[all-commits] [llvm/llvm-project] 321a71: [RISCV] Optimize BUILD_VECTOR sequences that revea...

[Fraser Cormack via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 321a71a77268c314c769a98d62c14609aff306e0
      https://github.com/llvm/llvm-project/commit/321a71a77268c314c769a98d62c14609aff306e0
  Author: Fraser Cormack <fraser at codeplay.com>
  Date:   2021-03-25 (Thu, 25 Mar 2021)

  Changed paths:
    M llvm/lib/Target/RISCV/RISCVISelLowering.cpp
    M llvm/test/CodeGen/RISCV/rvv/fixed-vectors-bitreverse.ll
    M llvm/test/CodeGen/RISCV/rvv/fixed-vectors-ctpop.ll
    M llvm/test/CodeGen/RISCV/rvv/fixed-vectors-int-buildvec.ll
    M llvm/test/CodeGen/RISCV/rvv/fixed-vectors-int.ll
    M llvm/test/CodeGen/RISCV/rvv/fixed-vectors-masked-gather.ll
    M llvm/test/CodeGen/RISCV/rvv/fixed-vectors-masked-scatter.ll

  Log Message:
  -----------
  [RISCV] Optimize BUILD_VECTOR sequences that reveal hidden splats

This patch adds further optimization techniques to RVV BUILD_VECTOR
lowering. It teaches the compiler to find splats of larger vector
element types "hidden" in smaller ones. For example, a v4i8 build_vector
(0x1, 0x2, 0x1, 0x2) could be splat as v2i16 0x0201. This is generally
more optimal than the dominant-element BUILD_VECTORs and so takes
priority.

This optimization is currently limited to all-constant-or-undef
BUILD_VECTORs as those were found to be the most common. There's no
reason this couldn't be extended to other BUILD_VECTORs, but the
additional bit-manipulation instructions may require more sophisticated
heuristics.

There are some cases where the materialization of the larger constant
takes more scalar instructions than it does to build the vector with
vector instructions. We could add heuristics to try and catch this.

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D99195