[DAGCombiner] Vector constant folding for comparisons

This patch adds support for vector constant folding of integer/float comparisons.

This requires FoldConstantVectorArithmetic to support scalar constant operands (in this case ISD::CONDCASE). In future we should be able to support other scalar constant types as necessary (and possibly start calling FoldConstantVectorArithmetic for all node creations)

Differential Revision: http://reviews.llvm.org/D14683

llvm-svn: 253504

1 files changed, 108 insertions, 1024 deletions

diff --git a/llvm/test/CodeGen/X86/vec_minmax_sint.ll b/llvm/test/CodeGen/X86/vec_minmax_sint.ll
index 80ceb80663a..c38d5a0c173 100644
--- a/llvm/test/CodeGen/X86/vec_minmax_sint.ll
+++ b/llvm/test/CodeGen/X86/vec_minmax_sint.ll
@@ -1601,64 +1601,14 @@ define <32 x i8> @min_le_v32i8(<32 x i8> %a, <32 x i8> %b) {
 ;
 
 define <2 x i64> @max_gt_v2i64c() {
-; SSE2-LABEL: max_gt_v2i64c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [18446744073709551609,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551615,1]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [2147483648,0,2147483648,0]
-; SSE2-NEXT:    movdqa %xmm0, %xmm3
-; SSE2-NEXT:    pxor %xmm2, %xmm3
-; SSE2-NEXT:    pxor %xmm1, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, %xmm4
-; SSE2-NEXT:    pcmpgtd %xmm3, %xmm4
-; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm3, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
-; SSE2-NEXT:    pand %xmm5, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[1,1,3,3]
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, %xmm3
-; SSE2-NEXT:    pandn %xmm2, %xmm3
-; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: max_gt_v2i64c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551609,7]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [18446744073709551615,1]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [2147483648,0,2147483648,0]
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    pxor %xmm1, %xmm3
-; SSE41-NEXT:    pxor %xmm2, %xmm0
-; SSE41-NEXT:    movdqa %xmm0, %xmm4
-; SSE41-NEXT:    pcmpgtd %xmm3, %xmm4
-; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
-; SSE41-NEXT:    pcmpeqd %xmm3, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
-; SSE41-NEXT:    pand %xmm5, %xmm3
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[1,1,3,3]
-; SSE41-NEXT:    por %xmm3, %xmm0
-; SSE41-NEXT:    blendvpd %xmm2, %xmm1
-; SSE41-NEXT:    movapd %xmm1, %xmm0
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: max_gt_v2i64c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551609,7]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm1 = [18446744073709551615,1]
-; SSE42-NEXT:    movdqa %xmm2, %xmm0
-; SSE42-NEXT:    pcmpgtq %xmm1, %xmm0
-; SSE42-NEXT:    blendvpd %xmm2, %xmm1
-; SSE42-NEXT:    movapd %xmm1, %xmm0
-; SSE42-NEXT:    retq
+; SSE-LABEL: max_gt_v2i64c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [18446744073709551615,7]
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: max_gt_v2i64c:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vmovdqa {{.*#+}} xmm0 = [18446744073709551609,7]
-; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709551615,1]
-; AVX-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm2
-; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [18446744073709551615,7]
 ; AVX-NEXT:    retq
   %1 = insertelement <2 x i64> <i64 -7, i64 7>, i64 -7, i32 0
   %2 = insertelement <2 x i64> <i64 -1, i64 1>, i64 -1, i32 0
@@ -1668,124 +1618,16 @@ define <2 x i64> @max_gt_v2i64c() {
 }
 
 define <4 x i64> @max_gt_v4i64c() {
-; SSE2-LABEL: max_gt_v4i64c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [18446744073709551609,18446744073709551615]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm8 = [1,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm5 = [18446744073709551615,18446744073709551609]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [7,1]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [2147483648,0,2147483648,0]
-; SSE2-NEXT:    movdqa %xmm0, %xmm1
-; SSE2-NEXT:    pxor %xmm3, %xmm1
-; SSE2-NEXT:    movdqa %xmm0, %xmm6
-; SSE2-NEXT:    pxor %xmm8, %xmm6
-; SSE2-NEXT:    movdqa %xmm6, %xmm7
-; SSE2-NEXT:    pcmpgtd %xmm1, %xmm7
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm7[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm1, %xmm6
-; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[1,1,3,3]
-; SSE2-NEXT:    pand %xmm2, %xmm6
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm7[1,1,3,3]
-; SSE2-NEXT:    por %xmm6, %xmm1
-; SSE2-NEXT:    movdqa %xmm0, %xmm2
-; SSE2-NEXT:    pxor %xmm5, %xmm2
-; SSE2-NEXT:    pxor %xmm4, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, %xmm6
-; SSE2-NEXT:    pcmpgtd %xmm2, %xmm6
-; SSE2-NEXT:    pshufd {{.*#+}} xmm7 = xmm6[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm2, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
-; SSE2-NEXT:    pand %xmm7, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm6[1,1,3,3]
-; SSE2-NEXT:    por %xmm2, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, %xmm2
-; SSE2-NEXT:    pandn %xmm5, %xmm2
-; SSE2-NEXT:    pand %xmm4, %xmm0
-; SSE2-NEXT:    por %xmm2, %xmm0
-; SSE2-NEXT:    movdqa %xmm1, %xmm2
-; SSE2-NEXT:    pandn %xmm3, %xmm2
-; SSE2-NEXT:    pand %xmm8, %xmm1
-; SSE2-NEXT:    por %xmm2, %xmm1
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: max_gt_v4i64c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movdqa {{.*#+}} xmm5 = [18446744073709551609,18446744073709551615]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm8 = [1,7]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551615,18446744073709551609]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [7,1]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [2147483648,0,2147483648,0]
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    pxor %xmm1, %xmm3
-; SSE41-NEXT:    movdqa %xmm0, %xmm6
-; SSE41-NEXT:    pxor %xmm8, %xmm6
-; SSE41-NEXT:    movdqa %xmm6, %xmm7
-; SSE41-NEXT:    pcmpgtd %xmm3, %xmm7
-; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm7[0,0,2,2]
-; SSE41-NEXT:    pcmpeqd %xmm3, %xmm6
-; SSE41-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[1,1,3,3]
-; SSE41-NEXT:    pand %xmm4, %xmm6
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm7[1,1,3,3]
-; SSE41-NEXT:    por %xmm6, %xmm3
-; SSE41-NEXT:    movdqa %xmm0, %xmm4
-; SSE41-NEXT:    pxor %xmm2, %xmm4
-; SSE41-NEXT:    pxor %xmm5, %xmm0
-; SSE41-NEXT:    movdqa %xmm0, %xmm6
-; SSE41-NEXT:    pcmpgtd %xmm4, %xmm6
-; SSE41-NEXT:    pshufd {{.*#+}} xmm7 = xmm6[0,0,2,2]
-; SSE41-NEXT:    pcmpeqd %xmm4, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm0[1,1,3,3]
-; SSE41-NEXT:    pand %xmm7, %xmm4
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm6[1,1,3,3]
-; SSE41-NEXT:    por %xmm4, %xmm0
-; SSE41-NEXT:    blendvpd %xmm5, %xmm2
-; SSE41-NEXT:    movdqa %xmm3, %xmm0
-; SSE41-NEXT:    blendvpd %xmm8, %xmm1
-; SSE41-NEXT:    movapd %xmm2, %xmm0
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: max_gt_v4i64c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movdqa {{.*#+}} xmm4 = [18446744073709551609,18446744073709551615]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm5 = [1,7]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551615,18446744073709551609]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm1 = [7,1]
-; SSE42-NEXT:    movdqa %xmm5, %xmm3
-; SSE42-NEXT:    pcmpgtq %xmm1, %xmm3
-; SSE42-NEXT:    movdqa %xmm4, %xmm0
-; SSE42-NEXT:    pcmpgtq %xmm2, %xmm0
-; SSE42-NEXT:    blendvpd %xmm4, %xmm2
-; SSE42-NEXT:    movdqa %xmm3, %xmm0
-; SSE42-NEXT:    blendvpd %xmm5, %xmm1
-; SSE42-NEXT:    movapd %xmm2, %xmm0
-; SSE42-NEXT:    retq
-;
-; AVX1-LABEL: max_gt_v4i64c:
-; AVX1:       # BB#0:
-; AVX1-NEXT:    vmovapd {{.*#+}} ymm0 = [18446744073709551615,18446744073709551609,7,1]
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709551609,18446744073709551615]
-; AVX1-NEXT:    vpcmpgtq {{.*}}(%rip), %xmm1, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [1,7]
-; AVX1-NEXT:    vpcmpgtq {{.*}}(%rip), %xmm2, %xmm2
-; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm1, %ymm1
-; AVX1-NEXT:    vblendvpd %ymm1, {{.*}}(%rip), %ymm0, %ymm0
-; AVX1-NEXT:    retq
-;
-; AVX2-LABEL: max_gt_v4i64c:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm0 = [18446744073709551609,18446744073709551615,1,7]
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = [18446744073709551615,18446744073709551609,7,1]
-; AVX2-NEXT:    vpcmpgtq %ymm1, %ymm0, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
-; AVX2-NEXT:    retq
+; SSE-LABEL: max_gt_v4i64c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [7,7]
+; SSE-NEXT:    pcmpeqd %xmm0, %xmm0
+; SSE-NEXT:    retq
 ;
-; AVX512-LABEL: max_gt_v4i64c:
-; AVX512:       # BB#0:
-; AVX512-NEXT:    vmovdqa {{.*#+}} ymm0 = [18446744073709551609,18446744073709551615,1,7]
-; AVX512-NEXT:    vmovdqa {{.*#+}} ymm1 = [18446744073709551615,18446744073709551609,7,1]
-; AVX512-NEXT:    vpcmpgtq %ymm1, %ymm0, %ymm2
-; AVX512-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
-; AVX512-NEXT:    retq
+; AVX-LABEL: max_gt_v4i64c:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovaps {{.*#+}} ymm0 = [18446744073709551615,18446744073709551615,7,7]
+; AVX-NEXT:    retq
   %1 = insertelement <4 x i64> <i64 -7, i64 -1, i64 1, i64 7>, i64 -7, i32 0
   %2 = insertelement <4 x i64> <i64 -1, i64 -7, i64 7, i64 1>, i64 -1, i32 0
   %3 = icmp sgt <4 x i64> %1, %2
@@ -1794,26 +1636,10 @@ define <4 x i64> @max_gt_v4i64c() {
 }
 
