %xmm4, %xmm6
+; SSE2-NEXT:    punpckhbw {{.*#+}} xmm6 = xmm6[8],xmm5[8],xmm6[9],xmm5[9],xmm6[10],xmm5[10],xmm6[11],xmm5[11],xmm6[12],xmm5[12],xmm6[13],xmm5[13],xmm6[14],xmm5[14],xmm6[15],xmm5[15]
+; SSE2-NEXT:    movdqa %xmm6, %xmm7
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm7 = xmm7[0],xmm5[0],xmm7[1],xmm5[1],xmm7[2],xmm5[2],xmm7[3],xmm5[3]
+; SSE2-NEXT:    psubd %xmm7, %xmm1
+; SSE2-NEXT:    punpckhwd {{.*#+}} xmm6 = xmm6[4],xmm5[4],xmm6[5],xmm5[5],xmm6[6],xmm5[6],xmm6[7],xmm5[7]
+; SSE2-NEXT:    psubd %xmm6, %xmm3
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1],xmm4[2],xmm5[2],xmm4[3],xmm5[3],xmm4[4],xmm5[4],xmm4[5],xmm5[5],xmm4[6],xmm5[6],xmm4[7],xmm5[7]
+; SSE2-NEXT:    movdqa %xmm4, %xmm6
+; SSE2-NEXT:    punpckhwd {{.*#+}} xmm6 = xmm6[4],xmm5[4],xmm6[5],xmm5[5],xmm6[6],xmm5[6],xmm6[7],xmm5[7]
+; SSE2-NEXT:    psubd %xmm6, %xmm2
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1],xmm4[2],xmm5[2],xmm4[3],xmm5[3]
+; SSE2-NEXT:    psubd %xmm4, %xmm0
+; SSE2-NEXT:    movdqa %xmm1, %xmm4
+; SSE2-NEXT:    psrad $31, %xmm4
+; SSE2-NEXT:    paddd %xmm4, %xmm1
+; SSE2-NEXT:    pxor %xmm4, %xmm1
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    psrad $31, %xmm4
+; SSE2-NEXT:    paddd %xmm4, %xmm3
+; SSE2-NEXT:    pxor %xmm4, %xmm3
+; SSE2-NEXT:    movdqa %xmm2, %xmm4
+; SSE2-NEXT:    psrad $31, %xmm4
+; SSE2-NEXT:    paddd %xmm4, %xmm2
+; SSE2-NEXT:    pxor %xmm4, %xmm2
+; SSE2-NEXT:    paddd %xmm3, %xmm2
+; SSE2-NEXT:    movdqa %xmm0, %xmm3
+; SSE2-NEXT:    psrad $31, %xmm3
+; SSE2-NEXT:    paddd %xmm3, %xmm0
+; SSE2-NEXT:    pxor %xmm3, %xmm0
+; SSE2-NEXT:    paddd %xmm1, %xmm0
+; SSE2-NEXT:    paddd %xmm2, %xmm0
+; SSE2-NEXT:    movdqu (%rdx), %xmm1
+; SSE2-NEXT:    movdqu (%rcx), %xmm2
+; SSE2-NEXT:    psadbw %xmm1, %xmm2
+; SSE2-NEXT:    paddd %xmm0, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[2,3,0,1]
+; SSE2-NEXT:    paddd %xmm2, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
+; SSE2-NEXT:    paddd %xmm0, %xmm1
+; SSE2-NEXT:    movd %xmm1, %eax
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: sad_double_reduction:
+; AVX1:       # %bb.0: # %bb
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpsubd %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpsubd %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpsubd %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpsubd %xmm4, %xmm3, %xmm3
+; AVX1-NEXT:    vpabsd %xmm0, %xmm0
+; AVX1-NEXT:    vpabsd %xmm1, %xmm1
+; AVX1-NEXT:    vpabsd %xmm2, %xmm2
+; AVX1-NEXT:    vpaddd %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vpabsd %xmm3, %xmm2
+; AVX1-NEXT:    vpaddd %xmm0, %xmm2, %xmm0
+; AVX1-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vmovdqu (%rdx), %xmm1
+; AVX1-NEXT:    vpsadbw (%rcx), %xmm1, %xmm1
+; AVX1-NEXT:    vpaddd %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
+; AVX1-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vphaddd %xmm0, %xmm0, %xmm0
+; AVX1-NEXT:    vmovd %xmm0, %eax
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: sad_double_reduction:
+; AVX2:       # %bb.0: # %bb
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
+; AVX2-NEXT:    vpsubd %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
+; AVX2-NEXT:    vpsubd %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpabsd %ymm0, %ymm0
+; AVX2-NEXT:    vpabsd %ymm1, %ymm1
+; AVX2-NEXT:    vpaddd %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vmovdqu (%rdx), %xmm1
+; AVX2-NEXT:    vpsadbw (%rcx), %xmm1, %xmm1
+; AVX2-NEXT:    vpaddd %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT:    vpaddd %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
+; AVX2-NEXT:    vpaddd %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vphaddd %ymm0, %ymm0, %ymm0
+; AVX2-NEXT:    vmovd %xmm0, %eax
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
+;
+; AVX512-LABEL: sad_double_reduction:
+; AVX512:       # %bb.0: # %bb
+; AVX512-NEXT:    vpmovzxbd {{.*#+}} zmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
+; AVX512-NEXT:    vpmovzxbd {{.*#+}} zmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
+; AVX512-NEXT:    vpsubd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpabsd %zmm0, %zmm0
+; AVX512-NEXT:    vmovdqu (%rdx), %xmm1
+; AVX512-NEXT:    vpsadbw (%rcx), %xmm1, %xmm1
+; AVX512-NEXT:    vpaddd %zmm0, %zmm1, %zmm0
+; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
+; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
+; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
+; AVX512-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vmovd %xmm0, %eax
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
+bb:
+  %tmp = load <16 x i8>, <16 x i8>* %arg, align 1
+  %tmp4 = load <16 x i8>, <16 x i8>* %arg1, align 1
+  %tmp5 = zext <16 x i8> %tmp to <16 x i32>
+  %tmp6 = zext <16 x i8> %tmp4 to <16 x i32>
+  %tmp7 = sub nsw <16 x i32> %tmp5, %tmp6
+  %tmp8 = icmp slt <16 x i32> %tmp7, zeroinitializer
+  %tmp9 = sub nsw <16 x i32> zeroinitializer, %tmp7
+  %tmp10 = select <16 x i1> %tmp8, <16 x i32> %tmp9, <16 x i32> %tmp7
+  %tmp11 = load <16 x i8>, <16 x i8>* %arg2, align 1
+  %tmp12 = load <16 x i8>, <16 x i8>* %arg3, align 1
+  %tmp13 = zext <16 x i8> %tmp11 to <16 x i32>
+  %tmp14 = zext <16 x i8> %tmp12 to <16 x i32>
+  %tmp15 = sub nsw <16 x i32> %tmp13, %tmp14
+  %tmp16 = icmp slt <16 x i32> %tmp15, zeroinitializer
+  %tmp17 = sub nsw <16 x i32> zeroinitializer, %tmp15
+  %tmp18 = select <16 x i1> %tmp16, <16 x i32> %tmp17, <16 x i32> %tmp15
+  %tmp19 = add nuw nsw <16 x i32> %tmp18, %tmp10
+  %tmp20 = shufflevector <16 x i32> %tmp19, <16 x i32> undef, <16 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+  %tmp21 = add <16 x i32> %tmp19, %tmp20
+  %tmp22 = shufflevector <16 x i32> %tmp21, <16 x i32> undef, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+  %tmp23 = add <16 x i32> %tmp21, %tmp22
+  %tmp24 = shufflevector <16 x i32> %tmp23, <16 x i32> undef, <16 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+  %tmp25 = add <16 x i32> %tmp23, %tmp24
+  %tmp26 = shufflevector <16 x i32> %tmp25, <16 x i32> undef, <16 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
+  %tmp27 = add <16 x i32> %tmp25, %tmp26
+  %tmp28 = extractelement <16 x i32> %tmp27, i64 0
+  ret i32 %tmp28
+}