diff options
Diffstat (limited to 'llvm/test/CodeGen/X86/combine-sdiv.ll')
| -rw-r--r-- | llvm/test/CodeGen/X86/combine-sdiv.ll | 75 |
1 files changed, 36 insertions, 39 deletions
diff --git a/llvm/test/CodeGen/X86/combine-sdiv.ll b/llvm/test/CodeGen/X86/combine-sdiv.ll index 39ce0c922e3..27d49cad966 100644 --- a/llvm/test/CodeGen/X86/combine-sdiv.ll +++ b/llvm/test/CodeGen/X86/combine-sdiv.ll @@ -1529,8 +1529,7 @@ define <2 x i64> @combine_vec_sdiv_by_pow2b_v2i64(<2 x i64> %x) { ; SSE2-NEXT: psrlq $2, %xmm1 ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3] ; SSE2-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1] -; SSE2-NEXT: movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1] -; SSE2-NEXT: movapd %xmm1, %xmm0 +; SSE2-NEXT: shufpd {{.*#+}} xmm0 = xmm0[0],xmm1[1] ; SSE2-NEXT: retq ; ; SSE41-LABEL: combine_vec_sdiv_by_pow2b_v2i64: @@ -1616,24 +1615,23 @@ define <4 x i64> @combine_vec_sdiv_by_pow2b_v4i64(<4 x i64> %x) { ; SSE2-NEXT: psrlq $2, %xmm2 ; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3] ; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1] -; SSE2-NEXT: movsd {{.*#+}} xmm2 = xmm0[0],xmm2[1] -; SSE2-NEXT: movdqa %xmm1, %xmm0 -; SSE2-NEXT: psrad $31, %xmm0 -; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3] -; SSE2-NEXT: movdqa %xmm3, %xmm0 -; SSE2-NEXT: psrlq $61, %xmm0 -; SSE2-NEXT: psrlq $60, %xmm3 -; SSE2-NEXT: movsd {{.*#+}} xmm3 = xmm0[0],xmm3[1] -; SSE2-NEXT: paddq %xmm1, %xmm3 -; SSE2-NEXT: movdqa %xmm3, %xmm0 -; SSE2-NEXT: psrlq $3, %xmm0 -; SSE2-NEXT: psrlq $4, %xmm3 -; SSE2-NEXT: movsd {{.*#+}} xmm3 = xmm0[0],xmm3[1] -; SSE2-NEXT: movapd {{.*#+}} xmm0 = [1152921504606846976,576460752303423488] -; SSE2-NEXT: xorpd %xmm0, %xmm3 -; SSE2-NEXT: psubq %xmm0, %xmm3 -; SSE2-NEXT: movapd %xmm2, %xmm0 -; SSE2-NEXT: movdqa %xmm3, %xmm1 +; SSE2-NEXT: shufpd {{.*#+}} xmm0 = xmm0[0],xmm2[1] +; SSE2-NEXT: movdqa %xmm1, %xmm2 +; SSE2-NEXT: psrad $31, %xmm2 +; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3] +; SSE2-NEXT: movdqa %xmm2, %xmm3 +; SSE2-NEXT: psrlq $61, %xmm3 +; SSE2-NEXT: psrlq $60, %xmm2 +; SSE2-NEXT: movsd {{.*#+}} xmm2 = xmm3[0],xmm2[1] +; SSE2-NEXT: paddq %xmm1, %xmm2 +; SSE2-NEXT: movdqa %xmm2, %xmm1 +; SSE2-NEXT: psrlq $3, %xmm1 +; SSE2-NEXT: psrlq $4, %xmm2 +; SSE2-NEXT: movsd {{.*#+}} xmm2 = xmm1[0],xmm2[1] +; SSE2-NEXT: movapd {{.*#+}} xmm1 = [1152921504606846976,576460752303423488] +; SSE2-NEXT: xorpd %xmm1, %xmm2 +; SSE2-NEXT: psubq %xmm1, %xmm2 +; SSE2-NEXT: movdqa %xmm2, %xmm1 ; SSE2-NEXT: retq ; ; SSE41-LABEL: combine_vec_sdiv_by_pow2b_v4i64: @@ -1745,29 +1743,28 @@ define <4 x i64> @combine_vec_sdiv_by_pow2b_v4i64(<4 x i64> %x) { define <8 x i64> @combine_vec_sdiv_by_pow2b_v8i64(<8 x i64> %x) { ; SSE2-LABEL: combine_vec_sdiv_by_pow2b_v8i64: ; SSE2: # %bb.0: -; SSE2-NEXT: movdqa %xmm2, %xmm4 -; SSE2-NEXT: movdqa %xmm0, %xmm2 -; SSE2-NEXT: psrad $31, %xmm0 -; SSE2-NEXT: psrlq $62, %xmm0 -; SSE2-NEXT: paddq %xmm2, %xmm0 -; SSE2-NEXT: movdqa %xmm0, %xmm5 +; SSE2-NEXT: movdqa %xmm0, %xmm4 +; SSE2-NEXT: psrad $31, %xmm4 +; SSE2-NEXT: psrlq $62, %xmm4 +; SSE2-NEXT: paddq %xmm0, %xmm4 +; SSE2-NEXT: movdqa %xmm4, %xmm5 ; SSE2-NEXT: psrad $2, %xmm5 ; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,3,2,3] -; SSE2-NEXT: psrlq $2, %xmm0 -; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] -; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1] -; SSE2-NEXT: movsd {{.*#+}} xmm0 = xmm2[0],xmm0[1] -; SSE2-NEXT: movdqa %xmm4, %xmm2 -; SSE2-NEXT: psrad $31, %xmm2 -; SSE2-NEXT: psrlq $62, %xmm2 -; SSE2-NEXT: paddq %xmm4, %xmm2 -; SSE2-NEXT: movdqa %xmm2, %xmm5 +; SSE2-NEXT: psrlq $2, %xmm4 +; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm4[0,2,2,3] +; SSE2-NEXT: punpckldq {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1] +; SSE2-NEXT: shufpd {{.*#+}} xmm0 = xmm0[0],xmm4[1] +; SSE2-NEXT: movdqa %xmm2, %xmm4 +; SSE2-NEXT: psrad $31, %xmm4 +; SSE2-NEXT: psrlq $62, %xmm4 +; SSE2-NEXT: paddq %xmm2, %xmm4 +; SSE2-NEXT: movdqa %xmm4, %xmm5 ; SSE2-NEXT: psrad $2, %xmm5 ; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,3,2,3] -; SSE2-NEXT: psrlq $2, %xmm2 -; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3] -; SSE2-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1] -; SSE2-NEXT: movsd {{.*#+}} xmm2 = xmm4[0],xmm2[1] +; SSE2-NEXT: psrlq $2, %xmm4 +; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm4[0,2,2,3] +; SSE2-NEXT: punpckldq {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1] +; SSE2-NEXT: shufpd {{.*#+}} xmm2 = xmm2[0],xmm4[1] ; SSE2-NEXT: movdqa %xmm1, %xmm4 ; SSE2-NEXT: psrad $31, %xmm4 ; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3] |
