diff options
Diffstat (limited to 'llvm/test/CodeGen/X86/scalarize-fp.ll')
| -rw-r--r-- | llvm/test/CodeGen/X86/scalarize-fp.ll | 42 |
1 files changed, 14 insertions, 28 deletions
diff --git a/llvm/test/CodeGen/X86/scalarize-fp.ll b/llvm/test/CodeGen/X86/scalarize-fp.ll index 40eed1ff38c..650b948d3ab 100644 --- a/llvm/test/CodeGen/X86/scalarize-fp.ll +++ b/llvm/test/CodeGen/X86/scalarize-fp.ll @@ -507,14 +507,14 @@ define <2 x i64> @add_splat_splat_v2i64(<2 x i64> %vx, <2 x i64> %vy) { define <2 x double> @fadd_splat_const_op1_v2f64(<2 x double> %vx) { ; SSE-LABEL: fadd_splat_const_op1_v2f64: ; SSE: # %bb.0: +; SSE-NEXT: addsd {{.*}}(%rip), %xmm0 ; SSE-NEXT: unpcklpd {{.*#+}} xmm0 = xmm0[0,0] -; SSE-NEXT: addpd {{.*}}(%rip), %xmm0 ; SSE-NEXT: retq ; ; AVX-LABEL: fadd_splat_const_op1_v2f64: ; AVX: # %bb.0: +; AVX-NEXT: vaddsd {{.*}}(%rip), %xmm0, %xmm0 ; AVX-NEXT: vmovddup {{.*#+}} xmm0 = xmm0[0,0] -; AVX-NEXT: vaddpd {{.*}}(%rip), %xmm0, %xmm0 ; AVX-NEXT: retq %splatx = shufflevector <2 x double> %vx, <2 x double> undef, <2 x i32> zeroinitializer %r = fadd <2 x double> %splatx, <double 42.0, double 42.0> @@ -548,14 +548,14 @@ define <4 x double> @fsub_const_op0_splat_v4f64(double %x) { define <4 x float> @fmul_splat_const_op1_v4f32(<4 x float> %vx, <4 x float> %vy) { ; SSE-LABEL: fmul_splat_const_op1_v4f32: ; SSE: # %bb.0: +; SSE-NEXT: mulss {{.*}}(%rip), %xmm0 ; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,0,0,0] -; SSE-NEXT: mulps {{.*}}(%rip), %xmm0 ; SSE-NEXT: retq ; ; AVX-LABEL: fmul_splat_const_op1_v4f32: ; AVX: # %bb.0: +; AVX-NEXT: vmulss {{.*}}(%rip), %xmm0, %xmm0 ; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,0,0,0] -; AVX-NEXT: vmulps {{.*}}(%rip), %xmm0, %xmm0 ; AVX-NEXT: retq %splatx = shufflevector <4 x float> %vx, <4 x float> undef, <4 x i32> zeroinitializer %r = fmul fast <4 x float> %splatx, <float 17.0, float 17.0, float 17.0, float 17.0> @@ -565,28 +565,18 @@ define <4 x float> @fmul_splat_const_op1_v4f32(<4 x float> %vx, <4 x float> %vy) define <8 x float> @fdiv_splat_const_op0_v8f32(<8 x float> %vy) { ; SSE-LABEL: fdiv_splat_const_op0_v8f32: ; SSE: # %bb.0: -; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,0,0,0] -; SSE-NEXT: rcpps %xmm0, %xmm2 -; SSE-NEXT: mulps %xmm2, %xmm0 -; SSE-NEXT: movaps {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] -; SSE-NEXT: subps %xmm0, %xmm1 -; SSE-NEXT: mulps %xmm2, %xmm1 -; SSE-NEXT: addps %xmm2, %xmm1 -; SSE-NEXT: mulps {{.*}}(%rip), %xmm1 +; SSE-NEXT: movss {{.*#+}} xmm1 = mem[0],zero,zero,zero +; SSE-NEXT: divss %xmm0, %xmm1 +; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,0,0,0] ; SSE-NEXT: movaps %xmm1, %xmm0 ; SSE-NEXT: retq ; ; AVX-LABEL: fdiv_splat_const_op0_v8f32: ; AVX: # %bb.0: +; AVX-NEXT: vmovss {{.*#+}} xmm1 = mem[0],zero,zero,zero +; AVX-NEXT: vdivss %xmm0, %xmm1, %xmm0 ; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,0,0,0] ; AVX-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0 -; AVX-NEXT: vrcpps %ymm0, %ymm1 -; AVX-NEXT: vmulps %ymm1, %ymm0, %ymm0 -; AVX-NEXT: vmovaps {{.*#+}} ymm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0] -; AVX-NEXT: vsubps %ymm0, %ymm2, %ymm0 -; AVX-NEXT: vmulps %ymm0, %ymm1, %ymm0 -; AVX-NEXT: vaddps %ymm0, %ymm1, %ymm0 -; AVX-NEXT: vmulps {{.*}}(%rip), %ymm0, %ymm0 ; AVX-NEXT: retq %splatx = shufflevector <8 x float> <float 4.5, float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0, float 7.0>, <8 x float> undef, <8 x i32> zeroinitializer %splaty = shufflevector <8 x float> %vy, <8 x float> undef, <8 x i32> zeroinitializer @@ -597,22 +587,18 @@ define <8 x float> @fdiv_splat_const_op0_v8f32(<8 x float> %vy) { define <8 x float> @fdiv_const_op1_splat_v8f32(<8 x float> %vx) { ; SSE-LABEL: fdiv_const_op1_splat_v8f32: ; SSE: # %bb.0: -; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,0,0,0] ; SSE-NEXT: xorps %xmm1, %xmm1 -; SSE-NEXT: rcpps %xmm1, %xmm1 -; SSE-NEXT: addps %xmm1, %xmm1 -; SSE-NEXT: mulps %xmm0, %xmm1 -; SSE-NEXT: movaps %xmm1, %xmm0 +; SSE-NEXT: divss %xmm1, %xmm0 +; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,0,0,0] +; SSE-NEXT: movaps %xmm0, %xmm1 ; SSE-NEXT: retq ; ; AVX-LABEL: fdiv_const_op1_splat_v8f32: ; AVX: # %bb.0: +; AVX-NEXT: vxorps %xmm1, %xmm1, %xmm1 +; AVX-NEXT: vdivss %xmm1, %xmm0, %xmm0 ; AVX-NEXT: vpermilps {{.*#+}} xmm0 = xmm0[0,0,0,0] ; AVX-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0 -; AVX-NEXT: vxorps %xmm1, %xmm1, %xmm1 -; AVX-NEXT: vrcpps %ymm1, %ymm1 -; AVX-NEXT: vaddps %ymm1, %ymm1, %ymm1 -; AVX-NEXT: vmulps %ymm1, %ymm0, %ymm0 ; AVX-NEXT: retq %splatx = shufflevector <8 x float> %vx, <8 x float> undef, <8 x i32> zeroinitializer %splaty = shufflevector <8 x float> <float 0.0, float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0, float 7.0>, <8 x float> undef, <8 x i32> zeroinitializer |
