[all-commits] [llvm/llvm-project] a24b02: [DSE] Remove stores in the same loop iteration

[David Green via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: a24b02193a305628066486a5eeee331054e06b5b
      https://github.com/llvm/llvm-project/commit/a24b02193a305628066486a5eeee331054e06b5b
  Author: David Green <david.green at arm.com>
  Date:   2021-06-20 (Sun, 20 Jun 2021)

  Changed paths:
    M llvm/lib/Transforms/Scalar/DeadStoreElimination.cpp
    M llvm/test/CodeGen/AMDGPU/opt-pipeline.ll
    M llvm/test/Other/opt-O2-pipeline.ll
    M llvm/test/Other/opt-O3-pipeline-enable-matrix.ll
    M llvm/test/Other/opt-O3-pipeline.ll
    M llvm/test/Other/opt-Os-pipeline.ll
    M llvm/test/Transforms/DeadStoreElimination/multiblock-loops.ll

  Log Message:
  -----------
  [DSE] Remove stores in the same loop iteration

DSE will currently only remove stores in the same block unless they can
be guaranteed to be loop invariant. This expands that to any stores that
are in the same Loop, at the same loop level. This should still account
for where AA/MSSA will not handle aliasing between loops, but allow the
dead stores to be removed where they overlap in the same loop iteration.
It requires adding loop info to DSE, but that looks fairly harmless.

The test case this helps is from code like this, which can come up in
certain matrix operations:
  for(i=..)
    dst[i] = 0;
    for(j=..)
      dst[i] += src[i*n+j];

After LICM, this becomes:
for(i=..)
  dst[i] = 0;
  sum = 0;
  for(j=..)
    sum += src[i*n+j];
  dst[i] = sum;

The first store is dead, and with this patch is now removed.

Differntial Revision: https://reviews.llvm.org/D100464