 define <4 x i32> @max_gt_v4i32c() {
-; SSE2-LABEL: max_gt_v4i32c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [4294967289,4294967295,1,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [4294967295,4294967289,7,1]
-; SSE2-NEXT:    movdqa %xmm1, %xmm0
-; SSE2-NEXT:    pcmpgtd %xmm2, %xmm0
-; SSE2-NEXT:    pand %xmm0, %xmm1
-; SSE2-NEXT:    pandn %xmm2, %xmm0
-; SSE2-NEXT:    por %xmm1, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: max_gt_v4i32c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [4294967295,4294967295,7,7]
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: max_gt_v4i32c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movaps {{.*#+}} xmm0 = [4294967295,4294967295,7,7]
-; SSE42-NEXT:    retq
+; SSE-LABEL: max_gt_v4i32c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [4294967295,4294967295,7,7]
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: max_gt_v4i32c:
 ; AVX:       # BB#0:
@@ -1827,35 +1653,11 @@ define <4 x i32> @max_gt_v4i32c() {
 }
 
 define <8 x i32> @max_gt_v8i32c() {
-; SSE2-LABEL: max_gt_v8i32c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [4294967289,4294967291,4294967293,4294967295]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [1,3,5,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [4294967295,4294967293,4294967291,4294967289]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm5 = [7,5,3,1]
-; SSE2-NEXT:    movdqa %xmm3, %xmm1
-; SSE2-NEXT:    pcmpgtd %xmm5, %xmm1
-; SSE2-NEXT:    movdqa %xmm2, %xmm0
-; SSE2-NEXT:    pcmpgtd %xmm4, %xmm0
-; SSE2-NEXT:    pand %xmm0, %xmm2
-; SSE2-NEXT:    pandn %xmm4, %xmm0
-; SSE2-NEXT:    por %xmm2, %xmm0
-; SSE2-NEXT:    pand %xmm1, %xmm3
-; SSE2-NEXT:    pandn %xmm5, %xmm1
-; SSE2-NEXT:    por %xmm3, %xmm1
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: max_gt_v8i32c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [4294967295,4294967293,4294967293,4294967295]
-; SSE41-NEXT:    movaps {{.*#+}} xmm1 = [7,5,5,7]
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: max_gt_v8i32c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movaps {{.*#+}} xmm0 = [4294967295,4294967293,4294967293,4294967295]
-; SSE42-NEXT:    movaps {{.*#+}} xmm1 = [7,5,5,7]
-; SSE42-NEXT:    retq
+; SSE-LABEL: max_gt_v8i32c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [4294967295,4294967293,4294967293,4294967295]
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [7,5,5,7]
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: max_gt_v8i32c:
 ; AVX:       # BB#0:
@@ -1904,26 +1706,10 @@ define <16 x i16> @max_gt_v16i16c() {
 }
 
 define <16 x i8> @max_gt_v16i8c() {
-; SSE2-LABEL: max_gt_v16i8c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [249,250,251,252,253,254,255,0,1,2,3,4,5,6,7,8]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,254,253,252,251,250,249,0,7,6,5,4,3,2,1,0]
-; SSE2-NEXT:    movdqa %xmm1, %xmm0
-; SSE2-NEXT:    pcmpgtb %xmm2, %xmm0
-; SSE2-NEXT:    pand %xmm0, %xmm1
-; SSE2-NEXT:    pandn %xmm2, %xmm0
-; SSE2-NEXT:    por %xmm1, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: max_gt_v16i8c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [255,254,253,252,253,254,255,0,7,6,5,4,5,6,7,8]
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: max_gt_v16i8c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movaps {{.*#+}} xmm0 = [255,254,253,252,253,254,255,0,7,6,5,4,5,6,7,8]
-; SSE42-NEXT:    retq
+; SSE-LABEL: max_gt_v16i8c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [255,254,253,252,253,254,255,0,7,6,5,4,5,6,7,8]
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: max_gt_v16i8c:
 ; AVX:       # BB#0:
@@ -1937,71 +1723,14 @@ define <16 x i8> @max_gt_v16i8c() {
 }
 
