[all-commits] [llvm/llvm-project] 3a6314: AMDGPU: Use LiveIntervals instead of LiveVariables...
Matt Arsenault via All-commits
all-commits at lists.llvm.org
Tue Sep 29 03:08:50 PDT 2026
Branch: refs/heads/users/arsenm/amdgpu/optimize-vgpr-live-range-pass-order
Home: https://github.com/llvm/llvm-project
Commit: 3a6314dcfd1fee6788a6e2f6466a7f69b5bcfb4f
https://github.com/llvm/llvm-project/commit/3a6314dcfd1fee6788a6e2f6466a7f69b5bcfb4f
Author: Matt Arsenault <Matthew.Arsenault at amd.com>
Date: 2026-09-29 (Tue, 29 Sep 2026)
Changed paths:
M llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
M llvm/lib/Target/AMDGPU/SIOptimizeVGPRLiveRange.cpp
M llvm/test/CodeGen/AMDGPU/GlobalISel/divergence-divergent-i1-used-outside-loop.ll
M llvm/test/CodeGen/AMDGPU/GlobalISel/divergence-structurizer.ll
M llvm/test/CodeGen/AMDGPU/GlobalISel/vni8-across-blocks.ll
M llvm/test/CodeGen/AMDGPU/a-v-flat-atomic-cmpxchg.ll
M llvm/test/CodeGen/AMDGPU/a-v-flat-atomicrmw.ll
M llvm/test/CodeGen/AMDGPU/absdiff.ll
M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.1024bit.ll
M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.128bit.ll
M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.256bit.ll
M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.320bit.ll
M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.32bit.ll
M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.48bit.ll
M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.512bit.ll
M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.640bit.ll
M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.64bit.ll
M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.704bit.ll
M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.768bit.ll
M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.832bit.ll
M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.896bit.ll
M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.960bit.ll
M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.96bit.ll
M llvm/test/CodeGen/AMDGPU/asyncmark-gfx12plus.ll
M llvm/test/CodeGen/AMDGPU/asyncmark-pregfx12.ll
M llvm/test/CodeGen/AMDGPU/atomicrmw-nand.ll
M llvm/test/CodeGen/AMDGPU/block-should-not-be-in-alive-blocks.mir
M llvm/test/CodeGen/AMDGPU/branch-folding-implicit-def-subreg.ll
M llvm/test/CodeGen/AMDGPU/branch-relaxation-gfx1250.ll
M llvm/test/CodeGen/AMDGPU/bypass-div.ll
M llvm/test/CodeGen/AMDGPU/div_i128.ll
M llvm/test/CodeGen/AMDGPU/div_v2i128.ll
M llvm/test/CodeGen/AMDGPU/flat_atomics_i64_noprivate.ll
M llvm/test/CodeGen/AMDGPU/flat_atomics_i64_system.ll
M llvm/test/CodeGen/AMDGPU/fptoi.i128.ll
M llvm/test/CodeGen/AMDGPU/fract-match.ll
M llvm/test/CodeGen/AMDGPU/frem.ll
M llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmax.ll
M llvm/test/CodeGen/AMDGPU/global_atomics_scan_fmin.ll
M llvm/test/CodeGen/AMDGPU/llc-pipeline-npm.ll
M llvm/test/CodeGen/AMDGPU/llc-pipeline.ll
M llvm/test/CodeGen/AMDGPU/llvm.amdgcn.init.whole.wave-w32.ll
M llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.fadd.ll
M llvm/test/CodeGen/AMDGPU/llvm.amdgcn.reduce.fsub.ll
M llvm/test/CodeGen/AMDGPU/local-atomicrmw-fadd.ll
M llvm/test/CodeGen/AMDGPU/local-atomicrmw-fmax.ll
M llvm/test/CodeGen/AMDGPU/local-atomicrmw-fmin.ll
M llvm/test/CodeGen/AMDGPU/local-atomicrmw-fsub.ll
M llvm/test/CodeGen/AMDGPU/lower-control-flow-live-variables-update.mir
R llvm/test/CodeGen/AMDGPU/lower-control-flow-live-variables-update.xfail.mir
M llvm/test/CodeGen/AMDGPU/mad_uint24.ll
M llvm/test/CodeGen/AMDGPU/rewrite-vgpr-mfma-to-agpr-spill-multi-store-codegen.ll
M llvm/test/CodeGen/AMDGPU/si-opt-vgpr-liverange-bug-deadlanes.mir
R llvm/test/CodeGen/AMDGPU/si-opt-vgpr-liverange-undef-use-livevars.mir
A llvm/test/CodeGen/AMDGPU/si-opt-vgpr-liverange-undef-use.mir
M llvm/test/CodeGen/AMDGPU/si-optimize-vgpr-live-range-dbg-instr.mir
M llvm/test/CodeGen/AMDGPU/splitkit-getsubrangeformask-phi-extend.ll
M llvm/test/CodeGen/AMDGPU/srem64.ll
M llvm/test/CodeGen/AMDGPU/urem64.ll
M llvm/test/CodeGen/AMDGPU/vgpr-liverange-ir.ll
M llvm/test/CodeGen/AMDGPU/vni8-across-blocks.ll
Log Message:
-----------
AMDGPU: Use LiveIntervals instead of LiveVariables in SIOptimizeVGPRLiveRange
Drop the LiveVariables dependency and the hand-written VarInfo
maintenance; the pass already knew how to recompute the affected
intervals with LiveIntervals, so make that the only path.
The legacy pass manager cannot schedule a pass requiring both
LiveIntervals and LiveVariables here, since LiveVariables (and its
UnreachableMachineBlockElim dependency) invalidates the LiveIntervals
just computed for it. In the AMDGPU pipeline, anchor the pass after
MachineLoopInfo instead of PHIElimination so LiveIntervals is computed
before PHIElimination, which is required anyway: the pass needs SSA and
introduces new PHIs. Preserving SlotIndexes keeps the transitive
last-user chain intact.
Since LiveVariables is no longer maintained past this point, PHIElimination
now splits critical edges using LiveIntervals, which accounts for most of
the test churn: LiveIntervalCalc drops kill flags and adds dead flags.
Co-Authored-By: Claude Opus 5 <noreply at anthropic.com>
To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications
More information about the All-commits
mailing list