| /llvm-project-15.0.7/llvm/test/CodeGen/X86/ |
| H A D | fma-commute-x86.ll | 11 ; FMA: # %bb.0: 15 ; FMA-NEXT: retq 22 ; FMA: # %bb.0: 25 ; FMA-NEXT: retq 32 ; FMA: # %bb.0: 35 ; FMA-NEXT: retq 46 ; FMA-NEXT: retq 56 ; FMA-NEXT: retq 66 ; FMA-NEXT: retq 77 ; FMA-NEXT: retq [all …]
|
| H A D | fma-intrinsics-x86-upgrade.ll | 9 ; CHECK-FMA: # %bb.0: 21 ; CHECK-FMA-WIN: # %bb.0: 34 ; CHECK-FMA: # %bb.0: 62 ; CHECK-FMA: # %bb.0: 87 ; CHECK-FMA: # %bb.0: 115 ; CHECK-FMA: # %bb.0: 140 ; CHECK-FMA: # %bb.0: 165 ; CHECK-FMA: # %bb.0: 190 ; CHECK-FMA: # %bb.0: 216 ; CHECK-FMA: # %bb.0: [all …]
|
| H A D | fma-intrinsics-x86.ll | 9 ; CHECK-FMA: # %bb.0: 21 ; CHECK-FMA-WIN: # %bb.0: 38 ; CHECK-FMA: # %bb.0: 69 ; CHECK-FMA: # %bb.0: 100 ; CHECK-FMA: # %bb.0: 129 ; CHECK-FMA: # %bb.0: 160 ; CHECK-FMA: # %bb.0: 184 ; CHECK-FMA: # %bb.0: 208 ; CHECK-FMA: # %bb.0: 232 ; CHECK-FMA: # %bb.0: [all …]
|
| H A D | fma-intrinsics-canonical.ll | 9 ; CHECK-FMA-LABEL: test_mm_fmadd_ps: 10 ; CHECK-FMA: # %bb.0: # %entry 34 ; CHECK-FMA-LABEL: test_mm_fmadd_pd: 35 ; CHECK-FMA: # %bb.0: # %entry 59 ; CHECK-FMA-LABEL: test_mm_fmadd_ss: 60 ; CHECK-FMA: # %bb.0: # %entry 89 ; CHECK-FMA-LABEL: test_mm_fmadd_sd: 90 ; CHECK-FMA: # %bb.0: # %entry 119 ; CHECK-FMA-LABEL: test_mm_fmsub_ps: 120 ; CHECK-FMA: # %bb.0: # %entry [all …]
|
| H A D | fma_patterns.ll | 17 ; FMA: # %bb.0: 19 ; FMA-NEXT: retq 37 ; FMA: # %bb.0: 39 ; FMA-NEXT: retq 59 ; FMA-NEXT: retq 79 ; FMA-NEXT: retq 99 ; FMA-NEXT: retq 119 ; FMA-NEXT: retq 143 ; FMA-NEXT: retq 163 ; FMA-NEXT: retq [all …]
|
| H A D | fma_patterns_wide.ll | 17 ; FMA: # %bb.0: 20 ; FMA-NEXT: retq 39 ; FMA: # %bb.0: 42 ; FMA-NEXT: retq 68 ; FMA-NEXT: retq 90 ; FMA-NEXT: retq 116 ; FMA-NEXT: retq 138 ; FMA-NEXT: retq 164 ; FMA-NEXT: retq 187 ; FMA-NEXT: retq [all …]
|
| H A D | complex-fastmath.ll | 36 ; FMA-LABEL: complex_square_f32: 37 ; FMA: # %bb.0: 44 ; FMA-NEXT: retq 83 ; FMA: # %bb.0: 90 ; FMA-NEXT: retq 135 ; FMA-LABEL: complex_mul_f32: 136 ; FMA: # %bb.0: 144 ; FMA-NEXT: retq 190 ; FMA-LABEL: complex_mul_f64: 191 ; FMA: # %bb.0: [all …]
|
