[X86] Add v2i64->v2i32/v2i16/v2i8 test cases to the trunc packus/ssat/usat tests...

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@374704 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/test/CodeGen/X86/vector-trunc-packus.ll b/test/CodeGen/X86/vector-trunc-packus.ll
index 8867aaaeddd6ffe788e752140a9d9c5944459afc..2f433d8b8f7422314e40e79972a8f6d295240d44 100644 (file)
--- a/test/CodeGen/X86/vector-trunc-packus.ll
+++ b/test/CodeGen/X86/vector-trunc-packus.ll
@@ -15,6 +15,318 @@
 ; PACKUS saturation truncation to vXi32
 ;
 
+define <2 x i32> @trunc_packus_v2i64_v2i32(<2 x i64> %a0) {
+; SSE2-LABEL: trunc_packus_v2i64_v2i32:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pxor %xmm1, %xmm2
+; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [2147483647,2147483647]
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm2, %xmm3
+; SSE2-NEXT:    pand %xmm3, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSE2-NEXT:    por %xmm0, %xmm3
+; SSE2-NEXT:    movdqa %xmm3, %xmm0
+; SSE2-NEXT:    pxor %xmm1, %xmm0
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSE2-NEXT:    por %xmm0, %xmm1
+; SSE2-NEXT:    pand %xmm3, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_packus_v2i64_v2i32:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pxor %xmm1, %xmm2
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [2147483647,2147483647]
+; SSSE3-NEXT:    movdqa %xmm3, %xmm4
+; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm5, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSSE3-NEXT:    por %xmm2, %xmm3
+; SSSE3-NEXT:    pand %xmm3, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSSE3-NEXT:    por %xmm0, %xmm3
+; SSSE3-NEXT:    movdqa %xmm3, %xmm0
+; SSSE3-NEXT:    pxor %xmm1, %xmm0
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    por %xmm0, %xmm1
+; SSSE3-NEXT:    pand %xmm3, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_packus_v2i64_v2i32:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [4294967295,4294967295]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648]
+; SSE41-NEXT:    pxor %xmm3, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [2147483647,2147483647]
+; SSE41-NEXT:    movdqa %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    xorpd %xmm1, %xmm1
+; SSE41-NEXT:    movapd %xmm2, %xmm4
+; SSE41-NEXT:    xorpd %xmm3, %xmm4
+; SSE41-NEXT:    movapd %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm3, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm3, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm2, %xmm1
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: trunc_packus_v2i64_v2i32:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [4294967295,4294967295]
+; AVX-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm1
+; AVX-NEXT:    vpand %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_packus_v2i64_v2i32:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [4294967295,4294967295]
+; AVX512F-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512F-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512VL-LABEL: trunc_packus_v2i64_v2i32:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512VL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512VL-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; AVX512VL-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_packus_v2i64_v2i32:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [4294967295,4294967295]
+; AVX512BW-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512BW-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_packus_v2i64_v2i32:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512BWVL-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_packus_v2i64_v2i32:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; SKX-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; SKX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SKX-NEXT:    retq
+  %1 = icmp slt <2 x i64> %a0, <i64 4294967295, i64 4294967295>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 4294967295, i64 4294967295>
+  %3 = icmp sgt <2 x i64> %2, zeroinitializer
+  %4 = select <2 x i1> %3, <2 x i64> %2, <2 x i64> zeroinitializer
+  %5 = trunc <2 x i64> %4 to <2 x i32>
+  ret <2 x i32> %5
+}
+
+define void @trunc_packus_v2i64_v2i32_store(<2 x i64> %a0, <2 x i32>* %p1) {
+; SSE2-LABEL: trunc_packus_v2i64_v2i32_store:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pxor %xmm1, %xmm2
+; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [2147483647,2147483647]
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm2, %xmm3
+; SSE2-NEXT:    pand %xmm3, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSE2-NEXT:    por %xmm0, %xmm3
+; SSE2-NEXT:    movdqa %xmm3, %xmm0
+; SSE2-NEXT:    pxor %xmm1, %xmm0
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSE2-NEXT:    por %xmm0, %xmm1
+; SSE2-NEXT:    pand %xmm3, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSE2-NEXT:    movq %xmm0, (%rdi)
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_packus_v2i64_v2i32_store:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pxor %xmm1, %xmm2
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [2147483647,2147483647]
+; SSSE3-NEXT:    movdqa %xmm3, %xmm4
+; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm5, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSSE3-NEXT:    por %xmm2, %xmm3
+; SSSE3-NEXT:    pand %xmm3, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSSE3-NEXT:    por %xmm0, %xmm3
+; SSSE3-NEXT:    movdqa %xmm3, %xmm0
+; SSSE3-NEXT:    pxor %xmm1, %xmm0
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    por %xmm0, %xmm1
+; SSSE3-NEXT:    pand %xmm3, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSSE3-NEXT:    movq %xmm0, (%rdi)
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_packus_v2i64_v2i32_store:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [4294967295,4294967295]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648]
+; SSE41-NEXT:    pxor %xmm3, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [2147483647,2147483647]
+; SSE41-NEXT:    movdqa %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    xorpd %xmm1, %xmm1
+; SSE41-NEXT:    movapd %xmm2, %xmm4
+; SSE41-NEXT:    xorpd %xmm3, %xmm4
+; SSE41-NEXT:    movapd %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm3, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm3, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm2, %xmm1
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSE41-NEXT:    movq %xmm0, (%rdi)
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: trunc_packus_v2i64_v2i32_store:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [4294967295,4294967295]
+; AVX-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm1
+; AVX-NEXT:    vpand %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX-NEXT:    vmovq %xmm0, (%rdi)
+; AVX-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_packus_v2i64_v2i32_store:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [4294967295,4294967295]
+; AVX512F-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512F-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512F-NEXT:    vmovq %xmm0, (%rdi)
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512VL-LABEL: trunc_packus_v2i64_v2i32_store:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512VL-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; AVX512VL-NEXT:    vpmovusqd %xmm0, (%rdi)
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_packus_v2i64_v2i32_store:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [4294967295,4294967295]
+; AVX512BW-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512BW-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512BW-NEXT:    vmovq %xmm0, (%rdi)
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_packus_v2i64_v2i32_store:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512BWVL-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpmovusqd %xmm0, (%rdi)
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_packus_v2i64_v2i32_store:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; SKX-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; SKX-NEXT:    vpmovusqd %xmm0, (%rdi)
+; SKX-NEXT:    retq
+  %1 = icmp slt <2 x i64> %a0, <i64 4294967295, i64 4294967295>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 4294967295, i64 4294967295>
+  %3 = icmp sgt <2 x i64> %2, zeroinitializer
+  %4 = select <2 x i1> %3, <2 x i64> %2, <2 x i64> zeroinitializer
+  %5 = trunc <2 x i64> %4 to <2 x i32>
+  store <2 x i32> %5, <2 x i32>* %p1
+  ret void
+}
+
 define <4 x i32> @trunc_packus_v4i64_v4i32(<4 x i64> %a0) {
 ; SSE2-LABEL: trunc_packus_v4i64_v4i32:
 ; SSE2:       # %bb.0:
@@ -660,35 +972,405 @@ define <8 x i32> @trunc_packus_v8i64_v8i32(<8 x i64>* %p0) "min-legal-vector-wid
 ; AVX2-FAST-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
 ; AVX2-FAST-NEXT:    retq
 ;
-; AVX512-LABEL: trunc_packus_v8i64_v8i32:
-; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpxor %xmm0, %xmm0, %xmm0
-; AVX512-NEXT:    vpmaxsq (%rdi), %zmm0, %zmm0
-; AVX512-NEXT:    vpmovusqd %zmm0, %ymm0
-; AVX512-NEXT:    retq
+; AVX512-LABEL: trunc_packus_v8i64_v8i32:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpxor %xmm0, %xmm0, %xmm0
+; AVX512-NEXT:    vpmaxsq (%rdi), %zmm0, %zmm0
+; AVX512-NEXT:    vpmovusqd %zmm0, %ymm0
+; AVX512-NEXT:    retq
+;
+; SKX-LABEL: trunc_packus_v8i64_v8i32:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpxor %xmm0, %xmm0, %xmm0
+; SKX-NEXT:    vpmaxsq (%rdi), %ymm0, %ymm1
+; SKX-NEXT:    vpmovusqd %ymm1, %xmm1
+; SKX-NEXT:    vpmaxsq 32(%rdi), %ymm0, %ymm0
+; SKX-NEXT:    vpmovusqd %ymm0, %xmm0
+; SKX-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
+; SKX-NEXT:    retq
+  %a0 = load <8 x i64>, <8 x i64>* %p0
+  %1 = icmp slt <8 x i64> %a0, <i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295>
+  %2 = select <8 x i1> %1, <8 x i64> %a0, <8 x i64> <i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295>
+  %3 = icmp sgt <8 x i64> %2, zeroinitializer
+  %4 = select <8 x i1> %3, <8 x i64> %2, <8 x i64> zeroinitializer
+  %5 = trunc <8 x i64> %4 to <8 x i32>
+  ret <8 x i32> %5
+}
+
+;
+; PACKUS saturation truncation to vXi16
+;
+
+define <2 x i16> @trunc_packus_v2i64_v2i16(<2 x i64> %a0) {
+; SSE2-LABEL: trunc_packus_v2i64_v2i16:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pxor %xmm1, %xmm2
+; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [2147549183,2147549183]
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm2, %xmm3
+; SSE2-NEXT:    pand %xmm3, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSE2-NEXT:    por %xmm0, %xmm3
+; SSE2-NEXT:    movdqa %xmm3, %xmm0
+; SSE2-NEXT:    pxor %xmm1, %xmm0
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSE2-NEXT:    por %xmm0, %xmm1
+; SSE2-NEXT:    pand %xmm3, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_packus_v2i64_v2i16:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pxor %xmm1, %xmm2
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [2147549183,2147549183]
+; SSSE3-NEXT:    movdqa %xmm3, %xmm4
+; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm5, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSSE3-NEXT:    por %xmm2, %xmm3
+; SSSE3-NEXT:    pand %xmm3, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSSE3-NEXT:    por %xmm0, %xmm3
+; SSSE3-NEXT:    movdqa %xmm3, %xmm0
+; SSSE3-NEXT:    pxor %xmm1, %xmm0
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    por %xmm0, %xmm1
+; SSSE3-NEXT:    pand %xmm3, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSSE3-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_packus_v2i64_v2i16:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [65535,65535]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648]
+; SSE41-NEXT:    pxor %xmm3, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [2147549183,2147549183]
