[all-commits] [llvm/llvm-project] 642daa: [AMDGPU] Fold compares of S_CSELECT of two differe...

Jay Foad via All-commits all-commits at lists.llvm.org
Wed Sep 30 14:04:42 PDT 2026 

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 642daaf27dd0cda7127d35bad1a3fe705a267918
      https://github.com/llvm/llvm-project/commit/642daaf27dd0cda7127d35bad1a3fe705a267918
  Author: Jay Foad <jay.foad at amd.com>
  Date:   2026-09-30 (Wed, 30 Sep 2026)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
    M llvm/test/CodeGen/AMDGPU/a-v-flat-atomicrmw.ll
    M llvm/test/CodeGen/AMDGPU/add.ll
    M llvm/test/CodeGen/AMDGPU/agpr-copy-no-free-registers.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.160bit.ll
    M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.16bit.ll
    M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.192bit.ll
    M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.224bit.ll
    M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.256bit.ll
    M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.288bit.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.352bit.ll
    M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.384bit.ll
    M llvm/test/CodeGen/AMDGPU/amdgcn.bitcast.448bit.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.576bit.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/amdgpu-miscellaneous-uniform-intrinsic.ll
    M llvm/test/CodeGen/AMDGPU/atomic_optimizations_pixelshader.ll
    M llvm/test/CodeGen/AMDGPU/av-split-dead-valno-crash.ll
    M llvm/test/CodeGen/AMDGPU/blender-no-live-segment-at-def-implicit-def.ll
    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/branch-relaxation.ll
    M llvm/test/CodeGen/AMDGPU/carryout-selection.ll
    M llvm/test/CodeGen/AMDGPU/coalescer_distribute.ll
    M llvm/test/CodeGen/AMDGPU/coexec-scheduler.ll
    M llvm/test/CodeGen/AMDGPU/combine-add-zext-xor.ll
    M llvm/test/CodeGen/AMDGPU/ctpop.ll
    M llvm/test/CodeGen/AMDGPU/ctpop16.ll
    M llvm/test/CodeGen/AMDGPU/ctpop64.ll
    M llvm/test/CodeGen/AMDGPU/dagcombine-select.ll
    M llvm/test/CodeGen/AMDGPU/dynamic_stackalloc.ll
    M llvm/test/CodeGen/AMDGPU/exec-mask-opt-cannot-create-empty-or-backward-segment.ll
    M llvm/test/CodeGen/AMDGPU/expand-scalar-carry-out-select-user.ll
    M llvm/test/CodeGen/AMDGPU/extract-subvector-16bit.ll
    M llvm/test/CodeGen/AMDGPU/fix-sgpr-copies-phi-regression-issue130646-issue130119.ll
    M llvm/test/CodeGen/AMDGPU/flat-saddr-atomics.ll
    M llvm/test/CodeGen/AMDGPU/flat_atomics_i64.ll
    M llvm/test/CodeGen/AMDGPU/flat_atomics_i64_system.ll
    M llvm/test/CodeGen/AMDGPU/frem.ll
    M llvm/test/CodeGen/AMDGPU/gep-const-address-space.ll
    M llvm/test/CodeGen/AMDGPU/indirect-addressing-si.ll
    M llvm/test/CodeGen/AMDGPU/infer-addrspace-flat-atomic.ll
    M llvm/test/CodeGen/AMDGPU/insert-delay-alu-bug.ll
    M llvm/test/CodeGen/AMDGPU/insert-waitcnts-merge.ll
    M llvm/test/CodeGen/AMDGPU/insert_vector_elt.ll
    M llvm/test/CodeGen/AMDGPU/issue176578.ll
    M llvm/test/CodeGen/AMDGPU/large-phi-search.ll
    M llvm/test/CodeGen/AMDGPU/llvm.amdgcn.is.private.ll
    M llvm/test/CodeGen/AMDGPU/llvm.amdgcn.is.shared.ll
    M llvm/test/CodeGen/AMDGPU/llvm.amdgcn.kill.ll
    M llvm/test/CodeGen/AMDGPU/llvm.amdgcn.set.inactive.ll
    M llvm/test/CodeGen/AMDGPU/llvm.amdgcn.tensor.load.store.ll
    M llvm/test/CodeGen/AMDGPU/long-branch-reserve-register.ll
    M llvm/test/CodeGen/AMDGPU/loop-header-align-gfx950.ll
    M llvm/test/CodeGen/AMDGPU/mad-combine.ll
    M llvm/test/CodeGen/AMDGPU/memcpy-crash-issue63986.ll
    M llvm/test/CodeGen/AMDGPU/mul.ll
    M llvm/test/CodeGen/AMDGPU/no-fold-accvgpr-mov.ll
    M llvm/test/CodeGen/AMDGPU/noclobber-barrier.ll
    A llvm/test/CodeGen/AMDGPU/optimize-compare-cselect.mir
    M llvm/test/CodeGen/AMDGPU/optimize-negated-cond.ll
    M llvm/test/CodeGen/AMDGPU/rewrite-vgpr-mfma-to-agpr-phi.ll
    M llvm/test/CodeGen/AMDGPU/rewrite-vgpr-mfma-to-agpr-reserved-wwm-vgpr.ll
    M llvm/test/CodeGen/AMDGPU/s-cluster-barrier.ll
    M llvm/test/CodeGen/AMDGPU/sdiv64.ll
    M llvm/test/CodeGen/AMDGPU/select-undef.ll
    M llvm/test/CodeGen/AMDGPU/setcc64.ll
    M llvm/test/CodeGen/AMDGPU/sgpr-control-flow.ll
    M llvm/test/CodeGen/AMDGPU/sgpr-copy.ll
    M llvm/test/CodeGen/AMDGPU/si-unify-exit-return-unreachable.ll
    M llvm/test/CodeGen/AMDGPU/simplifydemandedbits-recursion.ll
    M llvm/test/CodeGen/AMDGPU/sink-addr-memory-intrinsics.ll
    M llvm/test/CodeGen/AMDGPU/srem.ll
    M llvm/test/CodeGen/AMDGPU/srem64.ll
    M llvm/test/CodeGen/AMDGPU/structurize-hoist.ll
    M llvm/test/CodeGen/AMDGPU/tuple-allocation-failure.ll
    M llvm/test/CodeGen/AMDGPU/udiv64.ll
    M llvm/test/CodeGen/AMDGPU/undefined-subreg-liverange.ll
    M llvm/test/CodeGen/AMDGPU/unstructured-cfg-def-use-issue.ll
    M llvm/test/CodeGen/AMDGPU/urem64.ll
    M llvm/test/CodeGen/AMDGPU/wave32.ll
    M llvm/test/CodeGen/AMDGPU/wqm.ll
    M llvm/test/CodeGen/AMDGPU/xor.ll

  Log Message:
  -----------
  [AMDGPU] Fold compares of S_CSELECT of two different constants (#227772)

If sX = S_CSELECT* A, B with A != B, then sX == A exactly when SCC was
set, so comparing sX with A or B recomputes SCC or its inverse:

  s_cmp_eq_* sX, A => SCC      s_cmp_lg_* sX, A => !SCC
  s_cmp_eq_* sX, B => !SCC     s_cmp_lg_* sX, B => SCC

Generalize optimizeCmpSelect to handle this for any pair of distinct
constants, instead of only S_CSELECT* (non-zero imm), 0 compared with 0.
Also try optimizeCmpSelect for S_CMP_EQ_U64.

This removes thousands of "s_cselect_b32 sN, 1, 0; s_cmp_lg_u32 sN, 1"
sequences from the AMDGPU codegen tests. They come from branches on an
inverted i1: the setcc against true is promoted to a compare with 1.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply at anthropic.com>

Co-authored-by: Claude Opus 5.5 (1M context) <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