| H A D | recip-fastmath.ll | 61 ; FMA-RECIP: # %bb.0: 65 ; FMA-RECIP-NEXT: retq 144 ; FMA-RECIP: # %bb.0: 149 ; FMA-RECIP-NEXT: retq 249 ; FMA-RECIP-NEXT: retq 346 ; FMA-RECIP-NEXT: retq 414 ; FMA-RECIP-NEXT: retq 507 ; FMA-RECIP-NEXT: retq 607 ; FMA-RECIP-NEXT: retq 707 ; FMA-RECIP-NEXT: retq [all …]
|
| H A D | recip-fastmath2.ll | 57 ; FMA-RECIP: # %bb.0: 63 ; FMA-RECIP-NEXT: retq 155 ; FMA-RECIP: # %bb.0: 161 ; FMA-RECIP-NEXT: retq 276 ; FMA-RECIP-NEXT: retq 396 ; FMA-RECIP-NEXT: retq 494 ; FMA-RECIP-NEXT: retq 620 ; FMA-RECIP-NEXT: retq 749 ; FMA-RECIP-NEXT: retq 856 ; FMA-RECIP-NEXT: retq [all …]
|
| H A D | fp-intrinsics-fma.ll | 18 ; FMA-LABEL: f1: 21 ; FMA-NEXT: retq 46 ; FMA-LABEL: f2: 49 ; FMA-NEXT: retq 74 ; FMA-LABEL: f3: 77 ; FMA-NEXT: retq 102 ; FMA-LABEL: f4: 132 ; FMA-LABEL: f5: 163 ; FMA-LABEL: f6: 229 ; FMA-LABEL: f8: [all …]
|
| /llvm-project-15.0.7/lld/test/MachO/ |
| H A D | lto-mattrs.ll | 12 ; FMA: <_foo>: 13 ; FMA-NEXT: vrcpss %xmm0, %xmm0, %xmm1 16 ; FMA-NEXT: vfnmadd132ss %xmm1, %xmm1, %xmm0 17 ; FMA-NEXT: retq 19 ; NO-FMA: <_foo>: 20 ; NO-FMA-NEXT: vrcpss %xmm0, %xmm0, %xmm1 21 ; NO-FMA-NEXT: vmulss %xmm1, %xmm0, %xmm0 24 ; NO-FMA-NEXT: vsubss %xmm0, %xmm2, %xmm0 25 ; NO-FMA-NEXT: vmulss %xmm0, %xmm1, %xmm0 26 ; NO-FMA-NEXT: vaddss %xmm0, %xmm1, %xmm0 [all …]
|
| /llvm-project-15.0.7/llvm/test/Transforms/SLPVectorizer/X86/ |
| H A D | fma.ll | 24 ; FMA 28 ; NO-FMA-LABEL: @fma_2f64( 39 ; NO-FMA-NEXT: ret void 41 ; FMA-LABEL: @fma_2f64( 47 ; FMA-NEXT: ret void 86 ; FMA-LABEL: @fma_4f64( 92 ; FMA-NEXT: ret void 249 ; FMA-LABEL: @fma_4f32( 255 ; FMA-NEXT: ret void 324 ; FMA-LABEL: @fma_8f32( [all …]
|
| /llvm-project-15.0.7/llvm/test/CodeGen/PowerPC/ |
| H A D | register-pressure-reduction.ll | 30 ; CHECK-FMA-LABEL: foo_float: 31 ; CHECK-FMA: # %bb.0: 33 ; CHECK-FMA-NEXT: xsmulsp f1, f2, f1 39 ; CHECK-FMA-NEXT: blr 66 ; CHECK-FMA-LABEL: foo_double: 67 ; CHECK-FMA: # %bb.0: 69 ; CHECK-FMA-NEXT: xsmuldp f1, f2, f1 75 ; CHECK-FMA-NEXT: blr 115 ; CHECK-FMA: # %bb.0: 126 ; CHECK-FMA-NEXT: stfs f0, 0(r3) [all …]
|
