[PATCH] Improve DAG combine pass on certain IR vector patterns

[Fiona Glaser]

> On Jan 16, 2015, at 2:23 PM, Chandler Carruth <chandlerc at google.com> wrote:
> 
> 
> On Fri, Jan 16, 2015 at 2:22 PM, Fiona Glaser <fglaser at apple.com <mailto:fglaser at apple.com>> wrote:
>> Please use the same tetsing strategy as the vector-shuffle-{128,256,512}-*.ll test files. This has made it *much* easier to understand the feature sets tested and how we generate code for each operation.
> 
> So, going by vector-shuffle-256, I should do something like:
> 
> ; RUN: llc < %s -mcpu=x86-64 -mattr=+avx | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX1
> ; RUN: llc < %s -mcpu=x86-64 -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX2
> 
> target triple = "x86_64-unknown-unknown”
> 
> that?
> 
> That's my preference, yes.

Is this okay? I omitted the prefix stuff because both AVX and AVX2 should generate identical output in this test.

Fiona


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20150116/10a99aab/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: patch.diff
Type: application/octet-stream
Size: 5680 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20150116/10a99aab/attachment.obj>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20150116/10a99aab/attachment-0001.html>