diff options
Diffstat (limited to 'llvm/test/CodeGen/X86/vector-trunc-usat.ll')
| -rw-r--r-- | llvm/test/CodeGen/X86/vector-trunc-usat.ll | 2652 |
1 files changed, 1185 insertions, 1467 deletions
diff --git a/llvm/test/CodeGen/X86/vector-trunc-usat.ll b/llvm/test/CodeGen/X86/vector-trunc-usat.ll index 2f2fd688bea..30c3a53c7b0 100644 --- a/llvm/test/CodeGen/X86/vector-trunc-usat.ll +++ b/llvm/test/CodeGen/X86/vector-trunc-usat.ll @@ -207,206 +207,203 @@ define <4 x i32> @trunc_usat_v4i64_v4i32(<4 x i64> %a0) { define <8 x i32> @trunc_usat_v8i64_v8i32(<8 x i64> %a0) { ; SSE2-LABEL: trunc_usat_v8i64_v8i32: ; SSE2: # %bb.0: -; SSE2-NEXT: movdqa {{.*#+}} xmm7 = [2147483648,2147483648,2147483648,2147483648] -; SSE2-NEXT: movdqa %xmm0, %xmm4 -; SSE2-NEXT: pxor %xmm7, %xmm4 -; SSE2-NEXT: movdqa {{.*#+}} xmm8 = [9223372039002259455,9223372039002259455] -; SSE2-NEXT: movdqa %xmm8, %xmm5 -; SSE2-NEXT: pcmpgtd %xmm4, %xmm5 -; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2] -; SSE2-NEXT: pcmpeqd %xmm8, %xmm4 -; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3] -; SSE2-NEXT: pand %xmm6, %xmm4 -; SSE2-NEXT: pshufd {{.*#+}} xmm10 = xmm5[1,1,3,3] -; SSE2-NEXT: por %xmm4, %xmm10 -; SSE2-NEXT: movdqa %xmm1, %xmm4 -; SSE2-NEXT: pxor %xmm7, %xmm4 -; SSE2-NEXT: movdqa %xmm8, %xmm6 -; SSE2-NEXT: pcmpgtd %xmm4, %xmm6 -; SSE2-NEXT: pshufd {{.*#+}} xmm9 = xmm6[0,0,2,2] -; SSE2-NEXT: pcmpeqd %xmm8, %xmm4 -; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3] -; SSE2-NEXT: pand %xmm9, %xmm4 -; SSE2-NEXT: pshufd {{.*#+}} xmm11 = xmm6[1,1,3,3] -; SSE2-NEXT: por %xmm4, %xmm11 -; SSE2-NEXT: movdqa %xmm2, %xmm4 -; SSE2-NEXT: pxor %xmm7, %xmm4 -; SSE2-NEXT: movdqa %xmm8, %xmm5 -; SSE2-NEXT: pcmpgtd %xmm4, %xmm5 -; SSE2-NEXT: pshufd {{.*#+}} xmm9 = xmm5[0,0,2,2] -; SSE2-NEXT: pcmpeqd %xmm8, %xmm4 -; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm4[1,1,3,3] -; SSE2-NEXT: pand %xmm9, %xmm6 -; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm5[1,1,3,3] -; SSE2-NEXT: por %xmm6, %xmm4 -; SSE2-NEXT: pxor %xmm3, %xmm7 -; SSE2-NEXT: movdqa %xmm8, %xmm5 -; SSE2-NEXT: pcmpgtd %xmm7, %xmm5 -; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2] -; SSE2-NEXT: pcmpeqd %xmm8, %xmm7 +; SSE2-NEXT: movdqa {{.*#+}} xmm8 = [4294967295,4294967295] +; SSE2-NEXT: movdqa {{.*#+}} xmm5 = [2147483648,2147483648,2147483648,2147483648] +; SSE2-NEXT: movdqa %xmm3, %xmm7 +; SSE2-NEXT: pxor %xmm5, %xmm7 +; SSE2-NEXT: movdqa {{.*#+}} xmm9 = [9223372039002259455,9223372039002259455] +; SSE2-NEXT: movdqa %xmm9, %xmm6 +; SSE2-NEXT: pcmpgtd %xmm7, %xmm6 +; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm6[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm9, %xmm7 ; SSE2-NEXT: pshufd {{.*#+}} xmm7 = xmm7[1,1,3,3] -; SSE2-NEXT: pand %xmm6, %xmm7 -; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3] -; SSE2-NEXT: por %xmm7, %xmm5 -; SSE2-NEXT: movdqa {{.*#+}} xmm6 = [4294967295,4294967295] -; SSE2-NEXT: pand %xmm5, %xmm3 -; SSE2-NEXT: pandn %xmm6, %xmm5 -; SSE2-NEXT: por %xmm3, %xmm5 -; SSE2-NEXT: pand %xmm4, %xmm2 -; SSE2-NEXT: pandn %xmm6, %xmm4 +; SSE2-NEXT: pand %xmm4, %xmm7 +; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm6[1,1,3,3] +; SSE2-NEXT: por %xmm7, %xmm4 +; SSE2-NEXT: pand %xmm4, %xmm3 +; SSE2-NEXT: pandn %xmm8, %xmm4 +; SSE2-NEXT: por %xmm3, %xmm4 +; SSE2-NEXT: movdqa %xmm2, %xmm3 +; SSE2-NEXT: pxor %xmm5, %xmm3 +; SSE2-NEXT: movdqa %xmm9, %xmm6 +; SSE2-NEXT: pcmpgtd %xmm3, %xmm6 +; SSE2-NEXT: pshufd {{.*#+}} xmm10 = xmm6[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm9, %xmm3 +; SSE2-NEXT: pshufd {{.*#+}} xmm7 = xmm3[1,1,3,3] +; SSE2-NEXT: pand %xmm10, %xmm7 +; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm6[1,1,3,3] +; SSE2-NEXT: por %xmm7, %xmm3 +; SSE2-NEXT: pand %xmm3, %xmm2 +; SSE2-NEXT: pandn %xmm8, %xmm3 +; SSE2-NEXT: por %xmm2, %xmm3 +; SSE2-NEXT: shufps {{.*#+}} xmm3 = xmm3[0,2],xmm4[0,2] +; SSE2-NEXT: movdqa %xmm1, %xmm2 +; SSE2-NEXT: pxor %xmm5, %xmm2 +; SSE2-NEXT: movdqa %xmm9, %xmm4 +; SSE2-NEXT: pcmpgtd %xmm2, %xmm4 +; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm4[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm9, %xmm2 +; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3] +; SSE2-NEXT: pand %xmm6, %xmm2 +; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3] ; SSE2-NEXT: por %xmm2, %xmm4 -; SSE2-NEXT: shufps {{.*#+}} xmm4 = xmm4[0,2],xmm5[0,2] -; SSE2-NEXT: pand %xmm11, %xmm1 -; SSE2-NEXT: pandn %xmm6, %xmm11 -; SSE2-NEXT: por %xmm1, %xmm11 -; SSE2-NEXT: pand %xmm10, %xmm0 -; SSE2-NEXT: pandn %xmm6, %xmm10 -; SSE2-NEXT: por %xmm10, %xmm0 -; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2],xmm11[0,2] -; SSE2-NEXT: movaps %xmm4, %xmm1 +; SSE2-NEXT: pand %xmm4, %xmm1 +; SSE2-NEXT: pandn %xmm8, %xmm4 +; SSE2-NEXT: por %xmm1, %xmm4 +; SSE2-NEXT: pxor %xmm0, %xmm5 +; SSE2-NEXT: movdqa %xmm9, %xmm1 +; SSE2-NEXT: pcmpgtd %xmm5, %xmm1 +; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm1[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm9, %xmm5 +; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3] +; SSE2-NEXT: pand %xmm2, %xmm5 +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3] +; SSE2-NEXT: por %xmm5, %xmm1 +; SSE2-NEXT: pand %xmm1, %xmm0 +; SSE2-NEXT: pandn %xmm8, %xmm1 +; SSE2-NEXT: por %xmm1, %xmm0 +; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2],xmm4[0,2] +; SSE2-NEXT: movaps %xmm3, %xmm1 ; SSE2-NEXT: retq ; ; SSSE3-LABEL: trunc_usat_v8i64_v8i32: ; SSSE3: # %bb.0: -; SSSE3-NEXT: movdqa {{.*#+}} xmm7 = [2147483648,2147483648,2147483648,2147483648] -; SSSE3-NEXT: movdqa %xmm0, %xmm4 -; SSSE3-NEXT: pxor %xmm7, %xmm4 -; SSSE3-NEXT: movdqa {{.*#+}} xmm8 = [9223372039002259455,9223372039002259455] -; SSSE3-NEXT: movdqa %xmm8, %xmm5 -; SSSE3-NEXT: pcmpgtd %xmm4, %xmm5 -; SSSE3-NEXT: pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2] -; SSSE3-NEXT: pcmpeqd %xmm8, %xmm4 -; SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3] -; SSSE3-NEXT: pand %xmm6, %xmm4 -; SSSE3-NEXT: pshufd {{.*#+}} xmm10 = xmm5[1,1,3,3] -; SSSE3-NEXT: por %xmm4, %xmm10 -; SSSE3-NEXT: movdqa %xmm1, %xmm4 -; SSSE3-NEXT: pxor %xmm7, %xmm4 -; SSSE3-NEXT: movdqa %xmm8, %xmm6 -; SSSE3-NEXT: pcmpgtd %xmm4, %xmm6 -; SSSE3-NEXT: pshufd {{.*#+}} xmm9 = xmm6[0,0,2,2] -; SSSE3-NEXT: pcmpeqd %xmm8, %xmm4 -; SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3] -; SSSE3-NEXT: pand %xmm9, %xmm4 -; SSSE3-NEXT: pshufd {{.*#+}} xmm11 = xmm6[1,1,3,3] -; SSSE3-NEXT: por %xmm4, %xmm11 -; SSSE3-NEXT: movdqa %xmm2, %xmm4 -; SSSE3-NEXT: pxor %xmm7, %xmm4 -; SSSE3-NEXT: movdqa %xmm8, %xmm5 -; SSSE3-NEXT: pcmpgtd %xmm4, %xmm5 -; SSSE3-NEXT: pshufd {{.*#+}} xmm9 = xmm5[0,0,2,2] -; SSSE3-NEXT: pcmpeqd %xmm8, %xmm4 -; SSSE3-NEXT: pshufd {{.*#+}} xmm6 = xmm4[1,1,3,3] -; SSSE3-NEXT: pand %xmm9, %xmm6 -; SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm5[1,1,3,3] -; SSSE3-NEXT: por %xmm6, %xmm4 -; SSSE3-NEXT: pxor %xmm3, %xmm7 -; SSSE3-NEXT: movdqa %xmm8, %xmm5 -; SSSE3-NEXT: pcmpgtd %xmm7, %xmm5 -; SSSE3-NEXT: pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2] -; SSSE3-NEXT: pcmpeqd %xmm8, %xmm7 +; SSSE3-NEXT: movdqa {{.*#+}} xmm8 = [4294967295,4294967295] +; SSSE3-NEXT: movdqa {{.*#+}} xmm5 = [2147483648,2147483648,2147483648,2147483648] +; SSSE3-NEXT: movdqa %xmm3, %xmm7 +; SSSE3-NEXT: pxor %xmm5, %xmm7 +; SSSE3-NEXT: movdqa {{.*#+}} xmm9 = [9223372039002259455,9223372039002259455] +; SSSE3-NEXT: movdqa %xmm9, %xmm6 +; SSSE3-NEXT: pcmpgtd %xmm7, %xmm6 +; SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm6[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm9, %xmm7 ; SSSE3-NEXT: pshufd {{.*#+}} xmm7 = xmm7[1,1,3,3] -; SSSE3-NEXT: pand %xmm6, %xmm7 -; SSSE3-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3] -; SSSE3-NEXT: por %xmm7, %xmm5 -; SSSE3-NEXT: movdqa {{.*#+}} xmm6 = [4294967295,4294967295] -; SSSE3-NEXT: pand %xmm5, %xmm3 -; SSSE3-NEXT: pandn %xmm6, %xmm5 -; SSSE3-NEXT: por %xmm3, %xmm5 -; SSSE3-NEXT: pand %xmm4, %xmm2 -; SSSE3-NEXT: pandn %xmm6, %xmm4 +; SSSE3-NEXT: pand %xmm4, %xmm7 +; SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm6[1,1,3,3] +; SSSE3-NEXT: por %xmm7, %xmm4 +; SSSE3-NEXT: pand %xmm4, %xmm3 +; SSSE3-NEXT: pandn %xmm8, %xmm4 +; SSSE3-NEXT: por %xmm3, %xmm4 +; SSSE3-NEXT: movdqa %xmm2, %xmm3 +; SSSE3-NEXT: pxor %xmm5, %xmm3 +; SSSE3-NEXT: movdqa %xmm9, %xmm6 +; SSSE3-NEXT: pcmpgtd %xmm3, %xmm6 +; SSSE3-NEXT: pshufd {{.*#+}} xmm10 = xmm6[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm9, %xmm3 +; SSSE3-NEXT: pshufd {{.*#+}} xmm7 = xmm3[1,1,3,3] +; SSSE3-NEXT: pand %xmm10, %xmm7 +; SSSE3-NEXT: pshufd {{.*#+}} xmm3 = xmm6[1,1,3,3] +; SSSE3-NEXT: por %xmm7, %xmm3 +; SSSE3-NEXT: pand %xmm3, %xmm2 +; SSSE3-NEXT: pandn %xmm8, %xmm3 +; SSSE3-NEXT: por %xmm2, %xmm3 +; SSSE3-NEXT: shufps {{.*#+}} xmm3 = xmm3[0,2],xmm4[0,2] +; SSSE3-NEXT: movdqa %xmm1, %xmm2 +; SSSE3-NEXT: pxor %xmm5, %xmm2 +; SSSE3-NEXT: movdqa %xmm9, %xmm4 +; SSSE3-NEXT: pcmpgtd %xmm2, %xmm4 +; SSSE3-NEXT: pshufd {{.*#+}} xmm6 = xmm4[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm9, %xmm2 +; SSSE3-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3] +; SSSE3-NEXT: pand %xmm6, %xmm2 +; SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3] ; SSSE3-NEXT: por %xmm2, %xmm4 -; SSSE3-NEXT: shufps {{.*#+}} xmm4 = xmm4[0,2],xmm5[0,2] -; SSSE3-NEXT: pand %xmm11, %xmm1 -; SSSE3-NEXT: pandn %xmm6, %xmm11 -; SSSE3-NEXT: por %xmm1, %xmm11 -; SSSE3-NEXT: pand %xmm10, %xmm0 -; SSSE3-NEXT: pandn %xmm6, %xmm10 -; SSSE3-NEXT: por %xmm10, %xmm0 -; SSSE3-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2],xmm11[0,2] -; SSSE3-NEXT: movaps %xmm4, %xmm1 +; SSSE3-NEXT: pand %xmm4, %xmm1 +; SSSE3-NEXT: pandn %xmm8, %xmm4 +; SSSE3-NEXT: por %xmm1, %xmm4 +; SSSE3-NEXT: pxor %xmm0, %xmm5 +; SSSE3-NEXT: movdqa %xmm9, %xmm1 +; SSSE3-NEXT: pcmpgtd %xmm5, %xmm1 +; SSSE3-NEXT: pshufd {{.*#+}} xmm2 = xmm1[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm9, %xmm5 +; SSSE3-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3] +; SSSE3-NEXT: pand %xmm2, %xmm5 +; SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3] +; SSSE3-NEXT: por %xmm5, %xmm1 +; SSSE3-NEXT: pand %xmm1, %xmm0 +; SSSE3-NEXT: pandn %xmm8, %xmm1 +; SSSE3-NEXT: por %xmm1, %xmm0 +; SSSE3-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2],xmm4[0,2] +; SSSE3-NEXT: movaps %xmm3, %xmm1 ; SSSE3-NEXT: retq ; ; SSE41-LABEL: trunc_usat_v8i64_v8i32: ; SSE41: # %bb.0: ; SSE41-NEXT: movdqa %xmm0, %xmm8 -; SSE41-NEXT: movdqa {{.*#+}} xmm0 = [2147483648,2147483648,2147483648,2147483648] -; SSE41-NEXT: movdqa %xmm8, %xmm5 -; SSE41-NEXT: pxor %xmm0, %xmm5 -; SSE41-NEXT: movdqa {{.*#+}} xmm4 = [9223372039002259455,9223372039002259455] -; SSE41-NEXT: movdqa %xmm4, %xmm6 -; SSE41-NEXT: pcmpgtd %xmm5, %xmm6 -; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm6[0,0,2,2] -; SSE41-NEXT: pcmpeqd %xmm4, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3] -; SSE41-NEXT: pand %xmm7, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm9 = xmm6[1,1,3,3] -; SSE41-NEXT: por %xmm5, %xmm9 -; SSE41-NEXT: movdqa %xmm1, %xmm5 -; SSE41-NEXT: pxor %xmm0, %xmm5 -; SSE41-NEXT: movdqa %xmm4, %xmm6 -; SSE41-NEXT: pcmpgtd %xmm5, %xmm6 -; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm6[0,0,2,2] -; SSE41-NEXT: pcmpeqd %xmm4, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3] -; SSE41-NEXT: pand %xmm7, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm10 = xmm6[1,1,3,3] -; SSE41-NEXT: por %xmm5, %xmm10 -; SSE41-NEXT: movdqa %xmm2, %xmm5 -; SSE41-NEXT: pxor %xmm0, %xmm5 -; SSE41-NEXT: movdqa %xmm4, %xmm7 -; SSE41-NEXT: pcmpgtd %xmm5, %xmm7 -; SSE41-NEXT: pshufd {{.*#+}} xmm6 = xmm7[0,0,2,2] -; SSE41-NEXT: pcmpeqd %xmm4, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3] -; SSE41-NEXT: pand %xmm6, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm7[1,1,3,3] -; SSE41-NEXT: por %xmm5, %xmm7 -; SSE41-NEXT: pxor %xmm3, %xmm0 -; SSE41-NEXT: movdqa %xmm4, %xmm5 -; SSE41-NEXT: pcmpgtd %xmm0, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2] -; SSE41-NEXT: pcmpeqd %xmm4, %xmm0 +; SSE41-NEXT: movapd {{.*#+}} xmm6 = [4294967295,4294967295] +; SSE41-NEXT: movdqa {{.*#+}} xmm7 = [2147483648,2147483648,2147483648,2147483648] +; SSE41-NEXT: movdqa %xmm3, %xmm0 +; SSE41-NEXT: pxor %xmm7, %xmm0 +; SSE41-NEXT: movdqa {{.*#+}} xmm9 = [9223372039002259455,9223372039002259455] +; SSE41-NEXT: movdqa %xmm9, %xmm4 +; SSE41-NEXT: pcmpgtd %xmm0, %xmm4 +; SSE41-NEXT: pshufd {{.*#+}} xmm10 = xmm4[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm9, %xmm0 +; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm0[1,1,3,3] +; SSE41-NEXT: pand %xmm10, %xmm5 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm4[1,1,3,3] +; SSE41-NEXT: por %xmm5, %xmm0 +; SSE41-NEXT: movapd %xmm6, %xmm5 +; SSE41-NEXT: blendvpd %xmm0, %xmm3, %xmm5 +; SSE41-NEXT: movdqa %xmm2, %xmm0 +; SSE41-NEXT: pxor %xmm7, %xmm0 +; SSE41-NEXT: movdqa %xmm9, %xmm3 +; SSE41-NEXT: pcmpgtd %xmm0, %xmm3 +; SSE41-NEXT: pshufd {{.*#+}} xmm10 = xmm3[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm9, %xmm0 ; SSE41-NEXT: pshufd {{.*#+}} xmm4 = xmm0[1,1,3,3] -; SSE41-NEXT: pand %xmm6, %xmm4 -; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm5[1,1,3,3] +; SSE41-NEXT: pand %xmm10, %xmm4 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm3[1,1,3,3] ; SSE41-NEXT: por %xmm4, %xmm0 -; SSE41-NEXT: movapd {{.*#+}} xmm4 = [4294967295,4294967295] -; SSE41-NEXT: movapd %xmm4, %xmm5 -; SSE41-NEXT: blendvpd %xmm0, %xmm3, %xmm5 -; SSE41-NEXT: movapd %xmm4, %xmm3 -; SSE41-NEXT: movdqa %xmm7, %xmm0 -; SSE41-NEXT: blendvpd %xmm0, %xmm2, %xmm3 -; SSE41-NEXT: shufps {{.*#+}} xmm3 = xmm3[0,2],xmm5[0,2] -; SSE41-NEXT: movapd %xmm4, %xmm2 -; SSE41-NEXT: movdqa %xmm10, %xmm0 +; SSE41-NEXT: movapd %xmm6, %xmm4 +; SSE41-NEXT: blendvpd %xmm0, %xmm2, %xmm4 +; SSE41-NEXT: shufps {{.*#+}} xmm4 = xmm4[0,2],xmm5[0,2] +; SSE41-NEXT: movdqa %xmm1, %xmm0 +; SSE41-NEXT: pxor %xmm7, %xmm0 +; SSE41-NEXT: movdqa %xmm9, %xmm2 +; SSE41-NEXT: pcmpgtd %xmm0, %xmm2 +; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm9, %xmm0 +; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm0[1,1,3,3] +; SSE41-NEXT: pand %xmm3, %xmm5 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm2[1,1,3,3] +; SSE41-NEXT: por %xmm5, %xmm0 +; SSE41-NEXT: movapd %xmm6, %xmm2 ; SSE41-NEXT: blendvpd %xmm0, %xmm1, %xmm2 +; SSE41-NEXT: pxor %xmm8, %xmm7 ; SSE41-NEXT: movdqa %xmm9, %xmm0 -; SSE41-NEXT: blendvpd %xmm0, %xmm8, %xmm4 -; SSE41-NEXT: shufps {{.*#+}} xmm4 = xmm4[0,2],xmm2[0,2] -; SSE41-NEXT: movaps %xmm4, %xmm0 -; SSE41-NEXT: movaps %xmm3, %xmm1 +; SSE41-NEXT: pcmpgtd %xmm7, %xmm0 +; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm9, %xmm7 +; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm7[1,1,3,3] +; SSE41-NEXT: pand %xmm1, %xmm3 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] +; SSE41-NEXT: por %xmm3, %xmm0 +; SSE41-NEXT: blendvpd %xmm0, %xmm8, %xmm6 +; SSE41-NEXT: shufps {{.*#+}} xmm6 = xmm6[0,2],xmm2[0,2] +; SSE41-NEXT: movaps %xmm6, %xmm0 +; SSE41-NEXT: movaps %xmm4, %xmm1 ; SSE41-NEXT: retq ; ; AVX1-LABEL: trunc_usat_v8i64_v8i32: ; AVX1: # %bb.0: -; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm2 -; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [9223372036854775808,9223372036854775808] -; AVX1-NEXT: vpxor %xmm3, %xmm2, %xmm2 -; AVX1-NEXT: vmovdqa {{.*#+}} xmm4 = [9223372041149743103,9223372041149743103] -; AVX1-NEXT: vpcmpgtq %xmm2, %xmm4, %xmm2 -; AVX1-NEXT: vpxor %xmm3, %xmm1, %xmm5 -; AVX1-NEXT: vpcmpgtq %xmm5, %xmm4, %xmm5 -; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm5, %ymm2 -; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm5 -; AVX1-NEXT: vpxor %xmm3, %xmm5, %xmm5 -; AVX1-NEXT: vpcmpgtq %xmm5, %xmm4, %xmm5 -; AVX1-NEXT: vpxor %xmm3, %xmm0, %xmm3 -; AVX1-NEXT: vpcmpgtq %xmm3, %xmm4, %xmm3 -; AVX1-NEXT: vinsertf128 $1, %xmm5, %ymm3, %ymm3 -; AVX1-NEXT: vmovapd {{.