; SSE2: # BB#0:
; SSE2-NEXT: xorps %xmm1, %xmm0
; SSE2-NEXT: xorps %xmm1, %xmm1
-; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
+; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[2,3]
; SSE2-NEXT: retq
;
; SSSE3-LABEL: combine_bitwise_ops_test3c:
; SSSE3: # BB#0:
; SSSE3-NEXT: xorps %xmm1, %xmm0
; SSSE3-NEXT: xorps %xmm1, %xmm1
-; SSSE3-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
+; SSSE3-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[2,3]
; SSSE3-NEXT: retq
;
; SSE41-LABEL: combine_bitwise_ops_test3c:
; SSE: # BB#0:
; SSE-NEXT: xorps %xmm1, %xmm0
; SSE-NEXT: xorps %xmm1, %xmm1
-; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[1,3]
+; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,1],xmm0[1,3]
; SSE-NEXT: movaps %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX: # BB#0:
; AVX-NEXT: vxorps %xmm1, %xmm0, %xmm0
; AVX-NEXT: vxorps %xmm1, %xmm1, %xmm1
-; AVX-NEXT: vshufps {{.*#+}} xmm0 = xmm1[0,2],xmm0[1,3]
+; AVX-NEXT: vshufps {{.*#+}} xmm0 = xmm1[0,1],xmm0[1,3]
; AVX-NEXT: retq
%shuf1 = shufflevector <4 x i32> %c, <4 x i32> %a, <4 x i32><i32 0, i32 2, i32 5, i32 7>
%shuf2 = shufflevector <4 x i32> %c, <4 x i32> %b, <4 x i32><i32 0, i32 2, i32 5, i32 7>
}
define <4 x i32> @combine_nested_undef_test16(<4 x i32> %A, <4 x i32> %B) {
-; SSE-LABEL: combine_nested_undef_test16:
-; SSE: # BB#0:
-; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[1,3]
-; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; SSE-NEXT: retq
+; SSE2-LABEL: combine_nested_undef_test16:
+; SSE2: # BB#0:
+; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[1,3]
+; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; SSE2-NEXT: retq
+;
+; SSSE3-LABEL: combine_nested_undef_test16:
+; SSSE3: # BB#0:
+; SSSE3-NEXT: shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[1,3]
+; SSSE3-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; SSSE3-NEXT: retq
+;
+; SSE41-LABEL: combine_nested_undef_test16:
+; SSE41: # BB#0:
+; SSE41-NEXT: blendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
+; SSE41-NEXT: shufps {{.*#+}} xmm0 = xmm0[2,1,0,3]
+; SSE41-NEXT: retq
;
; AVX-LABEL: combine_nested_undef_test16:
; AVX: # BB#0:
-; AVX-NEXT: vshufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[1,3]
-; AVX-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
+; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[2,1,0,3]
; AVX-NEXT: retq
%1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
%2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 2, i32 1, i32 0, i32 3>
}
define <4 x i32> @combine_nested_undef_test17(<4 x i32> %A, <4 x i32> %B) {
-; SSE-LABEL: combine_nested_undef_test17:
-; SSE: # BB#0:
-; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,0]
-; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[3,1],xmm1[0,2]
-; SSE-NEXT: retq
+; SSE2-LABEL: combine_nested_undef_test17:
+; SSE2: # BB#0:
+; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,0]
+; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[3,1],xmm1[0,2]
+; SSE2-NEXT: retq
+;
+; SSSE3-LABEL: combine_nested_undef_test17:
+; SSSE3: # BB#0:
+; SSSE3-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,0]
+; SSSE3-NEXT: shufps {{.*#+}} xmm0 = xmm0[3,1],xmm1[0,2]
+; SSSE3-NEXT: retq
+;
+; SSE41-LABEL: combine_nested_undef_test17:
+; SSE41: # BB#0:
+; SSE41-NEXT: blendps {{.*#+}} xmm0 = xmm1[0],xmm0[1,2,3]
+; SSE41-NEXT: shufps {{.*#+}} xmm0 = xmm0[3,1,0,1]
+; SSE41-NEXT: retq
;
; AVX-LABEL: combine_nested_undef_test17:
; AVX: # BB#0:
-; AVX-NEXT: vshufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,0]
-; AVX-NEXT: vshufps {{.*#+}} xmm0 = xmm0[3,1],xmm1[0,2]
+; AVX-NEXT: vblendps {{.*#+}} xmm0 = xmm1[0],xmm0[1,2,3]
+; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[3,1,0,1]
; AVX-NEXT: retq
%1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 4, i32 1, i32 3, i32 1>
%2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 2, i32 1, i32 0, i32 3>
}
define <4 x i32> @combine_nested_undef_test19(<4 x i32> %A, <4 x i32> %B) {
-; SSE-LABEL: combine_nested_undef_test19:
-; SSE: # BB#0:
-; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
-; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[0,0]
-; SSE-NEXT: movaps %xmm1, %xmm0
-; SSE-NEXT: retq
+; SSE2-LABEL: combine_nested_undef_test19:
+; SSE2: # BB#0:
+; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
+; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[0,0]
+; SSE2-NEXT: movaps %xmm1, %xmm0
+; SSE2-NEXT: retq
+;
+; SSSE3-LABEL: combine_nested_undef_test19:
+; SSSE3: # BB#0:
+; SSSE3-NEXT: shufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
+; SSSE3-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[0,0]
+; SSSE3-NEXT: movaps %xmm1, %xmm0
+; SSSE3-NEXT: retq
+;
+; SSE41-LABEL: combine_nested_undef_test19:
+; SSE41: # BB#0:
+; SSE41-NEXT: blendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3]
+; SSE41-NEXT: shufps {{.*#+}} xmm0 = xmm0[1,0,0,0]