 define <2 x i64> @max_ge_v2i64c() {
-; SSE2-LABEL: max_ge_v2i64c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [18446744073709551609,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551615,1]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [2147483648,0,2147483648,0]
-; SSE2-NEXT:    movdqa %xmm0, %xmm3
-; SSE2-NEXT:    pxor %xmm1, %xmm3
-; SSE2-NEXT:    pxor %xmm2, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, %xmm4
-; SSE2-NEXT:    pcmpgtd %xmm3, %xmm4
-; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm3, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
-; SSE2-NEXT:    pand %xmm5, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
-; SSE2-NEXT:    por %xmm0, %xmm3
-; SSE2-NEXT:    pcmpeqd %xmm0, %xmm0
-; SSE2-NEXT:    pxor %xmm3, %xmm0
-; SSE2-NEXT:    pandn %xmm1, %xmm3
-; SSE2-NEXT:    pandn %xmm2, %xmm0
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: max_ge_v2i64c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551609,7]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [18446744073709551615,1]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [2147483648,0,2147483648,0]
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    pxor %xmm2, %xmm3
-; SSE41-NEXT:    pxor %xmm1, %xmm0
-; SSE41-NEXT:    movdqa %xmm0, %xmm4
-; SSE41-NEXT:    pcmpgtd %xmm3, %xmm4
-; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
-; SSE41-NEXT:    pcmpeqd %xmm3, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
-; SSE41-NEXT:    pand %xmm5, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
-; SSE41-NEXT:    por %xmm0, %xmm3
-; SSE41-NEXT:    pcmpeqd %xmm0, %xmm0
-; SSE41-NEXT:    pxor %xmm3, %xmm0
-; SSE41-NEXT:    blendvpd %xmm2, %xmm1
-; SSE41-NEXT:    movapd %xmm1, %xmm0
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: max_ge_v2i64c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551609,7]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm1 = [18446744073709551615,1]
-; SSE42-NEXT:    movdqa %xmm1, %xmm3
-; SSE42-NEXT:    pcmpgtq %xmm2, %xmm3
-; SSE42-NEXT:    pcmpeqd %xmm0, %xmm0
-; SSE42-NEXT:    pxor %xmm3, %xmm0
-; SSE42-NEXT:    blendvpd %xmm2, %xmm1
-; SSE42-NEXT:    movapd %xmm1, %xmm0
-; SSE42-NEXT:    retq
+; SSE-LABEL: max_ge_v2i64c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [18446744073709551615,7]
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: max_ge_v2i64c:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vmovdqa {{.*#+}} xmm0 = [18446744073709551609,7]
-; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709551615,1]
-; AVX-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
-; AVX-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
-; AVX-NEXT:    vpxor %xmm3, %xmm2, %xmm2
-; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [18446744073709551615,7]
 ; AVX-NEXT:    retq
   %1 = insertelement <2 x i64> <i64 -7, i64 7>, i64 -7, i32 0
   %2 = insertelement <2 x i64> <i64 -1, i64 1>, i64 -1, i32 0
@@ -2011,139 +1740,16 @@ define <2 x i64> @max_ge_v2i64c() {
 }
 
 define <4 x i64> @max_ge_v4i64c() {
-; SSE2-LABEL: max_ge_v4i64c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm10 = [18446744073709551609,18446744073709551615]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm8 = [1,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm5 = [18446744073709551615,18446744073709551609]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm9 = [7,1]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm7 = [2147483648,0,2147483648,0]
-; SSE2-NEXT:    movdqa %xmm7, %xmm0
-; SSE2-NEXT:    pxor %xmm8, %xmm0
-; SSE2-NEXT:    movdqa %xmm7, %xmm1
-; SSE2-NEXT:    pxor %xmm9, %xmm1
-; SSE2-NEXT:    movdqa %xmm1, %xmm6
-; SSE2-NEXT:    pcmpgtd %xmm0, %xmm6
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm6[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
-; SSE2-NEXT:    pand %xmm2, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[1,1,3,3]
-; SSE2-NEXT:    por %xmm0, %xmm6
-; SSE2-NEXT:    pcmpeqd %xmm0, %xmm0
-; SSE2-NEXT:    movdqa %xmm6, %xmm1
-; SSE2-NEXT:    pxor %xmm0, %xmm1
-; SSE2-NEXT:    movdqa %xmm7, %xmm2
-; SSE2-NEXT:    pxor %xmm10, %xmm2
-; SSE2-NEXT:    pxor %xmm5, %xmm7
-; SSE2-NEXT:    movdqa %xmm7, %xmm3
-; SSE2-NEXT:    pcmpgtd %xmm2, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm2, %xmm7
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm7[1,1,3,3]
-; SSE2-NEXT:    pand %xmm4, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
-; SSE2-NEXT:    por %xmm2, %xmm3
-; SSE2-NEXT:    pxor %xmm3, %xmm0
-; SSE2-NEXT:    pandn %xmm10, %xmm3
-; SSE2-NEXT:    pandn %xmm5, %xmm0
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    pandn %xmm8, %xmm6
-; SSE2-NEXT:    pandn %xmm9, %xmm1
-; SSE2-NEXT:    por %xmm6, %xmm1
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: max_ge_v4i64c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movdqa {{.*#+}} xmm9 = [18446744073709551609,18446744073709551615]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm8 = [1,7]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551615,18446744073709551609]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [7,1]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [2147483648,0,2147483648,0]
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    pxor %xmm8, %xmm3
-; SSE41-NEXT:    movdqa %xmm0, %xmm6
-; SSE41-NEXT:    pxor %xmm1, %xmm6
-; SSE41-NEXT:    movdqa %xmm6, %xmm7
-; SSE41-NEXT:    pcmpgtd %xmm3, %xmm7
-; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm7[0,0,2,2]
-; SSE41-NEXT:    pcmpeqd %xmm3, %xmm6
-; SSE41-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[1,1,3,3]
-; SSE41-NEXT:    pand %xmm4, %xmm6
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm7[1,1,3,3]
-; SSE41-NEXT:    por %xmm6, %xmm3
-; SSE41-NEXT:    pcmpeqd %xmm4, %xmm4
-; SSE41-NEXT:    pxor %xmm4, %xmm3
-; SSE41-NEXT:    movdqa %xmm0, %xmm6
-; SSE41-NEXT:    pxor %xmm9, %xmm6
-; SSE41-NEXT:    pxor %xmm2, %xmm0
-; SSE41-NEXT:    movdqa %xmm0, %xmm7
-; SSE41-NEXT:    pcmpgtd %xmm6, %xmm7
-; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm7[0,0,2,2]
-; SSE41-NEXT:    pcmpeqd %xmm6, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm6 = xmm0[1,1,3,3]
-; SSE41-NEXT:    pand %xmm5, %xmm6
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm7[1,1,3,3]
-; SSE41-NEXT:    por %xmm6, %xmm0
-; SSE41-NEXT:    pxor %xmm4, %xmm0
-; SSE41-NEXT:    blendvpd %xmm9, %xmm2
-; SSE41-NEXT:    movdqa %xmm3, %xmm0
-; SSE41-NEXT:    blendvpd %xmm8, %xmm1
-; SSE41-NEXT:    movapd %xmm2, %xmm0
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: max_ge_v4i64c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movdqa {{.*#+}} xmm4 = [18446744073709551609,18446744073709551615]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm5 = [1,7]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551615,18446744073709551609]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm1 = [7,1]
-; SSE42-NEXT:    movdqa %xmm1, %xmm3
-; SSE42-NEXT:    pcmpgtq %xmm5, %xmm3
-; SSE42-NEXT:    pcmpeqd %xmm6, %xmm6
-; SSE42-NEXT:    pxor %xmm6, %xmm3
-; SSE42-NEXT:    movdqa %xmm2, %xmm0
-; SSE42-NEXT:    pcmpgtq %xmm4, %xmm0
-; SSE42-NEXT:    pxor %xmm6, %xmm0
-; SSE42-NEXT:    blendvpd %xmm4, %xmm2
-; SSE42-NEXT:    movdqa %xmm3, %xmm0
-; SSE42-NEXT:    blendvpd %xmm5, %xmm1
-; SSE42-NEXT:    movapd %xmm2, %xmm0
-; SSE42-NEXT:    retq
-;
-; AVX1-LABEL: max_ge_v4i64c:
-; AVX1:       # BB#0:
-; AVX1-NEXT:    vmovapd {{.*#+}} ymm0 = [18446744073709551615,18446744073709551609,7,1]
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [7,1]
-; AVX1-NEXT:    vpcmpgtq {{.*}}(%rip), %xmm1, %xmm1
-; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpxor %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [18446744073709551615,18446744073709551609]
-; AVX1-NEXT:    vpcmpgtq {{.*}}(%rip), %xmm3, %xmm3
-; AVX1-NEXT:    vpxor %xmm2, %xmm3, %xmm2
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm2, %ymm1
-; AVX1-NEXT:    vblendvpd %ymm1, {{.*}}(%rip), %ymm0, %ymm0
-; AVX1-NEXT:    retq
-;
-; AVX2-LABEL: max_ge_v4i64c:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm0 = [18446744073709551609,18446744073709551615,1,7]
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = [18446744073709551615,18446744073709551609,7,1]
-; AVX2-NEXT:    vpcmpgtq %ymm0, %ymm1, %ymm2
-; AVX2-NEXT:    vpcmpeqd %ymm3, %ymm3, %ymm3
-; AVX2-NEXT:    vpxor %ymm3, %ymm2, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
-; AVX2-NEXT:    retq
+; SSE-LABEL: max_ge_v4i64c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [7,7]
+; SSE-NEXT:    pcmpeqd %xmm0, %xmm0
+; SSE-NEXT:    retq
 ;
-; AVX512-LABEL: max_ge_v4i64c:
-; AVX512:       # BB#0:
-; AVX512-NEXT:    vmovdqa {{.*#+}} ymm0 = [18446744073709551609,18446744073709551615,1,7]
-; AVX512-NEXT:    vmovdqa {{.*#+}} ymm1 = [18446744073709551615,18446744073709551609,7,1]
-; AVX512-NEXT:    vpcmpgtq %ymm0, %ymm1, %ymm2
-; AVX512-NEXT:    vpcmpeqd %ymm3, %ymm3, %ymm3
-; AVX512-NEXT:    vpxor %ymm3, %ymm2, %ymm2
-; AVX512-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
-; AVX512-NEXT:    retq
+; AVX-LABEL: max_ge_v4i64c:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovaps {{.*#+}} ymm0 = [18446744073709551615,18446744073709551615,7,7]
+; AVX-NEXT:    retq
   %1 = insertelement <4 x i64> <i64 -7, i64 -1, i64 1, i64 7>, i64 -7, i32 0
   %2 = insertelement <4 x i64> <i64 -1, i64 -7, i64 7, i64 1>, i64 -1, i32 0
   %3 = icmp sge <4 x i64> %1, %2
@@ -2152,28 +1758,10 @@ define <4 x i64> @max_ge_v4i64c() {
 }
 
