[all-commits] [llvm/llvm-project] 456ec1: [LoopInterchange] Remove 'S' Scalar Dependencies (...

[Sjoerd Meijer via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 456ec1c2f4e487de235c953e8f2832b97372e7b0
      https://github.com/llvm/llvm-project/commit/456ec1c2f4e487de235c953e8f2832b97372e7b0
  Author: Sjoerd Meijer <smeijer at nvidia.com>
  Date:   2025-01-20 (Mon, 20 Jan 2025)

  Changed paths:
    M llvm/lib/Transforms/Scalar/LoopInterchange.cpp
    M llvm/test/Transforms/LoopInterchange/gh54176-scalar-deps.ll
    M llvm/test/Transforms/LoopInterchange/inner-only-reductions.ll
    M llvm/test/Transforms/LoopInterchange/innermost-latch-uses-values-in-middle-header.ll
    M llvm/test/Transforms/LoopInterchange/interchange-flow-dep-outer.ll
    M llvm/test/Transforms/LoopInterchange/lcssa.ll
    M llvm/test/Transforms/LoopInterchange/outer-header-jump-to-inner-latch.ll
    M llvm/test/Transforms/LoopInterchange/pr43176-move-to-new-latch.ll
    M llvm/test/Transforms/LoopInterchange/pr43797-lcssa-for-multiple-outer-loop-blocks.ll
    M llvm/test/Transforms/LoopInterchange/profitability.ll
    M llvm/test/Transforms/LoopInterchange/reductions-across-inner-and-outer-loop.ll
    M llvm/test/Transforms/LoopInterchange/unique-dep-matrix.ll
    M llvm/test/Transforms/LoopInterchange/vector-gep-operand.ll

  Log Message:
  -----------
  [LoopInterchange] Remove 'S' Scalar Dependencies (#119345)

We are not handling 'S' scalar dependencies correctly and have at least
the following miscompiles related to that:

[LoopInterchange] incorrect handling of scalar dependencies and dependence vectors starting with ">" #54176
[LoopInterchange] Interchange breaks program correctness #46867
[LoopInterchange] Loops should not interchanged due to dependencies #47259
[LoopInterchange] Loops should not interchanged due to control flow #47401

This patch does no longer insert the "S" dependency/direction into the
dependency matrix, so a dependency is never "S". We seem to have
forgotten what the exact meaning is of this dependency type, and don't
see why it should be treated differently.

We prefer correctness over incorrect and more aggressive results. I.e.,
this prevents the miscompiles at the expense of handling less cases,
i.e. making interchange more pessimistic. However, some of the cases
that are now rejected for dependence analysis reasons, were rejected
before too but for other reasons (e.g. profitability). So at least for
the llvm regression tests, the number of regression are very reasonable.
This should be a stopgap. We would like to get interchange enabled by
default and thus prefer correctness over unsafe transforms, and later
see if we can get solve the regressions.



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