From: Kyrill Tkachov <kyrylo.tkachov@arm.com>
To: Uros Bizjak <ubizjak@gmail.com>
Cc: "gcc-patches@gcc.gnu.org" <gcc-patches@gcc.gnu.org>,
Jeff Law <law@redhat.com>, "H.J. Lu" <hjl.tools@gmail.com>
Subject: Re: [PATCH][RTL-ifcvt] Improve conditional select ops on immediates
Date: Mon, 03 Aug 2015 17:20:00 -0000 [thread overview]
Message-ID: <55BFA2DB.1040507@arm.com> (raw)
In-Reply-To: <55BF8E7F.2010108@arm.com>
On 03/08/15 16:53, Kyrill Tkachov wrote:
> On 03/08/15 16:45, Uros Bizjak wrote:
>> On Mon, Aug 3, 2015 at 5:37 PM, Kyrill Tkachov <kyrylo.tkachov@arm.com> wrote:
>>
>>>> Looking at the x86 movcc expansion code (ix86_expand_int_movcc) I
>>>> don't think this is a good idea. In the expander, there is already
>>>> quite some target-dependent code that goes great length to utilize sbb
>>>> insn as much as possible, before cmove is used.
>>>>
>>>> IMO, as far as x86 is concerned, the best solution would be to revert
>>>> the change. ix86_expand_int_movcc already does some tricks from your
>>>> patch in a target-efficient way. Generic change that was introduced by
>>>> your patch now interferes with this expansion.
>>> Well, technically the transformation was already there, it was just never
>>> reached for an x86 compilation because noce_try_cmove was tried in front of
>>> it
>>> and used a target-specific expansion.
>>> In any case, how's this proposal?
>>> The transformation noce_try_store_flag_constants
>>> /* if (test) x = a; else x = b;
>>> => x = (-(test != 0) & (b - a)) + a; */
>>>
>>> Is a catch-all-immediates transformation in noce_try_store_flag_constants.
>>> What if we moved it to noce_try_cmove and performed it only if the
>>> target-specific
>>> conditional move expansion there failed?
>>>
>>> That way we can try the x86_64-specific sequence first and still give the
>>> opportunity
>>> to noce_try_store_flag_constants to perform the transformations that can
>>> benefit targets
>>> that don't have highly specific conditional move expanders.
>> Yes, let's try this approach. As was found out, some targets (e.g.
>> x86) hide lots of different target-dependent expansion strategies into
>> movcc expander. Perhaps this fact should be documented in the comment
>> in the generic code?
> Ok, I'll work on that approach and add a comment.
I'm testing a patch that fix the testcases on x86_64 and does not
harm codegen on aarch64. Feel free to file a PR and assign it to me.
Thanks,
Kyrill
>
> Thanks,
> Kyrill
>
>
>> Uros.
>>
next prev parent reply other threads:[~2015-08-03 17:20 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-08-03 12:33 Uros Bizjak
2015-08-03 13:02 ` Kyrill Tkachov
2015-08-03 13:37 ` Uros Bizjak
2015-08-03 13:43 ` Kyrill Tkachov
2015-08-03 14:12 ` Kyrill Tkachov
2015-08-03 14:15 ` Uros Bizjak
2015-08-03 15:37 ` Kyrill Tkachov
2015-08-03 15:46 ` Uros Bizjak
2015-08-03 15:53 ` Kyrill Tkachov
2015-08-03 17:20 ` Kyrill Tkachov [this message]
2015-08-03 17:37 ` Uros Bizjak
2015-08-04 8:44 ` Kyrill Tkachov
2015-08-10 9:36 ` Kyrill Tkachov
2015-08-10 9:43 ` Uros Bizjak
2015-08-10 9:44 ` Kyrill Tkachov
2015-08-12 17:52 ` Jeff Law
2015-08-03 14:04 ` Uros Bizjak
2015-08-03 14:37 ` H.J. Lu
2015-08-03 15:50 ` Ilya Enkovich
2015-08-03 16:29 ` Jeff Law
-- strict thread matches above, loose matches on Subject: below --
2015-07-29 14:16 Kyrill Tkachov
2015-07-29 22:46 ` Jeff Law
2015-07-30 14:30 ` Kyrill Tkachov
2015-07-31 16:16 ` Jeff Law
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=55BFA2DB.1040507@arm.com \
--to=kyrylo.tkachov@arm.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=hjl.tools@gmail.com \
--cc=law@redhat.com \
--cc=ubizjak@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).