 define <4 x i32> @max_ge_v4i32c() {
-; SSE2-LABEL: max_ge_v4i32c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [4294967289,4294967295,1,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [4294967295,4294967289,7,1]
-; SSE2-NEXT:    movdqa %xmm2, %xmm3
-; SSE2-NEXT:    pcmpgtd %xmm1, %xmm3
-; SSE2-NEXT:    pcmpeqd %xmm0, %xmm0
-; SSE2-NEXT:    pxor %xmm3, %xmm0
-; SSE2-NEXT:    pandn %xmm1, %xmm3
-; SSE2-NEXT:    pandn %xmm2, %xmm0
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: max_ge_v4i32c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [4294967295,4294967295,7,7]
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: max_ge_v4i32c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movaps {{.*#+}} xmm0 = [4294967295,4294967295,7,7]
-; SSE42-NEXT:    retq
+; SSE-LABEL: max_ge_v4i32c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [4294967295,4294967295,7,7]
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: max_ge_v4i32c:
 ; AVX:       # BB#0:
@@ -2187,39 +1775,11 @@ define <4 x i32> @max_ge_v4i32c() {
 }
 
 define <8 x i32> @max_ge_v8i32c() {
-; SSE2-LABEL: max_ge_v8i32c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [4294967289,4294967291,4294967293,4294967295]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [1,3,5,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [4294967295,4294967293,4294967291,4294967289]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm5 = [7,5,3,1]
-; SSE2-NEXT:    movdqa %xmm5, %xmm6
-; SSE2-NEXT:    pcmpgtd %xmm3, %xmm6
-; SSE2-NEXT:    pcmpeqd %xmm0, %xmm0
-; SSE2-NEXT:    movdqa %xmm6, %xmm1
-; SSE2-NEXT:    pxor %xmm0, %xmm1
-; SSE2-NEXT:    movdqa %xmm4, %xmm7
-; SSE2-NEXT:    pcmpgtd %xmm2, %xmm7
-; SSE2-NEXT:    pxor %xmm7, %xmm0
-; SSE2-NEXT:    pandn %xmm2, %xmm7
-; SSE2-NEXT:    pandn %xmm4, %xmm0
-; SSE2-NEXT:    por %xmm7, %xmm0
-; SSE2-NEXT:    pandn %xmm3, %xmm6
-; SSE2-NEXT:    pandn %xmm5, %xmm1
-; SSE2-NEXT:    por %xmm6, %xmm1
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: max_ge_v8i32c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [4294967295,4294967293,4294967293,4294967295]
-; SSE41-NEXT:    movaps {{.*#+}} xmm1 = [7,5,5,7]
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: max_ge_v8i32c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movaps {{.*#+}} xmm0 = [4294967295,4294967293,4294967293,4294967295]
-; SSE42-NEXT:    movaps {{.*#+}} xmm1 = [7,5,5,7]
-; SSE42-NEXT:    retq
+; SSE-LABEL: max_ge_v8i32c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [4294967295,4294967293,4294967293,4294967295]
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [7,5,5,7]
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: max_ge_v8i32c:
 ; AVX:       # BB#0:
@@ -2268,28 +1828,10 @@ define <16 x i16> @max_ge_v16i16c() {
 }
 
 define <16 x i8> @max_ge_v16i8c() {
-; SSE2-LABEL: max_ge_v16i8c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [249,250,251,252,253,254,255,0,1,2,3,4,5,6,7,8]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,254,253,252,251,250,249,0,7,6,5,4,3,2,1,0]
-; SSE2-NEXT:    movdqa %xmm2, %xmm3
-; SSE2-NEXT:    pcmpgtb %xmm1, %xmm3
-; SSE2-NEXT:    pcmpeqd %xmm0, %xmm0
-; SSE2-NEXT:    pxor %xmm3, %xmm0
-; SSE2-NEXT:    pandn %xmm1, %xmm3
-; SSE2-NEXT:    pandn %xmm2, %xmm0
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: max_ge_v16i8c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [255,254,253,252,253,254,255,0,7,6,5,4,5,6,7,8]
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: max_ge_v16i8c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movaps {{.*#+}} xmm0 = [255,254,253,252,253,254,255,0,7,6,5,4,5,6,7,8]
-; SSE42-NEXT:    retq
+; SSE-LABEL: max_ge_v16i8c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [255,254,253,252,253,254,255,0,7,6,5,4,5,6,7,8]
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: max_ge_v16i8c:
 ; AVX:       # BB#0:
@@ -2303,64 +1845,14 @@ define <16 x i8> @max_ge_v16i8c() {
 }
 