+; SSE41-NEXT:    movdqa %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    xorpd %xmm1, %xmm1
+; SSE41-NEXT:    movapd %xmm2, %xmm4
+; SSE41-NEXT:    xorpd %xmm3, %xmm4
+; SSE41-NEXT:    movapd %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm3, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm3, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm2, %xmm1
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSE41-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: trunc_packus_v2i64_v2i16:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [65535,65535]
+; AVX1-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX1-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm1
+; AVX1-NEXT:    vpand %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-SLOW-LABEL: trunc_packus_v2i64_v2i16:
+; AVX2-SLOW:       # %bb.0:
+; AVX2-SLOW-NEXT:    vmovdqa {{.*#+}} xmm1 = [65535,65535]
+; AVX2-SLOW-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX2-SLOW-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX2-SLOW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX2-SLOW-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm1
+; AVX2-SLOW-NEXT:    vpand %xmm0, %xmm1, %xmm0
+; AVX2-SLOW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX2-SLOW-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX2-SLOW-NEXT:    retq
+;
+; AVX2-FAST-LABEL: trunc_packus_v2i64_v2i16:
+; AVX2-FAST:       # %bb.0:
+; AVX2-FAST-NEXT:    vmovdqa {{.*#+}} xmm1 = [65535,65535]
+; AVX2-FAST-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX2-FAST-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX2-FAST-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX2-FAST-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm1
+; AVX2-FAST-NEXT:    vpand %xmm0, %xmm1, %xmm0
+; AVX2-FAST-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX2-FAST-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_packus_v2i64_v2i16:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [65535,65535]
+; AVX512F-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512F-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512F-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512VL-LABEL: trunc_packus_v2i64_v2i16:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512VL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512VL-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_packus_v2i64_v2i16:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [65535,65535]
+; AVX512BW-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512BW-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_packus_v2i64_v2i16:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512BWVL-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_packus_v2i64_v2i16:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; SKX-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; SKX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; SKX-NEXT:    retq
+  %1 = icmp slt <2 x i64> %a0, <i64 65535, i64 65535>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 65535, i64 65535>
+  %3 = icmp sgt <2 x i64> %2, zeroinitializer
+  %4 = select <2 x i1> %3, <2 x i64> %2, <2 x i64> zeroinitializer
+  %5 = trunc <2 x i64> %4 to <2 x i16>
+  ret <2 x i16> %5
+}
+
+define void @trunc_packus_v2i64_v2i16_store(<2 x i64> %a0, <2 x i16> *%p1) {
+; SSE2-LABEL: trunc_packus_v2i64_v2i16_store:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pxor %xmm1, %xmm2
+; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [2147549183,2147549183]
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm2, %xmm3
+; SSE2-NEXT:    pand %xmm3, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSE2-NEXT:    por %xmm0, %xmm3
+; SSE2-NEXT:    movdqa %xmm3, %xmm0
+; SSE2-NEXT:    pxor %xmm1, %xmm0
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSE2-NEXT:    por %xmm0, %xmm1
+; SSE2-NEXT:    pand %xmm3, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSE2-NEXT:    movd %xmm0, (%rdi)
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_packus_v2i64_v2i16_store:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pxor %xmm1, %xmm2
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [2147549183,2147549183]
+; SSSE3-NEXT:    movdqa %xmm3, %xmm4
+; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm5, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSSE3-NEXT:    por %xmm2, %xmm3
+; SSSE3-NEXT:    pand %xmm3, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSSE3-NEXT:    por %xmm0, %xmm3
+; SSSE3-NEXT:    movdqa %xmm3, %xmm0
+; SSSE3-NEXT:    pxor %xmm1, %xmm0
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    por %xmm0, %xmm1
+; SSSE3-NEXT:    pand %xmm3, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSSE3-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSSE3-NEXT:    movd %xmm0, (%rdi)
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_packus_v2i64_v2i16_store:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [65535,65535]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648]
+; SSE41-NEXT:    pxor %xmm3, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [2147549183,2147549183]
+; SSE41-NEXT:    movdqa %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    xorpd %xmm1, %xmm1
+; SSE41-NEXT:    movapd %xmm2, %xmm4
+; SSE41-NEXT:    xorpd %xmm3, %xmm4
+; SSE41-NEXT:    movapd %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm3, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm3, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm2, %xmm1
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSE41-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSE41-NEXT:    movd %xmm0, (%rdi)
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: trunc_packus_v2i64_v2i16_store:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [65535,65535]
+; AVX1-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX1-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm1
+; AVX1-NEXT:    vpand %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX1-NEXT:    vmovd %xmm0, (%rdi)
+; AVX1-NEXT:    retq
+;
+; AVX2-SLOW-LABEL: trunc_packus_v2i64_v2i16_store:
+; AVX2-SLOW:       # %bb.0:
+; AVX2-SLOW-NEXT:    vmovdqa {{.*#+}} xmm1 = [65535,65535]
+; AVX2-SLOW-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX2-SLOW-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX2-SLOW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX2-SLOW-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm1
+; AVX2-SLOW-NEXT:    vpand %xmm0, %xmm1, %xmm0
+; AVX2-SLOW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX2-SLOW-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX2-SLOW-NEXT:    vmovd %xmm0, (%rdi)
+; AVX2-SLOW-NEXT:    retq
+;
+; AVX2-FAST-LABEL: trunc_packus_v2i64_v2i16_store:
+; AVX2-FAST:       # %bb.0:
+; AVX2-FAST-NEXT:    vmovdqa {{.*#+}} xmm1 = [65535,65535]
+; AVX2-FAST-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX2-FAST-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX2-FAST-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX2-FAST-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm1
+; AVX2-FAST-NEXT:    vpand %xmm0, %xmm1, %xmm0
+; AVX2-FAST-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX2-FAST-NEXT:    vmovd %xmm0, (%rdi)
+; AVX2-FAST-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_packus_v2i64_v2i16_store:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [65535,65535]
+; AVX512F-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512F-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512F-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512F-NEXT:    vmovd %xmm0, (%rdi)
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
 ;
-; SKX-LABEL: trunc_packus_v8i64_v8i32:
+; AVX512VL-LABEL: trunc_packus_v2i64_v2i16_store:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512VL-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; AVX512VL-NEXT:    vpmovusqw %xmm0, (%rdi)
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_packus_v2i64_v2i16_store:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [65535,65535]
+; AVX512BW-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512BW-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX512BW-NEXT:    vmovd %xmm0, (%rdi)
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_packus_v2i64_v2i16_store:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512BWVL-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpmovusqw %xmm0, (%rdi)
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_packus_v2i64_v2i16_store:
 ; SKX:       # %bb.0:
-; SKX-NEXT:    vpxor %xmm0, %xmm0, %xmm0
-; SKX-NEXT:    vpmaxsq (%rdi), %ymm0, %ymm1
-; SKX-NEXT:    vpmovusqd %ymm1, %xmm1
-; SKX-NEXT:    vpmaxsq 32(%rdi), %ymm0, %ymm0
-; SKX-NEXT:    vpmovusqd %ymm0, %xmm0
-; SKX-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
+; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; SKX-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; SKX-NEXT:    vpmovusqw %xmm0, (%rdi)
 ; SKX-NEXT:    retq
-  %a0 = load <8 x i64>, <8 x i64>* %p0
-  %1 = icmp slt <8 x i64> %a0, <i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295>
-  %2 = select <8 x i1> %1, <8 x i64> %a0, <8 x i64> <i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295, i64 4294967295>
-  %3 = icmp sgt <8 x i64> %2, zeroinitializer
-  %4 = select <8 x i1> %3, <8 x i64> %2, <8 x i64> zeroinitializer
-  %5 = trunc <8 x i64> %4 to <8 x i32>
-  ret <8 x i32> %5
+  %1 = icmp slt <2 x i64> %a0, <i64 65535, i64 65535>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 65535, i64 65535>
+  %3 = icmp sgt <2 x i64> %2, zeroinitializer
+  %4 = select <2 x i1> %3, <2 x i64> %2, <2 x i64> zeroinitializer
+  %5 = trunc <2 x i64> %4 to <2 x i16>
+  store <2 x i16> %5, <2 x i16> *%p1
+  ret void
 }
 
-;
-; PACKUS saturation truncation to vXi16
-;
-
 define <4 x i16> @trunc_packus_v4i64_v4i16(<4 x i64> %a0) {
 ; SSE2-LABEL: trunc_packus_v4i64_v4i16:
 ; SSE2:       # %bb.0:
@@ -2112,6 +2794,327 @@ define <16 x i16> @trunc_packus_v16i32_v16i16(<16 x i32>* %p0) "min-legal-vector
 ; PACKUS saturation truncation to vXi8
 ;
 
+define <2 x i8> @trunc_packus_v2i64_v2i8(<2 x i64> %a0) {
+; SSE2-LABEL: trunc_packus_v2i64_v2i8:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pxor %xmm1, %xmm2
+; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [2147483903,2147483903]
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm2, %xmm3
+; SSE2-NEXT:    pand %xmm3, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSE2-NEXT:    por %xmm0, %xmm3
+; SSE2-NEXT:    movdqa %xmm3, %xmm0
+; SSE2-NEXT:    pxor %xmm1, %xmm0
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[1,1,3,3]
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    pand %xmm3, %xmm0
+; SSE2-NEXT:    pand {{.*}}(%rip), %xmm0
+; SSE2-NEXT:    packuswb %xmm0, %xmm0
+; SSE2-NEXT:    packuswb %xmm0, %xmm0
+; SSE2-NEXT:    packuswb %xmm0, %xmm0
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_packus_v2i64_v2i8:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pxor %xmm1, %xmm2
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [2147483903,2147483903]
+; SSSE3-NEXT:    movdqa %xmm3, %xmm4
+; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm5, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSSE3-NEXT:    por %xmm2, %xmm3
+; SSSE3-NEXT:    pand %xmm3, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSSE3-NEXT:    por %xmm0, %xmm3
+; SSSE3-NEXT:    movdqa %xmm3, %xmm0
+; SSSE3-NEXT:    pxor %xmm1, %xmm0
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    por %xmm1, %xmm0
+; SSSE3-NEXT:    pand %xmm3, %xmm0
+; SSSE3-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_packus_v2i64_v2i8:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [255,255]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648]
+; SSE41-NEXT:    pxor %xmm3, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [2147483903,2147483903]
+; SSE41-NEXT:    movdqa %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    xorpd %xmm1, %xmm1
+; SSE41-NEXT:    movapd %xmm2, %xmm4
+; SSE41-NEXT:    xorpd %xmm3, %xmm4
+; SSE41-NEXT:    movapd %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm3, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm3, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm2, %xmm1
+; SSE41-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSE41-NEXT:    movdqa %xmm1, %xmm0
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: trunc_packus_v2i64_v2i8:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255]
+; AVX-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm1
+; AVX-NEXT:    vpand %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_packus_v2i64_v2i8:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255]
+; AVX512F-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512F-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512VL-LABEL: trunc_packus_v2i64_v2i8:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512VL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512VL-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_packus_v2i64_v2i8:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255]
+; AVX512BW-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512BW-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_packus_v2i64_v2i8:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512BWVL-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_packus_v2i64_v2i8:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; SKX-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; SKX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SKX-NEXT:    retq
+  %1 = icmp slt <2 x i64> %a0, <i64 255, i64 255>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 255, i64 255>
+  %3 = icmp sgt <2 x i64> %2, zeroinitializer
+  %4 = select <2 x i1> %3, <2 x i64> %2, <2 x i64> zeroinitializer
+  %5 = trunc <2 x i64> %4 to <2 x i8>
+  ret <2 x i8> %5
+}
+
+define void @trunc_packus_v2i64_v2i8_store(<2 x i64> %a0, <2 x i8> *%p1) {
+; SSE2-LABEL: trunc_packus_v2i64_v2i8_store:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pxor %xmm1, %xmm2
+; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [2147483903,2147483903]
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm2, %xmm3
+; SSE2-NEXT:    pand %xmm3, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSE2-NEXT:    por %xmm0, %xmm3
+; SSE2-NEXT:    movdqa %xmm3, %xmm0
+; SSE2-NEXT:    pxor %xmm1, %xmm0
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSE2-NEXT:    por %xmm0, %xmm1
+; SSE2-NEXT:    pand %xmm3, %xmm1
+; SSE2-NEXT:    pand {{.*}}(%rip), %xmm1
+; SSE2-NEXT:    packuswb %xmm1, %xmm1
+; SSE2-NEXT:    packuswb %xmm0, %xmm1
+; SSE2-NEXT:    packuswb %xmm0, %xmm1
+; SSE2-NEXT:    movd %xmm1, %eax
+; SSE2-NEXT:    movw %ax, (%rdi)
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_packus_v2i64_v2i8_store:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pxor %xmm1, %xmm2
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [2147483903,2147483903]
+; SSSE3-NEXT:    movdqa %xmm3, %xmm4
+; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm5, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSSE3-NEXT:    por %xmm2, %xmm3
+; SSSE3-NEXT:    pand %xmm3, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSSE3-NEXT:    por %xmm0, %xmm3
+; SSSE3-NEXT:    movdqa %xmm3, %xmm0
+; SSSE3-NEXT:    pxor %xmm1, %xmm0
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    por %xmm0, %xmm1
+; SSSE3-NEXT:    pand %xmm3, %xmm1
+; SSSE3-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSSE3-NEXT:    movd %xmm1, %eax
+; SSSE3-NEXT:    movw %ax, (%rdi)
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_packus_v2i64_v2i8_store:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [255,255]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648]
+; SSE41-NEXT:    pxor %xmm3, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [2147483903,2147483903]
+; SSE41-NEXT:    movdqa %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    xorpd %xmm1, %xmm1
+; SSE41-NEXT:    movapd %xmm2, %xmm4
+; SSE41-NEXT:    xorpd %xmm3, %xmm4
+; SSE41-NEXT:    movapd %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm3, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm3, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm2, %xmm1
+; SSE41-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSE41-NEXT:    pextrw $0, %xmm1, (%rdi)
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: trunc_packus_v2i64_v2i8_store:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255]
+; AVX-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm1
+; AVX-NEXT:    vpand %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX-NEXT:    vpextrw $0, %xmm0, (%rdi)
+; AVX-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_packus_v2i64_v2i8_store:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255]
+; AVX512F-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512F-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT:    vpextrw $0, %xmm0, (%rdi)
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512VL-LABEL: trunc_packus_v2i64_v2i8_store:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512VL-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; AVX512VL-NEXT:    vpmovusqb %xmm0, (%rdi)
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_packus_v2i64_v2i8_store:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255]
+; AVX512BW-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512BW-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512BW-NEXT:    vpextrw $0, %xmm0, (%rdi)
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_packus_v2i64_v2i8_store:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX512BWVL-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpmovusqb %xmm0, (%rdi)
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_packus_v2i64_v2i8_store:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; SKX-NEXT:    vpmaxsq %xmm1, %xmm0, %xmm0
+; SKX-NEXT:    vpmovusqb %xmm0, (%rdi)
+; SKX-NEXT:    retq
+  %1 = icmp slt <2 x i64> %a0, <i64 255, i64 255>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 255, i64 255>
+  %3 = icmp sgt <2 x i64> %2, zeroinitializer
+  %4 = select <2 x i1> %3, <2 x i64> %2, <2 x i64> zeroinitializer
+  %5 = trunc <2 x i64> %4 to <2 x i8>
+  store <2 x i8> %5, <2 x i8> *%p1
+  ret void
+}
+
 define <4 x i8> @trunc_packus_v4i64_v4i8(<4 x i64> %a0) {
 ; SSE2-LABEL: trunc_packus_v4i64_v4i8:
 ; SSE2:       # %bb.0:

diff --git a/test/CodeGen/X86/vector-trunc-ssat.ll b/test/CodeGen/X86/vector-trunc-ssat.ll
index 729ab00275cdcba615c69825565800ed2bd5e0b4..d15559e40312d47161d61c463f63380303bd0975 100644 (file)
--- a/test/CodeGen/X86/vector-trunc-ssat.ll
+++ b/test/CodeGen/X86/vector-trunc-ssat.ll
@@ -15,6 +15,317 @@
 ; Signed saturation truncation to vXi32
 ;
 
+define <2 x i32> @trunc_ssat_v2i64_v2i32(<2 x i64> %a0) {
+; SSE2-LABEL: trunc_ssat_v2i64_v2i32:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pxor %xmm1, %xmm2
+; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [4294967295,4294967295]
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm2, %xmm3
+; SSE2-NEXT:    pand %xmm3, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSE2-NEXT:    por %xmm0, %xmm3
+; SSE2-NEXT:    pxor %xmm3, %xmm1
+; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [18446744069414584320,18446744069414584320]
+; SSE2-NEXT:    movdqa %xmm1, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSE2-NEXT:    por %xmm0, %xmm1
+; SSE2-NEXT:    pand %xmm1, %xmm3
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm1
+; SSE2-NEXT:    por %xmm3, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_ssat_v2i64_v2i32:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pxor %xmm1, %xmm2
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [4294967295,4294967295]
+; SSSE3-NEXT:    movdqa %xmm3, %xmm4
+; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm5, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSSE3-NEXT:    por %xmm2, %xmm3
+; SSSE3-NEXT:    pand %xmm3, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSSE3-NEXT:    por %xmm0, %xmm3
+; SSSE3-NEXT:    pxor %xmm3, %xmm1
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm0 = [18446744069414584320,18446744069414584320]
+; SSSE3-NEXT:    movdqa %xmm1, %xmm2
+; SSSE3-NEXT:    pcmpgtd %xmm0, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    por %xmm0, %xmm1
+; SSSE3-NEXT:    pand %xmm1, %xmm3
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm1
+; SSSE3-NEXT:    por %xmm3, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_ssat_v2i64_v2i32:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [2147483647,2147483647]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648]
+; SSE41-NEXT:    pxor %xmm3, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [4294967295,4294967295]
+; SSE41-NEXT:    movdqa %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    movapd {{.*#+}} xmm1 = [18446744071562067968,18446744071562067968]
+; SSE41-NEXT:    pxor %xmm2, %xmm3
+; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [18446744069414584320,18446744069414584320]
+; SSE41-NEXT:    movdqa %xmm3, %xmm4
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm4
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm3
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[0,0,2,2]
+; SSE41-NEXT:    pand %xmm4, %xmm0
+; SSE41-NEXT:    por %xmm3, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm2, %xmm1
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: trunc_ssat_v2i64_v2i32:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [2147483647,2147483647]
+; AVX-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744071562067968,18446744071562067968]
+; AVX-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm2
+; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_ssat_v2i64_v2i32:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [2147483647,2147483647]
+; AVX512F-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744071562067968,18446744071562067968]
+; AVX512F-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512VL-LABEL: trunc_ssat_v2i64_v2i32:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512VL-NEXT:    vpmaxsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512VL-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_ssat_v2i64_v2i32:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [2147483647,2147483647]
+; AVX512BW-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744071562067968,18446744071562067968]
+; AVX512BW-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_ssat_v2i64_v2i32:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpmaxsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_ssat_v2i64_v2i32:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; SKX-NEXT:    vpmaxsq {{.*}}(%rip), %xmm0, %xmm0
+; SKX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SKX-NEXT:    retq
+  %1 = icmp slt <2 x i64> %a0, <i64 2147483647, i64 2147483647>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 2147483647, i64 2147483647>
+  %3 = icmp sgt <2 x i64> %2, <i64 -2147483648, i64 -2147483648>
+  %4 = select <2 x i1> %3, <2 x i64> %2, <2 x i64> <i64 -2147483648, i64 -2147483648>
+  %5 = trunc <2 x i64> %4 to <2 x i32>
+  ret <2 x i32> %5
+}
+
+define void @trunc_ssat_v2i64_v2i32_store(<2 x i64> %a0, <2 x i32>* %p1) {
+; SSE2-LABEL: trunc_ssat_v2i64_v2i32_store:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pxor %xmm1, %xmm2
+; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [4294967295,4294967295]
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm2, %xmm3
+; SSE2-NEXT:    pand %xmm3, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSE2-NEXT:    por %xmm0, %xmm3
+; SSE2-NEXT:    pxor %xmm3, %xmm1
+; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [18446744069414584320,18446744069414584320]
+; SSE2-NEXT:    movdqa %xmm1, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSE2-NEXT:    por %xmm0, %xmm1
+; SSE2-NEXT:    pand %xmm1, %xmm3
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm1
+; SSE2-NEXT:    por %xmm3, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSE2-NEXT:    movq %xmm0, (%rdi)
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_ssat_v2i64_v2i32_store:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pxor %xmm1, %xmm2
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [4294967295,4294967295]
+; SSSE3-NEXT:    movdqa %xmm3, %xmm4
+; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm5, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSSE3-NEXT:    por %xmm2, %xmm3
+; SSSE3-NEXT:    pand %xmm3, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSSE3-NEXT:    por %xmm0, %xmm3
+; SSSE3-NEXT:    pxor %xmm3, %xmm1
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm0 = [18446744069414584320,18446744069414584320]
+; SSSE3-NEXT:    movdqa %xmm1, %xmm2
+; SSSE3-NEXT:    pcmpgtd %xmm0, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    por %xmm0, %xmm1
+; SSSE3-NEXT:    pand %xmm1, %xmm3
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm1
+; SSSE3-NEXT:    por %xmm3, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSSE3-NEXT:    movq %xmm0, (%rdi)
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_ssat_v2i64_v2i32_store:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [2147483647,2147483647]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648]
+; SSE41-NEXT:    pxor %xmm3, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [4294967295,4294967295]
+; SSE41-NEXT:    movdqa %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    movapd {{.*#+}} xmm1 = [18446744071562067968,18446744071562067968]
+; SSE41-NEXT:    pxor %xmm2, %xmm3
+; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [18446744069414584320,18446744069414584320]
+; SSE41-NEXT:    movdqa %xmm3, %xmm4
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm4
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm3
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[0,0,2,2]
+; SSE41-NEXT:    pand %xmm4, %xmm0
+; SSE41-NEXT:    por %xmm3, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm2, %xmm1
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSE41-NEXT:    movq %xmm0, (%rdi)
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: trunc_ssat_v2i64_v2i32_store:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [2147483647,2147483647]
+; AVX-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744071562067968,18446744071562067968]
+; AVX-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm2
+; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX-NEXT:    vmovlpd %xmm0, (%rdi)
+; AVX-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_ssat_v2i64_v2i32_store:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [2147483647,2147483647]
+; AVX512F-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744071562067968,18446744071562067968]
+; AVX512F-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512F-NEXT:    vmovq %xmm0, (%rdi)
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512VL-LABEL: trunc_ssat_v2i64_v2i32_store:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpmovsqd %xmm0, (%rdi)
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_ssat_v2i64_v2i32_store:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [2147483647,2147483647]
+; AVX512BW-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744071562067968,18446744071562067968]
+; AVX512BW-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512BW-NEXT:    vmovq %xmm0, (%rdi)
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_ssat_v2i64_v2i32_store:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpmovsqd %xmm0, (%rdi)
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_ssat_v2i64_v2i32_store:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpmovsqd %xmm0, (%rdi)
+; SKX-NEXT:    retq
+  %1 = icmp slt <2 x i64> %a0, <i64 2147483647, i64 2147483647>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 2147483647, i64 2147483647>
+  %3 = icmp sgt <2 x i64> %2, <i64 -2147483648, i64 -2147483648>
+  %4 = select <2 x i1> %3, <2 x i64> %2, <2 x i64> <i64 -2147483648, i64 -2147483648>
+  %5 = trunc <2 x i64> %4 to <2 x i32>
+  store <2 x i32> %5, <2 x i32>* %p1
+  ret void
+}
+
 define <4 x i32> @trunc_ssat_v4i64_v4i32(<4 x i64> %a0) {
 ; SSE2-LABEL: trunc_ssat_v4i64_v4i32:
 ; SSE2:       # %bb.0:
@@ -639,74 +950,443 @@ define <8 x i32> @trunc_ssat_v8i64_v8i32(<8 x i64>* %p0) "min-legal-vector-width
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
-; AVX2-SLOW-LABEL: trunc_ssat_v8i64_v8i32:
+; AVX2-SLOW-LABEL: trunc_ssat_v8i64_v8i32:
+; AVX2-SLOW:       # %bb.0:
+; AVX2-SLOW-NEXT:    vmovdqa (%rdi), %ymm0
+; AVX2-SLOW-NEXT:    vmovdqa 32(%rdi), %ymm1
+; AVX2-SLOW-NEXT:    vpbroadcastq {{.*#+}} ymm2 = [2147483647,2147483647,2147483647,2147483647]
+; AVX2-SLOW-NEXT:    vpcmpgtq %ymm1, %ymm2, %ymm3
+; AVX2-SLOW-NEXT:    vblendvpd %ymm3, %ymm1, %ymm2, %ymm1
+; AVX2-SLOW-NEXT:    vpcmpgtq %ymm0, %ymm2, %ymm3
+; AVX2-SLOW-NEXT:    vblendvpd %ymm3, %ymm0, %ymm2, %ymm0
+; AVX2-SLOW-NEXT:    vpbroadcastq {{.*#+}} ymm2 = [18446744071562067968,18446744071562067968,18446744071562067968,18446744071562067968]
+; AVX2-SLOW-NEXT:    vpcmpgtq %ymm2, %ymm0, %ymm3
+; AVX2-SLOW-NEXT:    vblendvpd %ymm3, %ymm0, %ymm2, %ymm0
+; AVX2-SLOW-NEXT:    vpcmpgtq %ymm2, %ymm1, %ymm3
+; AVX2-SLOW-NEXT:    vblendvpd %ymm3, %ymm1, %ymm2, %ymm1
+; AVX2-SLOW-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX2-SLOW-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,2],xmm2[0,2]
+; AVX2-SLOW-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX2-SLOW-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[0,2]
+; AVX2-SLOW-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-SLOW-NEXT:    retq
+;
+; AVX2-FAST-LABEL: trunc_ssat_v8i64_v8i32:
+; AVX2-FAST:       # %bb.0:
+; AVX2-FAST-NEXT:    vmovdqa (%rdi), %ymm0
+; AVX2-FAST-NEXT:    vmovdqa 32(%rdi), %ymm1
+; AVX2-FAST-NEXT:    vpbroadcastq {{.*#+}} ymm2 = [2147483647,2147483647,2147483647,2147483647]
+; AVX2-FAST-NEXT:    vpcmpgtq %ymm0, %ymm2, %ymm3
+; AVX2-FAST-NEXT:    vblendvpd %ymm3, %ymm0, %ymm2, %ymm0
+; AVX2-FAST-NEXT:    vpcmpgtq %ymm1, %ymm2, %ymm3
+; AVX2-FAST-NEXT:    vblendvpd %ymm3, %ymm1, %ymm2, %ymm1
+; AVX2-FAST-NEXT:    vpbroadcastq {{.*#+}} ymm2 = [18446744071562067968,18446744071562067968,18446744071562067968,18446744071562067968]
+; AVX2-FAST-NEXT:    vpcmpgtq %ymm2, %ymm1, %ymm3
+; AVX2-FAST-NEXT:    vblendvpd %ymm3, %ymm1, %ymm2, %ymm1
+; AVX2-FAST-NEXT:    vpcmpgtq %ymm2, %ymm0, %ymm3
+; AVX2-FAST-NEXT:    vblendvpd %ymm3, %ymm0, %ymm2, %ymm0
+; AVX2-FAST-NEXT:    vmovapd {{.*#+}} ymm2 = [0,2,4,6,4,6,6,7]
+; AVX2-FAST-NEXT:    vpermps %ymm0, %ymm2, %ymm0
+; AVX2-FAST-NEXT:    vpermps %ymm1, %ymm2, %ymm1
+; AVX2-FAST-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-FAST-NEXT:    retq
+;
+; AVX512-LABEL: trunc_ssat_v8i64_v8i32:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vmovdqa64 (%rdi), %zmm0
+; AVX512-NEXT:    vpmovsqd %zmm0, %ymm0
+; AVX512-NEXT:    retq
+;
+; SKX-LABEL: trunc_ssat_v8i64_v8i32:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vmovdqa (%rdi), %ymm0
+; SKX-NEXT:    vmovdqa 32(%rdi), %ymm1
+; SKX-NEXT:    vpmovsqd %ymm0, %xmm0
+; SKX-NEXT:    vpmovsqd %ymm1, %xmm1
+; SKX-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; SKX-NEXT:    retq
+  %a0 = load <8 x i64>, <8 x i64>* %p0
+  %1 = icmp slt <8 x i64> %a0, <i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647>
+  %2 = select <8 x i1> %1, <8 x i64> %a0, <8 x i64> <i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647>
+  %3 = icmp sgt <8 x i64> %2, <i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648>
+  %4 = select <8 x i1> %3, <8 x i64> %2, <8 x i64> <i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648>
+  %5 = trunc <8 x i64> %4 to <8 x i32>
+  ret <8 x i32> %5
+}
+
+;
+; Signed saturation truncation to vXi16
+;
+
+define <2 x i16> @trunc_ssat_v2i64_v2i16(<2 x i64> %a0) {
+; SSE2-LABEL: trunc_ssat_v2i64_v2i16:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pxor %xmm1, %xmm2
+; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [2147516415,2147516415]
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm2, %xmm3
+; SSE2-NEXT:    pand %xmm3, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSE2-NEXT:    por %xmm0, %xmm3
+; SSE2-NEXT:    pxor %xmm3, %xmm1
+; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [18446744071562035200,18446744071562035200]
+; SSE2-NEXT:    movdqa %xmm1, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSE2-NEXT:    por %xmm0, %xmm1
+; SSE2-NEXT:    pand %xmm1, %xmm3
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm1
+; SSE2-NEXT:    por %xmm3, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_ssat_v2i64_v2i16:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pxor %xmm1, %xmm2
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [2147516415,2147516415]
+; SSSE3-NEXT:    movdqa %xmm3, %xmm4
+; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm5, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSSE3-NEXT:    por %xmm2, %xmm3
+; SSSE3-NEXT:    pand %xmm3, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSSE3-NEXT:    por %xmm0, %xmm3
+; SSSE3-NEXT:    pxor %xmm3, %xmm1
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm0 = [18446744071562035200,18446744071562035200]
+; SSSE3-NEXT:    movdqa %xmm1, %xmm2
+; SSSE3-NEXT:    pcmpgtd %xmm0, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    por %xmm0, %xmm1
+; SSSE3-NEXT:    pand %xmm1, %xmm3
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm1
+; SSSE3-NEXT:    por %xmm3, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSSE3-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_ssat_v2i64_v2i16:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [32767,32767]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648]
+; SSE41-NEXT:    pxor %xmm3, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [2147516415,2147516415]
+; SSE41-NEXT:    movdqa %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    movapd {{.*#+}} xmm1 = [18446744073709518848,18446744073709518848]
+; SSE41-NEXT:    pxor %xmm2, %xmm3
+; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [18446744071562035200,18446744071562035200]
+; SSE41-NEXT:    movdqa %xmm3, %xmm4
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm4
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm3
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[0,0,2,2]
+; SSE41-NEXT:    pand %xmm4, %xmm0
+; SSE41-NEXT:    por %xmm3, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm2, %xmm1
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSE41-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: trunc_ssat_v2i64_v2i16:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [32767,32767]
+; AVX1-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX1-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709518848,18446744073709518848]
+; AVX1-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm2
+; AVX1-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-SLOW-LABEL: trunc_ssat_v2i64_v2i16:
+; AVX2-SLOW:       # %bb.0:
+; AVX2-SLOW-NEXT:    vmovdqa {{.*#+}} xmm1 = [32767,32767]
+; AVX2-SLOW-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX2-SLOW-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX2-SLOW-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709518848,18446744073709518848]
+; AVX2-SLOW-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm2
