summaryrefslogtreecommitdiffstats
path: root/llvm/test/CodeGen/X86/recip-fastmath.ll
diff options
context:
space:
mode:
Diffstat (limited to 'llvm/test/CodeGen/X86/recip-fastmath.ll')
-rw-r--r--llvm/test/CodeGen/X86/recip-fastmath.ll86
1 files changed, 43 insertions, 43 deletions
diff --git a/llvm/test/CodeGen/X86/recip-fastmath.ll b/llvm/test/CodeGen/X86/recip-fastmath.ll
index 06447ab9d63..47484865693 100644
--- a/llvm/test/CodeGen/X86/recip-fastmath.ll
+++ b/llvm/test/CodeGen/X86/recip-fastmath.ll
@@ -283,32 +283,32 @@ define float @f32_two_step(float %x) #2 {
define <4 x float> @v4f32_no_estimate(<4 x float> %x) #0 {
; SSE-LABEL: v4f32_no_estimate:
; SSE: # %bb.0:
-; SSE-NEXT: movaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT: movaps {{.*#+}} xmm1 = [1,1,1,1]
; SSE-NEXT: divps %xmm0, %xmm1
; SSE-NEXT: movaps %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX-RECIP-LABEL: v4f32_no_estimate:
; AVX-RECIP: # %bb.0:
-; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm1 = [1,1,1,1]
; AVX-RECIP-NEXT: vdivps %xmm0, %xmm1, %xmm0
; AVX-RECIP-NEXT: retq
;
; FMA-RECIP-LABEL: v4f32_no_estimate:
; FMA-RECIP: # %bb.0:
-; FMA-RECIP-NEXT: vmovaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT: vmovaps {{.*#+}} xmm1 = [1,1,1,1]
; FMA-RECIP-NEXT: vdivps %xmm0, %xmm1, %xmm0
; FMA-RECIP-NEXT: retq
;
; BTVER2-LABEL: v4f32_no_estimate:
; BTVER2: # %bb.0:
-; BTVER2-NEXT: vmovaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT: vmovaps {{.*#+}} xmm1 = [1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vdivps %xmm0, %xmm1, %xmm0 # sched: [19:19.00]
; BTVER2-NEXT: retq # sched: [4:1.00]
;
; SANDY-LABEL: v4f32_no_estimate:
; SANDY: # %bb.0:
-; SANDY-NEXT: vmovaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [6:0.50]
+; SANDY-NEXT: vmovaps {{.*#+}} xmm1 = [1,1,1,1] sched: [6:0.50]
; SANDY-NEXT: vdivps %xmm0, %xmm1, %xmm0 # sched: [14:14.00]
; SANDY-NEXT: retq # sched: [1:1.00]
;
@@ -344,7 +344,7 @@ define <4 x float> @v4f32_one_step(<4 x float> %x) #1 {
; SSE: # %bb.0:
; SSE-NEXT: rcpps %xmm0, %xmm2
; SSE-NEXT: mulps %xmm2, %xmm0
-; SSE-NEXT: movaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT: movaps {{.*#+}} xmm1 = [1,1,1,1]
; SSE-NEXT: subps %xmm0, %xmm1
; SSE-NEXT: mulps %xmm2, %xmm1
; SSE-NEXT: addps %xmm2, %xmm1
@@ -355,7 +355,7 @@ define <4 x float> @v4f32_one_step(<4 x float> %x) #1 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %xmm0, %xmm1
; AVX-RECIP-NEXT: vmulps %xmm1, %xmm0, %xmm0
-; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1]
; AVX-RECIP-NEXT: vsubps %xmm0, %xmm2, %xmm0
; AVX-RECIP-NEXT: vmulps %xmm0, %xmm1, %xmm0
; AVX-RECIP-NEXT: vaddps %xmm0, %xmm1, %xmm0
@@ -370,7 +370,7 @@ define <4 x float> @v4f32_one_step(<4 x float> %x) #1 {
;
; BTVER2-LABEL: v4f32_one_step:
; BTVER2: # %bb.0:
-; BTVER2-NEXT: vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %xmm0, %xmm1 # sched: [2:1.00]
; BTVER2-NEXT: vmulps %xmm1, %xmm0, %xmm0 # sched: [2:1.00]
; BTVER2-NEXT: vsubps %xmm0, %xmm2, %xmm0 # sched: [3:1.00]
@@ -382,7 +382,7 @@ define <4 x float> @v4f32_one_step(<4 x float> %x) #1 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; SANDY-NEXT: vmulps %xmm1, %xmm0, %xmm0 # sched: [5:1.00]
