From: Jakub Jelinek <jakub@redhat.com>
To: Ilya Enkovich <enkovich.gnu@gmail.com>
Cc: "William J. Schmidt" <wschmidt@linux.vnet.ibm.com>,
gcc-patches@gcc.gnu.org
Subject: Re: [PATCH Atom][PR middle-end/44382] Tree reassociation improvement
Date: Wed, 13 Jul 2011 09:52:00 -0000 [thread overview]
Message-ID: <20110713090817.GV2687@tyan-ft48-01.lab.bos.redhat.com> (raw)
In-Reply-To: <CAMbmDYa64_sdXRgHHr5RvpRH+8DpdJDVsQ-B2NiTdTxzsQUeog@mail.gmail.com>
On Wed, Jul 13, 2011 at 01:01:59PM +0400, Ilya Enkovich wrote:
> > Well, if it is clearly a win to reassociate, you can always reassociate
> > them by doing arithmetics in corresponding unsigned type and afterwards
> > converting back to the signed type.
>
> You are right. But in this case we again make all operands have
> wrap-around type and thus disable some other optimization. It would be
> nice to have opportunity to reassociate and still have undefined
> behavior on overflow for optimizations. One way to do it for add/sub
> is to use wider type (long long instead of int).
I disagree. Widening would result in worse code in most cases, as you need
to sign extend all the operands. On the other side, I doubt you can
actually usefully use the undefinedness of signed overflow for a series of
3 or more operands of the associative operation.
Jakub
next prev parent reply other threads:[~2011-07-13 9:08 UTC|newest]
Thread overview: 47+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-07-12 12:35 Илья Энкович
2011-07-12 17:08 ` William J. Schmidt
2011-07-12 17:22 ` H.J. Lu
2011-07-13 8:28 ` Ilya Enkovich
2011-07-12 17:24 ` William J. Schmidt
2011-07-13 8:13 ` Ilya Enkovich
2011-07-13 8:30 ` Jakub Jelinek
2011-07-13 9:04 ` Ilya Enkovich
2011-07-13 9:52 ` Jakub Jelinek [this message]
2011-07-13 9:59 ` Ilya Enkovich
2011-07-13 11:01 ` Richard Guenther
2011-07-13 13:11 ` William J. Schmidt
2011-07-13 18:05 ` Michael Meissner
2011-07-13 20:18 ` Ilya Enkovich
2011-07-14 2:39 ` Michael Meissner
2011-07-14 9:40 ` Richard Guenther
2011-07-14 9:42 ` Richard Guenther
2011-07-14 10:04 ` Ilya Enkovich
2011-07-14 10:14 ` Richard Guenther
2011-07-14 11:03 ` Ilya Enkovich
2011-07-14 15:31 ` Ilya Enkovich
2011-07-19 12:10 ` Richard Guenther
2011-07-19 15:25 ` Ilya Enkovich
2011-08-05 11:46 ` Richard Guenther
2011-08-05 15:08 ` Ilya Enkovich
2011-08-10 15:19 ` Ilya Enkovich
2011-08-19 9:09 ` Ilya Enkovich
2011-08-26 11:45 ` Ilya Enkovich
2011-08-30 13:16 ` Richard Guenther
2011-08-31 15:38 ` Ilya Enkovich
2011-09-01 9:24 ` Richard Guenther
2011-09-01 10:28 ` Ilya Enkovich
2011-09-02 12:53 ` Uros Bizjak
2011-09-02 13:07 ` Richard Guenther
2011-09-02 13:46 ` Ilya Enkovich
2011-09-02 14:01 ` Richard Guenther
2011-09-02 14:13 ` Ilya Enkovich
2011-09-06 12:40 ` Ilya Enkovich
2011-09-06 15:53 ` Uros Bizjak
2011-09-06 16:18 ` Ilya Enkovich
2011-09-06 16:45 ` H.J. Lu
2011-07-26 9:54 ` Ilya Enkovich
2011-08-03 8:35 ` Ilya Enkovich
2011-07-21 20:14 ` Joseph S. Myers
2011-07-14 16:04 ` Michael Meissner
2011-07-14 16:08 ` Ilya Enkovich
2011-07-14 10:38 ` Ilya Enkovich
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=20110713090817.GV2687@tyan-ft48-01.lab.bos.redhat.com \
--to=jakub@redhat.com \
--cc=enkovich.gnu@gmail.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=wschmidt@linux.vnet.ibm.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).