+; AVX2-SLOW-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX2-SLOW-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX2-SLOW-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX2-SLOW-NEXT:    retq
+;
+; AVX2-FAST-LABEL: trunc_ssat_v2i64_v2i16:
+; AVX2-FAST:       # %bb.0:
+; AVX2-FAST-NEXT:    vmovdqa {{.*#+}} xmm1 = [32767,32767]
+; AVX2-FAST-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX2-FAST-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX2-FAST-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709518848,18446744073709518848]
+; AVX2-FAST-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm2
+; AVX2-FAST-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX2-FAST-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX2-FAST-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_ssat_v2i64_v2i16:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [32767,32767]
+; AVX512F-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709518848,18446744073709518848]
+; AVX512F-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512F-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512VL-LABEL: trunc_ssat_v2i64_v2i16:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512VL-NEXT:    vpmaxsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_ssat_v2i64_v2i16:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [32767,32767]
+; AVX512BW-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709518848,18446744073709518848]
+; AVX512BW-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_ssat_v2i64_v2i16:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpmaxsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_ssat_v2i64_v2i16:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; SKX-NEXT:    vpmaxsq {{.*}}(%rip), %xmm0, %xmm0
+; SKX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; SKX-NEXT:    retq
+  %1 = icmp slt <2 x i64> %a0, <i64 32767, i64 32767>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 32767, i64 32767>
+  %3 = icmp sgt <2 x i64> %2, <i64 -32768, i64 -32768>
+  %4 = select <2 x i1> %3, <2 x i64> %2, <2 x i64> <i64 -32768, i64 -32768>
+  %5 = trunc <2 x i64> %4 to <2 x i16>
+  ret <2 x i16> %5
+}
+
+define void @trunc_ssat_v2i64_v2i16_store(<2 x i64> %a0, <2 x i16> *%p1) {
+; SSE2-LABEL: trunc_ssat_v2i64_v2i16_store:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pxor %xmm1, %xmm2
+; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [2147516415,2147516415]
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm2, %xmm3
+; SSE2-NEXT:    pand %xmm3, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSE2-NEXT:    por %xmm0, %xmm3
+; SSE2-NEXT:    pxor %xmm3, %xmm1
+; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [18446744071562035200,18446744071562035200]
+; SSE2-NEXT:    movdqa %xmm1, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSE2-NEXT:    por %xmm0, %xmm1
+; SSE2-NEXT:    pand %xmm1, %xmm3
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm1
+; SSE2-NEXT:    por %xmm3, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSE2-NEXT:    movd %xmm0, (%rdi)
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_ssat_v2i64_v2i16_store:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pxor %xmm1, %xmm2
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [2147516415,2147516415]
+; SSSE3-NEXT:    movdqa %xmm3, %xmm4
+; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm5, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSSE3-NEXT:    por %xmm2, %xmm3
+; SSSE3-NEXT:    pand %xmm3, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSSE3-NEXT:    por %xmm0, %xmm3
+; SSSE3-NEXT:    pxor %xmm3, %xmm1
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm0 = [18446744071562035200,18446744071562035200]
+; SSSE3-NEXT:    movdqa %xmm1, %xmm2
+; SSSE3-NEXT:    pcmpgtd %xmm0, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    por %xmm0, %xmm1
+; SSSE3-NEXT:    pand %xmm1, %xmm3
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm1
+; SSSE3-NEXT:    por %xmm3, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSSE3-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSSE3-NEXT:    movd %xmm0, (%rdi)
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_ssat_v2i64_v2i16_store:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [32767,32767]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648]
+; SSE41-NEXT:    pxor %xmm3, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [2147516415,2147516415]
+; SSE41-NEXT:    movdqa %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    movapd {{.*#+}} xmm1 = [18446744073709518848,18446744073709518848]
+; SSE41-NEXT:    pxor %xmm2, %xmm3
+; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [18446744071562035200,18446744071562035200]
+; SSE41-NEXT:    movdqa %xmm3, %xmm4
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm4
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm3
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[0,0,2,2]
+; SSE41-NEXT:    pand %xmm4, %xmm0
+; SSE41-NEXT:    por %xmm3, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm2, %xmm1
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
+; SSE41-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSE41-NEXT:    movd %xmm0, (%rdi)
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: trunc_ssat_v2i64_v2i16_store:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [32767,32767]
+; AVX1-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX1-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709518848,18446744073709518848]
+; AVX1-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm2
+; AVX1-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX1-NEXT:    vmovd %xmm0, (%rdi)
+; AVX1-NEXT:    retq
+;
+; AVX2-SLOW-LABEL: trunc_ssat_v2i64_v2i16_store:
 ; AVX2-SLOW:       # %bb.0:
-; AVX2-SLOW-NEXT:    vmovdqa (%rdi), %ymm0
-; AVX2-SLOW-NEXT:    vmovdqa 32(%rdi), %ymm1
-; AVX2-SLOW-NEXT:    vpbroadcastq {{.*#+}} ymm2 = [2147483647,2147483647,2147483647,2147483647]
-; AVX2-SLOW-NEXT:    vpcmpgtq %ymm1, %ymm2, %ymm3
-; AVX2-SLOW-NEXT:    vblendvpd %ymm3, %ymm1, %ymm2, %ymm1
-; AVX2-SLOW-NEXT:    vpcmpgtq %ymm0, %ymm2, %ymm3
-; AVX2-SLOW-NEXT:    vblendvpd %ymm3, %ymm0, %ymm2, %ymm0
-; AVX2-SLOW-NEXT:    vpbroadcastq {{.*#+}} ymm2 = [18446744071562067968,18446744071562067968,18446744071562067968,18446744071562067968]
-; AVX2-SLOW-NEXT:    vpcmpgtq %ymm2, %ymm0, %ymm3
-; AVX2-SLOW-NEXT:    vblendvpd %ymm3, %ymm0, %ymm2, %ymm0
-; AVX2-SLOW-NEXT:    vpcmpgtq %ymm2, %ymm1, %ymm3
-; AVX2-SLOW-NEXT:    vblendvpd %ymm3, %ymm1, %ymm2, %ymm1
-; AVX2-SLOW-NEXT:    vextractf128 $1, %ymm1, %xmm2
-; AVX2-SLOW-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,2],xmm2[0,2]
-; AVX2-SLOW-NEXT:    vextractf128 $1, %ymm0, %xmm2
-; AVX2-SLOW-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[0,2]
-; AVX2-SLOW-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-SLOW-NEXT:    vmovdqa {{.*#+}} xmm1 = [32767,32767]
+; AVX2-SLOW-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX2-SLOW-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX2-SLOW-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709518848,18446744073709518848]
+; AVX2-SLOW-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm2
+; AVX2-SLOW-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX2-SLOW-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX2-SLOW-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX2-SLOW-NEXT:    vmovd %xmm0, (%rdi)
 ; AVX2-SLOW-NEXT:    retq
 ;
-; AVX2-FAST-LABEL: trunc_ssat_v8i64_v8i32:
+; AVX2-FAST-LABEL: trunc_ssat_v2i64_v2i16_store:
 ; AVX2-FAST:       # %bb.0:
-; AVX2-FAST-NEXT:    vmovdqa (%rdi), %ymm0
-; AVX2-FAST-NEXT:    vmovdqa 32(%rdi), %ymm1
-; AVX2-FAST-NEXT:    vpbroadcastq {{.*#+}} ymm2 = [2147483647,2147483647,2147483647,2147483647]
-; AVX2-FAST-NEXT:    vpcmpgtq %ymm0, %ymm2, %ymm3
-; AVX2-FAST-NEXT:    vblendvpd %ymm3, %ymm0, %ymm2, %ymm0
-; AVX2-FAST-NEXT:    vpcmpgtq %ymm1, %ymm2, %ymm3
-; AVX2-FAST-NEXT:    vblendvpd %ymm3, %ymm1, %ymm2, %ymm1
-; AVX2-FAST-NEXT:    vpbroadcastq {{.*#+}} ymm2 = [18446744071562067968,18446744071562067968,18446744071562067968,18446744071562067968]
-; AVX2-FAST-NEXT:    vpcmpgtq %ymm2, %ymm1, %ymm3
-; AVX2-FAST-NEXT:    vblendvpd %ymm3, %ymm1, %ymm2, %ymm1
-; AVX2-FAST-NEXT:    vpcmpgtq %ymm2, %ymm0, %ymm3
-; AVX2-FAST-NEXT:    vblendvpd %ymm3, %ymm0, %ymm2, %ymm0
-; AVX2-FAST-NEXT:    vmovapd {{.*#+}} ymm2 = [0,2,4,6,4,6,6,7]
-; AVX2-FAST-NEXT:    vpermps %ymm0, %ymm2, %ymm0
-; AVX2-FAST-NEXT:    vpermps %ymm1, %ymm2, %ymm1
-; AVX2-FAST-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX2-FAST-NEXT:    vmovdqa {{.*#+}} xmm1 = [32767,32767]
+; AVX2-FAST-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX2-FAST-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX2-FAST-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709518848,18446744073709518848]
+; AVX2-FAST-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm2
+; AVX2-FAST-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX2-FAST-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX2-FAST-NEXT:    vmovd %xmm0, (%rdi)
 ; AVX2-FAST-NEXT:    retq
 ;
-; AVX512-LABEL: trunc_ssat_v8i64_v8i32:
-; AVX512:       # %bb.0:
-; AVX512-NEXT:    vmovdqa64 (%rdi), %zmm0
-; AVX512-NEXT:    vpmovsqd %zmm0, %ymm0
-; AVX512-NEXT:    retq
+; AVX512F-LABEL: trunc_ssat_v2i64_v2i16_store:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [32767,32767]
+; AVX512F-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709518848,18446744073709518848]
+; AVX512F-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512F-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512F-NEXT:    vmovd %xmm0, (%rdi)
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
 ;
-; SKX-LABEL: trunc_ssat_v8i64_v8i32:
+; AVX512VL-LABEL: trunc_ssat_v2i64_v2i16_store:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpmovsqw %xmm0, (%rdi)
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_ssat_v2i64_v2i16_store:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [32767,32767]
+; AVX512BW-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709518848,18446744073709518848]
+; AVX512BW-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX512BW-NEXT:    vmovd %xmm0, (%rdi)
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_ssat_v2i64_v2i16_store:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpmovsqw %xmm0, (%rdi)
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_ssat_v2i64_v2i16_store:
 ; SKX:       # %bb.0:
-; SKX-NEXT:    vmovdqa (%rdi), %ymm0
-; SKX-NEXT:    vmovdqa 32(%rdi), %ymm1
-; SKX-NEXT:    vpmovsqd %ymm0, %xmm0
-; SKX-NEXT:    vpmovsqd %ymm1, %xmm1
-; SKX-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; SKX-NEXT:    vpmovsqw %xmm0, (%rdi)
 ; SKX-NEXT:    retq
-  %a0 = load <8 x i64>, <8 x i64>* %p0
-  %1 = icmp slt <8 x i64> %a0, <i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647>
-  %2 = select <8 x i1> %1, <8 x i64> %a0, <8 x i64> <i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647, i64 2147483647>
-  %3 = icmp sgt <8 x i64> %2, <i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648>
-  %4 = select <8 x i1> %3, <8 x i64> %2, <8 x i64> <i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648, i64 -2147483648>
-  %5 = trunc <8 x i64> %4 to <8 x i32>
-  ret <8 x i32> %5
+  %1 = icmp slt <2 x i64> %a0, <i64 32767, i64 32767>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 32767, i64 32767>
+  %3 = icmp sgt <2 x i64> %2, <i64 -32768, i64 -32768>
+  %4 = select <2 x i1> %3, <2 x i64> %2, <2 x i64> <i64 -32768, i64 -32768>
+  %5 = trunc <2 x i64> %4 to <2 x i16>
+  store <2 x i16> %5, <2 x i16> *%p1
+  ret void
 }
 
-;
-; Signed saturation truncation to vXi16
-;
-
 define <4 x i16> @trunc_ssat_v4i64_v4i16(<4 x i64> %a0) {
 ; SSE2-LABEL: trunc_ssat_v4i64_v4i16:
 ; SSE2:       # %bb.0:
@@ -1895,6 +2575,326 @@ define <16 x i16> @trunc_ssat_v16i32_v16i16(<16 x i32>* %p0) "min-legal-vector-w