 define <2 x i64> @min_lt_v2i64c() {
-; SSE2-LABEL: min_lt_v2i64c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [18446744073709551609,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551615,1]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [2147483648,0,2147483648,0]
-; SSE2-NEXT:    movdqa %xmm0, %xmm3
-; SSE2-NEXT:    pxor %xmm1, %xmm3
-; SSE2-NEXT:    pxor %xmm2, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, %xmm4
-; SSE2-NEXT:    pcmpgtd %xmm3, %xmm4
-; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm3, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
-; SSE2-NEXT:    pand %xmm5, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[1,1,3,3]
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, %xmm3
-; SSE2-NEXT:    pandn %xmm2, %xmm3
-; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: min_lt_v2i64c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551609,7]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [18446744073709551615,1]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [2147483648,0,2147483648,0]
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    pxor %xmm2, %xmm3
-; SSE41-NEXT:    pxor %xmm1, %xmm0
-; SSE41-NEXT:    movdqa %xmm0, %xmm4
-; SSE41-NEXT:    pcmpgtd %xmm3, %xmm4
-; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
-; SSE41-NEXT:    pcmpeqd %xmm3, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
-; SSE41-NEXT:    pand %xmm5, %xmm3
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[1,1,3,3]
-; SSE41-NEXT:    por %xmm3, %xmm0
-; SSE41-NEXT:    blendvpd %xmm2, %xmm1
-; SSE41-NEXT:    movapd %xmm1, %xmm0
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: min_lt_v2i64c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551609,7]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm1 = [18446744073709551615,1]
-; SSE42-NEXT:    movdqa %xmm1, %xmm0
-; SSE42-NEXT:    pcmpgtq %xmm2, %xmm0
-; SSE42-NEXT:    blendvpd %xmm2, %xmm1
-; SSE42-NEXT:    movapd %xmm1, %xmm0
-; SSE42-NEXT:    retq
+; SSE-LABEL: min_lt_v2i64c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [18446744073709551609,1]
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: min_lt_v2i64c:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vmovdqa {{.*#+}} xmm0 = [18446744073709551609,7]
-; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709551615,1]
-; AVX-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
-; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [18446744073709551609,1]
 ; AVX-NEXT:    retq
   %1 = insertelement <2 x i64> <i64 -7, i64 7>, i64 -7, i32 0
   %2 = insertelement <2 x i64> <i64 -1, i64 1>, i64 -1, i32 0
@@ -2370,124 +1862,16 @@ define <2 x i64> @min_lt_v2i64c() {
 }
 
 define <4 x i64> @min_lt_v4i64c() {
-; SSE2-LABEL: min_lt_v4i64c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [18446744073709551609,18446744073709551615]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm8 = [1,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm5 = [18446744073709551615,18446744073709551609]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [7,1]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [2147483648,0,2147483648,0]
-; SSE2-NEXT:    movdqa %xmm0, %xmm1
-; SSE2-NEXT:    pxor %xmm8, %xmm1
-; SSE2-NEXT:    movdqa %xmm0, %xmm6
-; SSE2-NEXT:    pxor %xmm3, %xmm6
-; SSE2-NEXT:    movdqa %xmm6, %xmm7
-; SSE2-NEXT:    pcmpgtd %xmm1, %xmm7
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm7[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm1, %xmm6
-; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[1,1,3,3]
-; SSE2-NEXT:    pand %xmm2, %xmm6
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm7[1,1,3,3]
-; SSE2-NEXT:    por %xmm6, %xmm1
-; SSE2-NEXT:    movdqa %xmm0, %xmm2
-; SSE2-NEXT:    pxor %xmm4, %xmm2
-; SSE2-NEXT:    pxor %xmm5, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, %xmm6
-; SSE2-NEXT:    pcmpgtd %xmm2, %xmm6
-; SSE2-NEXT:    pshufd {{.*#+}} xmm7 = xmm6[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm2, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
-; SSE2-NEXT:    pand %xmm7, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm6[1,1,3,3]
-; SSE2-NEXT:    por %xmm2, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, %xmm2
-; SSE2-NEXT:    pandn %xmm5, %xmm2
-; SSE2-NEXT:    pand %xmm4, %xmm0
-; SSE2-NEXT:    por %xmm2, %xmm0
-; SSE2-NEXT:    movdqa %xmm1, %xmm2
-; SSE2-NEXT:    pandn %xmm3, %xmm2
-; SSE2-NEXT:    pand %xmm8, %xmm1
-; SSE2-NEXT:    por %xmm2, %xmm1
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: min_lt_v4i64c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movdqa {{.*#+}} xmm5 = [18446744073709551609,18446744073709551615]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm8 = [1,7]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551615,18446744073709551609]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [7,1]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [2147483648,0,2147483648,0]
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    pxor %xmm8, %xmm3
-; SSE41-NEXT:    movdqa %xmm0, %xmm6
-; SSE41-NEXT:    pxor %xmm1, %xmm6
-; SSE41-NEXT:    movdqa %xmm6, %xmm7
-; SSE41-NEXT:    pcmpgtd %xmm3, %xmm7
-; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm7[0,0,2,2]
-; SSE41-NEXT:    pcmpeqd %xmm3, %xmm6
-; SSE41-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[1,1,3,3]
-; SSE41-NEXT:    pand %xmm4, %xmm6
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm7[1,1,3,3]
-; SSE41-NEXT:    por %xmm6, %xmm3
-; SSE41-NEXT:    movdqa %xmm0, %xmm4
-; SSE41-NEXT:    pxor %xmm5, %xmm4
-; SSE41-NEXT:    pxor %xmm2, %xmm0
-; SSE41-NEXT:    movdqa %xmm0, %xmm6
-; SSE41-NEXT:    pcmpgtd %xmm4, %xmm6
-; SSE41-NEXT:    pshufd {{.*#+}} xmm7 = xmm6[0,0,2,2]
-; SSE41-NEXT:    pcmpeqd %xmm4, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm0[1,1,3,3]
-; SSE41-NEXT:    pand %xmm7, %xmm4
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm6[1,1,3,3]
-; SSE41-NEXT:    por %xmm4, %xmm0
-; SSE41-NEXT:    blendvpd %xmm5, %xmm2
-; SSE41-NEXT:    movdqa %xmm3, %xmm0
-; SSE41-NEXT:    blendvpd %xmm8, %xmm1
-; SSE41-NEXT:    movapd %xmm2, %xmm0
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: min_lt_v4i64c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movdqa {{.*#+}} xmm4 = [18446744073709551609,18446744073709551615]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm5 = [1,7]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551615,18446744073709551609]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm1 = [7,1]
-; SSE42-NEXT:    movdqa %xmm1, %xmm3
-; SSE42-NEXT:    pcmpgtq %xmm5, %xmm3
-; SSE42-NEXT:    movdqa %xmm2, %xmm0
-; SSE42-NEXT:    pcmpgtq %xmm4, %xmm0
-; SSE42-NEXT:    blendvpd %xmm4, %xmm2
-; SSE42-NEXT:    movdqa %xmm3, %xmm0
-; SSE42-NEXT:    blendvpd %xmm5, %xmm1
-; SSE42-NEXT:    movapd %xmm2, %xmm0
-; SSE42-NEXT:    retq
-;
-; AVX1-LABEL: min_lt_v4i64c:
-; AVX1:       # BB#0:
-; AVX1-NEXT:    vmovapd {{.*#+}} ymm0 = [18446744073709551615,18446744073709551609,7,1]
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709551615,18446744073709551609]
-; AVX1-NEXT:    vpcmpgtq {{.*}}(%rip), %xmm1, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [7,1]
-; AVX1-NEXT:    vpcmpgtq {{.*}}(%rip), %xmm2, %xmm2
-; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm1, %ymm1
-; AVX1-NEXT:    vblendvpd %ymm1, {{.*}}(%rip), %ymm0, %ymm0
-; AVX1-NEXT:    retq
-;
-; AVX2-LABEL: min_lt_v4i64c:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm0 = [18446744073709551609,18446744073709551615,1,7]
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = [18446744073709551615,18446744073709551609,7,1]
-; AVX2-NEXT:    vpcmpgtq %ymm0, %ymm1, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
-; AVX2-NEXT:    retq
+; SSE-LABEL: min_lt_v4i64c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [18446744073709551609,18446744073709551609]
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1,1]
+; SSE-NEXT:    retq
 ;
-; AVX512-LABEL: min_lt_v4i64c:
-; AVX512:       # BB#0:
-; AVX512-NEXT:    vmovdqa {{.*#+}} ymm0 = [18446744073709551609,18446744073709551615,1,7]
-; AVX512-NEXT:    vmovdqa {{.*#+}} ymm1 = [18446744073709551615,18446744073709551609,7,1]
-; AVX512-NEXT:    vpcmpgtq %ymm0, %ymm1, %ymm2
-; AVX512-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
-; AVX512-NEXT:    retq
+; AVX-LABEL: min_lt_v4i64c:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovaps {{.*#+}} ymm0 = [18446744073709551609,18446744073709551609,1,1]
+; AVX-NEXT:    retq
   %1 = insertelement <4 x i64> <i64 -7, i64 -1, i64 1, i64 7>, i64 -7, i32 0
   %2 = insertelement <4 x i64> <i64 -1, i64 -7, i64 7, i64 1>, i64 -1, i32 0
   %3 = icmp slt <4 x i64> %1, %2
@@ -2496,26 +1880,10 @@ define <4 x i64> @min_lt_v4i64c() {
 }
 
