[all-commits] [llvm/llvm-project] 1e4544: DAG: Avoid forming shufflevector from a single ext...

[Matt Arsenault via All-commits]

  Branch: refs/heads/users/arsenm/dag-do-not-form-shuffle-for-single-extract-element
  Home:   https://github.com/llvm/llvm-project
  Commit: 1e4544e92f11844da838b95e3e0377de3867f41e
      https://github.com/llvm/llvm-project/commit/1e4544e92f11844da838b95e3e0377de3867f41e
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2025-01-16 (Thu, 16 Jan 2025)

  Changed paths:
    M llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
    M llvm/test/CodeGen/AMDGPU/insert_vector_dynelt.ll
    M llvm/test/CodeGen/X86/avx512-build-vector.ll
    M llvm/test/CodeGen/X86/insertelement-duplicates.ll
    M llvm/test/CodeGen/X86/sse-align-12.ll

  Log Message:
  -----------
  DAG: Avoid forming shufflevector from a single extract_vector_elt

This avoids regressions in a future AMDGPU commit. Previously we
would have a build_vector (extract_vector_elt x), undef with free
access to the elements bloated into a shuffle of one element + undef,
which has much worse combine support than the extract.

Alternatively could check aggressivelyPreferBuildVectorSources, but
I'm not sure it's really different than isExtractVecEltCheap.


  Commit: 3b7fe9f08fab5a7d0d8307e4f9b0f91f9d96d7cc
      https://github.com/llvm/llvm-project/commit/3b7fe9f08fab5a7d0d8307e4f9b0f91f9d96d7cc
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2025-01-16 (Thu, 16 Jan 2025)

  Changed paths:
    M llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
    M llvm/test/CodeGen/AMDGPU/insert_vector_dynelt.ll
    M llvm/test/CodeGen/X86/avx512-build-vector.ll
    M llvm/test/CodeGen/X86/sse-align-12.ll

  Log Message:
  -----------
  Check hasOneUse on the source vector. This gives up some of the x86 improvements


  Commit: 90e1346059d268899246c3b754efd8e3dfcdf441
      https://github.com/llvm/llvm-project/commit/90e1346059d268899246c3b754efd8e3dfcdf441
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2025-01-16 (Thu, 16 Jan 2025)

  Changed paths:
    M llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
    M llvm/test/CodeGen/AMDGPU/insert_vector_dynelt.ll
    M llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast.ll
    M llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast_from_memory.ll
    M llvm/test/CodeGen/X86/buildvec-extract.ll
    M llvm/test/CodeGen/X86/insertelement-duplicates.ll
    M llvm/test/CodeGen/X86/movmsk-bittest.ll
    M llvm/test/CodeGen/X86/split-extend-vector-inreg.ll
    M llvm/test/CodeGen/X86/sse41.ll
    M llvm/test/CodeGen/X86/vec_extract-avx.ll
    M llvm/test/CodeGen/X86/vector-interleaved-store-i32-stride-5.ll
    M llvm/test/CodeGen/X86/vector-narrow-binop.ll
    M llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast.ll
    M llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast_from_memory.ll

  Log Message:
  -----------
  Count the number of extract uses


  Commit: d1db74b158e93904e83d3ee4f790018e7a45e8b7
      https://github.com/llvm/llvm-project/commit/d1db74b158e93904e83d3ee4f790018e7a45e8b7
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2025-01-16 (Thu, 16 Jan 2025)

  Changed paths:
    M llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
    M llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast.ll
    M llvm/test/CodeGen/X86/any_extend_vector_inreg_of_broadcast_from_memory.ll
    M llvm/test/CodeGen/X86/buildvec-extract.ll
    M llvm/test/CodeGen/X86/movmsk-bittest.ll
    M llvm/test/CodeGen/X86/split-extend-vector-inreg.ll
    M llvm/test/CodeGen/X86/sse41.ll
    M llvm/test/CodeGen/X86/vec_extract-avx.ll
    M llvm/test/CodeGen/X86/vector-interleaved-store-i32-stride-5.ll
    M llvm/test/CodeGen/X86/vector-narrow-binop.ll
    M llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast.ll
    M llvm/test/CodeGen/X86/zero_extend_vector_inreg_of_broadcast_from_memory.ll

  Log Message:
  -----------
  Avoid arm regressions


  Commit: 3a4b2226a1bf2d9a7d187843f3f1d03b1b3be29e
      https://github.com/llvm/llvm-project/commit/3a4b2226a1bf2d9a7d187843f3f1d03b1b3be29e
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2025-01-16 (Thu, 16 Jan 2025)

  Changed paths:
    M llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp

  Log Message:
  -----------
  Cleanup and comment


Compare: https://github.com/llvm/llvm-project/compare/bb1a07439dfd...3a4b2226a1bf

To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications