Home
last modified time | relevance | path

Searched refs:fma (Results 1 – 25 of 193) sorted by relevance

12345678

/external/llvm/test/Transforms/InstSimplify/
Dfold-builtin-fma.ll7 declare double @llvm.fma.f64(double, double, double)
11 %1 = call double @llvm.fma.f64(double 7.0, double 8.0, double 0.0)
17 ; Test builtin fma with all finite non-zero constants.
19 %1 = call double @llvm.fma.f64(double 7.0, double 8.0, double 5.0)
25 ; Test builtin fma with a +/-NaN addend.
27 %1 = call double @llvm.fma.f64(double 7.0, double 8.0, double 0x7FF8000000000000)
34 %1 = call double @llvm.fma.f64(double 7.0, double 8.0, double 0xFFF8000000000000)
40 ; Test builtin fma with a +/-Inf addend.
42 %1 = call double @llvm.fma.f64(double 7.0, double 8.0, double 0x7FF0000000000000)
49 %1 = call double @llvm.fma.f64(double 7.0, double 8.0, double 0xFFF0000000000000)
[all …]
/external/llvm/test/CodeGen/AMDGPU/
Duse-sgpr-multiple-times.ll4 declare float @llvm.fma.f32(float, float, float) #1
5 declare double @llvm.fma.f64(double, double, double) #1
25 %fma = call float @llvm.fma.f32(float %a, float %a, float %a) #1
26 store float %fma, float addrspace(1)* %out, align 4
39 %fma = call float @llvm.fma.f32(float %a, float %a, float %b) #1
40 store float %fma, float addrspace(1)* %out, align 4
63 %fma0 = call float @llvm.fma.f32(float %a, float %va0, float %b) #1
64 %fma1 = call float @llvm.fma.f32(float %a, float %va1, float %b) #1
79 %fma = call float @llvm.fma.f32(float %a, float %b, float %a) #1
80 store float %fma, float addrspace(1)* %out, align 4
[all …]
Dcoalescer_remat.ll3 declare float @llvm.fma.f32(float, float, float)
23 %v0 = phi float [0.0, %entry], [%fma.0, %loop]
24 %v1 = phi float [0.0, %entry], [%fma.1, %loop]
25 %v2 = phi float [0.0, %entry], [%fma.2, %loop]
26 %v3 = phi float [0.0, %entry], [%fma.3, %loop]
37 %fma.0 = call float @llvm.fma.f32(float %v0, float %load.0, float %v0)
38 %fma.1 = call float @llvm.fma.f32(float %v1, float %load.1, float %v1)
39 %fma.2 = call float @llvm.fma.f32(float %v2, float %load.2, float %v2)
40 %fma.3 = call float @llvm.fma.f32(float %v3, float %load.3, float %v3)
47 %ev0 = phi float [0.0, %entry], [%fma.0, %loop]
[all …]
Dfma.ll4 declare float @llvm.fma.f32(float, float, float) nounwind readnone
5 declare <2 x float> @llvm.fma.v2f32(<2 x float>, <2 x float>, <2 x float>) nounwind readnone
6 declare <4 x float> @llvm.fma.v4f32(<4 x float>, <4 x float>, <4 x float>) nounwind readnone
20 %r3 = tail call float @llvm.fma.f32(float %r0, float %r1, float %r2)
37 %r3 = tail call <2 x float> @llvm.fma.v2f32(<2 x float> %r0, <2 x float> %r1, <2 x float> %r2)
58 %r3 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %r0, <4 x float> %r1, <4 x float> %r2)
74 %fma = call float @llvm.fma.f32(float %a, float 2.0, float %b)
75 store float %fma, float addrspace(1)* %out.gep, align 4
89 %fma = call float @llvm.fma.f32(float %a, float %b, float %c)
90 store float %fma, float addrspace(1)* %out.gep, align 4
/external/llvm/test/CodeGen/X86/