 ; Signed saturation truncation to vXi8
 ;
 
+define <2 x i8> @trunc_ssat_v2i64_v2i8(<2 x i64> %a0) {
+; SSE2-LABEL: trunc_ssat_v2i64_v2i8:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pxor %xmm1, %xmm2
+; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [2147483775,2147483775]
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm2, %xmm3
+; SSE2-NEXT:    pand %xmm3, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSE2-NEXT:    por %xmm0, %xmm3
+; SSE2-NEXT:    pxor %xmm3, %xmm1
+; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [18446744071562067840,18446744071562067840]
+; SSE2-NEXT:    movdqa %xmm1, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[1,1,3,3]
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    pand %xmm0, %xmm3
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm0
+; SSE2-NEXT:    por %xmm3, %xmm0
+; SSE2-NEXT:    pand {{.*}}(%rip), %xmm0
+; SSE2-NEXT:    packuswb %xmm0, %xmm0
+; SSE2-NEXT:    packuswb %xmm0, %xmm0
+; SSE2-NEXT:    packuswb %xmm0, %xmm0
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_ssat_v2i64_v2i8:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pxor %xmm1, %xmm2
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [2147483775,2147483775]
+; SSSE3-NEXT:    movdqa %xmm3, %xmm4
+; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm5, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSSE3-NEXT:    por %xmm2, %xmm3
+; SSSE3-NEXT:    pand %xmm3, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSSE3-NEXT:    por %xmm0, %xmm3
+; SSSE3-NEXT:    pxor %xmm3, %xmm1
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm0 = [18446744071562067840,18446744071562067840]
+; SSSE3-NEXT:    movdqa %xmm1, %xmm2
+; SSSE3-NEXT:    pcmpgtd %xmm0, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    por %xmm1, %xmm0
+; SSSE3-NEXT:    pand %xmm0, %xmm3
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm0
+; SSSE3-NEXT:    por %xmm3, %xmm0
+; SSSE3-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_ssat_v2i64_v2i8:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [127,127]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648]
+; SSE41-NEXT:    pxor %xmm3, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [2147483775,2147483775]
+; SSE41-NEXT:    movdqa %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    movapd {{.*#+}} xmm1 = [18446744073709551488,18446744073709551488]
+; SSE41-NEXT:    pxor %xmm2, %xmm3
+; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [18446744071562067840,18446744071562067840]
+; SSE41-NEXT:    movdqa %xmm3, %xmm4
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm4
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm3
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[0,0,2,2]
+; SSE41-NEXT:    pand %xmm4, %xmm0
+; SSE41-NEXT:    por %xmm3, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm2, %xmm1
+; SSE41-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSE41-NEXT:    movdqa %xmm1, %xmm0
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: trunc_ssat_v2i64_v2i8:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [127,127]
+; AVX-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709551488,18446744073709551488]
+; AVX-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm2
+; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_ssat_v2i64_v2i8:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [127,127]
+; AVX512F-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709551488,18446744073709551488]
+; AVX512F-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512VL-LABEL: trunc_ssat_v2i64_v2i8:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512VL-NEXT:    vpmaxsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_ssat_v2i64_v2i8:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [127,127]
+; AVX512BW-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709551488,18446744073709551488]
+; AVX512BW-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_ssat_v2i64_v2i8:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpmaxsq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_ssat_v2i64_v2i8:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpminsq {{.*}}(%rip), %xmm0, %xmm0
+; SKX-NEXT:    vpmaxsq {{.*}}(%rip), %xmm0, %xmm0
+; SKX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SKX-NEXT:    retq
+  %1 = icmp slt <2 x i64> %a0, <i64 127, i64 127>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 127, i64 127>
+  %3 = icmp sgt <2 x i64> %2, <i64 -128, i64 -128>
+  %4 = select <2 x i1> %3, <2 x i64> %2, <2 x i64> <i64 -128, i64 -128>
+  %5 = trunc <2 x i64> %4 to <2 x i8>
+  ret <2 x i8> %5
+}
+
+define void @trunc_ssat_v2i64_v2i8_store(<2 x i64> %a0, <2 x i8> *%p1) {
+; SSE2-LABEL: trunc_ssat_v2i64_v2i8_store:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    pxor %xmm1, %xmm2
+; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [2147483775,2147483775]
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm2, %xmm3
+; SSE2-NEXT:    pand %xmm3, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSE2-NEXT:    por %xmm0, %xmm3
+; SSE2-NEXT:    pxor %xmm3, %xmm1
+; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [18446744071562067840,18446744071562067840]
+; SSE2-NEXT:    movdqa %xmm1, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSE2-NEXT:    por %xmm0, %xmm1
+; SSE2-NEXT:    pand %xmm1, %xmm3
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm1
+; SSE2-NEXT:    por %xmm3, %xmm1
+; SSE2-NEXT:    pand {{.*}}(%rip), %xmm1
+; SSE2-NEXT:    packuswb %xmm1, %xmm1
+; SSE2-NEXT:    packuswb %xmm0, %xmm1
+; SSE2-NEXT:    packuswb %xmm0, %xmm1
+; SSE2-NEXT:    movd %xmm1, %eax
+; SSE2-NEXT:    movw %ax, (%rdi)
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_ssat_v2i64_v2i8_store:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    pxor %xmm1, %xmm2
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [2147483775,2147483775]
+; SSSE3-NEXT:    movdqa %xmm3, %xmm4
+; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm3, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm5, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSSE3-NEXT:    por %xmm2, %xmm3
+; SSSE3-NEXT:    pand %xmm3, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm3
+; SSSE3-NEXT:    por %xmm0, %xmm3
+; SSSE3-NEXT:    pxor %xmm3, %xmm1
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm0 = [18446744071562067840,18446744071562067840]
+; SSSE3-NEXT:    movdqa %xmm1, %xmm2
+; SSSE3-NEXT:    pcmpgtd %xmm0, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm0
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
+; SSSE3-NEXT:    por %xmm0, %xmm1
+; SSSE3-NEXT:    pand %xmm1, %xmm3
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm1
+; SSSE3-NEXT:    por %xmm3, %xmm1
+; SSSE3-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSSE3-NEXT:    movd %xmm1, %eax
+; SSSE3-NEXT:    movw %ax, (%rdi)
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_ssat_v2i64_v2i8_store:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [127,127]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648]
+; SSE41-NEXT:    pxor %xmm3, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [2147483775,2147483775]
+; SSE41-NEXT:    movdqa %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm5
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm4
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm4[0,0,2,2]
+; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    por %xmm4, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    movapd {{.*#+}} xmm1 = [18446744073709551488,18446744073709551488]
+; SSE41-NEXT:    pxor %xmm2, %xmm3
+; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [18446744071562067840,18446744071562067840]
+; SSE41-NEXT:    movdqa %xmm3, %xmm4
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm4
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm3
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[0,0,2,2]
+; SSE41-NEXT:    pand %xmm4, %xmm0
+; SSE41-NEXT:    por %xmm3, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm2, %xmm1
+; SSE41-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSE41-NEXT:    pextrw $0, %xmm1, (%rdi)
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: trunc_ssat_v2i64_v2i8_store:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [127,127]
+; AVX-NEXT:    vpcmpgtq %xmm0, %xmm1, %xmm2
+; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709551488,18446744073709551488]
+; AVX-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm2
+; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX-NEXT:    vpextrw $0, %xmm0, (%rdi)
+; AVX-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_ssat_v2i64_v2i8_store:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [127,127]
+; AVX512F-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709551488,18446744073709551488]
+; AVX512F-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT:    vpextrw $0, %xmm0, (%rdi)
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512VL-LABEL: trunc_ssat_v2i64_v2i8_store:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpmovsqb %xmm0, (%rdi)
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_ssat_v2i64_v2i8_store:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [127,127]
+; AVX512BW-NEXT:    vpminsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [18446744073709551488,18446744073709551488]
+; AVX512BW-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512BW-NEXT:    vpextrw $0, %xmm0, (%rdi)
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_ssat_v2i64_v2i8_store:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpmovsqb %xmm0, (%rdi)
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_ssat_v2i64_v2i8_store:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpmovsqb %xmm0, (%rdi)
+; SKX-NEXT:    retq
+  %1 = icmp slt <2 x i64> %a0, <i64 127, i64 127>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 127, i64 127>
+  %3 = icmp sgt <2 x i64> %2, <i64 -128, i64 -128>
+  %4 = select <2 x i1> %3, <2 x i64> %2, <2 x i64> <i64 -128, i64 -128>
+  %5 = trunc <2 x i64> %4 to <2 x i8>
+  store <2 x i8> %5, <2 x i8> *%p1
+  ret void
+}
+
 define <4 x i8> @trunc_ssat_v4i64_v4i8(<4 x i64> %a0) {
 ; SSE2-LABEL: trunc_ssat_v4i64_v4i8:
 ; SSE2:       # %bb.0:

diff --git a/test/CodeGen/X86/vector-trunc-usat.ll b/test/CodeGen/X86/vector-trunc-usat.ll
index e24def71513d90bcad3bdc4bf2ca244aed22f3ee..82488d7bc7bcef2c09172d9ec3230307ba246307 100644 (file)
--- a/test/CodeGen/X86/vector-trunc-usat.ll
+++ b/test/CodeGen/X86/vector-trunc-usat.ll
@@ -15,6 +15,224 @@
 ; Unsigned saturation truncation to vXi32
 ;
 
+define <2 x i32> @trunc_usat_v2i64_v2i32(<2 x i64> %a0) {
+; SSE2-LABEL: trunc_usat_v2i64_v2i32:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [9223372039002259456,9223372039002259456]
+; SSE2-NEXT:    pxor %xmm0, %xmm1
+; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [9223372039002259455,9223372039002259455]
+; SSE2-NEXT:    movdqa %xmm2, %xmm3
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
+; SSE2-NEXT:    por %xmm1, %xmm2
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm2
+; SSE2-NEXT:    por %xmm0, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_usat_v2i64_v2i32:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [9223372039002259456,9223372039002259456]
+; SSSE3-NEXT:    pxor %xmm0, %xmm1
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm2 = [9223372039002259455,9223372039002259455]
+; SSSE3-NEXT:    movdqa %xmm2, %xmm3
+; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
+; SSSE3-NEXT:    por %xmm1, %xmm2
+; SSSE3-NEXT:    pand %xmm2, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm2
+; SSSE3-NEXT:    por %xmm0, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_usat_v2i64_v2i32:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [4294967295,4294967295]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [9223372039002259456,9223372039002259456]
+; SSE41-NEXT:    pxor %xmm1, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [9223372039002259455,9223372039002259455]
+; SSE41-NEXT:    movdqa %xmm3, %xmm4
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm4
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm3
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[0,0,2,2]
+; SSE41-NEXT:    pand %xmm4, %xmm0
+; SSE41-NEXT:    por %xmm3, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: trunc_usat_v2i64_v2i32:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovapd {{.*#+}} xmm1 = [4294967295,4294967295]
+; AVX-NEXT:    vpxor {{.*}}(%rip), %xmm0, %xmm2
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm3 = [9223372041149743103,9223372041149743103]
+; AVX-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
+; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_usat_v2i64_v2i32:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [4294967295,4294967295]
+; AVX512F-NEXT:    vpminuq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512VL-LABEL: trunc_usat_v2i64_v2i32:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpminuq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512VL-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_usat_v2i64_v2i32:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [4294967295,4294967295]
+; AVX512BW-NEXT:    vpminuq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_usat_v2i64_v2i32:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpminuq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_usat_v2i64_v2i32:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpminuq {{.*}}(%rip), %xmm0, %xmm0
+; SKX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SKX-NEXT:    retq
+  %1 = icmp ult <2 x i64> %a0, <i64 4294967295, i64 4294967295>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 4294967295, i64 4294967295>
+  %3 = trunc <2 x i64> %2 to <2 x i32>
+  ret <2 x i32> %3
+}
+
+define void @trunc_usat_v2i64_v2i32_store(<2 x i64> %a0, <2 x i32>* %p1) {
+; SSE2-LABEL: trunc_usat_v2i64_v2i32_store:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [9223372039002259456,9223372039002259456]
+; SSE2-NEXT:    pxor %xmm0, %xmm1
+; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [9223372039002259455,9223372039002259455]
+; SSE2-NEXT:    movdqa %xmm2, %xmm3
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
+; SSE2-NEXT:    por %xmm1, %xmm2
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm2
+; SSE2-NEXT:    por %xmm0, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSE2-NEXT:    movq %xmm0, (%rdi)
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_usat_v2i64_v2i32_store:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [9223372039002259456,9223372039002259456]
+; SSSE3-NEXT:    pxor %xmm0, %xmm1
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm2 = [9223372039002259455,9223372039002259455]
+; SSSE3-NEXT:    movdqa %xmm2, %xmm3
+; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
+; SSSE3-NEXT:    por %xmm1, %xmm2
+; SSSE3-NEXT:    pand %xmm2, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm2
+; SSSE3-NEXT:    por %xmm0, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSSE3-NEXT:    movq %xmm0, (%rdi)
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_usat_v2i64_v2i32_store:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [4294967295,4294967295]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [9223372039002259456,9223372039002259456]
+; SSE41-NEXT:    pxor %xmm1, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [9223372039002259455,9223372039002259455]
+; SSE41-NEXT:    movdqa %xmm3, %xmm4
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm4
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm3
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[0,0,2,2]
+; SSE41-NEXT:    pand %xmm4, %xmm0
+; SSE41-NEXT:    por %xmm3, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSE41-NEXT:    movq %xmm0, (%rdi)
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: trunc_usat_v2i64_v2i32_store:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovapd {{.*#+}} xmm1 = [4294967295,4294967295]
+; AVX-NEXT:    vpxor {{.*}}(%rip), %xmm0, %xmm2
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm3 = [9223372041149743103,9223372041149743103]
+; AVX-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
+; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX-NEXT:    vmovlpd %xmm0, (%rdi)
+; AVX-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_usat_v2i64_v2i32_store:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [4294967295,4294967295]
+; AVX512F-NEXT:    vpminuq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512F-NEXT:    vmovq %xmm0, (%rdi)
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512VL-LABEL: trunc_usat_v2i64_v2i32_store:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpmovusqd %xmm0, (%rdi)
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_usat_v2i64_v2i32_store:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [4294967295,4294967295]
+; AVX512BW-NEXT:    vpminuq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512BW-NEXT:    vmovq %xmm0, (%rdi)
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_usat_v2i64_v2i32_store:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpmovusqd %xmm0, (%rdi)
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_usat_v2i64_v2i32_store:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpmovusqd %xmm0, (%rdi)
+; SKX-NEXT:    retq
+  %1 = icmp ult <2 x i64> %a0, <i64 4294967295, i64 4294967295>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 4294967295, i64 4294967295>
+  %3 = trunc <2 x i64> %2 to <2 x i32>
+  store <2 x i32> %3, <2 x i32>* %p1
+  ret void
+}
+
 define <4 x i32> @trunc_usat_v4i64_v4i32(<4 x i64> %a0) {
 ; SSE2-LABEL: trunc_usat_v4i64_v4i32:
 ; SSE2:       # %bb.0:
@@ -479,6 +697,278 @@ define <8 x i32> @trunc_usat_v8i64_v8i32(<8 x i64>* %p0) {
 ; Unsigned saturation truncation to vXi16
 ;
 
+define <2 x i16> @trunc_usat_v2i64_v2i16(<2 x i64> %a0) {
+; SSE2-LABEL: trunc_usat_v2i64_v2i16:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [9223372039002259456,9223372039002259456]
+; SSE2-NEXT:    pxor %xmm0, %xmm1
+; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [9223372039002324991,9223372039002324991]
+; SSE2-NEXT:    movdqa %xmm2, %xmm3
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
+; SSE2-NEXT:    por %xmm1, %xmm2
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm2
+; SSE2-NEXT:    por %xmm0, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_usat_v2i64_v2i16:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [9223372039002259456,9223372039002259456]
+; SSSE3-NEXT:    pxor %xmm0, %xmm1
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm2 = [9223372039002324991,9223372039002324991]
+; SSSE3-NEXT:    movdqa %xmm2, %xmm3
+; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
+; SSSE3-NEXT:    por %xmm1, %xmm2
+; SSSE3-NEXT:    pand %xmm2, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm2
+; SSSE3-NEXT:    por %xmm0, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSSE3-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_usat_v2i64_v2i16:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [65535,65535]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [9223372039002259456,9223372039002259456]
+; SSE41-NEXT:    pxor %xmm1, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [9223372039002324991,9223372039002324991]
+; SSE41-NEXT:    movdqa %xmm3, %xmm4
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm4
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm3
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[0,0,2,2]
+; SSE41-NEXT:    pand %xmm4, %xmm0
+; SSE41-NEXT:    por %xmm3, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSE41-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: trunc_usat_v2i64_v2i16:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovapd {{.*#+}} xmm1 = [65535,65535]
+; AVX1-NEXT:    vpxor {{.*}}(%rip), %xmm0, %xmm2
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [9223372036854841343,9223372036854841343]
+; AVX1-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-SLOW-LABEL: trunc_usat_v2i64_v2i16:
+; AVX2-SLOW:       # %bb.0:
+; AVX2-SLOW-NEXT:    vmovapd {{.*#+}} xmm1 = [65535,65535]
+; AVX2-SLOW-NEXT:    vpxor {{.*}}(%rip), %xmm0, %xmm2
+; AVX2-SLOW-NEXT:    vmovdqa {{.*#+}} xmm3 = [9223372036854841343,9223372036854841343]
+; AVX2-SLOW-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
+; AVX2-SLOW-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX2-SLOW-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX2-SLOW-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX2-SLOW-NEXT:    retq
+;
+; AVX2-FAST-LABEL: trunc_usat_v2i64_v2i16:
+; AVX2-FAST:       # %bb.0:
+; AVX2-FAST-NEXT:    vmovapd {{.*#+}} xmm1 = [65535,65535]
+; AVX2-FAST-NEXT:    vpxor {{.*}}(%rip), %xmm0, %xmm2
+; AVX2-FAST-NEXT:    vmovdqa {{.*#+}} xmm3 = [9223372036854841343,9223372036854841343]
+; AVX2-FAST-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
+; AVX2-FAST-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX2-FAST-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX2-FAST-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_usat_v2i64_v2i16:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [65535,65535]
+; AVX512F-NEXT:    vpminuq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512F-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512VL-LABEL: trunc_usat_v2i64_v2i16:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpminuq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_usat_v2i64_v2i16:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [65535,65535]
+; AVX512BW-NEXT:    vpminuq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_usat_v2i64_v2i16:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpminuq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_usat_v2i64_v2i16:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpminuq {{.*}}(%rip), %xmm0, %xmm0
+; SKX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; SKX-NEXT:    retq
+  %1 = icmp ult <2 x i64> %a0, <i64 65535, i64 65535>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 65535, i64 65535>
+  %3 = trunc <2 x i64> %2 to <2 x i16>
+  ret <2 x i16> %3
+}
+
+define void @trunc_usat_v2i64_v2i16_store(<2 x i64> %a0, <2 x i16>* %p1) {
+; SSE2-LABEL: trunc_usat_v2i64_v2i16_store:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [9223372039002259456,9223372039002259456]
+; SSE2-NEXT:    pxor %xmm0, %xmm1
+; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [9223372039002324991,9223372039002324991]
+; SSE2-NEXT:    movdqa %xmm2, %xmm3
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
+; SSE2-NEXT:    por %xmm1, %xmm2
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm2
+; SSE2-NEXT:    por %xmm0, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSE2-NEXT:    movd %xmm0, (%rdi)
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_usat_v2i64_v2i16_store:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [9223372039002259456,9223372039002259456]
+; SSSE3-NEXT:    pxor %xmm0, %xmm1
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm2 = [9223372039002324991,9223372039002324991]
+; SSSE3-NEXT:    movdqa %xmm2, %xmm3
+; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
+; SSSE3-NEXT:    por %xmm1, %xmm2
+; SSSE3-NEXT:    pand %xmm2, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm2
+; SSSE3-NEXT:    por %xmm0, %xmm2
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSSE3-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSSE3-NEXT:    movd %xmm0, (%rdi)
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_usat_v2i64_v2i16_store:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [65535,65535]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [9223372039002259456,9223372039002259456]