 define <4 x i32> @min_lt_v4i32c() {
-; SSE2-LABEL: min_lt_v4i32c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [4294967289,4294967295,1,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [4294967295,4294967289,7,1]
-; SSE2-NEXT:    movdqa %xmm2, %xmm0
-; SSE2-NEXT:    pcmpgtd %xmm1, %xmm0
-; SSE2-NEXT:    pand %xmm0, %xmm1
-; SSE2-NEXT:    pandn %xmm2, %xmm0
-; SSE2-NEXT:    por %xmm1, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: min_lt_v4i32c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [4294967289,4294967289,1,1]
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: min_lt_v4i32c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movaps {{.*#+}} xmm0 = [4294967289,4294967289,1,1]
-; SSE42-NEXT:    retq
+; SSE-LABEL: min_lt_v4i32c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [4294967289,4294967289,1,1]
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: min_lt_v4i32c:
 ; AVX:       # BB#0:
@@ -2529,35 +1897,11 @@ define <4 x i32> @min_lt_v4i32c() {
 }
 
 define <8 x i32> @min_lt_v8i32c() {
-; SSE2-LABEL: min_lt_v8i32c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [4294967289,4294967291,4294967293,4294967295]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [1,3,5,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [4294967295,4294967293,4294967291,4294967289]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm5 = [7,5,3,1]
-; SSE2-NEXT:    movdqa %xmm5, %xmm1
-; SSE2-NEXT:    pcmpgtd %xmm3, %xmm1
-; SSE2-NEXT:    movdqa %xmm4, %xmm0
-; SSE2-NEXT:    pcmpgtd %xmm2, %xmm0
-; SSE2-NEXT:    pand %xmm0, %xmm2
-; SSE2-NEXT:    pandn %xmm4, %xmm0
-; SSE2-NEXT:    por %xmm2, %xmm0
-; SSE2-NEXT:    pand %xmm1, %xmm3
-; SSE2-NEXT:    pandn %xmm5, %xmm1
-; SSE2-NEXT:    por %xmm3, %xmm1
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: min_lt_v8i32c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [4294967289,4294967291,4294967291,4294967289]
-; SSE41-NEXT:    movaps {{.*#+}} xmm1 = [1,3,3,1]
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: min_lt_v8i32c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movaps {{.*#+}} xmm0 = [4294967289,4294967291,4294967291,4294967289]
-; SSE42-NEXT:    movaps {{.*#+}} xmm1 = [1,3,3,1]
-; SSE42-NEXT:    retq
+; SSE-LABEL: min_lt_v8i32c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [4294967289,4294967291,4294967291,4294967289]
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1,3,3,1]
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: min_lt_v8i32c:
 ; AVX:       # BB#0:
@@ -2606,26 +1950,10 @@ define <16 x i16> @min_lt_v16i16c() {
 }
 
 define <16 x i8> @min_lt_v16i8c() {
-; SSE2-LABEL: min_lt_v16i8c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [249,250,251,252,253,254,255,0,1,2,3,4,5,6,7,8]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,254,253,252,251,250,249,0,7,6,5,4,3,2,1,0]
-; SSE2-NEXT:    movdqa %xmm2, %xmm0
-; SSE2-NEXT:    pcmpgtb %xmm1, %xmm0
-; SSE2-NEXT:    pand %xmm0, %xmm1
-; SSE2-NEXT:    pandn %xmm2, %xmm0
-; SSE2-NEXT:    por %xmm1, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: min_lt_v16i8c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [249,250,251,252,251,250,249,0,1,2,3,4,3,2,1,0]
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: min_lt_v16i8c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movaps {{.*#+}} xmm0 = [249,250,251,252,251,250,249,0,1,2,3,4,3,2,1,0]
-; SSE42-NEXT:    retq
+; SSE-LABEL: min_lt_v16i8c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [249,250,251,252,251,250,249,0,1,2,3,4,3,2,1,0]
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: min_lt_v16i8c:
 ; AVX:       # BB#0:
@@ -2639,71 +1967,14 @@ define <16 x i8> @min_lt_v16i8c() {
 }
 
 define <2 x i64> @min_le_v2i64c() {
-; SSE2-LABEL: min_le_v2i64c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [18446744073709551609,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551615,1]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [2147483648,0,2147483648,0]
-; SSE2-NEXT:    movdqa %xmm0, %xmm3
-; SSE2-NEXT:    pxor %xmm2, %xmm3
-; SSE2-NEXT:    pxor %xmm1, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, %xmm4
-; SSE2-NEXT:    pcmpgtd %xmm3, %xmm4
-; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm3, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
-; SSE2-NEXT:    pand %xmm5, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
-; SSE2-NEXT:    por %xmm0, %xmm3
-; SSE2-NEXT:    pcmpeqd %xmm0, %xmm0
-; SSE2-NEXT:    pxor %xmm3, %xmm0
-; SSE2-NEXT:    pandn %xmm1, %xmm3
-; SSE2-NEXT:    pandn %xmm2, %xmm0
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: min_le_v2i64c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551609,7]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [18446744073709551615,1]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [2147483648,0,2147483648,0]
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    pxor %xmm1, %xmm3
-; SSE41-NEXT:    pxor %xmm2, %xmm0
-; SSE41-NEXT:    movdqa %xmm0, %xmm4
-; SSE41-NEXT:    pcmpgtd %xmm3, %xmm4
-; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
-; SSE41-NEXT:    pcmpeqd %xmm3, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
-; SSE41-NEXT:    pand %xmm5, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
-; SSE41-NEXT:    por %xmm0, %xmm3
-; SSE41-NEXT:    pcmpeqd %xmm0, %xmm0
-; SSE41-NEXT:    pxor %xmm3, %xmm0
-; SSE41-NEXT:    blendvpd %xmm2, %xmm1
-; SSE41-NEXT:    movapd %xmm1, %xmm0
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: min_le_v2i64c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551609,7]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm1 = [18446744073709551615,1]
-; SSE42-NEXT:    movdqa %xmm2, %xmm3
-; SSE42-NEXT:    pcmpgtq %xmm1, %xmm3
-; SSE42-NEXT:    pcmpeqd %xmm0, %xmm0
-; SSE42-NEXT:    pxor %xmm3, %xmm0
-; SSE42-NEXT:    blendvpd %xmm2, %xmm1
-; SSE42-NEXT:    movapd %xmm1, %xmm0
-; SSE42-NEXT:    retq
+; SSE-LABEL: min_le_v2i64c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [18446744073709551609,1]
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: min_le_v2i64c:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vmovdqa {{.*#+}} xmm0 = [18446744073709551609,7]
-; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709551615,1]
-; AVX-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm2
-; AVX-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
-; AVX-NEXT:    vpxor %xmm3, %xmm2, %xmm2
-; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [18446744073709551609,1]
 ; AVX-NEXT:    retq
   %1 = insertelement <2 x i64> <i64 -7, i64 7>, i64 -7, i32 0
   %2 = insertelement <2 x i64> <i64 -1, i64 1>, i64 -1, i32 0
@@ -2713,139 +1984,16 @@ define <2 x i64> @min_le_v2i64c() {
 }
 
 define <4 x i64> @min_le_v4i64c() {
-; SSE2-LABEL: min_le_v4i64c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm10 = [18446744073709551609,18446744073709551615]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm8 = [1,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm5 = [18446744073709551615,18446744073709551609]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm9 = [7,1]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm7 = [2147483648,0,2147483648,0]
-; SSE2-NEXT:    movdqa %xmm7, %xmm0
-; SSE2-NEXT:    pxor %xmm9, %xmm0
-; SSE2-NEXT:    movdqa %xmm7, %xmm1
-; SSE2-NEXT:    pxor %xmm8, %xmm1
-; SSE2-NEXT:    movdqa %xmm1, %xmm6
-; SSE2-NEXT:    pcmpgtd %xmm0, %xmm6
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm6[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
-; SSE2-NEXT:    pand %xmm2, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[1,1,3,3]
-; SSE2-NEXT:    por %xmm0, %xmm6
-; SSE2-NEXT:    pcmpeqd %xmm0, %xmm0
-; SSE2-NEXT:    movdqa %xmm6, %xmm1
-; SSE2-NEXT:    pxor %xmm0, %xmm1
-; SSE2-NEXT:    movdqa %xmm7, %xmm2
-; SSE2-NEXT:    pxor %xmm5, %xmm2
-; SSE2-NEXT:    pxor %xmm10, %xmm7
-; SSE2-NEXT:    movdqa %xmm7, %xmm3
-; SSE2-NEXT:    pcmpgtd %xmm2, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm2, %xmm7
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm7[1,1,3,3]
-; SSE2-NEXT:    pand %xmm4, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
-; SSE2-NEXT:    por %xmm2, %xmm3
-; SSE2-NEXT:    pxor %xmm3, %xmm0
-; SSE2-NEXT:    pandn %xmm10, %xmm3
-; SSE2-NEXT:    pandn %xmm5, %xmm0
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    pandn %xmm8, %xmm6
-; SSE2-NEXT:    pandn %xmm9, %xmm1
-; SSE2-NEXT:    por %xmm6, %xmm1
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: min_le_v4i64c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movdqa {{.*#+}} xmm9 = [18446744073709551609,18446744073709551615]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm8 = [1,7]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551615,18446744073709551609]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [7,1]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [2147483648,0,2147483648,0]
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    pxor %xmm1, %xmm3
-; SSE41-NEXT:    movdqa %xmm0, %xmm6
-; SSE41-NEXT:    pxor %xmm8, %xmm6
-; SSE41-NEXT:    movdqa %xmm6, %xmm7
-; SSE41-NEXT:    pcmpgtd %xmm3, %xmm7
-; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm7[0,0,2,2]
-; SSE41-NEXT:    pcmpeqd %xmm3, %xmm6
-; SSE41-NEXT:    pshufd {{.*#+}} xmm6 = xmm6[1,1,3,3]
-; SSE41-NEXT:    pand %xmm4, %xmm6
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm7[1,1,3,3]
-; SSE41-NEXT:    por %xmm6, %xmm3
-; SSE41-NEXT:    pcmpeqd %xmm4, %xmm4
-; SSE41-NEXT:    pxor %xmm4, %xmm3
-; SSE41-NEXT:    movdqa %xmm0, %xmm6
-; SSE41-NEXT:    pxor %xmm2, %xmm6
-; SSE41-NEXT:    pxor %xmm9, %xmm0
-; SSE41-NEXT:    movdqa %xmm0, %xmm7
-; SSE41-NEXT:    pcmpgtd %xmm6, %xmm7
-; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm7[0,0,2,2]
-; SSE41-NEXT:    pcmpeqd %xmm6, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm6 = xmm0[1,1,3,3]
-; SSE41-NEXT:    pand %xmm5, %xmm6
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm7[1,1,3,3]
-; SSE41-NEXT:    por %xmm6, %xmm0
-; SSE41-NEXT:    pxor %xmm4, %xmm0
-; SSE41-NEXT:    blendvpd %xmm9, %xmm2
-; SSE41-NEXT:    movdqa %xmm3, %xmm0
-; SSE41-NEXT:    blendvpd %xmm8, %xmm1
-; SSE41-NEXT:    movapd %xmm2, %xmm0
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: min_le_v4i64c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movdqa {{.*#+}} xmm4 = [18446744073709551609,18446744073709551615]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm5 = [1,7]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm2 = [18446744073709551615,18446744073709551609]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm1 = [7,1]
-; SSE42-NEXT:    movdqa %xmm5, %xmm3
-; SSE42-NEXT:    pcmpgtq %xmm1, %xmm3
-; SSE42-NEXT:    pcmpeqd %xmm6, %xmm6
-; SSE42-NEXT:    pxor %xmm6, %xmm3
-; SSE42-NEXT:    movdqa %xmm4, %xmm0
-; SSE42-NEXT:    pcmpgtq %xmm2, %xmm0
-; SSE42-NEXT:    pxor %xmm6, %xmm0
-; SSE42-NEXT:    blendvpd %xmm4, %xmm2
-; SSE42-NEXT:    movdqa %xmm3, %xmm0
-; SSE42-NEXT:    blendvpd %xmm5, %xmm1
-; SSE42-NEXT:    movapd %xmm2, %xmm0
-; SSE42-NEXT:    retq
-;
-; AVX1-LABEL: min_le_v4i64c:
-; AVX1:       # BB#0:
-; AVX1-NEXT:    vmovapd {{.*#+}} ymm0 = [18446744073709551615,18446744073709551609,7,1]
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [1,7]
-; AVX1-NEXT:    vpcmpgtq {{.*}}(%rip), %xmm1, %xmm1
-; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpxor %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [18446744073709551609,18446744073709551615]
-; AVX1-NEXT:    vpcmpgtq {{.*}}(%rip), %xmm3, %xmm3
-; AVX1-NEXT:    vpxor %xmm2, %xmm3, %xmm2
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm2, %ymm1
-; AVX1-NEXT:    vblendvpd %ymm1, {{.*}}(%rip), %ymm0, %ymm0
-; AVX1-NEXT:    retq
-;
-; AVX2-LABEL: min_le_v4i64c:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm0 = [18446744073709551609,18446744073709551615,1,7]
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = [18446744073709551615,18446744073709551609,7,1]
-; AVX2-NEXT:    vpcmpgtq %ymm1, %ymm0, %ymm2
-; AVX2-NEXT:    vpcmpeqd %ymm3, %ymm3, %ymm3
-; AVX2-NEXT:    vpxor %ymm3, %ymm2, %ymm2
-; AVX2-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
-; AVX2-NEXT:    retq
+; SSE-LABEL: min_le_v4i64c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [18446744073709551609,18446744073709551609]
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1,1]
+; SSE-NEXT:    retq
 ;
-; AVX512-LABEL: min_le_v4i64c:
-; AVX512:       # BB#0:
-; AVX512-NEXT:    vmovdqa {{.*#+}} ymm0 = [18446744073709551609,18446744073709551615,1,7]
-; AVX512-NEXT:    vmovdqa {{.*#+}} ymm1 = [18446744073709551615,18446744073709551609,7,1]
-; AVX512-NEXT:    vpcmpgtq %ymm1, %ymm0, %ymm2
-; AVX512-NEXT:    vpcmpeqd %ymm3, %ymm3, %ymm3
-; AVX512-NEXT:    vpxor %ymm3, %ymm2, %ymm2
-; AVX512-NEXT:    vblendvpd %ymm2, %ymm0, %ymm1, %ymm0
-; AVX512-NEXT:    retq
+; AVX-LABEL: min_le_v4i64c:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovaps {{.*#+}} ymm0 = [18446744073709551609,18446744073709551609,1,1]
+; AVX-NEXT:    retq
   %1 = insertelement <4 x i64> <i64 -7, i64 -1, i64 1, i64 7>, i64 -7, i32 0
   %2 = insertelement <4 x i64> <i64 -1, i64 -7, i64 7, i64 1>, i64 -1, i32 0
   %3 = icmp sle <4 x i64> %1, %2
@@ -2854,28 +2002,10 @@ define <4 x i64> @min_le_v4i64c() {
 }
 
 define <4 x i32> @min_le_v4i32c() {
-; SSE2-LABEL: min_le_v4i32c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [4294967289,4294967295,1,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [4294967295,4294967289,7,1]
-; SSE2-NEXT:    movdqa %xmm1, %xmm3
-; SSE2-NEXT:    pcmpgtd %xmm2, %xmm3
-; SSE2-NEXT:    pcmpeqd %xmm0, %xmm0
-; SSE2-NEXT:    pxor %xmm3, %xmm0
-; SSE2-NEXT:    pandn %xmm1, %xmm3
-; SSE2-NEXT:    pandn %xmm2, %xmm0
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: min_le_v4i32c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [4294967289,4294967289,1,1]
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: min_le_v4i32c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movaps {{.*#+}} xmm0 = [4294967289,4294967289,1,1]
-; SSE42-NEXT:    retq
+; SSE-LABEL: min_le_v4i32c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [4294967289,4294967289,1,1]
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: min_le_v4i32c:
 ; AVX:       # BB#0:
@@ -2889,39 +2019,11 @@ define <4 x i32> @min_le_v4i32c() {
 }
 
 define <8 x i32> @min_le_v8i32c() {
-; SSE2-LABEL: min_le_v8i32c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [4294967289,4294967291,4294967293,4294967295]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [1,3,5,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [4294967295,4294967293,4294967291,4294967289]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm5 = [7,5,3,1]
-; SSE2-NEXT:    movdqa %xmm3, %xmm6
-; SSE2-NEXT:    pcmpgtd %xmm5, %xmm6
-; SSE2-NEXT:    pcmpeqd %xmm0, %xmm0
-; SSE2-NEXT:    movdqa %xmm6, %xmm1
-; SSE2-NEXT:    pxor %xmm0, %xmm1
-; SSE2-NEXT:    movdqa %xmm2, %xmm7
-; SSE2-NEXT:    pcmpgtd %xmm4, %xmm7
-; SSE2-NEXT:    pxor %xmm7, %xmm0
-; SSE2-NEXT:    pandn %xmm2, %xmm7
-; SSE2-NEXT:    pandn %xmm4, %xmm0
-; SSE2-NEXT:    por %xmm7, %xmm0
-; SSE2-NEXT:    pandn %xmm3, %xmm6
-; SSE2-NEXT:    pandn %xmm5, %xmm1
-; SSE2-NEXT:    por %xmm6, %xmm1
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: min_le_v8i32c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [4294967289,4294967291,4294967291,4294967289]
-; SSE41-NEXT:    movaps {{.*#+}} xmm1 = [1,3,3,1]
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: min_le_v8i32c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movaps {{.*#+}} xmm0 = [4294967289,4294967291,4294967291,4294967289]
-; SSE42-NEXT:    movaps {{.*#+}} xmm1 = [1,3,3,1]
-; SSE42-NEXT:    retq
+; SSE-LABEL: min_le_v8i32c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [4294967289,4294967291,4294967291,4294967289]
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1,3,3,1]
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: min_le_v8i32c:
 ; AVX:       # BB#0:
@@ -2970,28 +2072,10 @@ define <16 x i16> @min_le_v16i16c() {
 }
 
 define <16 x i8> @min_le_v16i8c() {
-; SSE2-LABEL: min_le_v16i8c:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [249,250,251,252,253,254,255,0,1,2,3,4,5,6,7,8]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,254,253,252,251,250,249,0,7,6,5,4,3,2,1,0]
-; SSE2-NEXT:    movdqa %xmm1, %xmm3
-; SSE2-NEXT:    pcmpgtb %xmm2, %xmm3
-; SSE2-NEXT:    pcmpeqd %xmm0, %xmm0
-; SSE2-NEXT:    pxor %xmm3, %xmm0
-; SSE2-NEXT:    pandn %xmm1, %xmm3
-; SSE2-NEXT:    pandn %xmm2, %xmm0
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: min_le_v16i8c:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [249,250,251,252,251,250,249,0,1,2,3,4,3,2,1,0]
-; SSE41-NEXT:    retq
-;
-; SSE42-LABEL: min_le_v16i8c:
-; SSE42:       # BB#0:
-; SSE42-NEXT:    movaps {{.*#+}} xmm0 = [249,250,251,252,251,250,249,0,1,2,3,4,3,2,1,0]
-; SSE42-NEXT:    retq
+; SSE-LABEL: min_le_v16i8c:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [249,250,251,252,251,250,249,0,1,2,3,4,3,2,1,0]
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: min_le_v16i8c:
 ; AVX:       # BB#0: