[PATCH] D53712: [DAGCombiner] Remove reduceBuildVecConvertToConvertBuildVec and rely on the vectorizers instead (PR35732)

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon created this revision.
RKSimon added reviewers: hfinkel, spatel, efriedma, ABataev, hliao, niravd, craig.topper, atanasyan.
Herald added subscribers: jrtc27, javed.absar, sdardis.

reduceBuildVecConvertToConvertBuildVec vectorizes int2float in the DAGCombiner, which means that even if the LV/SLP has decided to keep scalar code using the cost models, this will override this.

While there are cases where vectorization is necessary in the DAG (mainly due to legalization artefacts), I don't think this is the case here. I think we should get rid of this and instead assume that the vectorizers know what they are doing.


Repository:
  rL LLVM

https://reviews.llvm.org/D53712

Files:
  lib/CodeGen/SelectionDAG/DAGCombiner.cpp
  test/CodeGen/ARM/vdup.ll
  test/CodeGen/Mips/cconv/vector.ll
  test/CodeGen/X86/2009-02-26-MachineLICMBug.ll
  test/CodeGen/X86/cvtv2f32.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D53712.171122.patch
Type: text/x-patch
Size: 17402 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20181025/fec1b187/attachment.bin>