+; SSE41-NEXT:    pxor %xmm1, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [9223372039002324991,9223372039002324991]
+; SSE41-NEXT:    movdqa %xmm3, %xmm4
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm4
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm3
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[0,0,2,2]
+; SSE41-NEXT:    pand %xmm4, %xmm0
+; SSE41-NEXT:    por %xmm3, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
+; SSE41-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSE41-NEXT:    movd %xmm0, (%rdi)
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: trunc_usat_v2i64_v2i16_store:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovapd {{.*#+}} xmm1 = [65535,65535]
+; AVX1-NEXT:    vpxor {{.*}}(%rip), %xmm0, %xmm2
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [9223372036854841343,9223372036854841343]
+; AVX1-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX1-NEXT:    vmovd %xmm0, (%rdi)
+; AVX1-NEXT:    retq
+;
+; AVX2-SLOW-LABEL: trunc_usat_v2i64_v2i16_store:
+; AVX2-SLOW:       # %bb.0:
+; AVX2-SLOW-NEXT:    vmovapd {{.*#+}} xmm1 = [65535,65535]
+; AVX2-SLOW-NEXT:    vpxor {{.*}}(%rip), %xmm0, %xmm2
+; AVX2-SLOW-NEXT:    vmovdqa {{.*#+}} xmm3 = [9223372036854841343,9223372036854841343]
+; AVX2-SLOW-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
+; AVX2-SLOW-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX2-SLOW-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX2-SLOW-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX2-SLOW-NEXT:    vmovd %xmm0, (%rdi)
+; AVX2-SLOW-NEXT:    retq
+;
+; AVX2-FAST-LABEL: trunc_usat_v2i64_v2i16_store:
+; AVX2-FAST:       # %bb.0:
+; AVX2-FAST-NEXT:    vmovapd {{.*#+}} xmm1 = [65535,65535]
+; AVX2-FAST-NEXT:    vpxor {{.*}}(%rip), %xmm0, %xmm2
+; AVX2-FAST-NEXT:    vmovdqa {{.*#+}} xmm3 = [9223372036854841343,9223372036854841343]
+; AVX2-FAST-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
+; AVX2-FAST-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX2-FAST-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX2-FAST-NEXT:    vmovd %xmm0, (%rdi)
+; AVX2-FAST-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_usat_v2i64_v2i16_store:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [65535,65535]
+; AVX512F-NEXT:    vpminuq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX512F-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX512F-NEXT:    vmovd %xmm0, (%rdi)
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512VL-LABEL: trunc_usat_v2i64_v2i16_store:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpmovusqw %xmm0, (%rdi)
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_usat_v2i64_v2i16_store:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [65535,65535]
+; AVX512BW-NEXT:    vpminuq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
+; AVX512BW-NEXT:    vmovd %xmm0, (%rdi)
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_usat_v2i64_v2i16_store:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpmovusqw %xmm0, (%rdi)
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_usat_v2i64_v2i16_store:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpmovusqw %xmm0, (%rdi)
+; SKX-NEXT:    retq
+  %1 = icmp ult <2 x i64> %a0, <i64 65535, i64 65535>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 65535, i64 65535>
+  %3 = trunc <2 x i64> %2 to <2 x i16>
+  store <2 x i16> %3, <2 x i16>* %p1
+  ret void
+}
+
 define <4 x i16> @trunc_usat_v4i64_v4i16(<4 x i64> %a0) {
 ; SSE2-LABEL: trunc_usat_v4i64_v4i16:
 ; SSE2:       # %bb.0:
@@ -1592,6 +2082,234 @@ define <16 x i16> @trunc_usat_v16i32_v16i16(<16 x i32>* %p0) {
 ; Unsigned saturation truncation to vXi8
 ;
 
+define <2 x i8> @trunc_usat_v2i64_v2i8(<2 x i64> %a0) {
+; SSE2-LABEL: trunc_usat_v2i64_v2i8:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [9223372039002259456,9223372039002259456]
+; SSE2-NEXT:    pxor %xmm0, %xmm1
+; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [9223372039002259711,9223372039002259711]
+; SSE2-NEXT:    movdqa %xmm2, %xmm3
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[1,1,3,3]
+; SSE2-NEXT:    por %xmm2, %xmm1
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm1
+; SSE2-NEXT:    por %xmm0, %xmm1
+; SSE2-NEXT:    pand {{.*}}(%rip), %xmm1
+; SSE2-NEXT:    packuswb %xmm1, %xmm1
+; SSE2-NEXT:    packuswb %xmm1, %xmm1
+; SSE2-NEXT:    packuswb %xmm1, %xmm1
+; SSE2-NEXT:    movdqa %xmm1, %xmm0
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_usat_v2i64_v2i8:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [9223372039002259456,9223372039002259456]
+; SSSE3-NEXT:    pxor %xmm0, %xmm1
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm2 = [9223372039002259711,9223372039002259711]
+; SSSE3-NEXT:    movdqa %xmm2, %xmm3
+; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
+; SSSE3-NEXT:    por %xmm1, %xmm2
+; SSSE3-NEXT:    pand %xmm2, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm2
+; SSSE3-NEXT:    por %xmm2, %xmm0
+; SSSE3-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_usat_v2i64_v2i8:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [255,255]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [9223372039002259456,9223372039002259456]
+; SSE41-NEXT:    pxor %xmm1, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [9223372039002259711,9223372039002259711]
+; SSE41-NEXT:    movdqa %xmm3, %xmm4
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm4
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm3
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[0,0,2,2]
+; SSE41-NEXT:    pand %xmm4, %xmm0
+; SSE41-NEXT:    por %xmm3, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    pshufb {{.*#+}} xmm2 = xmm2[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSE41-NEXT:    movdqa %xmm2, %xmm0
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: trunc_usat_v2i64_v2i8:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovapd {{.*#+}} xmm1 = [255,255]
+; AVX-NEXT:    vpxor {{.*}}(%rip), %xmm0, %xmm2
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm3 = [9223372036854776063,9223372036854776063]
+; AVX-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
+; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_usat_v2i64_v2i8:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255]
+; AVX512F-NEXT:    vpminuq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512VL-LABEL: trunc_usat_v2i64_v2i8:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpminuq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_usat_v2i64_v2i8:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255]
+; AVX512BW-NEXT:    vpminuq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_usat_v2i64_v2i8:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpminuq {{.*}}(%rip), %xmm0, %xmm0
+; AVX512BWVL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_usat_v2i64_v2i8:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpminuq {{.*}}(%rip), %xmm0, %xmm0
+; SKX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SKX-NEXT:    retq
+  %1 = icmp ult <2 x i64> %a0, <i64 255, i64 255>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 255, i64 255>
+  %3 = trunc <2 x i64> %2 to <2 x i8>
+  ret <2 x i8> %3
+}
+
+define void @trunc_usat_v2i64_v2i8_store(<2 x i64> %a0, <2 x i8>* %p1) {
+; SSE2-LABEL: trunc_usat_v2i64_v2i8_store:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [9223372039002259456,9223372039002259456]
+; SSE2-NEXT:    pxor %xmm0, %xmm1
+; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [9223372039002259711,9223372039002259711]
+; SSE2-NEXT:    movdqa %xmm2, %xmm3
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
+; SSE2-NEXT:    por %xmm1, %xmm2
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pandn {{.*}}(%rip), %xmm2
+; SSE2-NEXT:    por %xmm0, %xmm2
+; SSE2-NEXT:    pand {{.*}}(%rip), %xmm2
+; SSE2-NEXT:    packuswb %xmm2, %xmm2
+; SSE2-NEXT:    packuswb %xmm0, %xmm2
+; SSE2-NEXT:    packuswb %xmm0, %xmm2
+; SSE2-NEXT:    movd %xmm2, %eax
+; SSE2-NEXT:    movw %ax, (%rdi)
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc_usat_v2i64_v2i8_store:
+; SSSE3:       # %bb.0:
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [9223372039002259456,9223372039002259456]
+; SSSE3-NEXT:    pxor %xmm0, %xmm1
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm2 = [9223372039002259711,9223372039002259711]
+; SSSE3-NEXT:    movdqa %xmm2, %xmm3
+; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
+; SSSE3-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSSE3-NEXT:    pand %xmm4, %xmm1
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
+; SSSE3-NEXT:    por %xmm1, %xmm2
+; SSSE3-NEXT:    pand %xmm2, %xmm0
+; SSSE3-NEXT:    pandn {{.*}}(%rip), %xmm2
+; SSSE3-NEXT:    por %xmm0, %xmm2
+; SSSE3-NEXT:    pshufb {{.*#+}} xmm2 = xmm2[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSSE3-NEXT:    movd %xmm2, %eax
+; SSSE3-NEXT:    movw %ax, (%rdi)
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_usat_v2i64_v2i8_store:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    movapd {{.*#+}} xmm2 = [255,255]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [9223372039002259456,9223372039002259456]
+; SSE41-NEXT:    pxor %xmm1, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [9223372039002259711,9223372039002259711]
+; SSE41-NEXT:    movdqa %xmm3, %xmm4
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm4
+; SSE41-NEXT:    pcmpgtd %xmm0, %xmm3
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[0,0,2,2]
+; SSE41-NEXT:    pand %xmm4, %xmm0
+; SSE41-NEXT:    por %xmm3, %xmm0
+; SSE41-NEXT:    blendvpd %xmm0, %xmm1, %xmm2
+; SSE41-NEXT:    pshufb {{.*#+}} xmm2 = xmm2[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSE41-NEXT:    pextrw $0, %xmm2, (%rdi)
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: trunc_usat_v2i64_v2i8_store:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovapd {{.*#+}} xmm1 = [255,255]
+; AVX-NEXT:    vpxor {{.*}}(%rip), %xmm0, %xmm2
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm3 = [9223372036854776063,9223372036854776063]
+; AVX-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
+; AVX-NEXT:    vblendvpd %xmm2, %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX-NEXT:    vpextrw $0, %xmm0, (%rdi)
+; AVX-NEXT:    retq
+;
+; AVX512F-LABEL: trunc_usat_v2i64_v2i8_store:
+; AVX512F:       # %bb.0:
+; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255]
+; AVX512F-NEXT:    vpminuq %zmm1, %zmm0, %zmm0
+; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT:    vpextrw $0, %xmm0, (%rdi)
+; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    retq
+;
+; AVX512VL-LABEL: trunc_usat_v2i64_v2i8_store:
+; AVX512VL:       # %bb.0:
+; AVX512VL-NEXT:    vpmovusqb %xmm0, (%rdi)
+; AVX512VL-NEXT:    retq
+;
+; AVX512BW-LABEL: trunc_usat_v2i64_v2i8_store:
+; AVX512BW:       # %bb.0:
+; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
+; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255]
+; AVX512BW-NEXT:    vpminuq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512BW-NEXT:    vpextrw $0, %xmm0, (%rdi)
+; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    retq
+;
+; AVX512BWVL-LABEL: trunc_usat_v2i64_v2i8_store:
+; AVX512BWVL:       # %bb.0:
+; AVX512BWVL-NEXT:    vpmovusqb %xmm0, (%rdi)
+; AVX512BWVL-NEXT:    retq
+;
+; SKX-LABEL: trunc_usat_v2i64_v2i8_store:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpmovusqb %xmm0, (%rdi)
+; SKX-NEXT:    retq
+  %1 = icmp ult <2 x i64> %a0, <i64 255, i64 255>
+  %2 = select <2 x i1> %1, <2 x i64> %a0, <2 x i64> <i64 255, i64 255>
+  %3 = trunc <2 x i64> %2 to <2 x i8>
+  store <2 x i8> %3, <2 x i8>* %p1
+  ret void
+}
+
 define <4 x i8> @trunc_usat_v4i64_v4i8(<4 x i64> %a0) {
 ; SSE2-LABEL: trunc_usat_v4i64_v4i8:
 ; SSE2:       # %bb.0: