[all-commits] [llvm/llvm-project] db7dea: [AMDGPU] Vectorize alloca thru bitcast

[Stanislav Mekhanoshin via All-commits]

  Branch: refs/heads/master
  Home:   https://github.com/llvm/llvm-project
  Commit: db7dea2b6f7f71bcb85962bba875b63232af2d64
      https://github.com/llvm/llvm-project/commit/db7dea2b6f7f71bcb85962bba875b63232af2d64
  Author: Stanislav Mekhanoshin <Stanislav.Mekhanoshin at amd.com>
  Date:   2020-05-08 (Fri, 08 May 2020)

  Changed paths:
    M llvm/lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp
    A llvm/test/CodeGen/AMDGPU/vector-alloca-bitcast.ll

  Log Message:
  -----------
  [AMDGPU] Vectorize alloca thru bitcast

This is mostly useful if alloca element type is not integer
and then casted to an integer for load or store. We now can
vectorize an [i32] alloca but cannot do so for [float].

There also a separate patch needed to properly lower 64 bit
types after they vectorized. At the moment these are lowered
via scratch anyway.

Differential Revision: https://reviews.llvm.org/D79641