+; SSE41-NEXT: retq
;
; AVX-LABEL: combine_nested_undef_test19:
; AVX: # BB#0:
-; AVX-NEXT: vshufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
-; AVX-NEXT: vshufps {{.*#+}} xmm0 = xmm1[0,2],xmm0[0,0]
+; AVX-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3]
+; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[1,0,0,0]
; AVX-NEXT: retq
%1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 0, i32 4, i32 5, i32 6>
%2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 2, i32 0, i32 0, i32 0>
}
define <4 x i32> @combine_nested_undef_test20(<4 x i32> %A, <4 x i32> %B) {
-; SSE-LABEL: combine_nested_undef_test20:
-; SSE: # BB#0:
-; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[2,3]
-; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,2,3,1]
-; SSE-NEXT: movaps %xmm1, %xmm0
-; SSE-NEXT: retq
+; SSE2-LABEL: combine_nested_undef_test20:
+; SSE2: # BB#0:
+; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[2,3]
+; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,2,3,1]
+; SSE2-NEXT: movaps %xmm1, %xmm0
+; SSE2-NEXT: retq
+;
+; SSSE3-LABEL: combine_nested_undef_test20:
+; SSSE3: # BB#0:
+; SSSE3-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[2,3]
+; SSSE3-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,2,3,1]
+; SSSE3-NEXT: movaps %xmm1, %xmm0
+; SSSE3-NEXT: retq
+;
+; SSE41-LABEL: combine_nested_undef_test20:
+; SSE41: # BB#0:
+; SSE41-NEXT: blendpd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
+; SSE41-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2,3,0]
+; SSE41-NEXT: retq
;
; AVX-LABEL: combine_nested_undef_test20:
; AVX: # BB#0:
-; AVX-NEXT: vshufps {{.*#+}} xmm0 = xmm1[0,0],xmm0[2,3]
-; AVX-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2,3,1]
+; AVX-NEXT: vblendpd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
+; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,2,3,0]
; AVX-NEXT: retq
%1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 3, i32 2, i32 4, i32 4>
%2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 2, i32 1, i32 0, i32 3>
}
define <4 x i32> @combine_nested_undef_test21(<4 x i32> %A, <4 x i32> %B) {
-; SSE-LABEL: combine_nested_undef_test21:
-; SSE: # BB#0:
-; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,1]
-; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,2,1,3]
-; SSE-NEXT: movaps %xmm1, %xmm0
-; SSE-NEXT: retq
+; SSE2-LABEL: combine_nested_undef_test21:
+; SSE2: # BB#0:
+; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,1]
+; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,2,1,3]
+; SSE2-NEXT: movaps %xmm1, %xmm0
+; SSE2-NEXT: retq
+;
+; SSSE3-LABEL: combine_nested_undef_test21:
+; SSSE3: # BB#0:
+; SSSE3-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,1]
+; SSSE3-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,2,1,3]
+; SSSE3-NEXT: movaps %xmm1, %xmm0
+; SSSE3-NEXT: retq
+;
+; SSE41-LABEL: combine_nested_undef_test21:
+; SSE41: # BB#0:
+; SSE41-NEXT: blendps {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2,3]
+; SSE41-NEXT: movddup {{.*#+}} xmm0 = xmm0[0,0]
+; SSE41-NEXT: retq
;
; AVX-LABEL: combine_nested_undef_test21:
; AVX: # BB#0:
-; AVX-NEXT: vshufps {{.*#+}} xmm0 = xmm1[0,0],xmm0[1,1]
-; AVX-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX-NEXT: vblendps {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2,3]
+; AVX-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0]
; AVX-NEXT: retq
%1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 4, i32 1, i32 3, i32 1>
%2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 3>
}
define <4 x float> @combine_test3b(<4 x float> %a, <4 x float> %b) {
-; SSE-LABEL: combine_test3b:
-; SSE: # BB#0:
-; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[3,0]
-; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[2,3]
-; SSE-NEXT: retq
+; SSE2-LABEL: combine_test3b:
+; SSE2: # BB#0:
+; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[3,0]
+; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[2,3]
+; SSE2-NEXT: retq
+;
+; SSSE3-LABEL: combine_test3b:
+; SSSE3: # BB#0:
+; SSSE3-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[3,0]
+; SSSE3-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[2,3]
+; SSSE3-NEXT: retq
+;
+; SSE41-LABEL: combine_test3b:
+; SSE41: # BB#0:
+; SSE41-NEXT: blendpd {{.*#+}} xmm0 = xmm0[0],xmm1[1]
+; SSE41-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,3,2,3]
+; SSE41-NEXT: retq
;
; AVX-LABEL: combine_test3b:
; AVX: # BB#0:
-; AVX-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[3,0]
-; AVX-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[2,3]
+; AVX-NEXT: vblendpd {{.*#+}} xmm0 = xmm0[0],xmm1[1]
+; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,3,2,3]
; AVX-NEXT: retq
%1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 0, i32 0, i32 6, i32 3>
%2 = shufflevector <4 x float> %1, <4 x float> %b, <4 x i32> <i32 0, i32 7, i32 2, i32 7>