| H A D | ppc-passname.ll | 50 …before=ppc-vsx-fma-mutate -o /dev/null 2>&1 | FileCheck %s -check-prefix=STOP-BEFORE-VSX-FMA-MUTATE 51 ; STOP-BEFORE-VSX-FMA-MUTATE-NOT: -ppc-vsx-fma-mutate 52 ; STOP-BEFORE-VSX-FMA-MUTATE-NOT: "ppc-vsx-fma-mutate" pass is not registered. 53 ; STOP-BEFORE-VSX-FMA-MUTATE-NOT: PowerPC VSX FMA Mutation 55 …p-after=ppc-vsx-fma-mutate -o /dev/null 2>&1 | FileCheck %s -check-prefix=STOP-AFTER-VSX-FMA-MUTATE 56 ; STOP-AFTER-VSX-FMA-MUTATE: -ppc-vsx-fma-mutate 57 ; STOP-AFTER-VSX-FMA-MUTATE-NOT: "ppc-vsx-fma-mutate" pass is not registered. 58 ; STOP-AFTER-VSX-FMA-MUTATE: PowerPC VSX FMA Mutation
|
| /llvm-project-15.0.7/llvm/test/CodeGen/AMDGPU/ |
| H A D | fmad-formation-fmul-distribute-denormal-mode.ll | 22 ; FMA: ; %bb.0: 25 ; FMA-NEXT: s_setpc_b64 s[30:31] 67 ; FMA: ; %bb.0: 70 ; FMA-NEXT: s_setpc_b64 s[30:31] 112 ; FMA: ; %bb.0: 116 ; FMA-NEXT: s_setpc_b64 s[30:31] 163 ; FMA: ; %bb.0: 167 ; FMA-NEXT: s_setpc_b64 s[30:31] 215 ; FMA: ; %bb.0: 218 ; FMA-NEXT: s_setpc_b64 s[30:31] [all …]
|
| H A D | madak.ll | 5 …chineinstrs -fp-contract=fast < %s | FileCheck -check-prefixes=GCN,GFX10,GFX8_9_10,FMA,GFX10-FMA %s 6 …chineinstrs -fp-contract=fast < %s | FileCheck -check-prefixes=GCN,GFX9,GFX8_9_10,FMA,GFX940-FMA %s 22 ; FMA: v_fmaak_f32 {{v[0-9]+}}, [[VA]], [[VB]], 0x41200000 54 ; FMA-DAG: v_fmaak_f32 {{v[0-9]+}}, [[VA]], [[VB]], 0x41200000 56 ; GFX10-FMA-DAG:v_fmaak_f32 {{v[0-9]+}}, [[VA]], [[VC]], 0x41200000 57 ; GFX940-FMA-DAG:v_fmac_f32_e32 [[VK]], [[VA]], [[VC]] 87 ; FMA: v_fmaak_f32 {{v[0-9]+}}, 4.0, [[VA]], 0x41200000 115 ; FMA: v_fma_f32 {{v[0-9]+}}, [[VA]], [[VB]], 4.0 140 ; GFX940-FMA: v_fmac_f32_e32 v{{[0-9]+}}, [[SB]], [[VA]] 162 ; GFX940-FMA: v_fmac_f32_e32 v{{[0-9]+}}, [[SB]], [[VA]] [all …]
|
| /llvm-project-15.0.7/llvm/test/Transforms/InstCombine/ |
| H A D | fma.ll | 22 ; CHECK-NEXT: ret float [[FMA]] 33 ; CHECK-NEXT: ret float [[FMA]] 77 ; CHECK-NEXT: ret float [[FMA]] 88 ; CHECK-NEXT: ret float [[FMA]] 99 ; CHECK-NEXT: ret float [[FMA]] 110 ; CHECK-NEXT: ret float [[FMA]] 120 ; CHECK-NEXT: ret float [[FMA]] 131 ; CHECK-NEXT: ret float [[FMA]] 143 ; CHECK-NEXT: ret float [[FMA]] 154 ; CHECK-NEXT: ret float [[FMA]] [all …]
|
| /llvm-project-15.0.7/llvm/test/CodeGen/AArch64/GlobalISel/ |
| H A D | legalize-fma.mir | 25 ; NO-FP16: [[FMA:%[0-9]+]]:_(s32) = G_FMA [[FPEXT]], [[FPEXT1]], [[FPEXT2]] 26 ; NO-FP16: [[FPTRUNC:%[0-9]+]]:_(s16) = G_FPTRUNC [[FMA]](s32) 51 ; FP16: $d0 = COPY [[FMA]](<4 x s16>) 82 ; NO-FP16: [[FPTRUNC:%[0-9]+]]:_(s16) = G_FPTRUNC [[FMA]](s32) 127 ; FP16: $q0 = COPY [[FMA]](<8 x s16>) 152 ; NO-FP16: $d0 = COPY [[FMA]](<2 x s32>) 160 ; FP16: $d0 = COPY [[FMA]](<2 x s32>) 185 ; NO-FP16: $q0 = COPY [[FMA]](<4 x s32>) 193 ; FP16: $q0 = COPY [[FMA]](<4 x s32>) 218 ; NO-FP16: $q0 = COPY [[FMA]](<2 x s64>) [all …]
|
| H A D | regbank-fma.mir | 21 …; CHECK: [[FMA:%[0-9]+]]:fpr(s32) = nnan ninf nsz arcp contract afn reassoc G_FMA [[COPY]], [[COPY… 22 ; CHECK: $s0 = COPY [[FMA]](s32) 47 …; CHECK: [[FMA:%[0-9]+]]:fpr(s64) = nnan ninf nsz arcp contract afn reassoc G_FMA [[COPY]], [[COPY… 48 ; CHECK: $d0 = COPY [[FMA]](s64)
|
| /llvm-project-15.0.7/llvm/test/CodeGen/AMDGPU/GlobalISel/ |
| H A D | legalize-fma.mir | 21 ; SI-NEXT: $vgpr0 = COPY [[FMA]](s32) 29 ; VI-NEXT: $vgpr0 = COPY [[FMA]](s32) 37 ; GFX9-NEXT: $vgpr0 = COPY [[FMA]](s32) 57 ; SI-NEXT: $vgpr0_vgpr1 = COPY [[FMA]](s64) 65 ; VI-NEXT: $vgpr0_vgpr1 = COPY [[FMA]](s64) 73 ; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[FMA]](s64) 113 ; VI-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FMA]](s16) 418 ; VI-NEXT: [[ZEXT:%[0-9]+]]:_(s32) = G_ZEXT [[FMA]](s16) 431 ; GFX9-NEXT: $vgpr0 = COPY [[FMA]](<2 x s16>) 547 ; VI-NEXT: [[ZEXT:%[0-9]+]]:_(s32) = G_ZEXT [[FMA]](s16) [all …]
|
| H A D | combine-fma-unmerge-values.mir | 20 ; GFX10-NEXT: [[FMA:%[0-9]+]]:_(s32) = G_FMA [[COPY]], [[COPY1]], %el1 21 ; GFX10-NEXT: $vgpr0 = COPY [[FMA]](s32) 49 ; GFX10-NEXT: [[FMA:%[0-9]+]]:_(s32) = G_FMA [[COPY]], [[COPY1]], %el1 50 ; GFX10-NEXT: $vgpr0 = COPY [[FMA]](s32) 82 ; GFX10-NEXT: [[FMA:%[0-9]+]]:_(s32) = G_FMA [[FPEXT]], [[FPEXT1]], %el1 83 ; GFX10-NEXT: $vgpr0 = COPY [[FMA]](s32) 118 ; GFX10-NEXT: [[FMA:%[0-9]+]]:_(s32) = G_FMA [[FPEXT]], [[FPEXT1]], %el1 119 ; GFX10-NEXT: $vgpr0 = COPY [[FMA]](s32) 149 ; GFX10-NEXT: [[FMA:%[0-9]+]]:_(s32) = G_FMA [[COPY2]], [[COPY3]], %el1 398 ; GFX10-NEXT: $vgpr0 = COPY [[FMA]](s32) [all …]
|
| H A D | regbankselect-fma.mir | 21 ; CHECK-NEXT: [[FMA:%[0-9]+]]:vgpr(s32) = G_FMA [[COPY3]], [[COPY4]], [[COPY5]] 42 ; CHECK-NEXT: [[FMA:%[0-9]+]]:vgpr(s32) = G_FMA [[COPY]], [[COPY3]], [[COPY4]] 63 ; CHECK-NEXT: [[FMA:%[0-9]+]]:vgpr(s32) = G_FMA [[COPY3]], [[COPY1]], [[COPY4]] 84 ; CHECK-NEXT: [[FMA:%[0-9]+]]:vgpr(s32) = G_FMA [[COPY3]], [[COPY4]], [[COPY2]] 104 ; CHECK-NEXT: [[FMA:%[0-9]+]]:vgpr(s32) = G_FMA [[COPY]], [[COPY1]], [[COPY3]] 124 ; CHECK-NEXT: [[FMA:%[0-9]+]]:vgpr(s32) = G_FMA [[COPY]], [[COPY3]], [[COPY2]] 144 ; CHECK-NEXT: [[FMA:%[0-9]+]]:vgpr(s32) = G_FMA [[COPY3]], [[COPY1]], [[COPY2]] 163 ; CHECK-NEXT: [[FMA:%[0-9]+]]:vgpr(s32) = G_FMA [[COPY]], [[COPY1]], [[COPY2]]
|
| /llvm-project-15.0.7/llvm/test/CodeGen/NVPTX/ |
| H A D | fma-disable.ll | 1 ; RUN: llc < %s -march=nvptx -mcpu=sm_20 -nvptx-fma-level=1 | FileCheck %s -check-prefix=FMA 3 ; RUN: llc < %s -march=nvptx64 -mcpu=sm_20 -nvptx-fma-level=1 | FileCheck %s -check-prefix=FMA 12 ; FMA: fma.rn.f32 22 ; FMA: fma.rn.f64
|
| /llvm-project-15.0.7/libc/docs/ |
| H A D | math.rst | 202 …-10, 10]` | Ryzen 1700 | Ubuntu 20.04 LTS x86_64 | Clang 12.0.0 | FMA | 204 …-10, 10]` | Ryzen 1700 | Ubuntu 20.04 LTS x86_64 | Clang 12.0.0 | FMA | 206 …-10, 10]` | Ryzen 1700 | Ubuntu 20.04 LTS x86_64 | Clang 12.0.0 | FMA | 218 …e^{-1}, e]` | Ryzen 1700 | Ubuntu 20.04 LTS x86_64 | Clang 12.0.0 | FMA | 220 …e^{-1}, e]` | Ryzen 1700 | Ubuntu 20.04 LTS x86_64 | Clang 12.0.0 | FMA | 222 …e^{-0.5} - 1, e^{0.5} - 1]` | Ryzen 1700 | Ubuntu 20.04 LTS x86_64 | Clang 12.0.0 | FMA | 224 …e^{-1}, e]` | Ryzen 1700 | Ubuntu 20.04 LTS x86_64 | Clang 12.0.0 | FMA | 226 …-\pi, \pi]` | Ryzen 1700 | Ubuntu 20.04 LTS x86_64 | Clang 12.0.0 | FMA |
|
| /llvm-project-15.0.7/llvm/lib/Target/X86/ |
| H A D | X86InstrFMA.td | 1 //===-- X86InstrFMA.td - FMA Instruction Set ---------------*- tablegen -*-===// 9 // This file describes FMA (Fused Multiply-Add) instructions. 23 // FMA*213*: 27 // FMA*132*: 31 // FMA*231*: 171 // would require an opcode change to FMA*231: 172 // FMA*132* reg1, reg2, reg3; // reg1 * reg3 + reg2; 174 // FMA*231* reg2, reg1, reg3; // reg1 * reg3 + reg2; 259 // All of the FMA*_Int opcodes are defined as commutable here. 286 // The FMA 213 form is created for lowering of scalar FMA intrinsics [all …]
|