Dfma.ll1 ; RUN: llc < %s -mtriple=i386-apple-darwin10 -mattr=+fma,-fma4 | FileCheck %s --check-prefix=CHEC…
2 ; RUN: llc < %s -mtriple=i386-apple-darwin10 -mattr=-fma,-fma4 | FileCheck %s --check-prefix=CHEC…
3 ; RUN: llc < %s -mtriple=x86_64-apple-darwin10 -mattr=+fma,-fma4 | FileCheck %s --check-prefix=CHEC…
4 ; RUN: llc < %s -mtriple=x86_64-apple-darwin10 -mattr=-fma,-fma4 | FileCheck %s --check-prefix=CHE…
5 ; RUN: llc < %s -mtriple=x86_64-apple-darwin10 -mattr=+avx512f,-fma,-fma4 | FileCheck %s --check-p…
7 ; RUN: llc < %s -march=x86 -mcpu=bdver2 -mattr=-fma,-fma4 | FileCheck %s --check-prefix=CHECK --che…
14 %call = call float @llvm.fma.f32(float %a, float %b, float %c)
20 ; CHECK-FMA-CALL: fma
23 %call = call double @llvm.fma.f64(double %a, double %b, double %c)
31 %call = call x86_fp80 @llvm.fma.f80(x86_fp80 %a, x86_fp80 %b, x86_fp80 %c)
[all …]
Dfma-commute-x86.ll2 ; RUN: llc < %s -mtriple=x86_64-pc-win32 -mattr=+fma | FileCheck %s
7 declare <4 x float> @llvm.x86.fma.vfmadd.ss(<4 x float>, <4 x float>, <4 x float>) nounwind readnone
15 …%res = call <4 x float> @llvm.x86.fma.vfmadd.ss(<4 x float> %b, <4 x float> %a, <4 x float> %a) no…
25 …%res = call <4 x float> @llvm.x86.fma.vfmadd.ss(<4 x float> %a, <4 x float> %b, <4 x float> %a) no…
35 …%res = call <4 x float> @llvm.x86.fma.vfmadd.ss(<4 x float> %b, <4 x float> %b, <4 x float> %a) no…
39 declare <4 x float> @llvm.x86.fma.vfmadd.ps(<4 x float>, <4 x float>, <4 x float>) nounwind readnone
46 …%res = call <4 x float> @llvm.x86.fma.vfmadd.ps(<4 x float> %b, <4 x float> %a, <4 x float> %a) no…
56 …%res = call <4 x float> @llvm.x86.fma.vfmadd.ps(<4 x float> %a, <4 x float> %b, <4 x float> %a) no…
66 …%res = call <4 x float> @llvm.x86.fma.vfmadd.ps(<4 x float> %b, <4 x float> %b, <4 x float> %a) no…
70 declare <8 x float> @llvm.x86.fma.vfmadd.ps.256(<8 x float>, <8 x float>, <8 x float>) nounwind rea…
[all …]
Dfma4-intrinsics-x86_64-folded-load.ll2 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=bdver2 -mattr=+avx,-fma | FileCheck %s
9 …%res = call < 4 x float > @llvm.x86.fma.vfmadd.ss(< 4 x float > %a0, < 4 x float > %a1, < 4 x floa…
16 …%res = call < 4 x float > @llvm.x86.fma.vfmadd.ss(< 4 x float > %a0, < 4 x float > %y, < 4 x float…
20 declare < 4 x float > @llvm.x86.fma.vfmadd.ss(< 4 x float >, < 4 x float >, < 4 x float >) nounwind…
26 …%res = call < 2 x double > @llvm.x86.fma.vfmadd.sd(< 2 x double > %a0, < 2 x double > %a1, < 2 x d…
33 …%res = call < 2 x double > @llvm.x86.fma.vfmadd.sd(< 2 x double > %a0, < 2 x double > %y, < 2 x do…
36 declare < 2 x double > @llvm.x86.fma.vfmadd.sd(< 2 x double >, < 2 x double >, < 2 x double >) noun…
40 …%res = call < 4 x float > @llvm.x86.fma.vfmadd.ps(< 4 x float > %a0, < 4 x float > %a1, < 4 x floa…
46 …%res = call < 4 x float > @llvm.x86.fma.vfmadd.ps(< 4 x float > %a0, < 4 x float > %x, < 4 x float…
49 declare < 4 x float > @llvm.x86.fma.vfmadd.ps(< 4 x float >, < 4 x float >, < 4 x float >) nounwind…
[all …]
Dfma-intrinsics-x86.ll1 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -march=x86-64 -mcpu=corei7-avx -mattr=+fma | FileCh…
2 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -march=x86-64 -mcpu=core-avx2 -mattr=+fma,+avx2 | F…
3 ; RUN: llc < %s -mtriple=x86_64-pc-windows -march=x86-64 -mcpu=core-avx2 -mattr=+fma,+avx2 | FileCh…
5 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=bdver2 -mattr=+avx,-fma | FileCheck %s --chec…
22 %res = call <4 x float> @llvm.x86.fma.vfmadd.ss(<4 x float> %a0, <4 x float> %a1, <4 x float> %a2)
39 %res = call <4 x float> @llvm.x86.fma.vfmadd.ss(<4 x float> %a1, <4 x float> %a0, <4 x float> %a2)
42 declare <4 x float> @llvm.x86.fma.vfmadd.ss(<4 x float>, <4 x float>, <4 x float>)
57 …%res = call <2 x double> @llvm.x86.fma.vfmadd.sd(<2 x double> %a0, <2 x double> %a1, <2 x double> …
75 …%res = call <2 x double> @llvm.x86.fma.vfmadd.sd(<2 x double> %a1, <2 x double> %a0, <2 x double> …
78 declare <2 x double> @llvm.x86.fma.vfmadd.sd(<2 x double>, <2 x double>, <2 x double>)
[all …]
Dfma-intrinsics-phi-213-to-231.ll1 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2,+fma | FileCheck %s
21 …%0 = call <2 x double> @llvm.x86.fma.vfmaddsub.pd(<2 x double> %a, <2 x double> %b, <2 x double> %…
47 …%0 = call <2 x double> @llvm.x86.fma.vfmsubadd.pd(<2 x double> %a, <2 x double> %b, <2 x double> %…
73 …%0 = call <2 x double> @llvm.x86.fma.vfmadd.pd(<2 x double> %a, <2 x double> %b, <2 x double> %c.a…
99 …%0 = call <2 x double> @llvm.x86.fma.vfmsub.pd(<2 x double> %a, <2 x double> %b, <2 x double> %c.a…
125 …%0 = call <2 x double> @llvm.x86.fma.vfnmadd.pd(<2 x double> %a, <2 x double> %b, <2 x double> %c.…
151 …%0 = call <2 x double> @llvm.x86.fma.vfnmsub.pd(<2 x double> %a, <2 x double> %b, <2 x double> %c.…
159 declare <2 x double> @llvm.x86.fma.vfmaddsub.pd(<2 x double>, <2 x double>, <2 x double>)
160 declare <2 x double> @llvm.x86.fma.vfmsubadd.pd(<2 x double>, <2 x double>, <2 x double>)
161 declare <2 x double> @llvm.x86.fma.vfmadd.pd(<2 x double>, <2 x double>, <2 x double>)
[all …]
Dfma-phi-213-to-231.ll1 ; RUN: llc < %s -mtriple=i386-apple-darwin10 -mattr=+fma,-fma4 | FileCheck %s
2 ; RUN: llc < %s -mtriple=x86_64-apple-darwin10 -mattr=+fma,-fma4 | FileCheck %s
14 %acc = call float @llvm.fma.f32(float %x, float %y, float %acc.01)
28 %add = call <4 x double> @llvm.fma.v4f64(<4 x double> %x, <4 x double> %y, <4 x double> %acc.04)
36 declare float @llvm.fma.f32(float, float, float)
37 declare <4 x double> @llvm.fma.v4f64(<4 x double>, <4 x double>, <4 x double>)
/external/llvm/test/CodeGen/AArch64/
Darm64-fmadd.ll7 %0 = tail call float @llvm.fma.f32(float %a, float %b, float %c)
15 %0 = tail call float @llvm.fma.f32(float %a, float %b, float %c)
25 %0 = tail call float @llvm.fma.f32(float %a, float %mul, float %c)
34 %0 = tail call float @llvm.fma.f32(float %mul, float %a, float %c)
43 %0 = tail call float @llvm.fma.f32(float %a, float %b, float %mul)
51 %0 = tail call double @llvm.fma.f64(double %a, double %b, double %c)
59 %0 = tail call double @llvm.fma.f64(double %a, double %b, double %c)
69 %0 = tail call double @llvm.fma.f64(double %a, double %mul, double %c)
78 %0 = tail call double @llvm.fma.f64(double %mul, double %a, double %c)
87 %0 = tail call double @llvm.fma.f64(double %a, double %b, double %mul)
[all …]
Dneon-scalar-by-elem-fma.ll3 declare float @llvm.fma.f32(float, float, float)
4 declare double @llvm.fma.f64(double, double, double)
10 %tmp2 = call float @llvm.fma.f32(float %b, float %tmp1, float %a)
18 %tmp2 = call float @llvm.fma.f32(float %tmp1, float %a, float %a)
26 %tmp2 = call float @llvm.fma.f32(float %b, float %tmp1, float %a)
34 %tmp2 = call double @llvm.fma.f64(double %b, double %tmp1, double %a)
42 %tmp2 = call double @llvm.fma.f64(double %b, double %tmp1, double %a)
50 %tmp2 = call double @llvm.fma.f64(double %tmp1, double %b, double %a)
59 %tmp3 = call float @llvm.fma.f32(float %tmp2, float %tmp1, float %a)
68 %tmp3 = call float @llvm.fma.f32(float %tmp1, float %tmp2, float %a)
[all …]
/external/llvm/test/CodeGen/ARM/
DfusedMAC.ll106 %tmp1 = tail call float @llvm.fma.f32(float %a, float %b, float %c) nounwind readnone
114 %tmp1 = tail call double @llvm.fma.f64(double %a, double %b, double %c) nounwind readnone
122 …%tmp1 = tail call <2 x float> @llvm.fma.v2f32(<2 x float> %a, <2 x float> %b, <2 x float> %c) noun…
131 %tmp2 = tail call double @llvm.fma.f64(double %tmp1, double %b, double %c) nounwind readnone
140 %tmp2 = tail call double @llvm.fma.f64(double %a, double %tmp1, double %c) nounwind readnone
149 %tmp3 = tail call float @llvm.fma.f32(float %a, float %b, float %tmp2) nounwind readnone
158 %tmp2 = tail call double @llvm.fma.f64(double %tmp1, double %b, double %c) nounwind readnone
168 %tmp2 = tail call double @llvm.fma.f64(double %a, double %tmp1, double %c) nounwind readnone
177 %tmp1 = tail call double @llvm.fma.f64(double %a, double %b, double %c) nounwind readnone
188 %tmp3 = tail call double @llvm.fma.f64(double %tmp1, double %b, double %tmp2) nounwind readnone
[all …]
Dfp-fast.ll10 %t2 = call float @llvm.fma.f32(float %x, float 2.0, float %t1)
20 %t2 = call float @llvm.fma.f32(float %t1, float 2.0, float %y)
29 %t2 = call float @llvm.fma.f32(float %x, float 1.0, float %y)
38 %t2 = call float @llvm.fma.f32(float %x, float -1.0, float %y)
47 %t2 = call float @llvm.fma.f32(float %x, float 2.0, float %x)
57 %t2 = call float @llvm.fma.f32(float %x, float 5.0, float %t1)
61 declare float @llvm.fma.f32(float, float, float)
/external/llvm/test/CodeGen/NVPTX/
Dfma-disable.ll1 ; RUN: llc < %s -march=nvptx -mcpu=sm_20 -nvptx-fma-level=1 | FileCheck %s -check-prefix=FMA
2 ; RUN: llc < %s -march=nvptx -mcpu=sm_20 -nvptx-fma-level=0 | FileCheck %s -check-prefix=MUL
3 ; RUN: llc < %s -march=nvptx64 -mcpu=sm_20 -nvptx-fma-level=1 | FileCheck %s -check-prefix=FMA
4 ; RUN: llc < %s -march=nvptx64 -mcpu=sm_20 -nvptx-fma-level=0 | FileCheck %s -check-prefix=MUL
8 ; FMA: fma.rn.f32
18 ; FMA: fma.rn.f64
Dfma.ll7 ; CHECK: fma.rn.f32 %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}};
15 ; CHECK: fma.rn.f32 %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}};
16 ; CHECK: fma.rn.f32 %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}};
26 ; CHECK: fma.rn.f64 %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}};
34 ; CHECK: fma.rn.f64 %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}};
35 ; CHECK: fma.rn.f64 %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}};
Dfma-assoc.ll5 ; CHECK: fma.rn.f32 %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}};
6 ; CHECK: fma.rn.f32 %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}}, %f{{[0-9]+}};
17 ; CHECK: fma.rn.f64 %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}};
18 ; CHECK: fma.rn.f64 %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}}, %fd{{[0-9]+}};
31 ; CHECK: fma.rn.f64
/external/swiftshader/third_party/LLVM/test/CodeGen/X86/
Dfma.ll9 %call = tail call float @llvm.fma.f32(float %a, float %b, float %c) nounwind readnone
18 %call = tail call double @llvm.fma.f64(double %a, double %b, double %c) nounwind readnone
27 %call = tail call x86_fp80 @llvm.fma.f80(x86_fp80 %a, x86_fp80 %b, x86_fp80 %c) nounwind readnone
31 declare float @llvm.fma.f32(float, float, float) nounwind readnone
32 declare double @llvm.fma.f64(double, double, double) nounwind readnone
33 declare x86_fp80 @llvm.fma.f80(x86_fp80, x86_fp80, x86_fp80) nounwind readnone
/external/llvm/test/CodeGen/PowerPC/
Dvsx-fma-m.ll5 ; Also run with -schedule-ppc-vsx-fma-mutation-early as a stress test for the
7 ; RUN: llc < %s -mcpu=pwr7 -mattr=+vsx -schedule-ppc-vsx-fma-mutation-early
13 %0 = tail call double @llvm.fma.f64(double %b, double %c, double %a)
15 %1 = tail call double @llvm.fma.f64(double %b, double %e, double %a)
40 %0 = tail call double @llvm.fma.f64(double %b, double %c, double %a)
42 %1 = tail call double @llvm.fma.f64(double %b, double %e, double %a)
45 %2 = tail call double @llvm.fma.f64(double %b, double %f, double %a)
78 %0 = tail call double @llvm.fma.f64(double %b, double %c, double %a)
80 %1 = tail call double @llvm.fma.f64(double %b, double %e, double %a)
81 %2 = tail call double @llvm.fma.f64(double %b, double %c, double %1)
[all …]
Dvsx-fma-sp.ll7 %0 = tail call float @llvm.fma.f32(float %b, float %c, float %a)
9 %1 = tail call float @llvm.fma.f32(float %b, float %e, float %a)
34 %0 = tail call float @llvm.fma.f32(float %b, float %c, float %a)
36 %1 = tail call float @llvm.fma.f32(float %b, float %e, float %a)
39 %2 = tail call float @llvm.fma.f32(float %b, float %f, float %a)
72 %0 = tail call float @llvm.fma.f32(float %b, float %c, float %a)
74 %1 = tail call float @llvm.fma.f32(float %b, float %e, float %a)
75 %2 = tail call float @llvm.fma.f32(float %b, float %c, float %1)
78 %3 = tail call float @llvm.fma.f32(float %b, float %f, float %a)
121 %0 = tail call float @llvm.fma.f32(float %b, float %c, float %a)
[all …]
/external/llvm/test/Transforms/SLPVectorizer/X86/