*#+}} ymm4 = [4294967295,4294967295,4294967295,4294967295] -; AVX1-NEXT: vblendvpd %ymm3, %ymm0, %ymm4, %ymm0 -; AVX1-NEXT: vblendvpd %ymm2, %ymm1, %ymm4, %ymm1 +; AVX1-NEXT: vmovapd {{.*#+}} ymm2 = [4294967295,4294967295,4294967295,4294967295] +; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm3 +; AVX1-NEXT: vmovdqa {{.*#+}} xmm4 = [9223372036854775808,9223372036854775808] +; AVX1-NEXT: vpxor %xmm4, %xmm3, %xmm3 +; AVX1-NEXT: vmovdqa {{.*#+}} xmm5 = [9223372041149743103,9223372041149743103] +; AVX1-NEXT: vpcmpgtq %xmm3, %xmm5, %xmm3 +; AVX1-NEXT: vpxor %xmm4, %xmm0, %xmm6 +; AVX1-NEXT: vpcmpgtq %xmm6, %xmm5, %xmm6 +; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm6, %ymm3 +; AVX1-NEXT: vblendvpd %ymm3, %ymm0, %ymm2, %ymm0 +; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm3 +; AVX1-NEXT: vpxor %xmm4, %xmm3, %xmm3 +; AVX1-NEXT: vpcmpgtq %xmm3, %xmm5, %xmm3 +; AVX1-NEXT: vpxor %xmm4, %xmm1, %xmm4 +; AVX1-NEXT: vpcmpgtq %xmm4, %xmm5, %xmm4 +; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm4, %ymm3 +; AVX1-NEXT: vblendvpd %ymm3, %ymm1, %ymm2, %ymm1 ; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm2 ; AVX1-NEXT: vshufps {{.*#+}} xmm1 = xmm1[0,2],xmm2[0,2] ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2 @@ -416,15 +413,15 @@ define <8 x i32> @trunc_usat_v8i64_v8i32(<8 x i64> %a0) { ; ; AVX2-SLOW-LABEL: trunc_usat_v8i64_v8i32: ; AVX2-SLOW: # %bb.0: -; AVX2-SLOW-NEXT: vpbroadcastq {{.*#+}} ymm2 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808] -; AVX2-SLOW-NEXT: vpxor %ymm2, %ymm0, %ymm3 -; AVX2-SLOW-NEXT: vpbroadcastq {{.*#+}} ymm4 = [9223372041149743103,9223372041149743103,9223372041149743103,9223372041149743103] -; AVX2-SLOW-NEXT: vpcmpgtq %ymm3, %ymm4, %ymm3 -; AVX2-SLOW-NEXT: vpxor %ymm2, %ymm1, %ymm2 -; AVX2-SLOW-NEXT: vpcmpgtq %ymm2, %ymm4, %ymm2 -; AVX2-SLOW-NEXT: vbroadcastsd {{.*#+}} ymm4 = [4294967295,4294967295,4294967295,4294967295] -; AVX2-SLOW-NEXT: vblendvpd %ymm2, %ymm1, %ymm4, %ymm1 -; AVX2-SLOW-NEXT: vblendvpd %ymm3, %ymm0, %ymm4, %ymm0 +; AVX2-SLOW-NEXT: vbroadcastsd {{.*#+}} ymm2 = [4294967295,4294967295,4294967295,4294967295] +; AVX2-SLOW-NEXT: vpbroadcastq {{.*#+}} ymm3 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808] +; AVX2-SLOW-NEXT: vpxor %ymm3, %ymm1, %ymm4 +; AVX2-SLOW-NEXT: vpbroadcastq {{.*#+}} ymm5 = [9223372041149743103,9223372041149743103,9223372041149743103,9223372041149743103] +; AVX2-SLOW-NEXT: vpcmpgtq %ymm4, %ymm5, %ymm4 +; AVX2-SLOW-NEXT: vblendvpd %ymm4, %ymm1, %ymm2, %ymm1 +; AVX2-SLOW-NEXT: vpxor %ymm3, %ymm0, %ymm3 +; AVX2-SLOW-NEXT: vpcmpgtq %ymm3, %ymm5, %ymm3 +; AVX2-SLOW-NEXT: vblendvpd %ymm3, %ymm0, %ymm2, %ymm0 ; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7] ; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,3] ; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7] @@ -434,15 +431,15 @@ define <8 x i32> @trunc_usat_v8i64_v8i32(<8 x i64> %a0) { ; ; AVX2-FAST-LABEL: trunc_usat_v8i64_v8i32: ; AVX2-FAST: # %bb.0: -; AVX2-FAST-NEXT: vpbroadcastq {{.*#+}} ymm2 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808] -; AVX2-FAST-NEXT: vpxor %ymm2, %ymm0, %ymm3 -; AVX2-FAST-NEXT: vpbroadcastq {{.*#+}} ymm4 = [9223372041149743103,9223372041149743103,9223372041149743103,9223372041149743103] -; AVX2-FAST-NEXT: vpcmpgtq %ymm3, %ymm4, %ymm3 -; AVX2-FAST-NEXT: vpxor %ymm2, %ymm1, %ymm2 -; AVX2-FAST-NEXT: vpcmpgtq %ymm2, %ymm4, %ymm2 -; AVX2-FAST-NEXT: vbroadcastsd {{.*#+}} ymm4 = [4294967295,4294967295,4294967295,4294967295] -; AVX2-FAST-NEXT: vblendvpd %ymm2, %ymm1, %ymm4, %ymm1 -; AVX2-FAST-NEXT: vblendvpd %ymm3, %ymm0, %ymm4, %ymm0 +; AVX2-FAST-NEXT: vbroadcastsd {{.*#+}} ymm2 = [4294967295,4294967295,4294967295,4294967295] +; AVX2-FAST-NEXT: vpbroadcastq {{.*#+}} ymm3 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808] +; AVX2-FAST-NEXT: vpxor %ymm3, %ymm1, %ymm4 +; AVX2-FAST-NEXT: vpbroadcastq {{.*#+}} ymm5 = [9223372041149743103,9223372041149743103,9223372041149743103,9223372041149743103] +; AVX2-FAST-NEXT: vpcmpgtq %ymm4, %ymm5, %ymm4 +; AVX2-FAST-NEXT: vblendvpd %ymm4, %ymm1, %ymm2, %ymm1 +; AVX2-FAST-NEXT: vpxor %ymm3, %ymm0, %ymm3 +; AVX2-FAST-NEXT: vpcmpgtq %ymm3, %ymm5, %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 @@ -466,68 +463,68 @@ define <8 x i32> @trunc_usat_v8i64_v8i32(<8 x i64> %a0) { define <8 x i16> @trunc_usat_v8i64_v8i16(<8 x i64> %a0) { ; SSE2-LABEL: trunc_usat_v8i64_v8i16: ; SSE2: # %bb.0: +; SSE2-NEXT: movdqa {{.*#+}} xmm8 = [65535,65535] ; SSE2-NEXT: movdqa {{.*#+}} xmm6 = [2147483648,2147483648,2147483648,2147483648] -; SSE2-NEXT: movdqa %xmm1, %xmm4 -; SSE2-NEXT: pxor %xmm6, %xmm4 -; SSE2-NEXT: movdqa {{.*#+}} xmm8 = [9223372039002324991,9223372039002324991] -; SSE2-NEXT: movdqa %xmm8, %xmm5 -; SSE2-NEXT: pcmpgtd %xmm4, %xmm5 -; SSE2-NEXT: pshufd {{.*#+}} xmm9 = xmm5[0,0,2,2] -; SSE2-NEXT: pcmpeqd %xmm8, %xmm4 -; SSE2-NEXT: pshufd {{.*#+}} xmm7 = xmm4[1,1,3,3] -; SSE2-NEXT: pand %xmm9, %xmm7 -; SSE2-NEXT: pshufd {{.*#+}} xmm10 = xmm5[1,1,3,3] -; SSE2-NEXT: por %xmm7, %xmm10 -; SSE2-NEXT: movdqa %xmm0, %xmm5 +; SSE2-NEXT: movdqa %xmm2, %xmm5 ; SSE2-NEXT: pxor %xmm6, %xmm5 -; SSE2-NEXT: movdqa %xmm8, %xmm7 +; SSE2-NEXT: movdqa {{.*#+}} xmm9 = [9223372039002324991,9223372039002324991] +; SSE2-NEXT: movdqa %xmm9, %xmm7 ; SSE2-NEXT: pcmpgtd %xmm5, %xmm7 -; SSE2-NEXT: pshufd {{.*#+}} xmm9 = xmm7[0,0,2,2] -; SSE2-NEXT: pcmpeqd %xmm8, %xmm5 +; SSE2-NEXT: pshufd {{.*#+}} xmm10 = xmm7[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm9, %xmm5 ; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm5[1,1,3,3] -; SSE2-NEXT: pand %xmm9, %xmm4 -; SSE2-NEXT: pshufd {{.*#+}} xmm11 = xmm7[1,1,3,3] -; SSE2-NEXT: por %xmm4, %xmm11 -; SSE2-NEXT: movdqa %xmm3, %xmm4 -; SSE2-NEXT: pxor %xmm6, %xmm4 -; SSE2-NEXT: movdqa %xmm8, %xmm7 -; SSE2-NEXT: pcmpgtd %xmm4, %xmm7 -; SSE2-NEXT: pshufd {{.*#+}} xmm9 = xmm7[0,0,2,2] -; SSE2-NEXT: pcmpeqd %xmm8, %xmm4 -; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3] -; SSE2-NEXT: pand %xmm9, %xmm4 -; SSE2-NEXT: pshufd {{.*#+}} xmm7 = xmm7[1,1,3,3] -; SSE2-NEXT: por %xmm4, %xmm7 -; SSE2-NEXT: pxor %xmm2, %xmm6 -; SSE2-NEXT: movdqa %xmm8, %xmm4 -; SSE2-NEXT: pcmpgtd %xmm6, %xmm4 -; SSE2-NEXT: pshufd {{.*#+}} xmm9 = xmm4[0,0,2,2] -; SSE2-NEXT: pcmpeqd %xmm8, %xmm6 -; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm6[1,1,3,3] -; SSE2-NEXT: pand %xmm9, %xmm5 -; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm4[1,1,3,3] -; SSE2-NEXT: por %xmm5, %xmm6 -; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [65535,65535] -; SSE2-NEXT: pand %xmm6, %xmm2 -; SSE2-NEXT: pandn %xmm4, %xmm6 -; SSE2-NEXT: por %xmm2, %xmm6 +; SSE2-NEXT: pand %xmm10, %xmm4 +; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm7[1,1,3,3] +; SSE2-NEXT: por %xmm4, %xmm5 +; SSE2-NEXT: pand %xmm5, %xmm2 +; SSE2-NEXT: pandn %xmm8, %xmm5 +; SSE2-NEXT: por %xmm2, %xmm5 +; SSE2-NEXT: movdqa %xmm3, %xmm2 +; SSE2-NEXT: pxor %xmm6, %xmm2 +; SSE2-NEXT: movdqa %xmm9, %xmm4 +; SSE2-NEXT: pcmpgtd %xmm2, %xmm4 +; SSE2-NEXT: pshufd {{.*#+}} xmm10 = xmm4[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm9, %xmm2 +; SSE2-NEXT: pshufd {{.*#+}} xmm7 = xmm2[1,1,3,3] +; SSE2-NEXT: pand %xmm10, %xmm7 +; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm4[1,1,3,3] +; SSE2-NEXT: por %xmm7, %xmm2 +; SSE2-NEXT: pand %xmm2, %xmm3 +; SSE2-NEXT: pandn %xmm8, %xmm2 +; SSE2-NEXT: por %xmm3, %xmm2 +; SSE2-NEXT: movdqa %xmm0, %xmm3 +; SSE2-NEXT: pxor %xmm6, %xmm3 +; SSE2-NEXT: movdqa %xmm9, %xmm4 +; SSE2-NEXT: pcmpgtd %xmm3, %xmm4 +; SSE2-NEXT: pshufd {{.*#+}} xmm7 = xmm4[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm9, %xmm3 +; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3] ; SSE2-NEXT: pand %xmm7, %xmm3 -; SSE2-NEXT: pandn %xmm4, %xmm7 -; SSE2-NEXT: por %xmm3, %xmm7 -; SSE2-NEXT: pand %xmm11, %xmm0 -; SSE2-NEXT: pandn %xmm4, %xmm11 -; SSE2-NEXT: por %xmm0, %xmm11 -; SSE2-NEXT: pand %xmm10, %xmm1 -; SSE2-NEXT: pandn %xmm4, %xmm10 -; SSE2-NEXT: por %xmm1, %xmm10 -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm10[0,2,2,3] +; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3] +; SSE2-NEXT: por %xmm3, %xmm4 +; SSE2-NEXT: pand %xmm4, %xmm0 +; SSE2-NEXT: pandn %xmm8, %xmm4 +; SSE2-NEXT: por %xmm0, %xmm4 +; SSE2-NEXT: pxor %xmm1, %xmm6 +; SSE2-NEXT: movdqa %xmm9, %xmm0 +; SSE2-NEXT: pcmpgtd %xmm6, %xmm0 +; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm0[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm9, %xmm6 +; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm6[1,1,3,3] +; SSE2-NEXT: pand %xmm3, %xmm6 +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] +; SSE2-NEXT: por %xmm6, %xmm0 +; SSE2-NEXT: pand %xmm0, %xmm1 +; SSE2-NEXT: pandn %xmm8, %xmm0 +; SSE2-NEXT: por %xmm1, %xmm0 +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7] -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm11[0,2,2,3] +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm4[0,2,2,3] ; SSE2-NEXT: pshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7] ; SSE2-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm7[0,2,2,3] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3] ; SSE2-NEXT: pshuflw {{.*#+}} xmm2 = xmm0[0,1,0,2,4,5,6,7] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm6[0,2,2,3] +; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm5[0,2,2,3] ; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,1,0,2,4,5,6,7] ; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1] ; SSE2-NEXT: movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1] @@ -535,68 +532,68 @@ define <8 x i16> @trunc_usat_v8i64_v8i16(<8 x i64> %a0) { ; ; SSSE3-LABEL: trunc_usat_v8i64_v8i16: ; SSSE3: # %bb.0: +; SSSE3-NEXT: movdqa {{.*#+}} xmm8 = [65535,65535] ; SSSE3-NEXT: movdqa {{.*#+}} xmm6 = [2147483648,2147483648,2147483648,2147483648] -; SSSE3-NEXT: movdqa %xmm1, %xmm4 -; SSSE3-NEXT: pxor %xmm6, %xmm4 -; SSSE3-NEXT: movdqa {{.*#+}} xmm8 = [9223372039002324991,9223372039002324991] -; SSSE3-NEXT: movdqa %xmm8, %xmm5 -; SSSE3-NEXT: pcmpgtd %xmm4, %xmm5 -; SSSE3-NEXT: pshufd {{.*#+}} xmm9 = xmm5[0,0,2,2] -; SSSE3-NEXT: pcmpeqd %xmm8, %xmm4 -; SSSE3-NEXT: pshufd {{.*#+}} xmm7 = xmm4[1,1,3,3] -; SSSE3-NEXT: pand %xmm9, %xmm7 -; SSSE3-NEXT: pshufd {{.*#+}} xmm10 = xmm5[1,1,3,3] -; SSSE3-NEXT: por %xmm7, %xmm10 -; SSSE3-NEXT: movdqa %xmm0, %xmm5 +; SSSE3-NEXT: movdqa %xmm2, %xmm5 ; SSSE3-NEXT: pxor %xmm6, %xmm5 -; SSSE3-NEXT: movdqa %xmm8, %xmm7 +; SSSE3-NEXT: movdqa {{.*#+}} xmm9 = [9223372039002324991,9223372039002324991] +; SSSE3-NEXT: movdqa %xmm9, %xmm7 ; SSSE3-NEXT: pcmpgtd %xmm5, %xmm7 -; SSSE3-NEXT: pshufd {{.*#+}} xmm9 = xmm7[0,0,2,2] -; SSSE3-NEXT: pcmpeqd %xmm8, %xmm5 +; SSSE3-NEXT: pshufd {{.*#+}} xmm10 = xmm7[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm9, %xmm5 ; SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm5[1,1,3,3] -; SSSE3-NEXT: pand %xmm9, %xmm4 -; SSSE3-NEXT: pshufd {{.*#+}} xmm11 = xmm7[1,1,3,3] -; SSSE3-NEXT: por %xmm4, %xmm11 -; SSSE3-NEXT: movdqa %xmm3, %xmm4 -; SSSE3-NEXT: pxor %xmm6, %xmm4 -; SSSE3-NEXT: movdqa %xmm8, %xmm7 -; SSSE3-NEXT: pcmpgtd %xmm4, %xmm7 -; SSSE3-NEXT: pshufd {{.*#+}} xmm9 = xmm7[0,0,2,2] -; SSSE3-NEXT: pcmpeqd %xmm8, %xmm4 -; SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3] -; SSSE3-NEXT: pand %xmm9, %xmm4 -; SSSE3-NEXT: pshufd {{.*#+}} xmm7 = xmm7[1,1,3,3] -; SSSE3-NEXT: por %xmm4, %xmm7 -; SSSE3-NEXT: pxor %xmm2, %xmm6 -; SSSE3-NEXT: movdqa %xmm8, %xmm4 -; SSSE3-NEXT: pcmpgtd %xmm6, %xmm4 -; SSSE3-NEXT: pshufd {{.*#+}} xmm9 = xmm4[0,0,2,2] -; SSSE3-NEXT: pcmpeqd %xmm8, %xmm6 -; SSSE3-NEXT: pshufd {{.*#+}} xmm5 = xmm6[1,1,3,3] -; SSSE3-NEXT: pand %xmm9, %xmm5 -; SSSE3-NEXT: pshufd {{.*#+}} xmm6 = xmm4[1,1,3,3] -; SSSE3-NEXT: por %xmm5, %xmm6 -; SSSE3-NEXT: movdqa {{.*#+}} xmm4 = [65535,65535] -; SSSE3-NEXT: pand %xmm6, %xmm2 -; SSSE3-NEXT: pandn %xmm4, %xmm6 -; SSSE3-NEXT: por %xmm2, %xmm6 +; SSSE3-NEXT: pand %xmm10, %xmm4 +; SSSE3-NEXT: pshufd {{.*#+}} xmm5 = xmm7[1,1,3,3] +; SSSE3-NEXT: por %xmm4, %xmm5 +; SSSE3-NEXT: pand %xmm5, %xmm2 +; SSSE3-NEXT: pandn %xmm8, %xmm5 +; SSSE3-NEXT: por %xmm2, %xmm5 +; SSSE3-NEXT: movdqa %xmm3, %xmm2 +; SSSE3-NEXT: pxor %xmm6, %xmm2 +; SSSE3-NEXT: movdqa %xmm9, %xmm4 +; SSSE3-NEXT: pcmpgtd %xmm2, %xmm4 +; SSSE3-NEXT: pshufd {{.*#+}} xmm10 = xmm4[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm9, %xmm2 +; SSSE3-NEXT: pshufd {{.*#+}} xmm7 = xmm2[1,1,3,3] +; SSSE3-NEXT: pand %xmm10, %xmm7 +; SSSE3-NEXT: pshufd {{.*#+}} xmm2 = xmm4[1,1,3,3] +; SSSE3-NEXT: por %xmm7, %xmm2 +; SSSE3-NEXT: pand %xmm2, %xmm3 +; SSSE3-NEXT: pandn %xmm8, %xmm2 +; SSSE3-NEXT: por %xmm3, %xmm2 +; SSSE3-NEXT: movdqa %xmm0, %xmm3 +; SSSE3-NEXT: pxor %xmm6, %xmm3 +; SSSE3-NEXT: movdqa %xmm9, %xmm4 +; SSSE3-NEXT: pcmpgtd %xmm3, %xmm4 +; SSSE3-NEXT: pshufd {{.*#+}} xmm7 = xmm4[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm9, %xmm3 +; SSSE3-NEXT: pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3] ; SSSE3-NEXT: pand %xmm7, %xmm3 -; SSSE3-NEXT: pandn %xmm4, %xmm7 -; SSSE3-NEXT: por %xmm3, %xmm7 -; SSSE3-NEXT: pand %xmm11, %xmm0 -; SSSE3-NEXT: pandn %xmm4, %xmm11 -; SSSE3-NEXT: por %xmm0, %xmm11 -; SSSE3-NEXT: pand %xmm10, %xmm1 -; SSSE3-NEXT: pandn %xmm4, %xmm10 -; SSSE3-NEXT: por %xmm1, %xmm10 -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm10[0,2,2,3] +; SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3] +; SSSE3-NEXT: por %xmm3, %xmm4 +; SSSE3-NEXT: pand %xmm4, %xmm0 +; SSSE3-NEXT: pandn %xmm8, %xmm4 +; SSSE3-NEXT: por %xmm0, %xmm4 +; SSSE3-NEXT: pxor %xmm1, %xmm6 +; SSSE3-NEXT: movdqa %xmm9, %xmm0 +; SSSE3-NEXT: pcmpgtd %xmm6, %xmm0 +; SSSE3-NEXT: pshufd {{.*#+}} xmm3 = xmm0[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm9, %xmm6 +; SSSE3-NEXT: pshufd {{.*#+}} xmm6 = xmm6[1,1,3,3] +; SSSE3-NEXT: pand %xmm3, %xmm6 +; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] +; SSSE3-NEXT: por %xmm6, %xmm0 +; SSSE3-NEXT: pand %xmm0, %xmm1 +; SSSE3-NEXT: pandn %xmm8, %xmm0 +; SSSE3-NEXT: por %xmm1, %xmm0 +; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] ; SSSE3-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7] -; SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm11[0,2,2,3] +; SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm4[0,2,2,3] ; SSSE3-NEXT: pshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7] ; SSSE3-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1] -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm7[0,2,2,3] +; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3] ; SSSE3-NEXT: pshuflw {{.*#+}} xmm2 = xmm0[0,1,0,2,4,5,6,7] -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm6[0,2,2,3] +; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm5[0,2,2,3] ; SSSE3-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,1,0,2,4,5,6,7] ; SSSE3-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1] ; SSSE3-NEXT: movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1] @@ -605,143 +602,105 @@ define <8 x i16> @trunc_usat_v8i64_v8i16(<8 x i64> %a0) { ; SSE41-LABEL: trunc_usat_v8i64_v8i16: ; SSE41: # %bb.0: ; SSE41-NEXT: movdqa %xmm0, %xmm8 -; SSE41-NEXT: movdqa {{.*#+}} xmm0 = [2147483648,2147483648,2147483648,2147483648] -; SSE41-NEXT: movdqa %xmm3, %xmm5 -; SSE41-NEXT: pxor %xmm0, %xmm5 -; SSE41-NEXT: movdqa {{.*#+}} xmm4 = [9223372039002324991,9223372039002324991] -; SSE41-NEXT: movdqa %xmm4, %xmm6 -; SSE41-NEXT: pcmpgtd %xmm5, %xmm6 -; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm6[0,0,2,2] -; SSE41-NEXT: pcmpeqd %xmm4, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3] -; SSE41-NEXT: pand %xmm7, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm9 = xmm6[1,1,3,3] -; SSE41-NEXT: por %xmm5, %xmm9 -; SSE41-NEXT: movdqa %xmm2, %xmm5 -; SSE41-NEXT: pxor %xmm0, %xmm5 -; SSE41-NEXT: movdqa %xmm4, %xmm6 -; SSE41-NEXT: pcmpgtd %xmm5, %xmm6 -; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm6[0,0,2,2] -; SSE41-NEXT: pcmpeqd %xmm4, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3] -; SSE41-NEXT: pand %xmm7, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm10 = xmm6[1,1,3,3] -; SSE41-NEXT: por %xmm5, %xmm10 -; SSE41-NEXT: movdqa %xmm1, %xmm5 -; SSE41-NEXT: pxor %xmm0, %xmm5 -; SSE41-NEXT: movdqa %xmm4, %xmm7 -; SSE41-NEXT: pcmpgtd %xmm5, %xmm7 -; SSE41-NEXT: pshufd {{.*#+}} xmm6 = xmm7[0,0,2,2] -; SSE41-NEXT: pcmpeqd %xmm4, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3] -; SSE41-NEXT: pand %xmm6, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm7[1,1,3,3] -; SSE41-NEXT: por %xmm5, %xmm7 -; SSE41-NEXT: pxor %xmm8, %xmm0 -; SSE41-NEXT: movdqa %xmm4, %xmm5 -; SSE41-NEXT: pcmpgtd %xmm0, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2] -; SSE41-NEXT: pcmpeqd %xmm4, %xmm0 +; SSE41-NEXT: movapd {{.*#+}} xmm5 = [65535,65535] +; SSE41-NEXT: movdqa {{.*#+}} xmm6 = [2147483648,2147483648,2147483648,2147483648] +; SSE41-NEXT: movdqa %xmm1, %xmm0 +; SSE41-NEXT: pxor %xmm6, %xmm0 +; SSE41-NEXT: movdqa {{.*#+}} xmm9 = [9223372039002324991,9223372039002324991] +; SSE41-NEXT: movdqa %xmm9, %xmm7 +; SSE41-NEXT: pcmpgtd %xmm0, %xmm7 +; SSE41-NEXT: pshufd {{.*#+}} xmm10 = xmm7[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm9, %xmm0 ; SSE41-NEXT: pshufd {{.*#+}} xmm4 = xmm0[1,1,3,3] -; SSE41-NEXT: pand %xmm6, %xmm4 -; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm5[1,1,3,3] +; SSE41-NEXT: pand %xmm10, %xmm4 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm7[1,1,3,3] ; SSE41-NEXT: por %xmm4, %xmm0 -; SSE41-NEXT: movapd {{.*#+}} xmm5 = [65535,65535] ; SSE41-NEXT: movapd %xmm5, %xmm4 -; SSE41-NEXT: blendvpd %xmm0, %xmm8, %xmm4 -; SSE41-NEXT: movapd %xmm5, %xmm6 -; SSE41-NEXT: movdqa %xmm7, %xmm0 -; SSE41-NEXT: blendvpd %xmm0, %xmm1, %xmm6 +; SSE41-NEXT: blendvpd %xmm0, %xmm1, %xmm4 +; SSE41-NEXT: movdqa %xmm8, %xmm0 +; SSE41-NEXT: pxor %xmm6, %xmm0 +; SSE41-NEXT: movdqa %xmm9, %xmm1 +; SSE41-NEXT: pcmpgtd %xmm0, %xmm1 +; SSE41-NEXT: pshufd {{.*#+}} xmm10 = xmm1[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm9, %xmm0 +; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm0[1,1,3,3] +; SSE41-NEXT: pand %xmm10, %xmm7 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3] +; SSE41-NEXT: por %xmm7, %xmm0 ; SSE41-NEXT: movapd %xmm5, %xmm1 -; SSE41-NEXT: movdqa %xmm10, %xmm0 -; SSE41-NEXT: blendvpd %xmm0, %xmm2, %xmm1 +; SSE41-NEXT: blendvpd %xmm0, %xmm8, %xmm1 +; SSE41-NEXT: packusdw %xmm4, %xmm1 +; SSE41-NEXT: movdqa %xmm3, %xmm0 +; SSE41-NEXT: pxor %xmm6, %xmm0 +; SSE41-NEXT: movdqa %xmm9, %xmm4 +; SSE41-NEXT: pcmpgtd %xmm0, %xmm4 +; SSE41-NEXT: pshufd {{.*#+}} xmm8 = xmm4[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm9, %xmm0 +; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm0[1,1,3,3] +; SSE41-NEXT: pand %xmm8, %xmm7 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm4[1,1,3,3] +; SSE41-NEXT: por %xmm7, %xmm0 +; SSE41-NEXT: movapd %xmm5, %xmm4 +; SSE41-NEXT: blendvpd %xmm0, %xmm3, %xmm4 +; SSE41-NEXT: pxor %xmm2, %xmm6 ; SSE41-NEXT: movdqa %xmm9, %xmm0 -; SSE41-NEXT: blendvpd %xmm0, %xmm3, %xmm5 -; SSE41-NEXT: pxor %xmm0, %xmm0 -; SSE41-NEXT: pblendw {{.*#+}} xmm5 = xmm5[0],xmm0[1,2,3],xmm5[4],xmm0[5,6,7] -; SSE41-NEXT: pblendw {{.*#+}} xmm1 = xmm1[0],xmm0[1,2,3],xmm1[4],xmm0[5,6,7] +; SSE41-NEXT: pcmpgtd %xmm6, %xmm0 +; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm0[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm9, %xmm6 +; SSE41-NEXT: pshufd {{.*#+}} xmm6 = xmm6[1,1,3,3] +; SSE41-NEXT: pand %xmm3, %xmm6 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] +; SSE41-NEXT: por %xmm6, %xmm0 +; SSE41-NEXT: blendvpd %xmm0, %xmm2, %xmm5 +; SSE41-NEXT: packusdw %xmm4, %xmm5 ; SSE41-NEXT: packusdw %xmm5, %xmm1 -; SSE41-NEXT: pblendw {{.*#+}} xmm6 = xmm6[0],xmm0[1,2,3],xmm6[4],xmm0[5,6,7] -; SSE41-NEXT: pblendw {{.*#+}} xmm4 = xmm4[0],xmm0[1,2,3],xmm4[4],xmm0[5,6,7] -; SSE41-NEXT: packusdw %xmm6, %xmm4 -; SSE41-NEXT: packusdw %xmm1, %xmm4 -; SSE41-NEXT: movdqa %xmm4, %xmm0 +; SSE41-NEXT: movdqa %xmm1, %xmm0 ; SSE41-NEXT: retq ; ; AVX1-LABEL: trunc_usat_v8i64_v8i16: ; AVX1: # %bb.0: +; AVX1-NEXT: vmovapd {{.*#+}} ymm2 = [65535,65535,65535,65535] +; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm3 +; AVX1-NEXT: vmovdqa {{.*#+}} xmm4 = [9223372036854775808,9223372036854775808] +; AVX1-NEXT: vpxor %xmm4, %xmm3, %xmm3 +; AVX1-NEXT: vmovdqa {{.*#+}} xmm5 = [9223372036854841343,9223372036854841343] +; AVX1-NEXT: vpcmpgtq %xmm3, %xmm5, %xmm3 +; AVX1-NEXT: vpxor %xmm4, %xmm0, %xmm6 +; AVX1-NEXT: vpcmpgtq %xmm6, %xmm5, %xmm6 +; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm6, %ymm3 +; AVX1-NEXT: vblendvpd %ymm3, %ymm0, %ymm2, %ymm0 +; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm3 +; AVX1-NEXT: vpxor %xmm4, %xmm3, %xmm3 +; AVX1-NEXT: vpcmpgtq %xmm3, %xmm5, %xmm3 +; AVX1-NEXT: vpxor %xmm4, %xmm1, %xmm4 +; AVX1-NEXT: vpcmpgtq %xmm4, %xmm5, %xmm4 +; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm4, %ymm3 +; AVX1-NEXT: vblendvpd %ymm3, %ymm1, %ymm2, %ymm1 ; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm2 -; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [9223372036854775808,9223372036854775808] -; AVX1-NEXT: vpxor %xmm3, %xmm2, %xmm2 -; AVX1-NEXT: vmovdqa {{.*#+}} xmm4 = [9223372036854841343,9223372036854841343] -; AVX1-NEXT: vpcmpgtq %xmm2, %xmm4, %xmm2 -; AVX1-NEXT: vpxor %xmm3, %xmm1, %xmm5 -; AVX1-NEXT: vpcmpgtq %xmm5, %xmm4, %xmm5 -; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm5, %ymm2 -; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm5 -; AVX1-NEXT: vpxor %xmm3, %xmm5, %xmm5 -; AVX1-NEXT: vpcmpgtq %xmm5, %xmm4, %xmm5 -; AVX1-NEXT: vpxor %xmm3, %xmm0, %xmm3 -; AVX1-NEXT: vpcmpgtq %xmm3, %xmm4, %xmm3 -; AVX1-NEXT: vinsertf128 $1, %xmm5, %ymm3, %ymm3 -; AVX1-NEXT: vmovapd {{.*#+}} ymm4 = [65535,65535,65535,65535] -; AVX1-NEXT: vblendvpd %ymm3, %ymm0, %ymm4, %ymm0 -; AVX1-NEXT: vblendvpd %ymm2, %ymm1, %ymm4, %ymm1 -; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm2 -; AVX1-NEXT: vxorpd %xmm3, %xmm3, %xmm3 -; AVX1-NEXT: vpblendw {{.*#+}} xmm2 = xmm2[0],xmm3[1,2,3],xmm2[4],xmm3[5,6,7] -; AVX1-NEXT: vpblendw {{.*#+}} xmm1 = xmm1[0],xmm3[1,2,3],xmm1[4],xmm3[5,6,7] ; AVX1-NEXT: vpackusdw %xmm2, %xmm1, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2 -; AVX1-NEXT: vpblendw {{.*#+}} xmm2 = xmm2[0],xmm3[1,2,3],xmm2[4],xmm3[5,6,7] -; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm3[1,2,3],xmm0[4],xmm3[5,6,7] ; AVX1-NEXT: vpackusdw %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: vpackusdw %xmm1, %xmm0, %xmm0 ; AVX1-NEXT: vzeroupper ; AVX1-NEXT: retq ; -; AVX2-SLOW-LABEL: trunc_usat_v8i64_v8i16: -; AVX2-SLOW: # %bb.0: -; AVX2-SLOW-NEXT: vpbroadcastq {{.*#+}} ymm2 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808] -; AVX2-SLOW-NEXT: vpxor %ymm2, %ymm0, %ymm3 -; AVX2-SLOW-NEXT: vpbroadcastq {{.*#+}} ymm4 = [9223372036854841343,9223372036854841343,9223372036854841343,9223372036854841343] -; AVX2-SLOW-NEXT: vpcmpgtq %ymm3, %ymm4, %ymm3 -; AVX2-SLOW-NEXT: vpxor %ymm2, %ymm1, %ymm2 -; AVX2-SLOW-NEXT: vpcmpgtq %ymm2, %ymm4, %ymm2 -; AVX2-SLOW-NEXT: vbroadcastsd {{.*#+}} ymm4 = [65535,65535,65535,65535] -; AVX2-SLOW-NEXT: vblendvpd %ymm2, %ymm1, %ymm4, %ymm1 -; AVX2-SLOW-NEXT: vblendvpd %ymm3, %ymm0, %ymm4, %ymm0 -; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7] -; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,3] -; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7] -; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,2,2,3] -; AVX2-SLOW-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 -; AVX2-SLOW-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31] -; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3] -; AVX2-SLOW-NEXT: # kill: def $xmm0 killed $xmm0 killed $ymm0 -; AVX2-SLOW-NEXT: vzeroupper -; AVX2-SLOW-NEXT: retq -; -; AVX2-FAST-LABEL: trunc_usat_v8i64_v8i16: -; AVX2-FAST: # %bb.0: -; AVX2-FAST-NEXT: vpbroadcastq {{.*#+}} ymm2 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808] -; AVX2-FAST-NEXT: vpxor %ymm2, %ymm0, %ymm3 -; AVX2-FAST-NEXT: vpbroadcastq {{.*#+}} ymm4 = [9223372036854841343,9223372036854841343,9223372036854841343,9223372036854841343] -; AVX2-FAST-NEXT: vpcmpgtq %ymm3, %ymm4, %ymm3 -; AVX2-FAST-NEXT: vpxor %ymm2, %ymm1, %ymm2 -; AVX2-FAST-NEXT: vpcmpgtq %ymm2, %ymm4, %ymm2 -; AVX2-FAST-NEXT: vbroadcastsd {{.*#+}} ymm4 = [65535,65535,65535,65535] -; AVX2-FAST-NEXT: vblendvpd %ymm2, %ymm1, %ymm4, %ymm1 -; AVX2-FAST-NEXT: vblendvpd %ymm3, %ymm0, %ymm4, %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: vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31] -; AVX2-FAST-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3] -; AVX2-FAST-NEXT: # kill: def $xmm0 killed $xmm0 killed $ymm0 -; AVX2-FAST-NEXT: vzeroupper -; AVX2-FAST-NEXT: retq +; AVX2-LABEL: trunc_usat_v8i64_v8i16: +; AVX2: # %bb.0: +; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm2 = [65535,65535,65535,65535] +; AVX2-NEXT: vpbroadcastq {{.*#+}} ymm3 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808] +; AVX2-NEXT: vpxor %ymm3, %ymm1, %ymm4 +; AVX2-NEXT: vpbroadcastq {{.*#+}} ymm5 = [9223372036854841343,9223372036854841343,9223372036854841343,9223372036854841343] +; AVX2-NEXT: vpcmpgtq %ymm4, %ymm5, %ymm4 +; AVX2-NEXT: vblendvpd %ymm4, %ymm1, %ymm2, %ymm1 +; AVX2-NEXT: vpxor %ymm3, %ymm0, %ymm3 +; AVX2-NEXT: vpcmpgtq %ymm3, %ymm5, %ymm3 +; AVX2-NEXT: vblendvpd %ymm3, %ymm0, %ymm2, %ymm0 +; AVX2-NEXT: vpackusdw %ymm1, %ymm0, %ymm0 +; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3] +; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 +; AVX2-NEXT: vpackusdw %xmm1, %xmm0, %xmm0 +; AVX2-NEXT: vzeroupper +; AVX2-NEXT: retq ; ; AVX512-LABEL: trunc_usat_v8i64_v8i16: ; AVX512: # %bb.0: @@ -757,18 +716,18 @@ define <8 x i16> @trunc_usat_v8i64_v8i16(<8 x i64> %a0) { define <8 x i16> @trunc_usat_v8i32_v8i16(<8 x i32> %a0) { ; SSE2-LABEL: trunc_usat_v8i32_v8i16: ; SSE2: # %bb.0: -; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [2147483648,2147483648,2147483648,2147483648] -; SSE2-NEXT: movdqa %xmm1, %xmm3 -; SSE2-NEXT: pxor %xmm2, %xmm3 -; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [2147549183,2147549183,2147549183,2147549183] -; SSE2-NEXT: movdqa %xmm4, %xmm5 -; SSE2-NEXT: pcmpgtd %xmm3, %xmm5 -; SSE2-NEXT: pxor %xmm0, %xmm2 -; SSE2-NEXT: pcmpgtd %xmm2, %xmm4 ; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [65535,65535,65535,65535] -; SSE2-NEXT: pand %xmm4, %xmm0 -; SSE2-NEXT: pandn %xmm2, %xmm4 -; SSE2-NEXT: por %xmm4, %xmm0 +; SSE2-NEXT: movdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648] +; SSE2-NEXT: movdqa %xmm0, %xmm4 +; SSE2-NEXT: pxor %xmm3, %xmm4 +; SSE2-NEXT: movdqa {{.*#+}} xmm5 = [2147549183,2147549183,2147549183,2147549183] +; SSE2-NEXT: movdqa %xmm5, %xmm6 +; SSE2-NEXT: pcmpgtd %xmm4, %xmm6 +; SSE2-NEXT: pand %xmm6, %xmm0 +; SSE2-NEXT: pandn %xmm2, %xmm6 +; SSE2-NEXT: por %xmm6, %xmm0 +; SSE2-NEXT: pxor %xmm1, %xmm3 +; SSE2-NEXT: pcmpgtd %xmm3, %xmm5 ; SSE2-NEXT: pand %xmm5, %xmm1 ; SSE2-NEXT: pandn %xmm2, %xmm5 ; SSE2-NEXT: por %xmm1, %xmm5 @@ -781,18 +740,18 @@ define <8 x i16> @trunc_usat_v8i32_v8i16(<8 x i32> %a0) { ; ; SSSE3-LABEL: trunc_usat_v8i32_v8i16: ; SSSE3: # %bb.0: -; SSSE3-NEXT: movdqa {{.*#+}} xmm2 = [2147483648,2147483648,2147483648,2147483648] -; SSSE3-NEXT: movdqa %xmm1, %xmm3 -; SSSE3-NEXT: pxor %xmm2, %xmm3 -; SSSE3-NEXT: movdqa {{.*#+}} xmm4 = [2147549183,2147549183,2147549183,2147549183] -; SSSE3-NEXT: movdqa %xmm4, %xmm5 -; SSSE3-NEXT: pcmpgtd %xmm3, %xmm5 -; SSSE3-NEXT: pxor %xmm0, %xmm2 -; SSSE3-NEXT: pcmpgtd %xmm2, %xmm4 ; SSSE3-NEXT: movdqa {{.*#+}} xmm2 = [65535,65535,65535,65535] -; SSSE3-NEXT: pand %xmm4, %xmm0 -; SSSE3-NEXT: pandn %xmm2, %xmm4 -; SSSE3-NEXT: por %xmm4, %xmm0 +; SSSE3-NEXT: movdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648] +; SSSE3-NEXT: movdqa %xmm0, %xmm4 +; SSSE3-NEXT: pxor %xmm3, %xmm4 +; SSSE3-NEXT: movdqa {{.*#+}} xmm5 = [2147549183,2147549183,2147549183,2147549183] +; SSSE3-NEXT: movdqa %xmm5, %xmm6 +; SSSE3-NEXT: pcmpgtd %xmm4, %xmm6 +; SSSE3-NEXT: pand %xmm6, %xmm0 +; SSSE3-NEXT: pandn %xmm2, %xmm6 +; SSSE3-NEXT: por %xmm6, %xmm0 +; SSSE3-NEXT: pxor %xmm1, %xmm3 +; SSSE3-NEXT: pcmpgtd %xmm3, %xmm5 ; SSSE3-NEXT: pand %xmm5, %xmm1 ; SSSE3-NEXT: pandn %xmm2, %xmm5 ; SSSE3-NEXT: por %xmm1, %xmm5 @@ -867,90 +826,90 @@ define <8 x i16> @trunc_usat_v8i32_v8i16(<8 x i32> %a0) { define <16 x i16> @trunc_usat_v16i32_v16i16(<16 x i32> %a0) { ; SSE2-LABEL: trunc_usat_v16i32_v16i16: ; SSE2: # %bb.0: -; SSE2-NEXT: movdqa %xmm1, %xmm8 +; SSE2-NEXT: movdqa %xmm1, %xmm4 +; SSE2-NEXT: movdqa {{.*#+}} xmm8 = [65535,65535,65535,65535] ; SSE2-NEXT: movdqa {{.*#+}} xmm7 = [2147483648,2147483648,2147483648,2147483648] -; SSE2-NEXT: movdqa %xmm1, %xmm5 -; SSE2-NEXT: pxor %xmm7, %xmm5 -; SSE2-NEXT: movdqa {{.*#+}} xmm1 = [2147549183,2147549183,2147549183,2147549183] -; SSE2-NEXT: movdqa %xmm1, %xmm6 -; SSE2-NEXT: pcmpgtd %xmm5, %xmm6 -; SSE2-NEXT: movdqa %xmm0, %xmm5 -; SSE2-NEXT: pxor %xmm7, %xmm5 -; SSE2-NEXT: movdqa %xmm1, %xmm9 -; SSE2-NEXT: pcmpgtd %xmm5, %xmm9 -; SSE2-NEXT: movdqa %xmm3, %xmm4 -; SSE2-NEXT: pxor %xmm7, %xmm4 -; SSE2-NEXT: movdqa %xmm1, %xmm5 -; SSE2-NEXT: pcmpgtd %xmm4, %xmm5 -; SSE2-NEXT: pxor %xmm2, %xmm7 -; SSE2-NEXT: pcmpgtd %xmm7, %xmm1 -; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [65535,65535,65535,65535] +; SSE2-NEXT: movdqa %xmm2, %xmm6 +; SSE2-NEXT: pxor %xmm7, %xmm6 +; SSE2-NEXT: movdqa {{.*#+}} xmm5 = [2147549183,2147549183,2147549183,2147549183] +; SSE2-NEXT: movdqa %xmm5, %xmm1 +; SSE2-NEXT: pcmpgtd %xmm6, %xmm1 ; SSE2-NEXT: pand %xmm1, %xmm2 -; SSE2-NEXT: pandn %xmm4, %xmm1 +; SSE2-NEXT: pandn %xmm8, %xmm1 ; SSE2-NEXT: por %xmm2, %xmm1 -; SSE2-NEXT: pand %xmm5, %xmm3 -; SSE2-NEXT: pandn %xmm4, %xmm5 -; SSE2-NEXT: por %xmm3, %xmm5 -; SSE2-NEXT: pand %xmm9, %xmm0 -; SSE2-NEXT: pandn %xmm4, %xmm9 -; SSE2-NEXT: por %xmm9, %xmm0 -; SSE2-NEXT: pand %xmm6, %xmm8 -; SSE2-NEXT: pandn %xmm4, %xmm6 -; SSE2-NEXT: por %xmm8, %xmm6 -; SSE2-NEXT: pslld $16, %xmm6 -; SSE2-NEXT: psrad $16, %xmm6 -; SSE2-NEXT: pslld $16, %xmm0 -; SSE2-NEXT: psrad $16, %xmm0 -; SSE2-NEXT: packssdw %xmm6, %xmm0 +; SSE2-NEXT: movdqa %xmm3, %xmm6 +; SSE2-NEXT: pxor %xmm7, %xmm6 +; SSE2-NEXT: movdqa %xmm5, %xmm2 +; SSE2-NEXT: pcmpgtd %xmm6, %xmm2 +; SSE2-NEXT: pand %xmm2, %xmm3 +; SSE2-NEXT: pandn %xmm8, %xmm2 +; SSE2-NEXT: por %xmm3, %xmm2 +; SSE2-NEXT: movdqa %xmm0, %xmm3 +; SSE2-NEXT: pxor %xmm7, %xmm3 +; SSE2-NEXT: movdqa %xmm5, %xmm6 +; SSE2-NEXT: pcmpgtd %xmm3, %xmm6 +; SSE2-NEXT: pand %xmm6, %xmm0 +; SSE2-NEXT: pandn %xmm8, %xmm6 +; SSE2-NEXT: por %xmm6, %xmm0 +; SSE2-NEXT: pxor %xmm4, %xmm7 +; SSE2-NEXT: pcmpgtd %xmm7, %xmm5 +; SSE2-NEXT: pand %xmm5, %xmm4 +; SSE2-NEXT: pandn %xmm8, %xmm5 +; SSE2-NEXT: por %xmm4, %xmm5 ; SSE2-NEXT: pslld $16, %xmm5 ; SSE2-NEXT: psrad $16, %xmm5 +; SSE2-NEXT: pslld $16, %xmm0 +; SSE2-NEXT: psrad $16, %xmm0 +; SSE2-NEXT: packssdw %xmm5, %xmm0 +; SSE2-NEXT: pslld $16, %xmm2 +; SSE2-NEXT: psrad $16, %xmm2 ; SSE2-NEXT: pslld $16, %xmm1 ; SSE2-NEXT: psrad $16, %xmm1 -; SSE2-NEXT: packssdw %xmm5, %xmm1 +; SSE2-NEXT: packssdw %xmm2, %xmm1 ; SSE2-NEXT: retq ; ; SSSE3-LABEL: trunc_usat_v16i32_v16i16: ; SSSE3: # %bb.0: -; SSSE3-NEXT: movdqa %xmm1, %xmm8 +; SSSE3-NEXT: movdqa %xmm1, %xmm4 +; SSSE3-NEXT: movdqa {{.*#+}} xmm8 = [65535,65535,65535,65535] ; SSSE3-NEXT: movdqa {{.*#+}} xmm7 = [2147483648,2147483648,2147483648,2147483648] -; SSSE3-NEXT: movdqa %xmm1, %xmm5 -; SSSE3-NEXT: pxor %xmm7, %xmm5 -; SSSE3-NEXT: movdqa {{.*#+}} xmm1 = [2147549183,2147549183,2147549183,2147549183] -; SSSE3-NEXT: movdqa %xmm1, %xmm6 -; SSSE3-NEXT: pcmpgtd %xmm5, %xmm6 -; SSSE3-NEXT: movdqa %xmm0, %xmm5 -; SSSE3-NEXT: pxor %xmm7, %xmm5 -; SSSE3-NEXT: movdqa %xmm1, %xmm9 -; SSSE3-NEXT: pcmpgtd %xmm5, %xmm9 -; SSSE3-NEXT: movdqa %xmm3, %xmm4 -; SSSE3-NEXT: pxor %xmm7, %xmm4 -; SSSE3-NEXT: movdqa %xmm1, %xmm5 -; SSSE3-NEXT: pcmpgtd %xmm4, %xmm5 -; SSSE3-NEXT: pxor %xmm2, %xmm7 -; SSSE3-NEXT: pcmpgtd %xmm7, %xmm1 -; SSSE3-NEXT: movdqa {{.*#+}} xmm4 = [65535,65535,65535,65535] +; SSSE3-NEXT: movdqa %xmm2, %xmm6 +; SSSE3-NEXT: pxor %xmm7, %xmm6 +; SSSE3-NEXT: movdqa {{.*#+}} xmm5 = [2147549183,2147549183,2147549183,2147549183] +; SSSE3-NEXT: movdqa %xmm5, %xmm1 +; SSSE3-NEXT: pcmpgtd %xmm6, %xmm1 ; SSSE3-NEXT: pand %xmm1, %xmm2 -; SSSE3-NEXT: pandn %xmm4, %xmm1 +; SSSE3-NEXT: pandn %xmm8, %xmm1 ; SSSE3-NEXT: por %xmm2, %xmm1 -; SSSE3-NEXT: pand %xmm5, %xmm3 -; SSSE3-NEXT: pandn %xmm4, %xmm5 -; SSSE3-NEXT: por %xmm3, %xmm5 -; SSSE3-NEXT: pand %xmm9, %xmm0 -; SSSE3-NEXT: pandn %xmm4, %xmm9 -; SSSE3-NEXT: por %xmm9, %xmm0 -; SSSE3-NEXT: pand %xmm6, %xmm8 -; SSSE3-NEXT: pandn %xmm4, %xmm6 -; SSSE3-NEXT: por %xmm8, %xmm6 -; SSSE3-NEXT: pslld $16, %xmm6 -; SSSE3-NEXT: psrad $16, %xmm6 -; SSSE3-NEXT: pslld $16, %xmm0 -; SSSE3-NEXT: psrad $16, %xmm0 -; SSSE3-NEXT: packssdw %xmm6, %xmm0 +; SSSE3-NEXT: movdqa %xmm3, %xmm6 +; SSSE3-NEXT: pxor %xmm7, %xmm6 +; SSSE3-NEXT: movdqa %xmm5, %xmm2 +; SSSE3-NEXT: pcmpgtd %xmm6, %xmm2 +; SSSE3-NEXT: pand %xmm2, %xmm3 +; SSSE3-NEXT: pandn %xmm8, %xmm2 +; SSSE3-NEXT: por %xmm3, %xmm2 +; SSSE3-NEXT: movdqa %xmm0, %xmm3 +; SSSE3-NEXT: pxor %xmm7, %xmm3 +; SSSE3-NEXT: movdqa %xmm5, %xmm6 +; SSSE3-NEXT: pcmpgtd %xmm3, %xmm6 +; SSSE3-NEXT: pand %xmm6, %xmm0 +; SSSE3-NEXT: pandn %xmm8, %xmm6 +; SSSE3-NEXT: por %xmm6, %xmm0 +; SSSE3-NEXT: pxor %xmm4, %xmm7 +; SSSE3-NEXT: pcmpgtd %xmm7, %xmm5 +; SSSE3-NEXT: pand %xmm5, %xmm4 +; SSSE3-NEXT: pandn %xmm8, %xmm5 +; SSSE3-NEXT: por %xmm4, %xmm5 ; SSSE3-NEXT: pslld $16, %xmm5 ; SSSE3-NEXT: psrad $16, %xmm5 +; SSSE3-NEXT: pslld $16, %xmm0 +; SSSE3-NEXT: psrad $16, %xmm0 +; SSSE3-NEXT: packssdw %xmm5, %xmm0 +; SSSE3-NEXT: pslld $16, %xmm2 +; SSSE3-NEXT: psrad $16, %xmm2 ; SSSE3-NEXT: pslld $16, %xmm1 ; SSSE3-NEXT: psrad $16, %xmm1 -; SSSE3-NEXT: packssdw %xmm5, %xmm1 +; SSSE3-NEXT: packssdw %xmm2, %xmm1 ; SSSE3-NEXT: retq ; ; SSE41-LABEL: trunc_usat_v16i32_v16i16: @@ -1005,282 +964,228 @@ define <16 x i16> @trunc_usat_v16i32_v16i16(<16 x i32> %a0) { define <8 x i8> @trunc_usat_v8i64_v8i8(<8 x i64> %a0) { ; SSE2-LABEL: trunc_usat_v8i64_v8i8: ; SSE2: # %bb.0: -; SSE2-NEXT: movdqa {{.*#+}} xmm6 = [2147483648,2147483648,2147483648,2147483648] -; SSE2-NEXT: movdqa %xmm1, %xmm4 -; SSE2-NEXT: pxor %xmm6, %xmm4 -; SSE2-NEXT: movdqa {{.*#+}} xmm8 = [9223372039002259711,9223372039002259711] -; SSE2-NEXT: movdqa %xmm8, %xmm5 -; SSE2-NEXT: pcmpgtd %xmm4, %xmm5 -; SSE2-NEXT: pshufd {{.*#+}} xmm9 = xmm5[0,0,2,2] -; SSE2-NEXT: pcmpeqd %xmm8, %xmm4 -; SSE2-NEXT: pshufd {{.*#+}} xmm7 = xmm4[1,1,3,3] -; SSE2-NEXT: pand %xmm9, %xmm7 -; SSE2-NEXT: pshufd {{.*#+}} xmm10 = xmm5[1,1,3,3] -; SSE2-NEXT: por %xmm7, %xmm10 -; SSE2-NEXT: movdqa %xmm0, %xmm5 -; SSE2-NEXT: pxor %xmm6, %xmm5 -; SSE2-NEXT: movdqa %xmm8, %xmm7 -; SSE2-NEXT: pcmpgtd %xmm5, %xmm7 -; SSE2-NEXT: pshufd {{.*#+}} xmm9 = xmm7[0,0,2,2] -; SSE2-NEXT: pcmpeqd %xmm8, %xmm5 -; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm5[1,1,3,3] -; SSE2-NEXT: pand %xmm9, %xmm4 -; SSE2-NEXT: pshufd {{.*#+}} xmm11 = xmm7[1,1,3,3] -; SSE2-NEXT: por %xmm4, %xmm11 -; SSE2-NEXT: movdqa %xmm3, %xmm4 -; SSE2-NEXT: pxor %xmm6, %xmm4 -; SSE2-NEXT: movdqa %xmm8, %xmm7 -; SSE2-NEXT: pcmpgtd %xmm4, %xmm7 -; SSE2-NEXT: pshufd {{.*#+}} xmm9 = xmm7[0,0,2,2] -; SSE2-NEXT: pcmpeqd %xmm8, %xmm4 -; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3] -; SSE2-NEXT: pand %xmm9, %xmm4 +; SSE2-NEXT: movdqa {{.*#+}} xmm8 = [255,255] +; SSE2-NEXT: movdqa {{.*#+}} xmm5 = [2147483648,2147483648,2147483648,2147483648] +; SSE2-NEXT: movdqa %xmm1, %xmm7 +; SSE2-NEXT: pxor %xmm5, %xmm7 +; SSE2-NEXT: movdqa {{.*#+}} xmm9 = [9223372039002259711,9223372039002259711] +; SSE2-NEXT: movdqa %xmm9, %xmm6 +; SSE2-NEXT: pcmpgtd %xmm7, %xmm6 +; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm6[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm9, %xmm7 ; SSE2-NEXT: pshufd {{.*#+}} xmm7 = xmm7[1,1,3,3] -; SSE2-NEXT: por %xmm4, %xmm7 -; SSE2-NEXT: pxor %xmm2, %xmm6 -; SSE2-NEXT: movdqa %xmm8, %xmm4 -; SSE2-NEXT: pcmpgtd %xmm6, %xmm4 -; SSE2-NEXT: pshufd {{.*#+}} xmm9 = xmm4[0,0,2,2] -; SSE2-NEXT: pcmpeqd %xmm8, %xmm6 -; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm6[1,1,3,3] -; SSE2-NEXT: pand %xmm9, %xmm5 -; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm4[1,1,3,3] -; SSE2-NEXT: por %xmm5, %xmm6 -; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [255,255] -; SSE2-NEXT: pand %xmm6, %xmm2 -; SSE2-NEXT: pandn %xmm4, %xmm6 -; SSE2-NEXT: por %xmm2, %xmm6 -; SSE2-NEXT: pand %xmm7, %xmm3 -; SSE2-NEXT: pandn %xmm4, %xmm7 -; SSE2-NEXT: por %xmm3, %xmm7 -; SSE2-NEXT: pand %xmm11, %xmm0 -; SSE2-NEXT: pandn %xmm4, %xmm11 -; SSE2-NEXT: por %xmm0, %xmm11 -; SSE2-NEXT: pand %xmm10, %xmm1 -; SSE2-NEXT: pandn %xmm4, %xmm10 -; SSE2-NEXT: por %xmm1, %xmm10 -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm10[0,2,2,3] -; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7] -; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm11[0,2,2,3] -; SSE2-NEXT: pshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7] -; SSE2-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm7[0,2,2,3] -; SSE2-NEXT: pshuflw {{.*#+}} xmm2 = xmm0[0,1,0,2,4,5,6,7] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm6[0,2,2,3] -; SSE2-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,1,0,2,4,5,6,7] -; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1] -; SSE2-NEXT: movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1] +; SSE2-NEXT: pand %xmm4, %xmm7 +; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm6[1,1,3,3] +; SSE2-NEXT: por %xmm7, %xmm4 +; SSE2-NEXT: pand %xmm4, %xmm1 +; SSE2-NEXT: pandn %xmm8, %xmm4 +; SSE2-NEXT: por %xmm1, %xmm4 +; SSE2-NEXT: movdqa %xmm0, %xmm1 +; SSE2-NEXT: pxor %xmm5, %xmm1 +; SSE2-NEXT: movdqa %xmm9, %xmm6 +; SSE2-NEXT: pcmpgtd %xmm1, %xmm6 +; SSE2-NEXT: pshufd {{.*#+}} xmm7 = xmm6[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm9, %xmm1 +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3] +; SSE2-NEXT: pand %xmm7, %xmm1 +; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm6[1,1,3,3] +; SSE2-NEXT: por %xmm1, %xmm6 +; SSE2-NEXT: pand %xmm6, %xmm0 +; SSE2-NEXT: pandn %xmm8, %xmm6 +; SSE2-NEXT: por %xmm6, %xmm0 +; SSE2-NEXT: packuswb %xmm4, %xmm0 +; SSE2-NEXT: movdqa %xmm3, %xmm1 +; SSE2-NEXT: pxor %xmm5, %xmm1 +; SSE2-NEXT: movdqa %xmm9, %xmm4 +; SSE2-NEXT: pcmpgtd %xmm1, %xmm4 +; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm4[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm9, %xmm1 +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3] +; SSE2-NEXT: pand %xmm6, %xmm1 +; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3] +; SSE2-NEXT: por %xmm1, %xmm4 +; SSE2-NEXT: pand %xmm4, %xmm3 +; SSE2-NEXT: pandn %xmm8, %xmm4 +; SSE2-NEXT: por %xmm3, %xmm4 +; SSE2-NEXT: pxor %xmm2, %xmm5 +; SSE2-NEXT: movdqa %xmm9, %xmm1 +; SSE2-NEXT: pcmpgtd %xmm5, %xmm1 +; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm1[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm9, %xmm5 +; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3] +; SSE2-NEXT: pand %xmm3, %xmm5 +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3] +; SSE2-NEXT: por %xmm5, %xmm1 +; SSE2-NEXT: pand %xmm1, %xmm2 +; SSE2-NEXT: pandn %xmm8, %xmm1 +; SSE2-NEXT: por %xmm2, %xmm1 +; SSE2-NEXT: packuswb %xmm4, %xmm1 +; SSE2-NEXT: packuswb %xmm1, %xmm0 ; SSE2-NEXT: retq ; ; SSSE3-LABEL: trunc_usat_v8i64_v8i8: ; SSSE3: # %bb.0: -; SSSE3-NEXT: movdqa {{.*#+}} xmm6 = [2147483648,2147483648,2147483648,2147483648] -; SSSE3-NEXT: movdqa %xmm1, %xmm4 -; SSSE3-NEXT: pxor %xmm6, %xmm4 -; SSSE3-NEXT: movdqa {{.*#+}} xmm8 = [9223372039002259711,9223372039002259711] -; SSSE3-NEXT: movdqa %xmm8, %xmm5 -; SSSE3-NEXT: pcmpgtd %xmm4, %xmm5 -; SSSE3-NEXT: pshufd {{.*#+}} xmm9 = xmm5[0,0,2,2] -; SSSE3-NEXT: pcmpeqd %xmm8, %xmm4 -; SSSE3-NEXT: pshufd {{.*#+}} xmm7 = xmm4[1,1,3,3] -; SSSE3-NEXT: pand %xmm9, %xmm7 -; SSSE3-NEXT: pshufd {{.*#+}} xmm10 = xmm5[1,1,3,3] -; SSSE3-NEXT: por %xmm7, %xmm10 -; SSSE3-NEXT: movdqa %xmm0, %xmm5 -; SSSE3-NEXT: pxor %xmm6, %xmm5 -; SSSE3-NEXT: movdqa %xmm8, %xmm7 -; SSSE3-NEXT: pcmpgtd %xmm5, %xmm7 -; SSSE3-NEXT: pshufd {{.*#+}} xmm9 = xmm7[0,0,2,2] -; SSSE3-NEXT: pcmpeqd %xmm8, %xmm5 -; SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm5[1,1,3,3] -; SSSE3-NEXT: pand %xmm9, %xmm4 -; SSSE3-NEXT: pshufd {{.*#+}} xmm11 = xmm7[1,1,3,3] -; SSSE3-NEXT: por %xmm4, %xmm11 -; SSSE3-NEXT: movdqa %xmm3, %xmm4 -; SSSE3-NEXT: pxor %xmm6, %xmm4 -; SSSE3-NEXT: movdqa %xmm8, %xmm7 -; SSSE3-NEXT: pcmpgtd %xmm4, %xmm7 -; SSSE3-NEXT: pshufd {{.*#+}} xmm9 = xmm7[0,0,2,2] -; SSSE3-NEXT: pcmpeqd %xmm8, %xmm4 -; SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3] -; SSSE3-NEXT: pand %xmm9, %xmm4 +; SSSE3-NEXT: movdqa {{.*#+}} xmm8 = [255,255] +; SSSE3-NEXT: movdqa {{.*#+}} xmm5 = [2147483648,2147483648,2147483648,2147483648] +; SSSE3-NEXT: movdqa %xmm1, %xmm7 +; SSSE3-NEXT: pxor %xmm5, %xmm7 +; SSSE3-NEXT: movdqa {{.*#+}} xmm9 = [9223372039002259711,9223372039002259711] +; SSSE3-NEXT: movdqa %xmm9, %xmm6 +; SSSE3-NEXT: pcmpgtd %xmm7, %xmm6 +; SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm6[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm9, %xmm7 ; SSSE3-NEXT: pshufd {{.*#+}} xmm7 = xmm7[1,1,3,3] -; SSSE3-NEXT: por %xmm4, %xmm7 -; SSSE3-NEXT: pxor %xmm2, %xmm6 -; SSSE3-NEXT: movdqa %xmm8, %xmm4 -; SSSE3-NEXT: pcmpgtd %xmm6, %xmm4 -; SSSE3-NEXT: pshufd {{.*#+}} xmm9 = xmm4[0,0,2,2] -; SSSE3-NEXT: pcmpeqd %xmm8, %xmm6 -; SSSE3-NEXT: pshufd {{.*#+}} xmm5 = xmm6[1,1,3,3] -; SSSE3-NEXT: pand %xmm9, %xmm5 -; SSSE3-NEXT: pshufd {{.*#+}} xmm6 = xmm4[1,1,3,3] -; SSSE3-NEXT: por %xmm5, %xmm6 -; SSSE3-NEXT: movdqa {{.*#+}} xmm4 = [255,255] -; SSSE3-NEXT: pand %xmm6, %xmm2 -; SSSE3-NEXT: pandn %xmm4, %xmm6 -; SSSE3-NEXT: por %xmm2, %xmm6 -; SSSE3-NEXT: pand %xmm7, %xmm3 -; SSSE3-NEXT: pandn %xmm4, %xmm7 -; SSSE3-NEXT: por %xmm3, %xmm7 -; SSSE3-NEXT: pand %xmm11, %xmm0 -; SSSE3-NEXT: pandn %xmm4, %xmm11 -; SSSE3-NEXT: por %xmm0, %xmm11 -; SSSE3-NEXT: pand %xmm10, %xmm1 -; SSSE3-NEXT: pandn %xmm4, %xmm10 -; SSSE3-NEXT: por %xmm1, %xmm10 -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm10[0,2,2,3] -; SSSE3-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7] -; SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm11[0,2,2,3] -; SSSE3-NEXT: pshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7] -; SSSE3-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1] -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm7[0,2,2,3] -; SSSE3-NEXT: pshuflw {{.*#+}} xmm2 = xmm0[0,1,0,2,4,5,6,7] -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm6[0,2,2,3] -; SSSE3-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,1,0,2,4,5,6,7] -; SSSE3-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1] -; SSSE3-NEXT: movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1] +; SSSE3-NEXT: pand %xmm4, %xmm7 +; SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm6[1,1,3,3] +; SSSE3-NEXT: por %xmm7, %xmm4 +; SSSE3-NEXT: pand %xmm4, %xmm1 +; SSSE3-NEXT: pandn %xmm8, %xmm4 +; SSSE3-NEXT: por %xmm1, %xmm4 +; SSSE3-NEXT: movdqa %xmm0, %xmm1 +; SSSE3-NEXT: pxor %xmm5, %xmm1 +; SSSE3-NEXT: movdqa %xmm9, %xmm6 +; SSSE3-NEXT: pcmpgtd %xmm1, %xmm6 +; SSSE3-NEXT: pshufd {{.*#+}} xmm7 = xmm6[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm9, %xmm1 +; SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3] +; SSSE3-NEXT: pand %xmm7, %xmm1 +; SSSE3-NEXT: pshufd {{.*#+}} xmm6 = xmm6[1,1,3,3] +; SSSE3-NEXT: por %xmm1, %xmm6 +; SSSE3-NEXT: pand %xmm6, %xmm0 +; SSSE3-NEXT: pandn %xmm8, %xmm6 +; SSSE3-NEXT: por %xmm6, %xmm0 +; SSSE3-NEXT: packuswb %xmm4, %xmm0 +; SSSE3-NEXT: movdqa %xmm3, %xmm1 +; SSSE3-NEXT: pxor %xmm5, %xmm1 +; SSSE3-NEXT: movdqa %xmm9, %xmm4 +; SSSE3-NEXT: pcmpgtd %xmm1, %xmm4 +; SSSE3-NEXT: pshufd {{.*#+}} xmm6 = xmm4[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm9, %xmm1 +; SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3] +; SSSE3-NEXT: pand %xmm6, %xmm1 +; SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3] +; SSSE3-NEXT: por %xmm1, %xmm4 +; SSSE3-NEXT: pand %xmm4, %xmm3 +; SSSE3-NEXT: pandn %xmm8, %xmm4 +; SSSE3-NEXT: por %xmm3, %xmm4 +; SSSE3-NEXT: pxor %xmm2, %xmm5 +; SSSE3-NEXT: movdqa %xmm9, %xmm1 +; SSSE3-NEXT: pcmpgtd %xmm5, %xmm1 +; SSSE3-NEXT: pshufd {{.*#+}} xmm3 = xmm1[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm9, %xmm5 +; SSSE3-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3] +; SSSE3-NEXT: pand %xmm3, %xmm5 +; SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3] +; SSSE3-NEXT: por %xmm5, %xmm1 +; SSSE3-NEXT: pand %xmm1, %xmm2 +; SSSE3-NEXT: pandn %xmm8, %xmm1 +; SSSE3-NEXT: por %xmm2, %xmm1 +; SSSE3-NEXT: packuswb %xmm4, %xmm1 +; SSSE3-NEXT: packuswb %xmm1, %xmm0 ; SSSE3-NEXT: retq ; ; SSE41-LABEL: trunc_usat_v8i64_v8i8: ; SSE41: # %bb.0: ; SSE41-NEXT: movdqa %xmm0, %xmm8 -; SSE41-NEXT: movdqa {{.*#+}} xmm0 = [2147483648,2147483648,2147483648,2147483648] -; SSE41-NEXT: movdqa %xmm3, %xmm5 -; SSE41-NEXT: pxor %xmm0, %xmm5 -; SSE41-NEXT: movdqa {{.*#+}} xmm4 = [9223372039002259711,9223372039002259711] -; SSE41-NEXT: movdqa %xmm4, %xmm6 -; SSE41-NEXT: pcmpgtd %xmm5, %xmm6 -; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm6[0,0,2,2] -; SSE41-NEXT: pcmpeqd %xmm4, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3] -; SSE41-NEXT: pand %xmm7, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm9 = xmm6[1,1,3,3] -; SSE41-NEXT: por %xmm5, %xmm9 -; SSE41-NEXT: movdqa %xmm2, %xmm5 -; SSE41-NEXT: pxor %xmm0, %xmm5 -; SSE41-NEXT: movdqa %xmm4, %xmm6 -; SSE41-NEXT: pcmpgtd %xmm5, %xmm6 -; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm6[0,0,2,2] -; SSE41-NEXT: pcmpeqd %xmm4, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3] -; SSE41-NEXT: pand %xmm7, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm10 = xmm6[1,1,3,3] -; SSE41-NEXT: por %xmm5, %xmm10 -; SSE41-NEXT: movdqa %xmm1, %xmm5 -; SSE41-NEXT: pxor %xmm0, %xmm5 -; SSE41-NEXT: movdqa %xmm4, %xmm7 -; SSE41-NEXT: pcmpgtd %xmm5, %xmm7 -; SSE41-NEXT: pshufd {{.*#+}} xmm6 = xmm7[0,0,2,2] -; SSE41-NEXT: pcmpeqd %xmm4, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3] -; SSE41-NEXT: pand %xmm6, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm7[1,1,3,3] -; SSE41-NEXT: por %xmm5, %xmm7 -; SSE41-NEXT: pxor %xmm8, %xmm0 -; SSE41-NEXT: movdqa %xmm4, %xmm5 -; SSE41-NEXT: pcmpgtd %xmm0, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2] -; SSE41-NEXT: pcmpeqd %xmm4, %xmm0 +; SSE41-NEXT: movapd {{.*#+}} xmm5 = [255,255] +; SSE41-NEXT: movdqa {{.*#+}} xmm6 = [2147483648,2147483648,2147483648,2147483648] +; SSE41-NEXT: movdqa %xmm1, %xmm0 +; SSE41-NEXT: pxor %xmm6, %xmm0 +; SSE41-NEXT: movdqa {{.*#+}} xmm9 = [9223372039002259711,9223372039002259711] +; SSE41-NEXT: movdqa %xmm9, %xmm7 +; SSE41-NEXT: pcmpgtd %xmm0, %xmm7 +; SSE41-NEXT: pshufd {{.*#+}} xmm10 = xmm7[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm9, %xmm0 ; SSE41-NEXT: pshufd {{.*#+}} xmm4 = xmm0[1,1,3,3] -; SSE41-NEXT: pand %xmm6, %xmm4 -; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm5[1,1,3,3] +; SSE41-NEXT: pand %xmm10, %xmm4 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm7[1,1,3,3] ; SSE41-NEXT: por %xmm4, %xmm0 -; SSE41-NEXT: movapd {{.*#+}} xmm5 = [255,255] ; SSE41-NEXT: movapd %xmm5, %xmm4 -; SSE41-NEXT: blendvpd %xmm0, %xmm8, %xmm4 -; SSE41-NEXT: movapd %xmm5, %xmm6 -; SSE41-NEXT: movdqa %xmm7, %xmm0 -; SSE41-NEXT: blendvpd %xmm0, %xmm1, %xmm6 +; SSE41-NEXT: blendvpd %xmm0, %xmm1, %xmm4 +; SSE41-NEXT: movdqa %xmm8, %xmm0 +; SSE41-NEXT: pxor %xmm6, %xmm0 +; SSE41-NEXT: movdqa %xmm9, %xmm1 +; SSE41-NEXT: pcmpgtd %xmm0, %xmm1 +; SSE41-NEXT: pshufd {{.*#+}} xmm10 = xmm1[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm9, %xmm0 +; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm0[1,1,3,3] +; SSE41-NEXT: pand %xmm10, %xmm7 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3] +; SSE41-NEXT: por %xmm7, %xmm0 ; SSE41-NEXT: movapd %xmm5, %xmm1 -; SSE41-NEXT: movdqa %xmm10, %xmm0 -; SSE41-NEXT: blendvpd %xmm0, %xmm2, %xmm1 +; SSE41-NEXT: blendvpd %xmm0, %xmm8, %xmm1 +; SSE41-NEXT: packssdw %xmm4, %xmm1 +; SSE41-NEXT: movdqa %xmm3, %xmm0 +; SSE41-NEXT: pxor %xmm6, %xmm0 +; SSE41-NEXT: movdqa %xmm9, %xmm4 +; SSE41-NEXT: pcmpgtd %xmm0, %xmm4 +; SSE41-NEXT: pshufd {{.*#+}} xmm8 = xmm4[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm9, %xmm0 +; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm0[1,1,3,3] +; SSE41-NEXT: pand %xmm8, %xmm7 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm4[1,1,3,3] +; SSE41-NEXT: por %xmm7, %xmm0 +; SSE41-NEXT: movapd %xmm5, %xmm4 +; SSE41-NEXT: blendvpd %xmm0, %xmm3, %xmm4 +; SSE41-NEXT: pxor %xmm2, %xmm6 ; SSE41-NEXT: movdqa %xmm9, %xmm0 -; SSE41-NEXT: blendvpd %xmm0, %xmm3, %xmm5 -; SSE41-NEXT: pxor %xmm0, %xmm0 -; SSE41-NEXT: pblendw {{.*#+}} xmm5 = xmm5[0],xmm0[1,2,3],xmm5[4],xmm0[5,6,7] -; SSE41-NEXT: pblendw {{.*#+}} xmm1 = xmm1[0],xmm0[1,2,3],xmm1[4],xmm0[5,6,7] +; SSE41-NEXT: pcmpgtd %xmm6, %xmm0 +; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm0[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm9, %xmm6 +; SSE41-NEXT: pshufd {{.*#+}} xmm6 = xmm6[1,1,3,3] +; SSE41-NEXT: pand %xmm3, %xmm6 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] +; SSE41-NEXT: por %xmm6, %xmm0 +; SSE41-NEXT: blendvpd %xmm0, %xmm2, %xmm5 +; SSE41-NEXT: packssdw %xmm4, %xmm5 ; SSE41-NEXT: packusdw %xmm5, %xmm1 -; SSE41-NEXT: pblendw {{.*#+}} xmm6 = xmm6[0],xmm0[1,2,3],xmm6[4],xmm0[5,6,7] -; SSE41-NEXT: pblendw {{.*#+}} xmm4 = xmm4[0],xmm0[1,2,3],xmm4[4],xmm0[5,6,7] -; SSE41-NEXT: packusdw %xmm6, %xmm4 -; SSE41-NEXT: packusdw %xmm1, %xmm4 -; SSE41-NEXT: movdqa %xmm4, %xmm0 +; SSE41-NEXT: movdqa %xmm1, %xmm0 ; SSE41-NEXT: retq ; ; AVX1-LABEL: trunc_usat_v8i64_v8i8: ; AVX1: # %bb.0: +; AVX1-NEXT: vmovapd {{.*#+}} ymm2 = [255,255,255,255] +; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm3 +; AVX1-NEXT: vmovdqa {{.*#+}} xmm4 = [9223372036854775808,9223372036854775808] +; AVX1-NEXT: vpxor %xmm4, %xmm3, %xmm3 +; AVX1-NEXT: vmovdqa {{.*#+}} xmm5 = [9223372036854776063,9223372036854776063] +; AVX1-NEXT: vpcmpgtq %xmm3, %xmm5, %xmm3 +; AVX1-NEXT: vpxor %xmm4, %xmm0, %xmm6 +; AVX1-NEXT: vpcmpgtq %xmm6, %xmm5, %xmm6 +; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm6, %ymm3 +; AVX1-NEXT: vblendvpd %ymm3, %ymm0, %ymm2, %ymm0 +; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm3 +; AVX1-NEXT: vpxor %xmm4, %xmm3, %xmm3 +; AVX1-NEXT: vpcmpgtq %xmm3, %xmm5, %xmm3 +; AVX1-NEXT: vpxor %xmm4, %xmm1, %xmm4 +; AVX1-NEXT: vpcmpgtq %xmm4, %xmm5, %xmm4 +; AVX1-NEXT: vinsertf128 $1, %xmm3, %ymm4, %ymm3 +; AVX1-NEXT: vblendvpd %ymm3, %ymm1, %ymm2, %ymm1 ; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm2 -; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [9223372036854775808,9223372036854775808] -; AVX1-NEXT: vpxor %xmm3, %xmm2, %xmm2 -; AVX1-NEXT: vmovdqa {{.*#+}} xmm4 = [9223372036854776063,9223372036854776063] -; AVX1-NEXT: vpcmpgtq %xmm2, %xmm4, %xmm2 -; AVX1-NEXT: vpxor %xmm3, %xmm1, %xmm5 -; AVX1-NEXT: vpcmpgtq %xmm5, %xmm4, %xmm5 -; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm5, %ymm2 -; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm5 -; AVX1-NEXT: vpxor %xmm3, %xmm5, %xmm5 -; AVX1-NEXT: vpcmpgtq %xmm5, %xmm4, %xmm5 -; AVX1-NEXT: vpxor %xmm3, %xmm0, %xmm3 -; AVX1-NEXT: vpcmpgtq %xmm3, %xmm4, %xmm3 -; AVX1-NEXT: vinsertf128 $1, %xmm5, %ymm3, %ymm3 -; AVX1-NEXT: vmovapd {{.*#+}} ymm4 = [255,255,255,255] -; AVX1-NEXT: vblendvpd %ymm3, %ymm0, %ymm4, %ymm0 -; AVX1-NEXT: vblendvpd %ymm2, %ymm1, %ymm4, %ymm1 -; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm2 -; AVX1-NEXT: vxorpd %xmm3, %xmm3, %xmm3 -; AVX1-NEXT: vpblendw {{.*#+}} xmm2 = xmm2[0],xmm3[1,2,3],xmm2[4],xmm3[5,6,7] -; AVX1-NEXT: vpblendw {{.*#+}} xmm1 = xmm1[0],xmm3[1,2,3],xmm1[4],xmm3[5,6,7] -; AVX1-NEXT: vpackusdw %xmm2, %xmm1, %xmm1 +; AVX1-NEXT: vpackssdw %xmm2, %xmm1, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm2 -; AVX1-NEXT: vpblendw {{.*#+}} xmm2 = xmm2[0],xmm3[1,2,3],xmm2[4],xmm3[5,6,7] -; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm3[1,2,3],xmm0[4],xmm3[5,6,7] -; AVX1-NEXT: vpackusdw %xmm2, %xmm0, %xmm0 +; AVX1-NEXT: vpackssdw %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: vpackusdw %xmm1, %xmm0, %xmm0 ; AVX1-NEXT: vzeroupper ; AVX1-NEXT: retq ; -; AVX2-SLOW-LABEL: trunc_usat_v8i64_v8i8: -; AVX2-SLOW: # %bb.0: -; AVX2-SLOW-NEXT: vpbroadcastq {{.*#+}} ymm2 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808] -; AVX2-SLOW-NEXT: vpxor %ymm2, %ymm0, %ymm3 -; AVX2-SLOW-NEXT: vpbroadcastq {{.*#+}} ymm4 = [9223372036854776063,9223372036854776063,9223372036854776063,9223372036854776063] -; AVX2-SLOW-NEXT: vpcmpgtq %ymm3, %ymm4, %ymm3 -; AVX2-SLOW-NEXT: vpxor %ymm2, %ymm1, %ymm2 -; AVX2-SLOW-NEXT: vpcmpgtq %ymm2, %ymm4, %ymm2 -; AVX2-SLOW-NEXT: vbroadcastsd {{.*#+}} ymm4 = [255,255,255,255] -; AVX2-SLOW-NEXT: vblendvpd %ymm2, %ymm1, %ymm4, %ymm1 -; AVX2-SLOW-NEXT: vblendvpd %ymm3, %ymm0, %ymm4, %ymm0 -; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7] -; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,3] -; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7] -; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,2,2,3] -; AVX2-SLOW-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 -; AVX2-SLOW-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31] -; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3] -; AVX2-SLOW-NEXT: # kill: def $xmm0 killed $xmm0 killed $ymm0 -; AVX2-SLOW-NEXT: vzeroupper -; AVX2-SLOW-NEXT: retq -; -; AVX2-FAST-LABEL: trunc_usat_v8i64_v8i8: -; AVX2-FAST: # %bb.0: -; AVX2-FAST-NEXT: vpbroadcastq {{.*#+}} ymm2 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808] -; AVX2-FAST-NEXT: vpxor %ymm2, %ymm0, %ymm3 -; AVX2-FAST-NEXT: vpbroadcastq {{.*#+}} ymm4 = [9223372036854776063,9223372036854776063,9223372036854776063,9223372036854776063] -; AVX2-FAST-NEXT: vpcmpgtq %ymm3, %ymm4, %ymm3 -; AVX2-FAST-NEXT: vpxor %ymm2, %ymm1, %ymm2 -; AVX2-FAST-NEXT: vpcmpgtq %ymm2, %ymm4, %ymm2 -; AVX2-FAST-NEXT: vbroadcastsd {{.*#+}} ymm4 = [255,255,255,255] -; AVX2-FAST-NEXT: vblendvpd %ymm2, %ymm1, %ymm4, %ymm1 -; AVX2-FAST-NEXT: vblendvpd %ymm3, %ymm0, %ymm4, %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: vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31] -; AVX2-FAST-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3] -; AVX2-FAST-NEXT: # kill: def $xmm0 killed $xmm0 killed $ymm0 -; AVX2-FAST-NEXT: vzeroupper -; AVX2-FAST-NEXT: retq +; AVX2-LABEL: trunc_usat_v8i64_v8i8: +; AVX2: # %bb.0: +; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm2 = [255,255,255,255] +; AVX2-NEXT: vpbroadcastq {{.*#+}} ymm3 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808] +; AVX2-NEXT: vpxor %ymm3, %ymm1, %ymm4 +; AVX2-NEXT: vpbroadcastq {{.*#+}} ymm5 = [9223372036854776063,9223372036854776063,9223372036854776063,9223372036854776063] +; AVX2-NEXT: vpcmpgtq %ymm4, %ymm5, %ymm4 +; AVX2-NEXT: vblendvpd %ymm4, %ymm1, %ymm2, %ymm1 +; AVX2-NEXT: vpxor %ymm3, %ymm0, %ymm3 +; AVX2-NEXT: vpcmpgtq %ymm3, %ymm5, %ymm3 +; AVX2-NEXT: vblendvpd %ymm3, %ymm0, %ymm2, %ymm0 +; AVX2-NEXT: vpackusdw %ymm1, %ymm0, %ymm0 +; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3] +; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 +; AVX2-NEXT: vpackusdw %xmm1, %xmm0, %xmm0 +; AVX2-NEXT: vzeroupper +; AVX2-NEXT: retq ; ; AVX512-LABEL: trunc_usat_v8i64_v8i8: ; AVX512: # %bb.0: @@ -1297,538 +1202,422 @@ define <8 x i8> @trunc_usat_v8i64_v8i8(<8 x i64> %a0) { define <16 x i8> @trunc_usat_v16i64_v16i8(<16 x i64> %a0) { ; SSE2-LABEL: trunc_usat_v16i64_v16i8: ; SSE2: # %bb.0: -; SSE2-NEXT: movdqa %xmm7, -{{[0-9]+}}(%rsp) # 16-byte Spill -; SSE2-NEXT: movdqa %xmm6, -{{[0-9]+}}(%rsp) # 16-byte Spill -; SSE2-NEXT: movdqa %xmm5, -{{[0-9]+}}(%rsp) # 16-byte Spill -; SSE2-NEXT: movdqa %xmm4, %xmm12 -; SSE2-NEXT: movdqa %xmm4, -{{[0-9]+}}(%rsp) # 16-byte Spill -; SSE2-NEXT: movdqa %xmm3, %xmm4 -; SSE2-NEXT: movdqa %xmm2, %xmm3 -; SSE2-NEXT: movdqa %xmm1, %xmm2 -; SSE2-NEXT: movdqa %xmm0, %xmm1 -; SSE2-NEXT: movdqa {{.*#+}} xmm14 = [2147483648,2147483648,2147483648,2147483648] -; SSE2-NEXT: movdqa %xmm7, %xmm0 -; SSE2-NEXT: pxor %xmm14, %xmm0 -; SSE2-NEXT: movdqa {{.*#+}} xmm15 = [9223372039002259711,9223372039002259711] -; SSE2-NEXT: movdqa %xmm15, %xmm8 -; SSE2-NEXT: pcmpgtd %xmm0, %xmm8 -; SSE2-NEXT: pshufd {{.*#+}} xmm9 = xmm8[0,0,2,2] -; SSE2-NEXT: pcmpeqd %xmm15, %xmm0 -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSE2-NEXT: pand %xmm9, %xmm0 -; SSE2-NEXT: pshufd {{.*#+}} xmm8 = xmm8[1,1,3,3] -; SSE2-NEXT: por %xmm0, %xmm8 -; SSE2-NEXT: movdqa %xmm6, %xmm0 -; SSE2-NEXT: pxor %xmm14, %xmm0 -; SSE2-NEXT: movdqa %xmm15, %xmm9 -; SSE2-NEXT: pcmpgtd %xmm0, %xmm9 -; SSE2-NEXT: pshufd {{.*#+}} xmm10 = xmm9[0,0,2,2] -; SSE2-NEXT: pcmpeqd %xmm15, %xmm0 -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSE2-NEXT: pand %xmm10, %xmm0 -; SSE2-NEXT: pshufd {{.*#+}} xmm9 = xmm9[1,1,3,3] -; SSE2-NEXT: por %xmm0, %xmm9 -; SSE2-NEXT: movdqa %xmm5, %xmm0 -; SSE2-NEXT: pxor %xmm14, %xmm0 -; SSE2-NEXT: movdqa %xmm15, %xmm10 -; SSE2-NEXT: pcmpgtd %xmm0, %xmm10 -; SSE2-NEXT: pshufd {{.*#+}} xmm11 = xmm10[0,0,2,2] -; SSE2-NEXT: pcmpeqd %xmm15, %xmm0 -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSE2-NEXT: pand %xmm11, %xmm0 -; SSE2-NEXT: pshufd {{.*#+}} xmm11 = xmm10[1,1,3,3] -; SSE2-NEXT: por %xmm0, %xmm11 -; SSE2-NEXT: movdqa %xmm12, %xmm0 -; SSE2-NEXT: pxor %xmm14, %xmm0 -; SSE2-NEXT: movdqa %xmm15, %xmm10 -; SSE2-NEXT: pcmpgtd %xmm0, %xmm10 -; SSE2-NEXT: pshufd {{.*#+}} xmm12 = xmm10[0,0,2,2] -; SSE2-NEXT: pcmpeqd %xmm15, %xmm0 -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSE2-NEXT: pand %xmm12, %xmm0 -; SSE2-NEXT: pshufd {{.*#+}} xmm10 = xmm10[1,1,3,3] -; SSE2-NEXT: por %xmm0, %xmm10 -; SSE2-NEXT: movdqa %xmm4, %xmm0 -; SSE2-NEXT: pxor %xmm14, %xmm0 -; SSE2-NEXT: movdqa %xmm15, %xmm12 -; SSE2-NEXT: pcmpgtd %xmm0, %xmm12 -; SSE2-NEXT: pcmpeqd %xmm15, %xmm0 -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm7 = xmm12[0,0,2,2] -; SSE2-NEXT: pand %xmm7, %xmm0 +; SSE2-NEXT: movdqa {{.*#+}} xmm8 = [255,255] +; SSE2-NEXT: movdqa {{.*#+}} xmm9 = [2147483648,2147483648,2147483648,2147483648] +; SSE2-NEXT: movdqa %xmm1, %xmm11 +; SSE2-NEXT: pxor %xmm9, %xmm11 +; SSE2-NEXT: movdqa {{.*#+}} xmm10 = [9223372039002259711,9223372039002259711] +; SSE2-NEXT: movdqa %xmm10, %xmm12 +; SSE2-NEXT: pcmpgtd %xmm11, %xmm12 +; SSE2-NEXT: pshufd {{.*#+}} xmm13 = xmm12[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm10, %xmm11 +; SSE2-NEXT: pshufd {{.*#+}} xmm11 = xmm11[1,1,3,3] +; SSE2-NEXT: pand %xmm13, %xmm11 ; SSE2-NEXT: pshufd {{.*#+}} xmm12 = xmm12[1,1,3,3] -; SSE2-NEXT: por %xmm0, %xmm12 -; SSE2-NEXT: movdqa %xmm3, %xmm0 -; SSE2-NEXT: pxor %xmm14, %xmm0 -; SSE2-NEXT: movdqa %xmm15, %xmm7 -; SSE2-NEXT: pcmpgtd %xmm0, %xmm7 -; SSE2-NEXT: pcmpeqd %xmm15, %xmm0 -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm7[0,0,2,2] -; SSE2-NEXT: pand %xmm6, %xmm0 -; SSE2-NEXT: pshufd {{.*#+}} xmm13 = xmm7[1,1,3,3] -; SSE2-NEXT: por %xmm0, %xmm13 -; SSE2-NEXT: movdqa %xmm2, %xmm0 -; SSE2-NEXT: pxor %xmm14, %xmm0 -; SSE2-NEXT: movdqa %xmm15, %xmm6 -; SSE2-NEXT: pcmpgtd %xmm0, %xmm6 -; SSE2-NEXT: pcmpeqd %xmm15, %xmm0 -; SSE2-NEXT: pshufd {{.*#+}} xmm7 = xmm0[1,1,3,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm6[0,0,2,2] -; SSE2-NEXT: pand %xmm0, %xmm7 -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm6[1,1,3,3] -; SSE2-NEXT: por %xmm7, %xmm0 -; SSE2-NEXT: pxor %xmm1, %xmm14 -; SSE2-NEXT: movdqa %xmm15, %xmm6 -; SSE2-NEXT: pcmpgtd %xmm14, %xmm6 -; SSE2-NEXT: pcmpeqd %xmm15, %xmm14 -; SSE2-NEXT: pshufd {{.*#+}} xmm7 = xmm14[1,1,3,3] -; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm6[0,0,2,2] -; SSE2-NEXT: pand %xmm5, %xmm7 -; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm6[1,1,3,3] -; SSE2-NEXT: por %xmm7, %xmm5 -; SSE2-NEXT: pand %xmm5, %xmm1 -; SSE2-NEXT: movdqa {{.*#+}} xmm14 = [255,255] -; SSE2-NEXT: pandn %xmm14, %xmm5 +; SSE2-NEXT: por %xmm11, %xmm12 +; SSE2-NEXT: pand %xmm12, %xmm1 +; SSE2-NEXT: pandn %xmm8, %xmm12 +; SSE2-NEXT: por %xmm1, %xmm12 +; SSE2-NEXT: movdqa %xmm0, %xmm1 +; SSE2-NEXT: pxor %xmm9, %xmm1 +; SSE2-NEXT: movdqa %xmm10, %xmm11 +; SSE2-NEXT: pcmpgtd %xmm1, %xmm11 +; SSE2-NEXT: pshufd {{.*#+}} xmm13 = xmm11[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm10, %xmm1 +; SSE2-NEXT: pshufd {{.*#+}} xmm14 = xmm1[1,1,3,3] +; SSE2-NEXT: pand %xmm13, %xmm14 +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm11[1,1,3,3] +; SSE2-NEXT: por %xmm14, %xmm1 +; SSE2-NEXT: pand %xmm1, %xmm0 +; SSE2-NEXT: pandn %xmm8, %xmm1 +; SSE2-NEXT: por %xmm1, %xmm0 +; SSE2-NEXT: packuswb %xmm12, %xmm0 +; SSE2-NEXT: movdqa %xmm3, %xmm1 +; SSE2-NEXT: pxor %xmm9, %xmm1 +; SSE2-NEXT: movdqa %xmm10, %xmm11 +; SSE2-NEXT: pcmpgtd %xmm1, %xmm11 +; SSE2-NEXT: pshufd {{.*#+}} xmm12 = xmm11[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm10, %xmm1 +; SSE2-NEXT: pshufd {{.*#+}} xmm13 = xmm1[1,1,3,3] +; SSE2-NEXT: pand %xmm12, %xmm13 +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm11[1,1,3,3] +; SSE2-NEXT: por %xmm13, %xmm1 +; SSE2-NEXT: pand %xmm1, %xmm3 +; SSE2-NEXT: pandn %xmm8, %xmm1 +; SSE2-NEXT: por %xmm3, %xmm1 +; SSE2-NEXT: movdqa %xmm2, %xmm3 +; SSE2-NEXT: pxor %xmm9, %xmm3 +; SSE2-NEXT: movdqa %xmm10, %xmm11 +; SSE2-NEXT: pcmpgtd %xmm3, %xmm11 +; SSE2-NEXT: pshufd {{.*#+}} xmm12 = xmm11[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm10, %xmm3 +; SSE2-NEXT: pshufd {{.*#+}} xmm13 = xmm3[1,1,3,3] +; SSE2-NEXT: pand %xmm12, %xmm13 +; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm11[1,1,3,3] +; SSE2-NEXT: por %xmm13, %xmm3 +; SSE2-NEXT: pand %xmm3, %xmm2 +; SSE2-NEXT: pandn %xmm8, %xmm3 +; SSE2-NEXT: por %xmm2, %xmm3 +; SSE2-NEXT: packuswb %xmm1, %xmm3 +; SSE2-NEXT: packuswb %xmm3, %xmm0 +; SSE2-NEXT: movdqa %xmm5, %xmm1 +; SSE2-NEXT: pxor %xmm9, %xmm1 +; SSE2-NEXT: movdqa %xmm10, %xmm2 +; SSE2-NEXT: pcmpgtd %xmm1, %xmm2 +; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm10, %xmm1 +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3] +; SSE2-NEXT: pand %xmm3, %xmm1 +; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3] +; SSE2-NEXT: por %xmm1, %xmm2 +; SSE2-NEXT: pand %xmm2, %xmm5 +; SSE2-NEXT: pandn %xmm8, %xmm2 +; SSE2-NEXT: por %xmm5, %xmm2 +; SSE2-NEXT: movdqa %xmm4, %xmm1 +; SSE2-NEXT: pxor %xmm9, %xmm1 +; SSE2-NEXT: movdqa %xmm10, %xmm3 +; SSE2-NEXT: pcmpgtd %xmm1, %xmm3 +; SSE2-NEXT: pshufd {{.*#+}} xmm11 = xmm3[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm10, %xmm1 +; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm1[1,1,3,3] +; SSE2-NEXT: pand %xmm11, %xmm5 +; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm3[1,1,3,3] ; SSE2-NEXT: por %xmm5, %xmm1 -; SSE2-NEXT: pand %xmm0, %xmm2 -; SSE2-NEXT: pandn %xmm14, %xmm0 -; SSE2-NEXT: por %xmm2, %xmm0 -; SSE2-NEXT: pand %xmm13, %xmm3 -; SSE2-NEXT: pandn %xmm14, %xmm13 -; SSE2-NEXT: por %xmm3, %xmm13 -; SSE2-NEXT: pand %xmm12, %xmm4 -; SSE2-NEXT: pandn %xmm14, %xmm12 -; SSE2-NEXT: por %xmm4, %xmm12 -; SSE2-NEXT: movdqa -{{[0-9]+}}(%rsp), %xmm2 # 16-byte Reload -; SSE2-NEXT: pand %xmm10, %xmm2 -; SSE2-NEXT: pandn %xmm14, %xmm10 -; SSE2-NEXT: por %xmm2, %xmm10 -; SSE2-NEXT: movdqa -{{[0-9]+}}(%rsp), %xmm2 # 16-byte Reload -; SSE2-NEXT: pand %xmm11, %xmm2 -; SSE2-NEXT: pandn %xmm14, %xmm11 -; SSE2-NEXT: por %xmm2, %xmm11 -; SSE2-NEXT: movdqa -{{[0-9]+}}(%rsp), %xmm2 # 16-byte Reload -; SSE2-NEXT: pand %xmm9, %xmm2 -; SSE2-NEXT: pandn %xmm14, %xmm9 -; SSE2-NEXT: por %xmm2, %xmm9 -; SSE2-NEXT: movdqa -{{[0-9]+}}(%rsp), %xmm2 # 16-byte Reload -; SSE2-NEXT: pand %xmm8, %xmm2 -; SSE2-NEXT: pandn %xmm14, %xmm8 -; SSE2-NEXT: por %xmm2, %xmm8 -; SSE2-NEXT: pand %xmm14, %xmm8 -; SSE2-NEXT: pand %xmm14, %xmm9 -; SSE2-NEXT: packuswb %xmm8, %xmm9 -; SSE2-NEXT: pand %xmm14, %xmm11 -; SSE2-NEXT: pand %xmm14, %xmm10 -; SSE2-NEXT: packuswb %xmm11, %xmm10 -; SSE2-NEXT: packuswb %xmm9, %xmm10 -; SSE2-NEXT: pand %xmm14, %xmm12 -; SSE2-NEXT: pand %xmm14, %xmm13 -; SSE2-NEXT: packuswb %xmm12, %xmm13 -; SSE2-NEXT: pand %xmm14, %xmm0 -; SSE2-NEXT: pand %xmm14, %xmm1 -; SSE2-NEXT: packuswb %xmm0, %xmm1 -; SSE2-NEXT: packuswb %xmm13, %xmm1 -; SSE2-NEXT: packuswb %xmm10, %xmm1 -; SSE2-NEXT: movdqa %xmm1, %xmm0 +; SSE2-NEXT: pand %xmm1, %xmm4 +; SSE2-NEXT: pandn %xmm8, %xmm1 +; SSE2-NEXT: por %xmm4, %xmm1 +; SSE2-NEXT: packuswb %xmm2, %xmm1 +; SSE2-NEXT: movdqa %xmm7, %xmm2 +; SSE2-NEXT: pxor %xmm9, %xmm2 +; SSE2-NEXT: movdqa %xmm10, %xmm3 +; SSE2-NEXT: pcmpgtd %xmm2, %xmm3 +; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm10, %xmm2 +; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3] +; SSE2-NEXT: pand %xmm4, %xmm2 +; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3] +; SSE2-NEXT: por %xmm2, %xmm3 +; SSE2-NEXT: pand %xmm3, %xmm7 +; SSE2-NEXT: pandn %xmm8, %xmm3 +; SSE2-NEXT: por %xmm7, %xmm3 +; SSE2-NEXT: pxor %xmm6, %xmm9 +; SSE2-NEXT: movdqa %xmm10, %xmm2 +; SSE2-NEXT: pcmpgtd %xmm9, %xmm2 +; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2] +; SSE2-NEXT: pcmpeqd %xmm10, %xmm9 +; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm9[1,1,3,3] +; SSE2-NEXT: pand %xmm4, %xmm5 +; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3] +; SSE2-NEXT: por %xmm5, %xmm2 +; SSE2-NEXT: pand %xmm2, %xmm6 +; SSE2-NEXT: pandn %xmm8, %xmm2 +; SSE2-NEXT: por %xmm6, %xmm2 +; SSE2-NEXT: packuswb %xmm3, %xmm2 +; SSE2-NEXT: packuswb %xmm2, %xmm1 +; SSE2-NEXT: packuswb %xmm1, %xmm0 ; SSE2-NEXT: retq ; ; SSSE3-LABEL: trunc_usat_v16i64_v16i8: ; SSSE3: # %bb.0: -; SSSE3-NEXT: movdqa %xmm7, -{{[0-9]+}}(%rsp) # 16-byte Spill -; SSSE3-NEXT: movdqa %xmm6, -{{[0-9]+}}(%rsp) # 16-byte Spill -; SSSE3-NEXT: movdqa %xmm5, -{{[0-9]+}}(%rsp) # 16-byte Spill -; SSSE3-NEXT: movdqa %xmm4, %xmm12 -; SSSE3-NEXT: movdqa %xmm4, -{{[0-9]+}}(%rsp) # 16-byte Spill -; SSSE3-NEXT: movdqa %xmm3, %xmm4 -; SSSE3-NEXT: movdqa %xmm2, %xmm3 -; SSSE3-NEXT: movdqa %xmm1, %xmm2 -; SSSE3-NEXT: movdqa %xmm0, %xmm1 -; SSSE3-NEXT: movdqa {{.*#+}} xmm14 = [2147483648,2147483648,2147483648,2147483648] -; SSSE3-NEXT: movdqa %xmm7, %xmm0 -; SSSE3-NEXT: pxor %xmm14, %xmm0 -; SSSE3-NEXT: movdqa {{.*#+}} xmm15 = [9223372039002259711,9223372039002259711] -; SSSE3-NEXT: movdqa %xmm15, %xmm8 -; SSSE3-NEXT: pcmpgtd %xmm0, %xmm8 -; SSSE3-NEXT: pshufd {{.*#+}} xmm9 = xmm8[0,0,2,2] -; SSSE3-NEXT: pcmpeqd %xmm15, %xmm0 -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSSE3-NEXT: pand %xmm9, %xmm0 -; SSSE3-NEXT: pshufd {{.*#+}} xmm8 = xmm8[1,1,3,3] -; SSSE3-NEXT: por %xmm0, %xmm8 -; SSSE3-NEXT: movdqa %xmm6, %xmm0 -; SSSE3-NEXT: pxor %xmm14, %xmm0 -; SSSE3-NEXT: movdqa %xmm15, %xmm9 -; SSSE3-NEXT: pcmpgtd %xmm0, %xmm9 -; SSSE3-NEXT: pshufd {{.*#+}} xmm10 = xmm9[0,0,2,2] -; SSSE3-NEXT: pcmpeqd %xmm15, %xmm0 -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSSE3-NEXT: pand %xmm10, %xmm0 -; SSSE3-NEXT: pshufd {{.*#+}} xmm9 = xmm9[1,1,3,3] -; SSSE3-NEXT: por %xmm0, %xmm9 -; SSSE3-NEXT: movdqa %xmm5, %xmm0 -; SSSE3-NEXT: pxor %xmm14, %xmm0 -; SSSE3-NEXT: movdqa %xmm15, %xmm10 -; SSSE3-NEXT: pcmpgtd %xmm0, %xmm10 -; SSSE3-NEXT: pshufd {{.*#+}} xmm11 = xmm10[0,0,2,2] -; SSSE3-NEXT: pcmpeqd %xmm15, %xmm0 -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSSE3-NEXT: pand %xmm11, %xmm0 -; SSSE3-NEXT: pshufd {{.*#+}} xmm11 = xmm10[1,1,3,3] -; SSSE3-NEXT: por %xmm0, %xmm11 -; SSSE3-NEXT: movdqa %xmm12, %xmm0 -; SSSE3-NEXT: pxor %xmm14, %xmm0 -; SSSE3-NEXT: movdqa %xmm15, %xmm10 -; SSSE3-NEXT: pcmpgtd %xmm0, %xmm10 -; SSSE3-NEXT: pshufd {{.*#+}} xmm12 = xmm10[0,0,2,2] -; SSSE3-NEXT: pcmpeqd %xmm15, %xmm0 -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSSE3-NEXT: pand %xmm12, %xmm0 -; SSSE3-NEXT: pshufd {{.*#+}} xmm10 = xmm10[1,1,3,3] -; SSSE3-NEXT: por %xmm0, %xmm10 -; SSSE3-NEXT: movdqa %xmm4, %xmm0 -; SSSE3-NEXT: pxor %xmm14, %xmm0 -; SSSE3-NEXT: movdqa %xmm15, %xmm12 -; SSSE3-NEXT: pcmpgtd %xmm0, %xmm12 -; SSSE3-NEXT: pcmpeqd %xmm15, %xmm0 -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSSE3-NEXT: pshufd {{.*#+}} xmm7 = xmm12[0,0,2,2] -; SSSE3-NEXT: pand %xmm7, %xmm0 +; SSSE3-NEXT: movdqa {{.*#+}} xmm8 = [255,255] +; SSSE3-NEXT: movdqa {{.*#+}} xmm9 = [2147483648,2147483648,2147483648,2147483648] +; SSSE3-NEXT: movdqa %xmm1, %xmm11 +; SSSE3-NEXT: pxor %xmm9, %xmm11 +; SSSE3-NEXT: movdqa {{.*#+}} xmm10 = [9223372039002259711,9223372039002259711] +; SSSE3-NEXT: movdqa %xmm10, %xmm12 +; SSSE3-NEXT: pcmpgtd %xmm11, %xmm12 +; SSSE3-NEXT: pshufd {{.*#+}} xmm13 = xmm12[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm10, %xmm11 +; SSSE3-NEXT: pshufd {{.*#+}} xmm11 = xmm11[1,1,3,3] +; SSSE3-NEXT: pand %xmm13, %xmm11 ; SSSE3-NEXT: pshufd {{.*#+}} xmm12 = xmm12[1,1,3,3] -; SSSE3-NEXT: por %xmm0, %xmm12 -; SSSE3-NEXT: movdqa %xmm3, %xmm0 -; SSSE3-NEXT: pxor %xmm14, %xmm0 -; SSSE3-NEXT: movdqa %xmm15, %xmm7 -; SSSE3-NEXT: pcmpgtd %xmm0, %xmm7 -; SSSE3-NEXT: pcmpeqd %xmm15, %xmm0 -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSSE3-NEXT: pshufd {{.*#+}} xmm6 = xmm7[0,0,2,2] -; SSSE3-NEXT: pand %xmm6, %xmm0 -; SSSE3-NEXT: pshufd {{.*#+}} xmm13 = xmm7[1,1,3,3] -; SSSE3-NEXT: por %xmm0, %xmm13 -; SSSE3-NEXT: movdqa %xmm2, %xmm0 -; SSSE3-NEXT: pxor %xmm14, %xmm0 -; SSSE3-NEXT: movdqa %xmm15, %xmm6 -; SSSE3-NEXT: pcmpgtd %xmm0, %xmm6 -; SSSE3-NEXT: pcmpeqd %xmm15, %xmm0 -; SSSE3-NEXT: pshufd {{.*#+}} xmm7 = xmm0[1,1,3,3] -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm6[0,0,2,2] -; SSSE3-NEXT: pand %xmm0, %xmm7 -; SSSE3-NEXT: pshufd {{.*#+}} xmm0 = xmm6[1,1,3,3] -; SSSE3-NEXT: por %xmm7, %xmm0 -; SSSE3-NEXT: pxor %xmm1, %xmm14 -; SSSE3-NEXT: movdqa %xmm15, %xmm6 -; SSSE3-NEXT: pcmpgtd %xmm14, %xmm6 -; SSSE3-NEXT: pcmpeqd %xmm15, %xmm14 -; SSSE3-NEXT: pshufd {{.*#+}} xmm7 = xmm14[1,1,3,3] -; SSSE3-NEXT: pshufd {{.*#+}} xmm5 = xmm6[0,0,2,2] -; SSSE3-NEXT: pand %xmm5, %xmm7 -; SSSE3-NEXT: pshufd {{.*#+}} xmm5 = xmm6[1,1,3,3] -; SSSE3-NEXT: por %xmm7, %xmm5 -; SSSE3-NEXT: pand %xmm5, %xmm1 -; SSSE3-NEXT: movdqa {{.*#+}} xmm14 = [255,255] -; SSSE3-NEXT: pandn %xmm14, %xmm5 +; SSSE3-NEXT: por %xmm11, %xmm12 +; SSSE3-NEXT: pand %xmm12, %xmm1 +; SSSE3-NEXT: pandn %xmm8, %xmm12 +; SSSE3-NEXT: por %xmm1, %xmm12 +; SSSE3-NEXT: movdqa %xmm0, %xmm1 +; SSSE3-NEXT: pxor %xmm9, %xmm1 +; SSSE3-NEXT: movdqa %xmm10, %xmm11 +; SSSE3-NEXT: pcmpgtd %xmm1, %xmm11 +; SSSE3-NEXT: pshufd {{.*#+}} xmm13 = xmm11[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm10, %xmm1 +; SSSE3-NEXT: pshufd {{.*#+}} xmm14 = xmm1[1,1,3,3] +; SSSE3-NEXT: pand %xmm13, %xmm14 +; SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm11[1,1,3,3] +; SSSE3-NEXT: por %xmm14, %xmm1 +; SSSE3-NEXT: pand %xmm1, %xmm0 +; SSSE3-NEXT: pandn %xmm8, %xmm1 +; SSSE3-NEXT: por %xmm1, %xmm0 +; SSSE3-NEXT: packuswb %xmm12, %xmm0 +; SSSE3-NEXT: movdqa %xmm3, %xmm1 +; SSSE3-NEXT: pxor %xmm9, %xmm1 +; SSSE3-NEXT: movdqa %xmm10, %xmm11 +; SSSE3-NEXT: pcmpgtd %xmm1, %xmm11 +; SSSE3-NEXT: pshufd {{.*#+}} xmm12 = xmm11[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm10, %xmm1 +; SSSE3-NEXT: pshufd {{.*#+}} xmm13 = xmm1[1,1,3,3] +; SSSE3-NEXT: pand %xmm12, %xmm13 +; SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm11[1,1,3,3] +; SSSE3-NEXT: por %xmm13, %xmm1 +; SSSE3-NEXT: pand %xmm1, %xmm3 +; SSSE3-NEXT: pandn %xmm8, %xmm1 +; SSSE3-NEXT: por %xmm3, %xmm1 +; SSSE3-NEXT: movdqa %xmm2, %xmm3 +; SSSE3-NEXT: pxor %xmm9, %xmm3 +; SSSE3-NEXT: movdqa %xmm10, %xmm11 +; SSSE3-NEXT: pcmpgtd %xmm3, %xmm11 +; SSSE3-NEXT: pshufd {{.*#+}} xmm12 = xmm11[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm10, %xmm3 +; SSSE3-NEXT: pshufd {{.*#+}} xmm13 = xmm3[1,1,3,3] +; SSSE3-NEXT: pand %xmm12, %xmm13 +; SSSE3-NEXT: pshufd {{.*#+}} xmm3 = xmm11[1,1,3,3] +; SSSE3-NEXT: por %xmm13, %xmm3 +; SSSE3-NEXT: pand %xmm3, %xmm2 +; SSSE3-NEXT: pandn %xmm8, %xmm3 +; SSSE3-NEXT: por %xmm2, %xmm3 +; SSSE3-NEXT: packuswb %xmm1, %xmm3 +; SSSE3-NEXT: packuswb %xmm3, %xmm0 +; SSSE3-NEXT: movdqa %xmm5, %xmm1 +; SSSE3-NEXT: pxor %xmm9, %xmm1 +; SSSE3-NEXT: movdqa %xmm10, %xmm2 +; SSSE3-NEXT: pcmpgtd %xmm1, %xmm2 +; SSSE3-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm10, %xmm1 +; SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3] +; SSSE3-NEXT: pand %xmm3, %xmm1 +; SSSE3-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3] +; SSSE3-NEXT: por %xmm1, %xmm2 +; SSSE3-NEXT: pand %xmm2, %xmm5 +; SSSE3-NEXT: pandn %xmm8, %xmm2 +; SSSE3-NEXT: por %xmm5, %xmm2 +; SSSE3-NEXT: movdqa %xmm4, %xmm1 +; SSSE3-NEXT: pxor %xmm9, %xmm1 +; SSSE3-NEXT: movdqa %xmm10, %xmm3 +; SSSE3-NEXT: pcmpgtd %xmm1, %xmm3 +; SSSE3-NEXT: pshufd {{.*#+}} xmm11 = xmm3[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm10, %xmm1 +; SSSE3-NEXT: pshufd {{.*#+}} xmm5 = xmm1[1,1,3,3] +; SSSE3-NEXT: pand %xmm11, %xmm5 +; SSSE3-NEXT: pshufd {{.*#+}} xmm1 = xmm3[1,1,3,3] ; SSSE3-NEXT: por %xmm5, %xmm1 -; SSSE3-NEXT: pand %xmm0, %xmm2 -; SSSE3-NEXT: pandn %xmm14, %xmm0 -; SSSE3-NEXT: por %xmm2, %xmm0 -; SSSE3-NEXT: pand %xmm13, %xmm3 -; SSSE3-NEXT: pandn %xmm14, %xmm13 -; SSSE3-NEXT: por %xmm3, %xmm13 -; SSSE3-NEXT: pand %xmm12, %xmm4 -; SSSE3-NEXT: pandn %xmm14, %xmm12 -; SSSE3-NEXT: por %xmm4, %xmm12 -; SSSE3-NEXT: movdqa -{{[0-9]+}}(%rsp), %xmm2 # 16-byte Reload -; SSSE3-NEXT: pand %xmm10, %xmm2 -; SSSE3-NEXT: pandn %xmm14, %xmm10 -; SSSE3-NEXT: por %xmm2, %xmm10 -; SSSE3-NEXT: movdqa -{{[0-9]+}}(%rsp), %xmm2 # 16-byte Reload -; SSSE3-NEXT: pand %xmm11, %xmm2 -; SSSE3-NEXT: pandn %xmm14, %xmm11 -; SSSE3-NEXT: por %xmm2, %xmm11 -; SSSE3-NEXT: movdqa -{{[0-9]+}}(%rsp), %xmm2 # 16-byte Reload -; SSSE3-NEXT: pand %xmm9, %xmm2 -; SSSE3-NEXT: pandn %xmm14, %xmm9 -; SSSE3-NEXT: por %xmm2, %xmm9 -; SSSE3-NEXT: movdqa -{{[0-9]+}}(%rsp), %xmm2 # 16-byte Reload -; SSSE3-NEXT: pand %xmm8, %xmm2 -; SSSE3-NEXT: pandn %xmm14, %xmm8 -; SSSE3-NEXT: por %xmm2, %xmm8 -; SSSE3-NEXT: pand %xmm14, %xmm8 -; SSSE3-NEXT: pand %xmm14, %xmm9 -; SSSE3-NEXT: packuswb %xmm8, %xmm9 -; SSSE3-NEXT: pand %xmm14, %xmm11 -; SSSE3-NEXT: pand %xmm14, %xmm10 -; SSSE3-NEXT: packuswb %xmm11, %xmm10 -; SSSE3-NEXT: packuswb %xmm9, %xmm10 -; SSSE3-NEXT: pand %xmm14, %xmm12 -; SSSE3-NEXT: pand %xmm14, %xmm13 -; SSSE3-NEXT: packuswb %xmm12, %xmm13 -; SSSE3-NEXT: pand %xmm14, %xmm0 -; SSSE3-NEXT: pand %xmm14, %xmm1 -; SSSE3-NEXT: packuswb %xmm0, %xmm1 -; SSSE3-NEXT: packuswb %xmm13, %xmm1 -; SSSE3-NEXT: packuswb %xmm10, %xmm1 -; SSSE3-NEXT: movdqa %xmm1, %xmm0 +; SSSE3-NEXT: pand %xmm1, %xmm4 +; SSSE3-NEXT: pandn %xmm8, %xmm1 +; SSSE3-NEXT: por %xmm4, %xmm1 +; SSSE3-NEXT: packuswb %xmm2, %xmm1 +; SSSE3-NEXT: movdqa %xmm7, %xmm2 +; SSSE3-NEXT: pxor %xmm9, %xmm2 +; SSSE3-NEXT: movdqa %xmm10, %xmm3 +; SSSE3-NEXT: pcmpgtd %xmm2, %xmm3 +; SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm10, %xmm2 +; SSSE3-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3] +; SSSE3-NEXT: pand %xmm4, %xmm2 +; SSSE3-NEXT: pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3] +; SSSE3-NEXT: por %xmm2, %xmm3 +; SSSE3-NEXT: pand %xmm3, %xmm7 +; SSSE3-NEXT: pandn %xmm8, %xmm3 +; SSSE3-NEXT: por %xmm7, %xmm3 +; SSSE3-NEXT: pxor %xmm6, %xmm9 +; SSSE3-NEXT: movdqa %xmm10, %xmm2 +; SSSE3-NEXT: pcmpgtd %xmm9, %xmm2 +; SSSE3-NEXT: pshufd {{.*#+}} xmm4 = xmm2[0,0,2,2] +; SSSE3-NEXT: pcmpeqd %xmm10, %xmm9 +; SSSE3-NEXT: pshufd {{.*#+}} xmm5 = xmm9[1,1,3,3] +; SSSE3-NEXT: pand %xmm4, %xmm5 +; SSSE3-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3] +; SSSE3-NEXT: por %xmm5, %xmm2 +; SSSE3-NEXT: pand %xmm2, %xmm6 +; SSSE3-NEXT: pandn %xmm8, %xmm2 +; SSSE3-NEXT: por %xmm6, %xmm2 +; SSSE3-NEXT: packuswb %xmm3, %xmm2 +; SSSE3-NEXT: packuswb %xmm2, %xmm1 +; SSSE3-NEXT: packuswb %xmm1, %xmm0 ; SSSE3-NEXT: retq ; ; SSE41-LABEL: trunc_usat_v16i64_v16i8: ; SSE41: # %bb.0: -; SSE41-NEXT: movdqa %xmm7, -{{[0-9]+}}(%rsp) # 16-byte Spill -; SSE41-NEXT: movdqa %xmm6, %xmm11 -; SSE41-NEXT: movdqa %xmm6, -{{[0-9]+}}(%rsp) # 16-byte Spill -; SSE41-NEXT: movdqa %xmm5, %xmm12 -; SSE41-NEXT: movdqa %xmm5, -{{[0-9]+}}(%rsp) # 16-byte Spill -; SSE41-NEXT: movdqa %xmm4, %xmm13 -; SSE41-NEXT: movdqa %xmm4, -{{[0-9]+}}(%rsp) # 16-byte Spill -; SSE41-NEXT: movdqa %xmm3, %xmm6 -; SSE41-NEXT: movdqa %xmm2, %xmm5 -; SSE41-NEXT: movdqa %xmm1, %xmm4 ; SSE41-NEXT: movdqa %xmm0, %xmm8 -; SSE41-NEXT: movdqa {{.*#+}} xmm9 = [2147483648,2147483648,2147483648,2147483648] -; SSE41-NEXT: movdqa %xmm7, %xmm10 -; SSE41-NEXT: pxor %xmm9, %xmm10 -; SSE41-NEXT: pxor %xmm9, %xmm11 -; SSE41-NEXT: pxor %xmm9, %xmm12 -; SSE41-NEXT: pxor %xmm9, %xmm13 -; SSE41-NEXT: movdqa %xmm3, %xmm14 -; SSE41-NEXT: pxor %xmm9, %xmm14 -; SSE41-NEXT: movdqa %xmm2, %xmm3 -; SSE41-NEXT: pxor %xmm9, %xmm3 -; SSE41-NEXT: movdqa %xmm1, %xmm2 -; SSE41-NEXT: pxor %xmm9, %xmm2 -; SSE41-NEXT: pxor %xmm0, %xmm9 -; SSE41-NEXT: movdqa {{.*#+}} xmm1 = [9223372039002259711,9223372039002259711] -; SSE41-NEXT: movdqa %xmm1, %xmm0 -; SSE41-NEXT: pcmpgtd %xmm9, %xmm0 -; SSE41-NEXT: pcmpeqd %xmm1, %xmm9 -; SSE41-NEXT: pshufd {{.*#+}} xmm9 = xmm9[1,1,3,3] -; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm0[0,0,2,2] -; SSE41-NEXT: pand %xmm7, %xmm9 -; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSE41-NEXT: por %xmm9, %xmm0 -; SSE41-NEXT: movapd {{.*#+}} xmm15 = [255,255] -; SSE41-NEXT: movapd %xmm15, %xmm9 -; SSE41-NEXT: blendvpd %xmm0, %xmm8, %xmm9 -; SSE41-NEXT: movdqa %xmm1, %xmm0 -; SSE41-NEXT: pcmpgtd %xmm2, %xmm0 -; SSE41-NEXT: pcmpeqd %xmm1, %xmm2 -; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3] -; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm0[0,0,2,2] -; SSE41-NEXT: pand %xmm7, %xmm2 -; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSE41-NEXT: por %xmm2, %xmm0 -; SSE41-NEXT: movapd %xmm15, %xmm8 -; SSE41-NEXT: blendvpd %xmm0, %xmm4, %xmm8 +; SSE41-NEXT: movapd {{.*#+}} xmm9 = [255,255] +; SSE41-NEXT: movdqa {{.*#+}} xmm10 = [2147483648,2147483648,2147483648,2147483648] ; SSE41-NEXT: movdqa %xmm1, %xmm0 -; SSE41-NEXT: pcmpgtd %xmm3, %xmm0 -; SSE41-NEXT: pcmpeqd %xmm1, %xmm3 -; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3] -; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm0[0,0,2,2] -; SSE41-NEXT: pand %xmm3, %xmm2 -; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSE41-NEXT: por %xmm2, %xmm0 -; SSE41-NEXT: movapd %xmm15, %xmm2 -; SSE41-NEXT: blendvpd %xmm0, %xmm5, %xmm2 -; SSE41-NEXT: movdqa %xmm1, %xmm0 -; SSE41-NEXT: pcmpgtd %xmm14, %xmm0 -; SSE41-NEXT: pcmpeqd %xmm1, %xmm14 -; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm14[1,1,3,3] -; SSE41-NEXT: pshufd {{.*#+}} xmm4 = xmm0[0,0,2,2] -; SSE41-NEXT: pand %xmm4, %xmm3 -; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] +; SSE41-NEXT: pxor %xmm10, %xmm0 +; SSE41-NEXT: movdqa {{.*#+}} xmm11 = [9223372039002259711,9223372039002259711] +; SSE41-NEXT: movdqa %xmm11, %xmm12 +; SSE41-NEXT: pcmpgtd %xmm0, %xmm12 +; SSE41-NEXT: pshufd {{.*#+}} xmm13 = xmm12[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm11, %xmm0 +; SSE41-NEXT: pshufd {{.