-; SANDY-NEXT: vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [6:0.50]
+; SANDY-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; SANDY-NEXT: vsubps %xmm0, %xmm2, %xmm0 # sched: [3:1.00]
; SANDY-NEXT: vmulps %xmm0, %xmm1, %xmm0 # sched: [5:1.00]
; SANDY-NEXT: vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
@@ -430,7 +430,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
; SSE-NEXT: rcpps %xmm0, %xmm2
; SSE-NEXT: movaps %xmm0, %xmm3
; SSE-NEXT: mulps %xmm2, %xmm3
-; SSE-NEXT: movaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT: movaps {{.*#+}} xmm1 = [1,1,1,1]
; SSE-NEXT: movaps %xmm1, %xmm4
; SSE-NEXT: subps %xmm3, %xmm4
; SSE-NEXT: mulps %xmm2, %xmm4
@@ -446,7 +446,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %xmm0, %xmm1
; AVX-RECIP-NEXT: vmulps %xmm1, %xmm0, %xmm2
-; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm3 = [1,1,1,1]
; AVX-RECIP-NEXT: vsubps %xmm2, %xmm3, %xmm2
; AVX-RECIP-NEXT: vmulps %xmm2, %xmm1, %xmm2
; AVX-RECIP-NEXT: vaddps %xmm2, %xmm1, %xmm1
@@ -459,7 +459,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
; FMA-RECIP-LABEL: v4f32_two_step:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %xmm0, %xmm1
-; FMA-RECIP-NEXT: vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1]
; FMA-RECIP-NEXT: vmovaps %xmm1, %xmm3
; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} xmm3 = -(xmm0 * xmm3) + xmm2
; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} xmm3 = (xmm3 * xmm1) + xmm1
@@ -469,7 +469,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
;
; BTVER2-LABEL: v4f32_two_step:
; BTVER2: # %bb.0:
-; BTVER2-NEXT: vmovaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT: vmovaps {{.*#+}} xmm3 = [1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %xmm0, %xmm1 # sched: [2:1.00]
; BTVER2-NEXT: vmulps %xmm1, %xmm0, %xmm2 # sched: [2:1.00]
; BTVER2-NEXT: vsubps %xmm2, %xmm3, %xmm2 # sched: [3:1.00]
@@ -485,7 +485,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; SANDY-NEXT: vmulps %xmm1, %xmm0, %xmm2 # sched: [5:1.00]
-; SANDY-NEXT: vmovaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [6:0.50]
+; SANDY-NEXT: vmovaps {{.*#+}} xmm3 = [1,1,1,1] sched: [6:0.50]
; SANDY-NEXT: vsubps %xmm2, %xmm3, %xmm2 # sched: [3:1.00]
; SANDY-NEXT: vmulps %xmm2, %xmm1, %xmm2 # sched: [5:1.00]
; SANDY-NEXT: vaddps %xmm2, %xmm1, %xmm1 # sched: [3:1.00]
@@ -548,7 +548,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
define <8 x float> @v8f32_no_estimate(<8 x float> %x) #0 {
; SSE-LABEL: v8f32_no_estimate:
; SSE: # %bb.0:
-; SSE-NEXT: movaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT: movaps {{.*#+}} xmm2 = [1,1,1,1]
; SSE-NEXT: movaps %xmm2, %xmm3
; SSE-NEXT: divps %xmm0, %xmm3
; SSE-NEXT: divps %xmm1, %xmm2
@@ -558,25 +558,25 @@ define <8 x float> @v8f32_no_estimate(<8 x float> %x) #0 {
;
; AVX-RECIP-LABEL: v8f32_no_estimate:
; AVX-RECIP: # %bb.0:
-; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
; AVX-RECIP-NEXT: vdivps %ymm0, %ymm1, %ymm0
; AVX-RECIP-NEXT: retq
;
; FMA-RECIP-LABEL: v8f32_no_estimate:
; FMA-RECIP: # %bb.0:
-; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
; FMA-RECIP-NEXT: vdivps %ymm0, %ymm1, %ymm0
; FMA-RECIP-NEXT: retq
;
; BTVER2-LABEL: v8f32_no_estimate:
; BTVER2: # %bb.0:
-; BTVER2-NEXT: vmovaps {{.*#+}} ymm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT: vmovaps {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vdivps %ymm0, %ymm1, %ymm0 # sched: [38:38.00]
; BTVER2-NEXT: retq # sched: [4:1.00]
;
; SANDY-LABEL: v8f32_no_estimate:
; SANDY: # %bb.0:
-; SANDY-NEXT: vmovaps {{.*#+}} ymm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT: vmovaps {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-NEXT: vdivps %ymm0, %ymm1, %ymm0 # sched: [29:28.00]
; SANDY-NEXT: retq # sched: [1:1.00]
;
@@ -612,7 +612,7 @@ define <8 x float> @v8f32_one_step(<8 x float> %x) #1 {
; SSE: # %bb.0:
; SSE-NEXT: rcpps %xmm0, %xmm4
; SSE-NEXT: mulps %xmm4, %xmm0
-; SSE-NEXT: movaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT: movaps {{.*#+}} xmm2 = [1,1,1,1]
; SSE-NEXT: movaps %xmm2, %xmm3
; SSE-NEXT: subps %xmm0, %xmm3
; SSE-NEXT: mulps %xmm4, %xmm3
@@ -630,7 +630,7 @@ define <8 x float> @v8f32_one_step(<8 x float> %x) #1 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %ymm0, %ymm1
; AVX-RECIP-NEXT: vmulps %ymm1, %ymm0, %ymm0
-; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
; AVX-RECIP-NEXT: vsubps %ymm0, %ymm2, %ymm0
; AVX-RECIP-NEXT: vmulps %ymm0, %ymm1, %ymm0
; AVX-RECIP-NEXT: vaddps %ymm0, %ymm1, %ymm0
@@ -645,7 +645,7 @@ define <8 x float> @v8f32_one_step(<8 x float> %x) #1 {
;
; BTVER2-LABEL: v8f32_one_step:
; BTVER2: # %bb.0:
-; BTVER2-NEXT: vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %ymm0, %ymm1 # sched: [2:2.00]
; BTVER2-NEXT: vmulps %ymm1, %ymm0, %ymm0 # sched: [2:2.00]
; BTVER2-NEXT: vsubps %ymm0, %ymm2, %ymm0 # sched: [3:2.00]
@@ -657,7 +657,7 @@ define <8 x float> @v8f32_one_step(<8 x float> %x) #1 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %ymm0, %ymm1 # sched: [7:2.00]
; SANDY-NEXT: vmulps %ymm1, %ymm0, %ymm0 # sched: [5:1.00]
-; SANDY-NEXT: vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-NEXT: vsubps %ymm0, %ymm2, %ymm0 # sched: [3:1.00]
; SANDY-NEXT: vmulps %ymm0, %ymm1, %ymm0 # sched: [5:1.00]
; SANDY-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
@@ -706,7 +706,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
; SSE-NEXT: rcpps %xmm0, %xmm3
; SSE-NEXT: movaps %xmm0, %xmm4
; SSE-NEXT: mulps %xmm3, %xmm4
-; SSE-NEXT: movaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT: movaps {{.*#+}} xmm1 = [1,1,1,1]
; SSE-NEXT: movaps %xmm1, %xmm5
; SSE-NEXT: subps %xmm4, %xmm5
; SSE-NEXT: mulps %xmm3, %xmm5
@@ -734,7 +734,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %ymm0, %ymm1
; AVX-RECIP-NEXT: vmulps %ymm1, %ymm0, %ymm2
-; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; AVX-RECIP-NEXT: vsubps %ymm2, %ymm3, %ymm2
; AVX-RECIP-NEXT: vmulps %ymm2, %ymm1, %ymm2
; AVX-RECIP-NEXT: vaddps %ymm2, %ymm1, %ymm1
@@ -747,7 +747,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
; FMA-RECIP-LABEL: v8f32_two_step:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm1
-; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
; FMA-RECIP-NEXT: vmovaps %ymm1, %ymm3
; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm3 = -(ymm0 * ymm3) + ymm2
; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm3 = (ymm3 * ymm1) + ymm1
@@ -757,7 +757,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
;
; BTVER2-LABEL: v8f32_two_step:
; BTVER2: # %bb.0:
-; BTVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %ymm0, %ymm1 # sched: [2:2.00]
; BTVER2-NEXT: vmulps %ymm1, %ymm0, %ymm2 # sched: [2:2.00]
; BTVER2-NEXT: vsubps %ymm2, %ymm3, %ymm2 # sched: [3:2.00]
@@ -773,7 +773,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %ymm0, %ymm1 # sched: [7:2.00]
; SANDY-NEXT: vmulps %ymm1, %ymm0, %ymm2 # sched: [5:1.00]
-; SANDY-NEXT: vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-NEXT: vsubps %ymm2, %ymm3, %ymm2 # sched: [3:1.00]
; SANDY-NEXT: vmulps %ymm2, %ymm1, %ymm2 # sched: [5:1.00]
; SANDY-NEXT: vaddps %ymm2, %ymm1, %ymm1 # sched: [3:1.00]
@@ -836,7 +836,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
define <16 x float> @v16f32_no_estimate(<16 x float> %x) #0 {
; SSE-LABEL: v16f32_no_estimate:
; SSE: # %bb.0:
-; SSE-NEXT: movaps {{.*#+}} xmm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT: movaps {{.*#+}} xmm4 = [1,1,1,1]
; SSE-NEXT: movaps %xmm4, %xmm5
; SSE-NEXT: divps %xmm0, %xmm5
; SSE-NEXT: movaps %xmm4, %xmm6
@@ -852,28 +852,28 @@ define <16 x float> @v16f32_no_estimate(<16 x float> %x) #0 {
;
; AVX-RECIP-LABEL: v16f32_no_estimate:
; AVX-RECIP: # %bb.0:
-; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
; AVX-RECIP-NEXT: vdivps %ymm0, %ymm2, %ymm0
; AVX-RECIP-NEXT: vdivps %ymm1, %ymm2, %ymm1
; AVX-RECIP-NEXT: retq
;
; FMA-RECIP-LABEL: v16f32_no_estimate:
; FMA-RECIP: # %bb.0:
-; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
; FMA-RECIP-NEXT: vdivps %ymm0, %ymm2, %ymm0
; FMA-RECIP-NEXT: vdivps %ymm1, %ymm2, %ymm1
; FMA-RECIP-NEXT: retq
;
; BTVER2-LABEL: v16f32_no_estimate:
; BTVER2: # %bb.0:
-; BTVER2-NEXT: vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vdivps %ymm0, %ymm2, %ymm0 # sched: [38:38.00]
; BTVER2-NEXT: vdivps %ymm1, %ymm2, %ymm1 # sched: [38:38.00]
; BTVER2-NEXT: retq # sched: [4:1.00]
;
; SANDY-LABEL: v16f32_no_estimate:
; SANDY: # %bb.0:
-; SANDY-NEXT: vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-NEXT: vdivps %ymm0, %ymm2, %ymm0 # sched: [29:28.00]
; SANDY-NEXT: vdivps %ymm1, %ymm2, %ymm1 # sched: [29:28.00]
; SANDY-NEXT: retq # sched: [1:1.00]
@@ -914,7 +914,7 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
; SSE-NEXT: movaps %xmm0, %xmm5
; SSE-NEXT: rcpps %xmm0, %xmm6
; SSE-NEXT: mulps %xmm6, %xmm5
-; SSE-NEXT: movaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT: movaps {{.*#+}} xmm3 = [1,1,1,1]
; SSE-NEXT: movaps %xmm3, %xmm0
; SSE-NEXT: subps %xmm5, %xmm0
; SSE-NEXT: mulps %xmm6, %xmm0
@@ -944,7 +944,7 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %ymm0, %ymm2
; AVX-RECIP-NEXT: vmulps %ymm2, %ymm0, %ymm0
-; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; AVX-RECIP-NEXT: vsubps %ymm0, %ymm3, %ymm0
; AVX-RECIP-NEXT: vmulps %ymm0, %ymm2, %ymm0
; AVX-RECIP-NEXT: vaddps %ymm0, %ymm2, %ymm0
@@ -958,7 +958,7 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
; FMA-RECIP-LABEL: v16f32_one_step:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm2
-; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm2 * ymm0) + ymm3
; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm2) + ymm2
; FMA-RECIP-NEXT: vrcpps %ymm1, %ymm2
@@ -968,7 +968,7 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
;
; BTVER2-LABEL: v16f32_one_step:
; BTVER2: # %bb.0:
-; BTVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %ymm0, %ymm2 # sched: [2:2.00]
; BTVER2-NEXT: vrcpps %ymm1, %ymm4 # sched: [2:2.00]
; BTVER2-NEXT: vmulps %ymm2, %ymm0, %ymm0 # sched: [2:2.00]
@@ -985,7 +985,7 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %ymm0, %ymm2 # sched: [7:2.00]
; SANDY-NEXT: vmulps %ymm2, %ymm0, %ymm0 # sched: [5:1.00]
-; SANDY-NEXT: vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-NEXT: vsubps %ymm0, %ymm3, %ymm0 # sched: [3:1.00]
; SANDY-NEXT: vmulps %ymm0, %ymm2, %ymm0 # sched: [5:1.00]
; SANDY-NEXT: vaddps %ymm0, %ymm2, %ymm0 # sched: [3:1.00]
@@ -1048,7 +1048,7 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
; SSE-NEXT: rcpps %xmm0, %xmm0
; SSE-NEXT: movaps %xmm1, %xmm6
; SSE-NEXT: mulps %xmm0, %xmm6
-; SSE-NEXT: movaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT: movaps {{.*#+}} xmm3 = [1,1,1,1]
; SSE-NEXT: movaps %xmm3, %xmm7
; SSE-NEXT: subps %xmm6, %xmm7
; SSE-NEXT: mulps %xmm0, %xmm7
@@ -1100,7 +1100,7 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %ymm0, %ymm2
; AVX-RECIP-NEXT: vmulps %ymm2, %ymm0, %ymm3
-; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
; AVX-RECIP-NEXT: vsubps %ymm3, %ymm4, %ymm3
; AVX-RECIP-NEXT: vmulps %ymm3, %ymm2, %ymm3
; AVX-RECIP-NEXT: vaddps %ymm3, %ymm2, %ymm2
@@ -1122,7 +1122,7 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
; FMA-RECIP-LABEL: v16f32_two_step:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm2
-; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; FMA-RECIP-NEXT: vmovaps %ymm2, %ymm4
; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm4 = -(ymm0 * ymm4) + ymm3
; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm4 = (ymm4 * ymm2) + ymm2
@@ -1138,7 +1138,7 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
;
; BTVER2-LABEL: v16f32_two_step:
; BTVER2: # %bb.0:
-; BTVER2-NEXT: vmovaps {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT: vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %ymm0, %ymm2 # sched: [2:2.00]
; BTVER2-NEXT: vmulps %ymm2, %ymm0, %ymm3 # sched: [2:2.00]
; BTVER2-NEXT: vsubps %ymm3, %ymm4, %ymm3 # sched: [3:2.00]
@@ -1163,7 +1163,7 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %ymm0, %ymm2 # sched: [7:2.00]
; SANDY-NEXT: vmulps %ymm2, %ymm0, %ymm3 # sched: [5:1.00]
-; SANDY-NEXT: vmovaps {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT: vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-NEXT: vsubps %ymm3, %ymm4, %ymm3 # sched: [3:1.00]
; SANDY-NEXT: vmulps %ymm3, %ymm2, %ymm3 # sched: [5:1.00]
; SANDY-NEXT: vaddps %ymm3, %ymm2, %ymm2 # sched: [3:1.00]
OpenPOWER on IntegriCloud