mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2025-02-01 13:11:39 +01:00
3f23d4b8c3
tryLatency compares two sched candidates. For the top zone it prefers the one with lesser depth, but only if that depth is greater than the total latency of the instructions we've already scheduled -- otherwise its latency would be hidden and there would be no stall. Unfortunately it only tests the depth of one of the candidates. This can lead to situations where the TopDepthReduce heuristic does not kick in, but a lower priority heuristic chooses the other candidate, whose depth *is* greater than the already scheduled latency, which causes a stall. The fix is to apply the heuristic if the depth of *either* candidate is greater than the already scheduled latency. All this also applies to the BotHeightReduce heuristic in the bottom zone. Differential Revision: https://reviews.llvm.org/D72392
70 lines
3.6 KiB
YAML
70 lines
3.6 KiB
YAML
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
|
|
# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx906 -verify-machineinstrs -run-pass=machine-scheduler -verify-misched -o - %s | FileCheck %s
|
|
|
|
# This would assert that a dead def should have no uses, but the dead
|
|
# def and use have different subreg indices.
|
|
|
|
---
|
|
name: multi_def_dead_reg_subreg_check
|
|
tracksRegLiveness: true
|
|
machineFunctionInfo:
|
|
isEntryFunction: true
|
|
scratchRSrcReg: '$sgpr24_sgpr25_sgpr26_sgpr27'
|
|
frameOffsetReg: '$sgpr32'
|
|
stackPtrOffsetReg: '$sgpr32'
|
|
argumentInfo:
|
|
privateSegmentBuffer: { reg: '$sgpr0_sgpr1_sgpr2_sgpr3' }
|
|
privateSegmentWaveByteOffset: { reg: '$sgpr33' }
|
|
body: |
|
|
; CHECK-LABEL: name: multi_def_dead_reg_subreg_check
|
|
; CHECK: bb.0:
|
|
; CHECK: successors: %bb.1(0x80000000)
|
|
; CHECK: liveins: $sgpr6_sgpr7
|
|
; CHECK: undef %0.sub3:vreg_512 = V_MOV_B32_e32 0, implicit $exec
|
|
; CHECK: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
|
|
; CHECK: [[V_ADD_U32_e32_:%[0-9]+]]:vgpr_32 = V_ADD_U32_e32 0, [[V_MOV_B32_e32_]], implicit $exec
|
|
; CHECK: [[V_MOV_B32_e32_1:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
|
|
; CHECK: [[COPY:%[0-9]+]]:vreg_512 = COPY %0
|
|
; CHECK: bb.1:
|
|
; CHECK: successors: %bb.1(0x80000000)
|
|
; CHECK: BUFFER_STORE_DWORD_OFFEN %0.sub3, undef %5:vgpr_32, $sgpr24_sgpr25_sgpr26_sgpr27, $sgpr32, 0, 0, 0, 0, 0, 0, implicit $exec :: (store 4, align 8, addrspace 5)
|
|
; CHECK: dead %6:vgpr_32 = DS_READ_B32_gfx9 undef %7:vgpr_32, 0, 0, implicit $exec
|
|
; CHECK: dead %8:vreg_64 = DS_READ_B64_gfx9 [[V_MOV_B32_e32_]], 0, 0, implicit $exec
|
|
; CHECK: dead %9:vreg_128 = DS_READ_B128_gfx9 [[V_ADD_U32_e32_]], 0, 0, implicit $exec
|
|
; CHECK: [[COPY1:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub0
|
|
; CHECK: undef %11.sub1:vreg_512 = COPY [[COPY]].sub1
|
|
; CHECK: INLINEASM &"", 1 /* sideeffect attdialect */, 851978 /* regdef:VGPR_LO16 */, def dead [[COPY1]], 851978 /* regdef:VGPR_LO16 */, def dead [[COPY]].sub1, 2147483657 /* reguse tiedto:$0 */, [[COPY1]], 2147549193 /* reguse tiedto:$1 */, [[COPY]].sub1
|
|
; CHECK: %11.sub0:vreg_512 = COPY [[COPY]].sub0
|
|
; CHECK: %11.sub3:vreg_512 = COPY [[COPY]].sub3
|
|
; CHECK: %11.sub2:vreg_512 = COPY undef [[V_MOV_B32_e32_]]
|
|
; CHECK: %11.sub5:vreg_512 = COPY undef [[V_MOV_B32_e32_]]
|
|
; CHECK: [[COPY2:%[0-9]+]]:vreg_512 = COPY %11
|
|
; CHECK: dead %10:vgpr_32 = V_ADD_CO_U32_e32 4, [[V_MOV_B32_e32_1]], implicit-def dead $vcc, implicit $exec
|
|
; CHECK: S_BRANCH %bb.1
|
|
bb.0:
|
|
liveins: $sgpr6_sgpr7
|
|
|
|
undef %0.sub3:vreg_512 = V_MOV_B32_e32 0, implicit $exec
|
|
%1:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
|
|
%2:vgpr_32 = V_ADD_U32_e32 0, %1, implicit $exec
|
|
%3:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
|
|
%4:vreg_512 = COPY %0
|
|
|
|
bb.1:
|
|
BUFFER_STORE_DWORD_OFFEN %0.sub3, undef %5:vgpr_32, $sgpr24_sgpr25_sgpr26_sgpr27, $sgpr32, 0, 0, 0, 0, 0, 0, implicit $exec :: (store 4, align 8, addrspace 5)
|
|
%6:vgpr_32 = DS_READ_B32_gfx9 undef %7:vgpr_32, 0, 0, implicit $exec
|
|
%8:vreg_64 = DS_READ_B64_gfx9 %1, 0, 0, implicit $exec
|
|
%9:vreg_128 = DS_READ_B128_gfx9 %2, 0, 0, implicit $exec
|
|
%10:vgpr_32 = V_ADD_CO_U32_e32 4, %3, implicit-def dead $vcc, implicit $exec
|
|
undef %11.sub0:vreg_512 = COPY %4.sub0
|
|
%12:vgpr_32 = COPY %4.sub0
|
|
%11.sub1:vreg_512 = COPY %4.sub1
|
|
INLINEASM &"", 1, 851978, def dead %12, 851978, def dead %4.sub1, 2147483657, %12, 2147549193, %4.sub1
|
|
%11.sub2:vreg_512 = COPY undef %1
|
|
%11.sub3:vreg_512 = COPY %4.sub3
|
|
%11.sub5:vreg_512 = COPY undef %1
|
|
%4:vreg_512 = COPY %11
|
|
S_BRANCH %bb.1
|
|
|
|
...
|