Dfma.ll19 declare float @llvm.fma.f32(float, float, float)
20 declare double @llvm.fma.f64(double, double, double)
34 ; NO-FMA-NEXT: [[FMA0:%.*]] = call double @llvm.fma.f64(double [[A0]], double [[B0]], double [[C…
35 ; NO-FMA-NEXT: [[FMA1:%.*]] = call double @llvm.fma.f64(double [[A1]], double [[B1]], double [[C…
44 ; FMA-NEXT: [[TMP4:%.*]] = call <2 x double> @llvm.fma.v2f64(<2 x double> [[TMP1]], <2 x double>…
54 %fma0 = call double @llvm.fma.f64(double %a0, double %b0, double %c0)
55 %fma1 = call double @llvm.fma.f64(double %a1, double %b1, double %c1)
75 ; NO-FMA-NEXT: [[FMA0:%.*]] = call double @llvm.fma.f64(double [[A0]], double [[B0]], double [[C…
76 ; NO-FMA-NEXT: [[FMA1:%.*]] = call double @llvm.fma.f64(double [[A1]], double [[B1]], double [[C…
77 ; NO-FMA-NEXT: [[FMA2:%.*]] = call double @llvm.fma.f64(double [[A2]], double [[B2]], double [[C…
[all …]
/external/llvm/test/ExecutionEngine/
Dfma3-jit.ll10 %fma = tail call double @llvm.fma.f64(double 3.0, double 3.0, double 3.0) nounwind readnone
13 call i32 (i8*,...) @printf(i8* %ptr1, double %fma)
18 declare double @llvm.fma.f64(double, double, double) nounwind readnone
/external/llvm/test/CodeGen/SystemZ/
Dfp-mul-06.ll3 declare float @llvm.fma.f32(float %f1, float %f2, float %f3)
10 %res = call float @llvm.fma.f32 (float %f1, float %f2, float %acc)
20 %res = call float @llvm.fma.f32 (float %f1, float %f2, float %acc)
31 %res = call float @llvm.fma.f32 (float %f1, float %f2, float %acc)
46 %res = call float @llvm.fma.f32 (float %f1, float %f2, float %acc)
61 %res = call float @llvm.fma.f32 (float %f1, float %f2, float %acc)
73 %res = call float @llvm.fma.f32 (float %f1, float %f2, float %acc)
86 %res = call float @llvm.fma.f32 (float %f1, float %f2, float %acc)
100 %res = call float @llvm.fma.f32 (float %f1, float %f2, float %acc)
Dfp-mul-08.ll3 declare float @llvm.fma.f32(float %f1, float %f2, float %f3)
11 %res = call float @llvm.fma.f32 (float %f1, float %f2, float %negacc)
22 %res = call float @llvm.fma.f32 (float %f1, float %f2, float %negacc)
34 %res = call float @llvm.fma.f32 (float %f1, float %f2, float %negacc)
50 %res = call float @llvm.fma.f32 (float %f1, float %f2, float %negacc)
66 %res = call float @llvm.fma.f32 (float %f1, float %f2, float %negacc)
79 %res = call float @llvm.fma.f32 (float %f1, float %f2, float %negacc)
93 %res = call float @llvm.fma.f32 (float %f1, float %f2, float %negacc)
108 %res = call float @llvm.fma.f32 (float %f1, float %f2, float %negacc)
/external/llvm/test/Transforms/InstCombine/
Dconstant-fold-math.ll3 declare float @llvm.fma.f32(float, float, float) #0
5 declare <4 x float> @llvm.fma.v4f32(<4 x float>, <4 x float>, <4 x float>) #0
7 declare double @llvm.fma.f64(double, double, double) #0
16 %x = call float @llvm.fma.f32(float 1.0, float 2.0, float 4.0) #0
23 …%x = call <4 x float> @llvm.fma.v4f32(<4 x float> <float 1.0, float 2.0, float 3.0, float 4.0>, <4…
37 %x = call double @llvm.fma.f64(double 1.0, double 2.0, double 4.0) #0

12345678