This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Fix up AVX512F 128/256-bit shifts from using EVEX counts (PR target/84786)
- From: Kirill Yukhin <kirill dot yukhin at gmail dot com>
- To: Jakub Jelinek <jakub at redhat dot com>
- Cc: Uros Bizjak <ubizjak at gmail dot com>, gcc-patches at gcc dot gnu dot org
- Date: Mon, 25 Jun 2018 14:27:54 +0300
- Subject: Re: [PATCH] Fix up AVX512F 128/256-bit shifts from using EVEX counts (PR target/84786)
- References: <20180622214718.GY7166@tucnak>
Hello Jakub,
On 22 июн 23:47, Jakub Jelinek wrote:
> Hi!
>
> The following testcase got fixed in 8/trunk with r253924 part of
> PR82370 enhancements, but that is not IMHO something we should backport.
> So instead the following patch adds something simpler, use Yv constraint
> for the DImode shift count in instructions that need AVX512VL when EVEX
> encoded if AVX512VL is not enabled instead of v. Bootstrapped/regtested
> on 7.x branch on x86_64-linux and i686-linux, ok for 7.x?
>
> Is the testcase alone ok also for trunk/8.2?
Patch is ok for trunk and ports.
--
Regards, Kirill Yukhin