1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-23 03:02:36 +01:00
llvm-mirror/test/CodeGen/AMDGPU/GlobalISel/inst-select-fma.s32.mir
Joe Nash 521d6a1785 [AMDGPU] Add _e64 suffix to VOP3 Insts
Previously, instructions which could be
expressed as VOP3 in addition to another
encoding had a _e64 suffix on the tablegen
record name, while those
only available as VOP3 did not. With this
patch, all VOP3s will have the _e64 suffix.
The assembly does not change, only  the mir.

Reviewed By: foad

Differential Revision: https://reviews.llvm.org/D94341

Change-Id: Ia8ec8890d47f8f94bbbdac43745b4e9dd2b03423
2021-01-12 18:33:18 -05:00

227 lines
8.9 KiB
YAML

# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
# RUN: llc -march=amdgcn -mcpu=fiji -run-pass=instruction-select -verify-machineinstrs -o - %s | FileCheck -check-prefix=GFX6 %s
# RUN: llc -march=amdgcn -mcpu=gfx906 -run-pass=instruction-select -verify-machineinstrs -o - %s | FileCheck -check-prefix=GFX9-DL %s
# RUN: llc -march=amdgcn -mcpu=gfx1010 -run-pass=instruction-select -verify-machineinstrs -o - %s | FileCheck -check-prefix=GFX10 %s
---
name: fma_f32
legalized: true
regBankSelected: true
body: |
bb.0:
liveins: $vgpr0, $vgpr1, $vgpr2
; GFX6-LABEL: name: fma_f32
; GFX6: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX6: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX6: %3:vgpr_32 = nofpexcept V_FMA_F32_e64 0, [[COPY]], 0, [[COPY1]], 0, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX6: S_ENDPGM 0, implicit %3
; GFX9-DL-LABEL: name: fma_f32
; GFX9-DL: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-DL: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX9-DL: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX9-DL: %3:vgpr_32 = nofpexcept V_FMAC_F32_e64 0, [[COPY]], 0, [[COPY1]], 0, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX9-DL: S_ENDPGM 0, implicit %3
; GFX10-LABEL: name: fma_f32
; GFX10: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX10: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX10: %3:vgpr_32 = nofpexcept V_FMAC_F32_e64 0, [[COPY]], 0, [[COPY1]], 0, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX10: S_ENDPGM 0, implicit %3
%0:vgpr(s32) = COPY $vgpr0
%1:vgpr(s32) = COPY $vgpr1
%2:vgpr(s32) = COPY $vgpr2
%3:vgpr(s32) = G_FMA %0, %1, %2
S_ENDPGM 0, implicit %3
...
---
name: fma_f32_fneg_src0
legalized: true
regBankSelected: true
body: |
bb.0:
liveins: $vgpr0, $vgpr1, $vgpr2
; GFX6-LABEL: name: fma_f32_fneg_src0
; GFX6: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX6: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX6: %4:vgpr_32 = nofpexcept V_FMA_F32_e64 1, [[COPY]], 0, [[COPY1]], 0, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX6: S_ENDPGM 0, implicit %4
; GFX9-DL-LABEL: name: fma_f32_fneg_src0
; GFX9-DL: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-DL: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX9-DL: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX9-DL: %4:vgpr_32 = nofpexcept V_FMAC_F32_e64 1, [[COPY]], 0, [[COPY1]], 0, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX9-DL: S_ENDPGM 0, implicit %4
; GFX10-LABEL: name: fma_f32_fneg_src0
; GFX10: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX10: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX10: %4:vgpr_32 = nofpexcept V_FMAC_F32_e64 1, [[COPY]], 0, [[COPY1]], 0, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX10: S_ENDPGM 0, implicit %4
%0:vgpr(s32) = COPY $vgpr0
%1:vgpr(s32) = COPY $vgpr1
%2:vgpr(s32) = COPY $vgpr2
%3:vgpr(s32) = G_FNEG %0
%4:vgpr(s32) = G_FMA %3, %1, %2
S_ENDPGM 0, implicit %4
...
---
name: fma_f32_fneg_src1
legalized: true
regBankSelected: true
body: |
bb.0:
liveins: $vgpr0, $vgpr1, $vgpr2
; GFX6-LABEL: name: fma_f32_fneg_src1
; GFX6: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX6: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX6: %4:vgpr_32 = nofpexcept V_FMA_F32_e64 0, [[COPY]], 1, [[COPY1]], 0, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX6: S_ENDPGM 0, implicit %4
; GFX9-DL-LABEL: name: fma_f32_fneg_src1
; GFX9-DL: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-DL: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX9-DL: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX9-DL: %4:vgpr_32 = nofpexcept V_FMAC_F32_e64 0, [[COPY]], 1, [[COPY1]], 0, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX9-DL: S_ENDPGM 0, implicit %4
; GFX10-LABEL: name: fma_f32_fneg_src1
; GFX10: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX10: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX10: %4:vgpr_32 = nofpexcept V_FMAC_F32_e64 0, [[COPY]], 1, [[COPY1]], 0, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX10: S_ENDPGM 0, implicit %4
%0:vgpr(s32) = COPY $vgpr0
%1:vgpr(s32) = COPY $vgpr1
%2:vgpr(s32) = COPY $vgpr2
%3:vgpr(s32) = G_FNEG %1
%4:vgpr(s32) = G_FMA %0, %3, %2
S_ENDPGM 0, implicit %4
...
---
name: fma_f32_fneg_src2
legalized: true
regBankSelected: true
body: |
bb.0:
liveins: $vgpr0, $vgpr1, $vgpr2
; GFX6-LABEL: name: fma_f32_fneg_src2
; GFX6: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX6: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX6: %4:vgpr_32 = nofpexcept V_FMA_F32_e64 0, [[COPY]], 0, [[COPY1]], 1, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX6: S_ENDPGM 0, implicit %4
; GFX9-DL-LABEL: name: fma_f32_fneg_src2
; GFX9-DL: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-DL: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX9-DL: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX9-DL: %4:vgpr_32 = nofpexcept V_FMA_F32_e64 0, [[COPY]], 0, [[COPY1]], 1, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX9-DL: S_ENDPGM 0, implicit %4
; GFX10-LABEL: name: fma_f32_fneg_src2
; GFX10: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX10: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX10: %4:vgpr_32 = nofpexcept V_FMA_F32_e64 0, [[COPY]], 0, [[COPY1]], 1, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX10: S_ENDPGM 0, implicit %4
%0:vgpr(s32) = COPY $vgpr0
%1:vgpr(s32) = COPY $vgpr1
%2:vgpr(s32) = COPY $vgpr2
%3:vgpr(s32) = G_FNEG %2
%4:vgpr(s32) = G_FMA %0, %1, %3
S_ENDPGM 0, implicit %4
...
---
name: fma_f32_fabs_src2
legalized: true
regBankSelected: true
body: |
bb.0:
liveins: $vgpr0, $vgpr1, $vgpr2
; GFX6-LABEL: name: fma_f32_fabs_src2
; GFX6: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX6: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX6: %4:vgpr_32 = nofpexcept V_FMA_F32_e64 0, [[COPY]], 0, [[COPY1]], 2, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX6: S_ENDPGM 0, implicit %4
; GFX9-DL-LABEL: name: fma_f32_fabs_src2
; GFX9-DL: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-DL: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX9-DL: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX9-DL: %4:vgpr_32 = nofpexcept V_FMA_F32_e64 0, [[COPY]], 0, [[COPY1]], 2, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX9-DL: S_ENDPGM 0, implicit %4
; GFX10-LABEL: name: fma_f32_fabs_src2
; GFX10: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX10: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX10: %4:vgpr_32 = nofpexcept V_FMA_F32_e64 0, [[COPY]], 0, [[COPY1]], 2, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX10: S_ENDPGM 0, implicit %4
%0:vgpr(s32) = COPY $vgpr0
%1:vgpr(s32) = COPY $vgpr1
%2:vgpr(s32) = COPY $vgpr2
%3:vgpr(s32) = G_FABS %2
%4:vgpr(s32) = G_FMA %0, %1, %3
S_ENDPGM 0, implicit %4
...
---
name: fma_f32_copy_fneg_src2
legalized: true
regBankSelected: true
body: |
bb.0:
liveins: $vgpr0, $vgpr1, $vgpr2
; GFX6-LABEL: name: fma_f32_copy_fneg_src2
; GFX6: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX6: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX6: %5:vgpr_32 = nofpexcept V_FMA_F32_e64 0, [[COPY]], 0, [[COPY1]], 1, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX6: S_ENDPGM 0, implicit %5
; GFX9-DL-LABEL: name: fma_f32_copy_fneg_src2
; GFX9-DL: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-DL: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX9-DL: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX9-DL: %5:vgpr_32 = nofpexcept V_FMA_F32_e64 0, [[COPY]], 0, [[COPY1]], 1, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX9-DL: S_ENDPGM 0, implicit %5
; GFX10-LABEL: name: fma_f32_copy_fneg_src2
; GFX10: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX10: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX10: %5:vgpr_32 = nofpexcept V_FMA_F32_e64 0, [[COPY]], 0, [[COPY1]], 1, [[COPY2]], 0, 0, implicit $mode, implicit $exec
; GFX10: S_ENDPGM 0, implicit %5
%0:vgpr(s32) = COPY $vgpr0
%1:vgpr(s32) = COPY $vgpr1
%2:vgpr(s32) = COPY $vgpr2
%3:vgpr(s32) = G_FNEG %2
%4:vgpr(s32) = COPY %3
%5:vgpr(s32) = G_FMA %0, %1, %4
S_ENDPGM 0, implicit %5
...