*#+}} xmm14 = xmm0[1,1,3,3] +; SSE41-NEXT: pand %xmm13, %xmm14 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm12[1,1,3,3] +; SSE41-NEXT: por %xmm14, %xmm0 +; SSE41-NEXT: movapd %xmm9, %xmm12 +; SSE41-NEXT: blendvpd %xmm0, %xmm1, %xmm12 +; SSE41-NEXT: movdqa %xmm8, %xmm0 +; SSE41-NEXT: pxor %xmm10, %xmm0 +; SSE41-NEXT: movdqa %xmm11, %xmm1 +; SSE41-NEXT: pcmpgtd %xmm0, %xmm1 +; SSE41-NEXT: pshufd {{.*#+}} xmm13 = xmm1[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm11, %xmm0 +; SSE41-NEXT: pshufd {{.*#+}} xmm14 = xmm0[1,1,3,3] +; SSE41-NEXT: pand %xmm13, %xmm14 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3] +; SSE41-NEXT: por %xmm14, %xmm0 +; SSE41-NEXT: movapd %xmm9, %xmm13 +; SSE41-NEXT: blendvpd %xmm0, %xmm8, %xmm13 +; SSE41-NEXT: packssdw %xmm12, %xmm13 +; SSE41-NEXT: movdqa %xmm3, %xmm0 +; SSE41-NEXT: pxor %xmm10, %xmm0 +; SSE41-NEXT: movdqa %xmm11, %xmm1 +; SSE41-NEXT: pcmpgtd %xmm0, %xmm1 +; SSE41-NEXT: pshufd {{.*#+}} xmm8 = xmm1[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm11, %xmm0 +; SSE41-NEXT: pshufd {{.*#+}} xmm12 = xmm0[1,1,3,3] +; SSE41-NEXT: pand %xmm8, %xmm12 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3] +; SSE41-NEXT: por %xmm12, %xmm0 +; SSE41-NEXT: movapd %xmm9, %xmm12 +; SSE41-NEXT: blendvpd %xmm0, %xmm3, %xmm12 +; SSE41-NEXT: movdqa %xmm2, %xmm0 +; SSE41-NEXT: pxor %xmm10, %xmm0 +; SSE41-NEXT: movdqa %xmm11, %xmm3 +; SSE41-NEXT: pcmpgtd %xmm0, %xmm3 +; SSE41-NEXT: pshufd {{.*#+}} xmm8 = xmm3[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm11, %xmm0 +; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,3,3] +; SSE41-NEXT: pand %xmm8, %xmm1 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm3[1,1,3,3] +; SSE41-NEXT: por %xmm1, %xmm0 +; SSE41-NEXT: movapd %xmm9, %xmm1 +; SSE41-NEXT: blendvpd %xmm0, %xmm2, %xmm1 +; SSE41-NEXT: packssdw %xmm12, %xmm1 +; SSE41-NEXT: packusdw %xmm1, %xmm13 +; SSE41-NEXT: movdqa %xmm5, %xmm0 +; SSE41-NEXT: pxor %xmm10, %xmm0 +; SSE41-NEXT: movdqa %xmm11, %xmm1 +; SSE41-NEXT: pcmpgtd %xmm0, %xmm1 +; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm1[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm11, %xmm0 +; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3] +; SSE41-NEXT: pand %xmm2, %xmm3 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3] ; SSE41-NEXT: por %xmm3, %xmm0 -; SSE41-NEXT: movapd %xmm15, %xmm3 -; SSE41-NEXT: blendvpd %xmm0, %xmm6, %xmm3 -; SSE41-NEXT: movdqa %xmm1, %xmm0 -; SSE41-NEXT: pcmpgtd %xmm13, %xmm0 -; SSE41-NEXT: pshufd {{.*#+}} xmm4 = xmm0[0,0,2,2] -; SSE41-NEXT: pcmpeqd %xmm1, %xmm13 -; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm13[1,1,3,3] -; SSE41-NEXT: pand %xmm4, %xmm5 -; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] +; SSE41-NEXT: movapd %xmm9, %xmm1 +; SSE41-NEXT: blendvpd %xmm0, %xmm5, %xmm1 +; SSE41-NEXT: movdqa %xmm4, %xmm0 +; SSE41-NEXT: pxor %xmm10, %xmm0 +; SSE41-NEXT: movdqa %xmm11, %xmm2 +; SSE41-NEXT: pcmpgtd %xmm0, %xmm2 +; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm2[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm11, %xmm0 +; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm0[1,1,3,3] +; SSE41-NEXT: pand %xmm3, %xmm5 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm2[1,1,3,3] ; SSE41-NEXT: por %xmm5, %xmm0 -; SSE41-NEXT: movapd %xmm15, %xmm4 -; SSE41-NEXT: blendvpd %xmm0, -{{[0-9]+}}(%rsp), %xmm4 # 16-byte Folded Reload -; SSE41-NEXT: movdqa %xmm1, %xmm0 -; SSE41-NEXT: pcmpgtd %xmm12, %xmm0 -; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm0[0,0,2,2] -; SSE41-NEXT: pcmpeqd %xmm1, %xmm12 -; SSE41-NEXT: pshufd {{.*#+}} xmm6 = xmm12[1,1,3,3] -; SSE41-NEXT: pand %xmm5, %xmm6 -; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSE41-NEXT: por %xmm6, %xmm0 -; SSE41-NEXT: movapd %xmm15, %xmm5 -; SSE41-NEXT: blendvpd %xmm0, -{{[0-9]+}}(%rsp), %xmm5 # 16-byte Folded Reload -; SSE41-NEXT: movdqa %xmm1, %xmm0 -; SSE41-NEXT: pcmpgtd %xmm11, %xmm0 -; SSE41-NEXT: pshufd {{.*#+}} xmm6 = xmm0[0,0,2,2] -; SSE41-NEXT: pcmpeqd %xmm1, %xmm11 -; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm11[1,1,3,3] -; SSE41-NEXT: pand %xmm6, %xmm7 -; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSE41-NEXT: por %xmm7, %xmm0 -; SSE41-NEXT: movapd %xmm15, %xmm6 -; SSE41-NEXT: blendvpd %xmm0, -{{[0-9]+}}(%rsp), %xmm6 # 16-byte Folded Reload -; SSE41-NEXT: movdqa %xmm1, %xmm0 +; SSE41-NEXT: movapd %xmm9, %xmm2 +; SSE41-NEXT: blendvpd %xmm0, %xmm4, %xmm2 +; SSE41-NEXT: packssdw %xmm1, %xmm2 +; SSE41-NEXT: movdqa %xmm7, %xmm0 +; SSE41-NEXT: pxor %xmm10, %xmm0 +; SSE41-NEXT: movdqa %xmm11, %xmm1 +; SSE41-NEXT: pcmpgtd %xmm0, %xmm1 +; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm1[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm11, %xmm0 +; SSE41-NEXT: pshufd {{.*#+}} xmm4 = xmm0[1,1,3,3] +; SSE41-NEXT: pand %xmm3, %xmm4 +; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3] +; SSE41-NEXT: por %xmm4, %xmm0 +; SSE41-NEXT: movapd %xmm9, %xmm1 +; SSE41-NEXT: blendvpd %xmm0, %xmm7, %xmm1 +; SSE41-NEXT: pxor %xmm6, %xmm10 +; SSE41-NEXT: movdqa %xmm11, %xmm0 ; SSE41-NEXT: pcmpgtd %xmm10, %xmm0 -; SSE41-NEXT: pcmpeqd %xmm1, %xmm10 -; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm0[0,0,2,2] -; SSE41-NEXT: pshufd {{.*#+}} xmm7 = xmm10[1,1,3,3] -; SSE41-NEXT: pand %xmm1, %xmm7 +; SSE41-NEXT: pshufd {{.*#+}} xmm3 = xmm0[0,0,2,2] +; SSE41-NEXT: pcmpeqd %xmm11, %xmm10 +; SSE41-NEXT: pshufd {{.*#+}} xmm4 = xmm10[1,1,3,3] +; SSE41-NEXT: pand %xmm3, %xmm4 ; SSE41-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3] -; SSE41-NEXT: por %xmm7, %xmm0 -; SSE41-NEXT: movapd %xmm15, %xmm1 -; SSE41-NEXT: blendvpd %xmm0, -{{[0-9]+}}(%rsp), %xmm1 # 16-byte Folded Reload -; SSE41-NEXT: andpd %xmm15, %xmm1 -; SSE41-NEXT: andpd %xmm15, %xmm6 -; SSE41-NEXT: packuswb %xmm1, %xmm6 -; SSE41-NEXT: andpd %xmm15, %xmm5 -; SSE41-NEXT: andpd %xmm15, %xmm4 -; SSE41-NEXT: packuswb %xmm5, %xmm4 -; SSE41-NEXT: packuswb %xmm6, %xmm4 -; SSE41-NEXT: andpd %xmm15, %xmm3 -; SSE41-NEXT: andpd %xmm15, %xmm2 -; SSE41-NEXT: packuswb %xmm3, %xmm2 -; SSE41-NEXT: andpd %xmm15, %xmm8 -; SSE41-NEXT: andpd %xmm15, %xmm9 -; SSE41-NEXT: packuswb %xmm8, %xmm9 -; SSE41-NEXT: packuswb %xmm2, %xmm9 -; SSE41-NEXT: packuswb %xmm4, %xmm9 -; SSE41-NEXT: movdqa %xmm9, %xmm0 +; SSE41-NEXT: por %xmm4, %xmm0 +; SSE41-NEXT: blendvpd %xmm0, %xmm6, %xmm9 +; SSE41-NEXT: packssdw %xmm1, %xmm9 +; SSE41-NEXT: packusdw %xmm9, %xmm2 +; SSE41-NEXT: packuswb %xmm2, %xmm13 +; SSE41-NEXT: movdqa %xmm13, %xmm0 ; SSE41-NEXT: retq ; ; AVX1-LABEL: trunc_usat_v16i64_v16i8: ; AVX1: # %bb.0: +; AVX1-NEXT: vmovapd {{.*#+}} ymm8 = [255,255,255,255] +; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm5 +; AVX1-NEXT: vmovdqa {{.*#+}} xmm6 = [9223372036854775808,9223372036854775808] +; AVX1-NEXT: vpxor %xmm6, %xmm5, %xmm5 +; AVX1-NEXT: vmovdqa {{.*#+}} xmm7 = [9223372036854776063,9223372036854776063] +; AVX1-NEXT: vpcmpgtq %xmm5, %xmm7, %xmm5 +; AVX1-NEXT: vpxor %xmm6, %xmm0, %xmm4 +; AVX1-NEXT: vpcmpgtq %xmm4, %xmm7, %xmm4 +; AVX1-NEXT: vinsertf128 $1, %xmm5, %ymm4, %ymm4 +; AVX1-NEXT: vblendvpd %ymm4, %ymm0, %ymm8, %ymm0 +; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm4 +; AVX1-NEXT: vpxor %xmm6, %xmm4, %xmm4 +; AVX1-NEXT: vpcmpgtq %xmm4, %xmm7, %xmm4 +; AVX1-NEXT: vpxor %xmm6, %xmm1, %xmm5 +; AVX1-NEXT: vpcmpgtq %xmm5, %xmm7, %xmm5 +; AVX1-NEXT: vinsertf128 $1, %xmm4, %ymm5, %ymm4 +; AVX1-NEXT: vblendvpd %ymm4, %ymm1, %ymm8, %ymm1 +; AVX1-NEXT: vextractf128 $1, %ymm2, %xmm4 +; AVX1-NEXT: vpxor %xmm6, %xmm4, %xmm4 +; AVX1-NEXT: vpcmpgtq %xmm4, %xmm7, %xmm4 +; AVX1-NEXT: vpxor %xmm6, %xmm2, %xmm5 +; AVX1-NEXT: vpcmpgtq %xmm5, %xmm7, %xmm5 +; AVX1-NEXT: vinsertf128 $1, %xmm4, %ymm5, %ymm4 +; AVX1-NEXT: vblendvpd %ymm4, %ymm2, %ymm8, %ymm2 ; AVX1-NEXT: vextractf128 $1, %ymm3, %xmm4 -; AVX1-NEXT: vmovdqa {{.*#+}} xmm5 = [9223372036854775808,9223372036854775808] -; AVX1-NEXT: vpxor %xmm5, %xmm4, %xmm4 -; AVX1-NEXT: vmovdqa {{.*#+}} xmm6 = [9223372036854776063,9223372036854776063] -; AVX1-NEXT: vpcmpgtq %xmm4, %xmm6, %xmm4 -; AVX1-NEXT: vpxor %xmm5, %xmm3, %xmm7 -; AVX1-NEXT: vpcmpgtq %xmm7, %xmm6, %xmm7 -; AVX1-NEXT: vinsertf128 $1, %xmm4, %ymm7, %ymm8 -; AVX1-NEXT: vextractf128 $1, %ymm2, %xmm7 -; AVX1-NEXT: vpxor %xmm5, %xmm7, %xmm7 -; AVX1-NEXT: vpcmpgtq %xmm7, %xmm6, %xmm7 -; AVX1-NEXT: vpxor %xmm5, %xmm2, %xmm4 -; AVX1-NEXT: vpcmpgtq %xmm4, %xmm6, %xmm4 -; AVX1-NEXT: vinsertf128 $1, %xmm7, %ymm4, %ymm9 -; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm7 -; AVX1-NEXT: vpxor %xmm5, %xmm7, %xmm7 -; AVX1-NEXT: vpcmpgtq %xmm7, %xmm6, %xmm7 -; AVX1-NEXT: vpxor %xmm5, %xmm1, %xmm4 -; AVX1-NEXT: vpcmpgtq %xmm4, %xmm6, %xmm4 -; AVX1-NEXT: vinsertf128 $1, %xmm7, %ymm4, %ymm4 -; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm7 -; AVX1-NEXT: vpxor %xmm5, %xmm7, %xmm7 -; AVX1-NEXT: vpcmpgtq %xmm7, %xmm6, %xmm7 -; AVX1-NEXT: vpxor %xmm5, %xmm0, %xmm5 -; AVX1-NEXT: vpcmpgtq %xmm5, %xmm6, %xmm5 -; AVX1-NEXT: vinsertf128 $1, %xmm7, %ymm5, %ymm5 -; AVX1-NEXT: vmovapd {{.*#+}} ymm6 = [255,255,255,255] -; AVX1-NEXT: vblendvpd %ymm5, %ymm0, %ymm6, %ymm0 -; AVX1-NEXT: vblendvpd %ymm4, %ymm1, %ymm6, %ymm1 -; AVX1-NEXT: vblendvpd %ymm9, %ymm2, %ymm6, %ymm2 -; AVX1-NEXT: vblendvpd %ymm8, %ymm3, %ymm6, %ymm3 +; AVX1-NEXT: vpxor %xmm6, %xmm4, %xmm4 +; AVX1-NEXT: vpcmpgtq %xmm4, %xmm7, %xmm4 +; AVX1-NEXT: vpxor %xmm6, %xmm3, %xmm5 +; AVX1-NEXT: vpcmpgtq %xmm5, %xmm7, %xmm5 +; AVX1-NEXT: vinsertf128 $1, %xmm4, %ymm5, %ymm4 +; AVX1-NEXT: vblendvpd %ymm4, %ymm3, %ymm8, %ymm3 ; AVX1-NEXT: vextractf128 $1, %ymm3, %xmm4 -; AVX1-NEXT: vmovapd {{.*#+}} xmm5 = [255,0,0,0,0,0,0,0,255,0,0,0,0,0,0,0] -; AVX1-NEXT: vandpd %xmm5, %xmm4, %xmm4 -; AVX1-NEXT: vandpd %xmm5, %xmm3, %xmm3 -; AVX1-NEXT: vpackuswb %xmm4, %xmm3, %xmm3 +; AVX1-NEXT: vpackssdw %xmm4, %xmm3, %xmm3 ; AVX1-NEXT: vextractf128 $1, %ymm2, %xmm4 -; AVX1-NEXT: vandpd %xmm5, %xmm4, %xmm4 -; AVX1-NEXT: vandpd %xmm5, %xmm2, %xmm2 -; AVX1-NEXT: vpackuswb %xmm4, %xmm2, %xmm2 -; AVX1-NEXT: vpackuswb %xmm3, %xmm2, %xmm2 +; AVX1-NEXT: vpackssdw %xmm4, %xmm2, %xmm2 +; AVX1-NEXT: vpackusdw %xmm3, %xmm2, %xmm2 ; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm3 -; AVX1-NEXT: vandpd %xmm5, %xmm3, %xmm3 -; AVX1-NEXT: vandpd %xmm5, %xmm1, %xmm1 -; AVX1-NEXT: vpackuswb %xmm3, %xmm1, %xmm1 +; AVX1-NEXT: vpackssdw %xmm3, %xmm1, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm3 -; AVX1-NEXT: vandpd %xmm5, %xmm3, %xmm3 -; AVX1-NEXT: vandpd %xmm5, %xmm0, %xmm0 -; AVX1-NEXT: vpackuswb %xmm3, %xmm0, %xmm0 -; AVX1-NEXT: vpackuswb %xmm1, %xmm0, %xmm0 +; AVX1-NEXT: vpackssdw %xmm3, %xmm0, %xmm0 +; AVX1-NEXT: vpackusdw %xmm1, %xmm0, %xmm0 ; AVX1-NEXT: vpackuswb %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: vzeroupper ; AVX1-NEXT: retq ; -; AVX2-SLOW-LABEL: trunc_usat_v16i64_v16i8: -; AVX2-SLOW: # %bb.0: -; AVX2-SLOW-NEXT: vpbroadcastq {{.*#+}} ymm4 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808] -; AVX2-SLOW-NEXT: vpxor %ymm4, %ymm2, %ymm5 -; AVX2-SLOW-NEXT: vpbroadcastq {{.*#+}} ymm6 = [9223372036854776063,9223372036854776063,9223372036854776063,9223372036854776063] -; AVX2-SLOW-NEXT: vpcmpgtq %ymm5, %ymm6, %ymm5 -; AVX2-SLOW-NEXT: vpxor %ymm4, %ymm3, %ymm7 -; AVX2-SLOW-NEXT: vpcmpgtq %ymm7, %ymm6, %ymm7 -; AVX2-SLOW-NEXT: vpxor %ymm4, %ymm0, %ymm8 -; AVX2-SLOW-NEXT: vpcmpgtq %ymm8, %ymm6, %ymm8 -; AVX2-SLOW-NEXT: vpxor %ymm4, %ymm1, %ymm4 -; AVX2-SLOW-NEXT: vpcmpgtq %ymm4, %ymm6, %ymm4 -; AVX2-SLOW-NEXT: vbroadcastsd {{.*#+}} ymm6 = [255,255,255,255] -; AVX2-SLOW-NEXT: vblendvpd %ymm4, %ymm1, %ymm6, %ymm1 -; AVX2-SLOW-NEXT: vblendvpd %ymm8, %ymm0, %ymm6, %ymm0 -; AVX2-SLOW-NEXT: vblendvpd %ymm7, %ymm3, %ymm6, %ymm3 -; AVX2-SLOW-NEXT: vblendvpd %ymm5, %ymm2, %ymm6, %ymm2 -; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm2 = ymm2[0,2,2,3,4,6,6,7] -; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm2 = ymm2[0,2,2,3] -; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm3 = ymm3[0,2,2,3,4,6,6,7] -; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm3 = ymm3[0,2,2,3] -; AVX2-SLOW-NEXT: vinsertf128 $1, %xmm3, %ymm2, %ymm2 -; AVX2-SLOW-NEXT: vmovdqa {{.*#+}} ymm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31] -; AVX2-SLOW-NEXT: vpshufb %ymm3, %ymm2, %ymm2 -; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3] -; AVX2-SLOW-NEXT: vmovdqa {{.*#+}} xmm4 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u> -; AVX2-SLOW-NEXT: vpshufb %xmm4, %xmm2, %xmm2 -; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7] -; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,3] -; AVX2-SLOW-NEXT: vpermilps {{.*#+}} ymm1 = ymm1[0,2,2,3,4,6,6,7] -; AVX2-SLOW-NEXT: vpermpd {{.*#+}} ymm1 = ymm1[0,2,2,3] -; AVX2-SLOW-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 -; AVX2-SLOW-NEXT: vpshufb %ymm3, %ymm0, %ymm0 -; AVX2-SLOW-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3] -; AVX2-SLOW-NEXT: vpshufb %xmm4, %xmm0, %xmm0 -; AVX2-SLOW-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0] -; AVX2-SLOW-NEXT: vzeroupper -; AVX2-SLOW-NEXT: retq -; -; AVX2-FAST-LABEL: trunc_usat_v16i64_v16i8: -; AVX2-FAST: # %bb.0: -; AVX2-FAST-NEXT: vpbroadcastq {{.*#+}} ymm4 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808] -; AVX2-FAST-NEXT: vpxor %ymm4, %ymm2, %ymm5 -; AVX2-FAST-NEXT: vpbroadcastq {{.*#+}} ymm6 = [9223372036854776063,9223372036854776063,9223372036854776063,9223372036854776063] -; AVX2-FAST-NEXT: vpcmpgtq %ymm5, %ymm6, %ymm5 -; AVX2-FAST-NEXT: vpxor %ymm4, %ymm3, %ymm7 -; AVX2-FAST-NEXT: vpcmpgtq %ymm7, %ymm6, %ymm7 -; AVX2-FAST-NEXT: vpxor %ymm4, %ymm0, %ymm8 -; AVX2-FAST-NEXT: vpcmpgtq %ymm8, %ymm6, %ymm8 -; AVX2-FAST-NEXT: vpxor %ymm4, %ymm1, %ymm4 -; AVX2-FAST-NEXT: vpcmpgtq %ymm4, %ymm6, %ymm4 -; AVX2-FAST-NEXT: vbroadcastsd {{.*#+}} ymm6 = [255,255,255,255] -; AVX2-FAST-NEXT: vblendvpd %ymm4, %ymm1, %ymm6, %ymm1 -; AVX2-FAST-NEXT: vblendvpd %ymm8, %ymm0, %ymm6, %ymm0 -; AVX2-FAST-NEXT: vblendvpd %ymm7, %ymm3, %ymm6, %ymm3 -; AVX2-FAST-NEXT: vblendvpd %ymm5, %ymm2, %ymm6, %ymm2 -; AVX2-FAST-NEXT: vmovapd {{.*#+}} ymm4 = [0,2,4,6,4,6,6,7] -; AVX2-FAST-NEXT: vpermps %ymm2, %ymm4, %ymm2 -; AVX2-FAST-NEXT: vpermps %ymm3, %ymm4, %ymm3 -; AVX2-FAST-NEXT: vinsertf128 $1, %xmm3, %ymm2, %ymm2 -; AVX2-FAST-NEXT: vmovdqa {{.*#+}} ymm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31] -; AVX2-FAST-NEXT: vpshufb %ymm3, %ymm2, %ymm2 -; AVX2-FAST-NEXT: vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3] -; AVX2-FAST-NEXT: vmovdqa {{.*#+}} xmm5 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u> -; AVX2-FAST-NEXT: vpshufb %xmm5, %xmm2, %xmm2 -; AVX2-FAST-NEXT: vpermps %ymm0, %ymm4, %ymm0 -; AVX2-FAST-NEXT: vpermps %ymm1, %ymm4, %ymm1 -; AVX2-FAST-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0 -; AVX2-FAST-NEXT: vpshufb %ymm3, %ymm0, %ymm0 -; AVX2-FAST-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3] -; AVX2-FAST-NEXT: vpshufb %xmm5, %xmm0, %xmm0 -; AVX2-FAST-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0] -; AVX2-FAST-NEXT: vzeroupper -; AVX2-FAST-NEXT: retq +; AVX2-LABEL: trunc_usat_v16i64_v16i8: +; AVX2: # %bb.0: +; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm4 = [255,255,255,255] +; AVX2-NEXT: vpbroadcastq {{.*#+}} ymm5 = [9223372036854775808,9223372036854775808,9223372036854775808,9223372036854775808] +; AVX2-NEXT: vpxor %ymm5, %ymm1, %ymm6 +; AVX2-NEXT: vpbroadcastq {{.*#+}} ymm7 = [9223372036854776063,9223372036854776063,9223372036854776063,9223372036854776063] +; AVX2-NEXT: vpcmpgtq %ymm6, %ymm7, %ymm6 +; AVX2-NEXT: vblendvpd %ymm6, %ymm1, %ymm4, %ymm1 +; AVX2-NEXT: vpxor %ymm5, %ymm0, %ymm6 +; AVX2-NEXT: vpcmpgtq %ymm6, %ymm7, %ymm6 +; AVX2-NEXT: vblendvpd %ymm6, %ymm0, %ymm4, %ymm0 +; AVX2-NEXT: vpackusdw %ymm1, %ymm0, %ymm0 +; AVX2-NEXT: vpxor %ymm5, %ymm3, %ymm1 +; AVX2-NEXT: vpcmpgtq %ymm1, %ymm7, %ymm1 +; AVX2-NEXT: vblendvpd %ymm1, %ymm3, %ymm4, %ymm1 +; AVX2-NEXT: vpxor %ymm5, %ymm2, %ymm3 +; AVX2-NEXT: vpcmpgtq %ymm3, %ymm7, %ymm3 +; AVX2-NEXT: vblendvpd %ymm3, %ymm2, %ymm4, %ymm2 +; AVX2-NEXT: vpackusdw %ymm1, %ymm2, %ymm1 +; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,2,1,3] +; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3] +; AVX2-NEXT: vpackuswb %ymm1, %ymm0, %ymm0 +; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3] +; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 +; AVX2-NEXT: vpackuswb %xmm1, %xmm0, %xmm0 +; AVX2-NEXT: vzeroupper +; AVX2-NEXT: retq ; ; AVX512-LABEL: trunc_usat_v16i64_v16i8: ; AVX512: # %bb.0: @@ -1850,49 +1639,42 @@ define <16 x i8> @trunc_usat_v16i64_v16i8(<16 x i64> %a0) { define <8 x i8> @trunc_usat_v8i32_v8i8(<8 x i32> %a0) { ; SSE2-LABEL: trunc_usat_v8i32_v8i8: ; SSE2: # %bb.0: -; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [2147483648,2147483648,2147483648,2147483648] -; SSE2-NEXT: movdqa %xmm1, %xmm3 -; SSE2-NEXT: pxor %xmm2, %xmm3 -; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [2147483903,2147483903,2147483903,2147483903] -; SSE2-NEXT: movdqa %xmm4, %xmm5 -; SSE2-NEXT: pcmpgtd %xmm3, %xmm5 -; SSE2-NEXT: pxor %xmm0, %xmm2 -; SSE2-NEXT: pcmpgtd %xmm2, %xmm4 ; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [255,255,255,255] -; SSE2-NEXT: pand %xmm4, %xmm0 -; SSE2-NEXT: pandn %xmm2, %xmm4 -; SSE2-NEXT: por %xmm4, %xmm0 -; SSE2-NEXT: pand %xmm5, %xmm1 +; SSE2-NEXT: movdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648] +; SSE2-NEXT: movdqa %xmm1, %xmm4 +; SSE2-NEXT: pxor %xmm3, %xmm4 +; SSE2-NEXT: movdqa {{.*#+}} xmm5 = [2147483903,2147483903,2147483903,2147483903] +; SSE2-NEXT: movdqa %xmm5, %xmm6 +; SSE2-NEXT: pcmpgtd %xmm4, %xmm6 +; SSE2-NEXT: pand %xmm6, %xmm1 +; SSE2-NEXT: pandn %xmm2, %xmm6 +; SSE2-NEXT: por %xmm1, %xmm6 +; SSE2-NEXT: pxor %xmm0, %xmm3 +; SSE2-NEXT: pcmpgtd %xmm3, %xmm5 +; SSE2-NEXT: pand %xmm5, %xmm0 ; SSE2-NEXT: pandn %xmm2, %xmm5 -; SSE2-NEXT: por %xmm1, %xmm5 -; SSE2-NEXT: pslld $16, %xmm5 -; SSE2-NEXT: psrad $16, %xmm5 -; SSE2-NEXT: pslld $16, %xmm0 -; SSE2-NEXT: psrad $16, %xmm0 -; SSE2-NEXT: packssdw %xmm5, %xmm0 +; SSE2-NEXT: por %xmm5, %xmm0 +; SSE2-NEXT: packuswb %xmm6, %xmm0 ; SSE2-NEXT: retq ; ; SSSE3-LABEL: trunc_usat_v8i32_v8i8: ; SSSE3: # %bb.0: -; SSSE3-NEXT: movdqa {{.*#+}} xmm2 = [2147483648,2147483648,2147483648,2147483648] -; SSSE3-NEXT: movdqa %xmm1, %xmm3 -; SSSE3-NEXT: pxor %xmm2, %xmm3 -; SSSE3-NEXT: movdqa {{.*#+}} xmm4 = [2147483903,2147483903,2147483903,2147483903] -; SSSE3-NEXT: movdqa %xmm4, %xmm5 -; SSSE3-NEXT: pcmpgtd %xmm3, %xmm5 -; SSSE3-NEXT: pxor %xmm0, %xmm2 -; SSSE3-NEXT: pcmpgtd %xmm2, %xmm4 ; SSSE3-NEXT: movdqa {{.*#+}} xmm2 = [255,255,255,255] -; SSSE3-NEXT: pand %xmm4, %xmm0 -; SSSE3-NEXT: pandn %xmm2, %xmm4 -; SSSE3-NEXT: por %xmm4, %xmm0 -; SSSE3-NEXT: pand %xmm5, %xmm1 +; SSSE3-NEXT: movdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648] +; SSSE3-NEXT: movdqa %xmm1, %xmm4 +; SSSE3-NEXT: pxor %xmm3, %xmm4 +; SSSE3-NEXT: movdqa {{.*#+}} xmm5 = [2147483903,2147483903,2147483903,2147483903] +; SSSE3-NEXT: movdqa %xmm5, %xmm6 +; SSSE3-NEXT: pcmpgtd %xmm4, %xmm6 +; SSSE3-NEXT: pand %xmm6, %xmm1 +; SSSE3-NEXT: pandn %xmm2, %xmm6 +; SSSE3-NEXT: por %xmm1, %xmm6 +; SSSE3-NEXT: pxor %xmm0, %xmm3 +; SSSE3-NEXT: pcmpgtd %xmm3, %xmm5 +; SSSE3-NEXT: pand %xmm5, %xmm0 ; SSSE3-NEXT: pandn %xmm2, %xmm5 -; SSSE3-NEXT: por %xmm1, %xmm5 -; SSSE3-NEXT: movdqa {{.*#+}} xmm1 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15] -; SSSE3-NEXT: pshufb %xmm1, %xmm5 -; SSSE3-NEXT: pshufb %xmm1, %xmm0 -; SSSE3-NEXT: punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm5[0] +; SSSE3-NEXT: por %xmm5, %xmm0 +; SSSE3-NEXT: packuswb %xmm6, %xmm0 ; SSSE3-NEXT: retq ; ; SSE41-LABEL: trunc_usat_v8i32_v8i8: @@ -1962,82 +1744,74 @@ define <8 x i8> @trunc_usat_v8i32_v8i8(<8 x i32> %a0) { define <16 x i8> @trunc_usat_v16i32_v16i8(<16 x i32> %a0) { ; SSE2-LABEL: trunc_usat_v16i32_v16i8: ; SSE2: # %bb.0: -; SSE2-NEXT: movdqa {{.*#+}} xmm7 = [2147483648,2147483648,2147483648,2147483648] -; SSE2-NEXT: movdqa %xmm3, %xmm4 -; SSE2-NEXT: pxor %xmm7, %xmm4 +; SSE2-NEXT: movdqa {{.*#+}} xmm8 = [255,255,255,255] +; SSE2-NEXT: movdqa {{.*#+}} xmm6 = [2147483648,2147483648,2147483648,2147483648] +; SSE2-NEXT: movdqa %xmm1, %xmm7 +; SSE2-NEXT: pxor %xmm6, %xmm7 ; SSE2-NEXT: movdqa {{.*#+}} xmm5 = [2147483903,2147483903,2147483903,2147483903] -; SSE2-NEXT: movdqa %xmm5, %xmm8 -; SSE2-NEXT: pcmpgtd %xmm4, %xmm8 -; SSE2-NEXT: movdqa %xmm2, %xmm6 -; SSE2-NEXT: pxor %xmm7, %xmm6 ; SSE2-NEXT: movdqa %xmm5, %xmm4 -; SSE2-NEXT: pcmpgtd %xmm6, %xmm4 -; SSE2-NEXT: movdqa %xmm1, %xmm9 -; SSE2-NEXT: pxor %xmm7, %xmm9 -; SSE2-NEXT: movdqa %xmm5, %xmm6 -; SSE2-NEXT: pcmpgtd %xmm9, %xmm6 -; SSE2-NEXT: pxor %xmm0, %xmm7 -; SSE2-NEXT: pcmpgtd %xmm7, %xmm5 -; SSE2-NEXT: movdqa {{.*#+}} xmm7 = [255,255,255,255] -; SSE2-NEXT: pand %xmm5, %xmm0 -; SSE2-NEXT: pandn %xmm7, %xmm5 -; SSE2-NEXT: por %xmm5, %xmm0 -; SSE2-NEXT: pand %xmm6, %xmm1 -; SSE2-NEXT: pandn %xmm7, %xmm6 -; SSE2-NEXT: por %xmm1, %xmm6 -; SSE2-NEXT: pand %xmm4, %xmm2 -; SSE2-NEXT: pandn %xmm7, %xmm4 -; SSE2-NEXT: por %xmm2, %xmm4 -; SSE2-NEXT: pand %xmm8, %xmm3 -; SSE2-NEXT: pandn %xmm7, %xmm8 -; SSE2-NEXT: por %xmm3, %xmm8 -; SSE2-NEXT: pand %xmm7, %xmm8 -; SSE2-NEXT: pand %xmm7, %xmm4 -; SSE2-NEXT: packuswb %xmm8, %xmm4 -; SSE2-NEXT: pand %xmm7, %xmm6 +; SSE2-NEXT: pcmpgtd %xmm7, %xmm4 +; SSE2-NEXT: pand %xmm4, %xmm1 +; SSE2-NEXT: pandn %xmm8, %xmm4 +; SSE2-NEXT: por %xmm1, %xmm4 +; SSE2-NEXT: movdqa %xmm0, %xmm1 +; SSE2-NEXT: pxor %xmm6, %xmm1 +; SSE2-NEXT: movdqa %xmm5, %xmm7 +; SSE2-NEXT: pcmpgtd %xmm1, %xmm7 ; SSE2-NEXT: pand %xmm7, %xmm0 -; SSE2-NEXT: packuswb %xmm6, %xmm0 +; SSE2-NEXT: pandn %xmm8, %xmm7 +; SSE2-NEXT: por %xmm7, %xmm0 ; SSE2-NEXT: packuswb %xmm4, %xmm0 +; SSE2-NEXT: movdqa %xmm3, %xmm1 +; SSE2-NEXT: pxor %xmm6, %xmm1 +; SSE2-NEXT: movdqa %xmm5, %xmm4 +; SSE2-NEXT: pcmpgtd %xmm1, %xmm4 +; SSE2-NEXT: pand %xmm4, %xmm3 +; SSE2-NEXT: pandn %xmm8, %xmm4 +; SSE2-NEXT: por %xmm3, %xmm4 +; SSE2-NEXT: pxor %xmm2, %xmm6 +; SSE2-NEXT: pcmpgtd %xmm6, %xmm5 +; SSE2-NEXT: pand %xmm5, %xmm2 +; SSE2-NEXT: pandn %xmm8, %xmm5 +; SSE2-NEXT: por %xmm2, %xmm5 +; SSE2-NEXT: packuswb %xmm4, %xmm5 +; SSE2-NEXT: packuswb %xmm5, %xmm0 ; SSE2-NEXT: retq ; ; SSSE3-LABEL: trunc_usat_v16i32_v16i8: ; SSSE3: # %bb.0: -; SSSE3-NEXT: movdqa {{.*#+}} xmm7 = [2147483648,2147483648,2147483648,2147483648] -; SSSE3-NEXT: movdqa %xmm3, %xmm4 -; SSSE3-NEXT: pxor %xmm7, %xmm4 +; SSSE3-NEXT: movdqa {{.*#+}} xmm8 = [255,255,255,255] +; SSSE3-NEXT: movdqa {{.*#+}} xmm6 = [2147483648,2147483648,2147483648,2147483648] +; SSSE3-NEXT: movdqa %xmm1, %xmm7 +; SSSE3-NEXT: pxor %xmm6, %xmm7 ; SSSE3-NEXT: movdqa {{.*#+}} xmm5 = [2147483903,2147483903,2147483903,2147483903] -; SSSE3-NEXT: movdqa %xmm5, %xmm8 -; SSSE3-NEXT: pcmpgtd %xmm4, %xmm8 -; SSSE3-NEXT: movdqa %xmm2, %xmm6 -; SSSE3-NEXT: pxor %xmm7, %xmm6 ; SSSE3-NEXT: movdqa %xmm5, %xmm4 -; SSSE3-NEXT: pcmpgtd %xmm6, %xmm4 -; SSSE3-NEXT: movdqa %xmm1, %xmm9 -; SSSE3-NEXT: pxor %xmm7, %xmm9 -; SSSE3-NEXT: movdqa %xmm5, %xmm6 -; SSSE3-NEXT: pcmpgtd %xmm9, %xmm6 -; SSSE3-NEXT: pxor %xmm0, %xmm7 -; SSSE3-NEXT: pcmpgtd %xmm7, %xmm5 -; SSSE3-NEXT: movdqa {{.*#+}} xmm7 = [255,255,255,255] -; SSSE3-NEXT: pand %xmm5, %xmm0 -; SSSE3-NEXT: pandn %xmm7, %xmm5 -; SSSE3-NEXT: por %xmm5, %xmm0 -; SSSE3-NEXT: pand %xmm6, %xmm1 -; SSSE3-NEXT: pandn %xmm7, %xmm6 -; SSSE3-NEXT: por %xmm1, %xmm6 -; SSSE3-NEXT: pand %xmm4, %xmm2 -; SSSE3-NEXT: pandn %xmm7, %xmm4 -; SSSE3-NEXT: por %xmm2, %xmm4 -; SSSE3-NEXT: pand %xmm8, %xmm3 -; SSSE3-NEXT: pandn %xmm7, %xmm8 -; SSSE3-NEXT: por %xmm3, %xmm8 -; SSSE3-NEXT: pand %xmm7, %xmm8 -; SSSE3-NEXT: pand %xmm7, %xmm4 -; SSSE3-NEXT: packuswb %xmm8, %xmm4 -; SSSE3-NEXT: pand %xmm7, %xmm6 +; SSSE3-NEXT: pcmpgtd %xmm7, %xmm4 +; SSSE3-NEXT: pand %xmm4, %xmm1 +; SSSE3-NEXT: pandn %xmm8, %xmm4 +; SSSE3-NEXT: por %xmm1, %xmm4 +; SSSE3-NEXT: movdqa %xmm0, %xmm1 +; SSSE3-NEXT: pxor %xmm6, %xmm1 +; SSSE3-NEXT: movdqa %xmm5, %xmm7 +; SSSE3-NEXT: pcmpgtd %xmm1, %xmm7 ; SSSE3-NEXT: pand %xmm7, %xmm0 -; SSSE3-NEXT: packuswb %xmm6, %xmm0 +; SSSE3-NEXT: pandn %xmm8, %xmm7 +; SSSE3-NEXT: por %xmm7, %xmm0 ; SSSE3-NEXT: packuswb %xmm4, %xmm0 +; SSSE3-NEXT: movdqa %xmm3, %xmm1 +; SSSE3-NEXT: pxor %xmm6, %xmm1 +; SSSE3-NEXT: movdqa %xmm5, %xmm4 +; SSSE3-NEXT: pcmpgtd %xmm1, %xmm4 +; SSSE3-NEXT: pand %xmm4, %xmm3 +; SSSE3-NEXT: pandn %xmm8, %xmm4 +; SSSE3-NEXT: por %xmm3, %xmm4 +; SSSE3-NEXT: pxor %xmm2, %xmm6 +; SSSE3-NEXT: pcmpgtd %xmm6, %xmm5 +; SSSE3-NEXT: pand %xmm5, %xmm2 +; SSSE3-NEXT: pandn %xmm8, %xmm5 +; SSSE3-NEXT: por %xmm2, %xmm5 +; SSSE3-NEXT: packuswb %xmm4, %xmm5 +; SSSE3-NEXT: packuswb %xmm5, %xmm0 ; SSSE3-NEXT: retq ; ; SSE41-LABEL: trunc_usat_v16i32_v16i8: @@ -2093,49 +1867,28 @@ define <16 x i8> @trunc_usat_v16i32_v16i8(<16 x i32> %a0) { define <16 x i8> @trunc_usat_v16i16_v16i8(<16 x i16> %a0) { ; SSE2-LABEL: trunc_usat_v16i16_v16i8: ; SSE2: # %bb.0: -; SSE2-NEXT: movdqa {{.*#+}} xmm3 = [254,254,254,254,254,254,254,254] -; SSE2-NEXT: movdqa %xmm1, %xmm4 -; SSE2-NEXT: psubusw %xmm3, %xmm4 -; SSE2-NEXT: pxor %xmm5, %xmm5 -; SSE2-NEXT: pcmpeqw %xmm5, %xmm4 -; SSE2-NEXT: movdqa %xmm0, %xmm2 -; SSE2-NEXT: psubusw %xmm3, %xmm2 -; SSE2-NEXT: pcmpeqw %xmm5, %xmm2 -; SSE2-NEXT: movdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255] -; SSE2-NEXT: pand %xmm2, %xmm0 -; SSE2-NEXT: pandn %xmm3, %xmm2 -; SSE2-NEXT: por %xmm0, %xmm2 -; SSE2-NEXT: pand %xmm4, %xmm1 -; SSE2-NEXT: pandn %xmm3, %xmm4 -; SSE2-NEXT: por %xmm1, %xmm4 -; SSE2-NEXT: pand %xmm3, %xmm4 -; SSE2-NEXT: pand %xmm3, %xmm2 -; SSE2-NEXT: packuswb %xmm4, %xmm2 -; SSE2-NEXT: movdqa %xmm2, %xmm0 +; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [32768,32768,32768,32768,32768,32768,32768,32768] +; SSE2-NEXT: pxor %xmm2, %xmm1 +; SSE2-NEXT: movdqa {{.*#+}} xmm3 = [33023,33023,33023,33023,33023,33023,33023,33023] +; SSE2-NEXT: pminsw %xmm3, %xmm1 +; SSE2-NEXT: pxor %xmm2, %xmm1 +; SSE2-NEXT: pxor %xmm2, %xmm0 +; SSE2-NEXT: pminsw %xmm3, %xmm0 +; SSE2-NEXT: pxor %xmm2, %xmm0 +; SSE2-NEXT: packuswb %xmm1, %xmm0 ; SSE2-NEXT: retq ; ; SSSE3-LABEL: trunc_usat_v16i16_v16i8: ; SSSE3: # %bb.0: -; SSSE3-NEXT: movdqa {{.*#+}} xmm3 = [254,254,254,254,254,254,254,254] -; SSSE3-NEXT: movdqa %xmm1, %xmm4 -; SSSE3-NEXT: psubusw %xmm3, %xmm4 -; SSSE3-NEXT: pxor %xmm5, %xmm5 -; SSSE3-NEXT: pcmpeqw %xmm5, %xmm4 -; SSSE3-NEXT: movdqa %xmm0, %xmm2 -; SSSE3-NEXT: psubusw %xmm3, %xmm2 -; SSSE3-NEXT: pcmpeqw %xmm5, %xmm2 -; SSSE3-NEXT: movdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255] -; SSSE3-NEXT: pand %xmm2, %xmm0 -; SSSE3-NEXT: pandn %xmm3, %xmm2 -; SSSE3-NEXT: por %xmm0, %xmm2 -; SSSE3-NEXT: pand %xmm4, %xmm1 -; SSSE3-NEXT: pandn %xmm3, %xmm4 -; SSSE3-NEXT: por %xmm1, %xmm4 -; SSSE3-NEXT: movdqa {{.*#+}} xmm0 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u> -; SSSE3-NEXT: pshufb %xmm0, %xmm4 -; SSSE3-NEXT: pshufb %xmm0, %xmm2 -; SSSE3-NEXT: punpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm4[0] -; SSSE3-NEXT: movdqa %xmm2, %xmm0 +; SSSE3-NEXT: movdqa {{.*#+}} xmm2 = [32768,32768,32768,32768,32768,32768,32768,32768] +; SSSE3-NEXT: pxor %xmm2, %xmm1 +; SSSE3-NEXT: movdqa {{.*#+}} xmm3 = [33023,33023,33023,33023,33023,33023,33023,33023] +; SSSE3-NEXT: pminsw %xmm3, %xmm1 +; SSSE3-NEXT: pxor %xmm2, %xmm1 +; SSSE3-NEXT: pxor %xmm2, %xmm0 +; SSSE3-NEXT: pminsw %xmm3, %xmm0 +; SSSE3-NEXT: pxor %xmm2, %xmm0 +; SSSE3-NEXT: packuswb %xmm1, %xmm0 ; SSSE3-NEXT: retq ; ; SSE41-LABEL: trunc_usat_v16i16_v16i8: @@ -2202,79 +1955,44 @@ define <16 x i8> @trunc_usat_v16i16_v16i8(<16 x i16> %a0) { define <32 x i8> @trunc_usat_v32i16_v32i8(<32 x i16> %a0) { ; SSE2-LABEL: trunc_usat_v32i16_v32i8: ; SSE2: # %bb.0: -; SSE2-NEXT: movdqa %xmm0, %xmm8 -; SSE2-NEXT: movdqa {{.*#+}} xmm9 = [254,254,254,254,254,254,254,254] -; SSE2-NEXT: movdqa %xmm1, %xmm6 -; SSE2-NEXT: psubusw %xmm9, %xmm6 -; SSE2-NEXT: pxor %xmm5, %xmm5 -; SSE2-NEXT: pcmpeqw %xmm5, %xmm6 -; SSE2-NEXT: psubusw %xmm9, %xmm0 -; SSE2-NEXT: pcmpeqw %xmm5, %xmm0 -; SSE2-NEXT: movdqa %xmm3, %xmm7 -; SSE2-NEXT: psubusw %xmm9, %xmm7 -; SSE2-NEXT: pcmpeqw %xmm5, %xmm7 -; SSE2-NEXT: movdqa %xmm2, %xmm4 -; SSE2-NEXT: psubusw %xmm9, %xmm4 -; SSE2-NEXT: pcmpeqw %xmm5, %xmm4 -; SSE2-NEXT: movdqa {{.*#+}} xmm5 = [255,255,255,255,255,255,255,255] -; SSE2-NEXT: pand %xmm4, %xmm2 -; SSE2-NEXT: pandn %xmm5, %xmm4 -; SSE2-NEXT: por %xmm2, %xmm4 -; SSE2-NEXT: pand %xmm7, %xmm3 -; SSE2-NEXT: pandn %xmm5, %xmm7 -; SSE2-NEXT: por %xmm3, %xmm7 -; SSE2-NEXT: pand %xmm0, %xmm8 -; SSE2-NEXT: pandn %xmm5, %xmm0 -; SSE2-NEXT: por %xmm8, %xmm0 -; SSE2-NEXT: pand %xmm6, %xmm1 -; SSE2-NEXT: pandn %xmm5, %xmm6 -; SSE2-NEXT: por %xmm1, %xmm6 -; SSE2-NEXT: pand %xmm5, %xmm6 -; SSE2-NEXT: pand %xmm5, %xmm0 -; SSE2-NEXT: packuswb %xmm6, %xmm0 -; SSE2-NEXT: pand %xmm5, %xmm7 -; SSE2-NEXT: pand %xmm4, %xmm5 -; SSE2-NEXT: packuswb %xmm7, %xmm5 -; SSE2-NEXT: movdqa %xmm5, %xmm1 +; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [32768,32768,32768,32768,32768,32768,32768,32768] +; SSE2-NEXT: pxor %xmm4, %xmm3 +; SSE2-NEXT: movdqa {{.*#+}} xmm5 = [33023,33023,33023,33023,33023,33023,33023,33023] +; SSE2-NEXT: pminsw %xmm5, %xmm3 +; SSE2-NEXT: pxor %xmm4, %xmm3 +; SSE2-NEXT: pxor %xmm4, %xmm2 +; SSE2-NEXT: pminsw %xmm5, %xmm2 +; SSE2-NEXT: pxor %xmm4, %xmm2 +; SSE2-NEXT: packuswb %xmm3, %xmm2 +; SSE2-NEXT: pxor %xmm4, %xmm1 +; SSE2-NEXT: pminsw %xmm5, %xmm1 +; SSE2-NEXT: pxor %xmm4, %xmm1 +; SSE2-NEXT: pxor %xmm4, %xmm0 +; SSE2-NEXT: pminsw %xmm5, %xmm0 +; SSE2-NEXT: pxor %xmm4, %xmm0 +; SSE2-NEXT: packuswb %xmm1, %xmm0 +; SSE2-NEXT: movdqa %xmm2, %xmm1 ; SSE2-NEXT: retq ; ; SSSE3-LABEL: trunc_usat_v32i16_v32i8: ; SSSE3: # %bb.0: -; SSSE3-NEXT: movdqa %xmm0, %xmm4 -; SSSE3-NEXT: movdqa {{.*#+}} xmm8 = [254,254,254,254,254,254,254,254] -; SSSE3-NEXT: movdqa %xmm1, %xmm6 -; SSSE3-NEXT: psubusw %xmm8, %xmm6 -; SSSE3-NEXT: pxor %xmm9, %xmm9 -; SSSE3-NEXT: pcmpeqw %xmm9, %xmm6 -; SSSE3-NEXT: psubusw %xmm8, %xmm0 -; SSSE3-NEXT: pcmpeqw %xmm9, %xmm0 -; SSSE3-NEXT: movdqa %xmm3, %xmm7 -; SSSE3-NEXT: psubusw %xmm8, %xmm7 -; SSSE3-NEXT: pcmpeqw %xmm9, %xmm7 -; SSSE3-NEXT: movdqa %xmm2, %xmm5 -; SSSE3-NEXT: psubusw %xmm8, %xmm5 -; SSSE3-NEXT: pcmpeqw %xmm9, %xmm5 -; SSSE3-NEXT: movdqa {{.*#+}} xmm8 = [255,255,255,255,255,255,255,255] -; SSSE3-NEXT: pand %xmm5, %xmm2 -; SSSE3-NEXT: pandn %xmm8, %xmm5 -; SSSE3-NEXT: por %xmm2, %xmm5 -; SSSE3-NEXT: pand %xmm7, %xmm3 -; SSSE3-NEXT: pandn %xmm8, %xmm7 -; SSSE3-NEXT: por %xmm3, %xmm7 -; SSSE3-NEXT: pand %xmm0, %xmm4 -; SSSE3-NEXT: pandn %xmm8, %xmm0 -; SSSE3-NEXT: por %xmm4, %xmm0 -; SSSE3-NEXT: pand %xmm6, %xmm1 -; SSSE3-NEXT: pandn %xmm8, %xmm6 -; SSSE3-NEXT: por %xmm1, %xmm6 -; SSSE3-NEXT: movdqa {{.*#+}} xmm1 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u> -; SSSE3-NEXT: pshufb %xmm1, %xmm6 -; SSSE3-NEXT: pshufb %xmm1, %xmm0 -; SSSE3-NEXT: punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm6[0] -; SSSE3-NEXT: pshufb %xmm1, %xmm7 -; SSSE3-NEXT: pshufb %xmm1, %xmm5 -; SSSE3-NEXT: punpcklqdq {{.*#+}} xmm5 = xmm5[0],xmm7[0] -; SSSE3-NEXT: movdqa %xmm5, %xmm1 +; SSSE3-NEXT: movdqa {{.*#+}} xmm4 = [32768,32768,32768,32768,32768,32768,32768,32768] +; SSSE3-NEXT: pxor %xmm4, %xmm3 +; SSSE3-NEXT: movdqa {{.*#+}} xmm5 = [33023,33023,33023,33023,33023,33023,33023,33023] +; SSSE3-NEXT: pminsw %xmm5, %xmm3 +; SSSE3-NEXT: pxor %xmm4, %xmm3 +; SSSE3-NEXT: pxor %xmm4, %xmm2 +; SSSE3-NEXT: pminsw %xmm5, %xmm2 +; SSSE3-NEXT: pxor %xmm4, %xmm2 +; SSSE3-NEXT: packuswb %xmm3, %xmm2 +; SSSE3-NEXT: pxor %xmm4, %xmm1 +; SSSE3-NEXT: pminsw %xmm5, %xmm1 +; SSSE3-NEXT: pxor %xmm4, %xmm1 +; SSSE3-NEXT: pxor %xmm4, %xmm0 +; SSSE3-NEXT: pminsw %xmm5, %xmm0 +; SSSE3-NEXT: pxor %xmm4, %xmm0 +; SSSE3-NEXT: packuswb %xmm1, %xmm0 +; SSSE3-NEXT: movdqa %xmm2, %xmm1 ; SSSE3-NEXT: retq ; ; SSE41-LABEL: trunc_usat_v32i16_v32i8: |
