[llvm-dev] IR canonicalization: shufflevector or vector trunc?

[Sanjay Patel via llvm-dev]

On Thu, Jan 26, 2017 at 10:22 AM, Sanjay Patel <spatel at rotateright.com>
wrote:

>
> I think this is irrelevant. If there was a DAG canonicalization providing
> a backstop, then we might just say, "Who cares?", but there isn't:
>
>
> $ ./opt -instcombine shtr.ll -S | ./llc -o - -mtriple=powerpc64
> A:                                      # @A
>     mr 3, 5
>     blr
> B:                                      # @B
>     rotldi     3, 5, 32
>     rlwimi 3, 6, 0, 0, 31
>     blr
>

Disregard that example - I forgot to put the layout with big 'E' in the IR
file. But that doesn't change that @A is canonical. :)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20170126/b17c84fd/attachment.html>