[all-commits] [llvm/llvm-project] 0c866a: [LoopVec] Support non-instructions as argument to ...

[Philip Reames via All-commits]

  Branch: refs/heads/master
  Home:   https://github.com/llvm/llvm-project
  Commit: 0c866a3d6aa492b01c29a2c582c56c0fd75c2970
      https://github.com/llvm/llvm-project/commit/0c866a3d6aa492b01c29a2c582c56c0fd75c2970
  Author: Philip Reames <listmail at philipreames.com>
  Date:   2020-12-03 (Thu, 03 Dec 2020)

  Changed paths:
    M llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
    M llvm/test/Transforms/LoopVectorize/X86/cost-model-assert.ll
    M llvm/test/Transforms/LoopVectorize/X86/uniform_mem_op.ll
    M llvm/test/Transforms/LoopVectorize/pr44488-predication.ll

  Log Message:
  -----------
  [LoopVec] Support non-instructions as argument to uniform mem ops

The initial step of the uniform-after-vectorization (lane-0 demanded only) analysis was very awkwardly written. It would revisit use list of each pointer operand of a widened load/store. As a result, it was in the worst case O(N^2) where N was the number of instructions in a loop, and had restricted operand Value types to reduce the size of use lists.

This patch replaces the original algorithm with one which is at most O(2N) in the number of instructions in the loop. (The key observation is that each use of a potentially interesting pointer is visited at most twice, once on first scan, once in the use list of *it's* operand. Only instructions within the loop have their uses scanned.)

In the process, we remove a restriction which required the operand of the uniform mem op to itself be an instruction.  This allows detection of uniform mem ops involving global addresses.

Differential Revision: https://reviews.llvm.org/D92056