[all-commits] [llvm/llvm-project] c59129: [RISCV] Recursively split concat_vector into small...

[Luke Lau via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: c59129a7c79448837d665de8f2743ad4b14666f6
      https://github.com/llvm/llvm-project/commit/c59129a7c79448837d665de8f2743ad4b14666f6
  Author: Luke Lau <luke at igalia.com>
  Date:   2024-03-07 (Thu, 07 Mar 2024)

  Changed paths:
    M llvm/lib/Target/RISCV/RISCVISelLowering.cpp
    M llvm/test/CodeGen/RISCV/rvv/active_lane_mask.ll
    M llvm/test/CodeGen/RISCV/rvv/combine-store-extract-crash.ll
    M llvm/test/CodeGen/RISCV/rvv/extract-subvector.ll
    M llvm/test/CodeGen/RISCV/rvv/fixed-vectors-shuffle-concat.ll
    M llvm/test/CodeGen/RISCV/rvv/fixed-vectors-strided-load-combine.ll
    M llvm/test/CodeGen/RISCV/rvv/fpclamptosat_vec.ll
    M llvm/test/CodeGen/RISCV/rvv/mgather-sdnode.ll
    M llvm/test/CodeGen/RISCV/rvv/pr63596.ll

  Log Message:
  -----------
  [RISCV] Recursively split concat_vector into smaller LMULs (#83035)

This is the concat_vector equivalent of #81312, in that we recursively
split concat_vectors with more than two operands into smaller
concat_vectors.

This allows us to break up the chain of vslideups, as well as perform
the vslideups at a smaller LMUL, which in turn reduces register pressure
as the previous lowering performed N vslideups at the highest result
LMUL. For now, it stops splitting past MF2.

This is done as a DAG combine so that any undef operands are combined
away: If we do this during lowering then we end up with unnecessary
vslideups of undefs.



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