public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
From: "joseph at codesourcery dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug middle-end/34678] Optimization generates incorrect code with -frounding-math option (#pragma STDC FENV_ACCESS not implemented)
Date: Mon, 11 Nov 2013 16:38:00 -0000 [thread overview]
Message-ID: <bug-34678-4-SeVOgDsgJl@http.gcc.gnu.org/bugzilla/> (raw)
In-Reply-To: <bug-34678-4@http.gcc.gnu.org/bugzilla/>
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34678
--- Comment #22 from joseph at codesourcery dot com <joseph at codesourcery dot com> ---
On Mon, 11 Nov 2013, nmm1 at cam dot ac.uk wrote:
> There are also very strong grounds for not wanting IEEE 754 support by
> default, anyway, because of the performance impact and because a lot of
> programs won't reset the state before calling external functions (and
> hence may well give wrong answers). That is especially true if the code
> is used within a C++ program or uses GPUs or some SIMD units - let alone
> OpenMP :-(
Note also that the documented default is -ftrapping-math
-fno-rounding-math. I suspect that if -ftrapping-math actually
implemented everything required for the floating-point exceptions aspects
of FENV_ACCESS, it would be just as bad for optimization as
-frounding-math - it would disallow constant-folding inexact
floating-point expressions because that would eliminate the side effect of
raising the "inexact" exception, for example (just as -frounding-math does
disable such constant folding, although not in all cases it should,
because the result depends on the rounding mode), and would mean a value
computed before a function call can't be reused for the same computation
after that call because the computation might raise exceptions that the
function call could have cleared (just as -frounding-math should prevent
such reuse because the call might change the rounding mode).
So a key part of actually making rounding modes and exceptions work
reliably would be working out a definition of GCC's default mode that
allows more or less the same optimizations as at present, while allowing
users wanting the full support (and consequent optimization cost) to
specify the appropriate command-line options or FENV_ACCESS pragma to
enable it.
next prev parent reply other threads:[~2013-11-11 16:38 UTC|newest]
Thread overview: 37+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <bug-34678-4@http.gcc.gnu.org/bugzilla/>
2010-10-26 10:33 ` paolo.carlini at oracle dot com
2011-02-08 2:02 ` pinskia at gcc dot gnu.org
2011-04-01 19:22 ` pinskia at gcc dot gnu.org
2013-11-11 9:30 ` nmm1 at cam dot ac.uk
2013-11-11 13:37 ` vincent-gcc at vinc17 dot net
2013-11-11 16:38 ` joseph at codesourcery dot com [this message]
2013-11-11 17:32 ` nmm1 at cam dot ac.uk
2014-01-10 11:40 ` vincent-gcc at vinc17 dot net
2014-01-14 14:58 ` vincent-gcc at vinc17 dot net
2014-01-14 15:13 ` nmm1 at cam dot ac.uk
2014-01-14 15:52 ` vincent-gcc at vinc17 dot net
2014-01-14 17:22 ` nmm1 at cam dot ac.uk
2014-01-14 23:16 ` vincent-gcc at vinc17 dot net
2014-02-16 10:03 ` jackie.rosen at hushmail dot com
2020-04-16 15:35 ` vincent-gcc at vinc17 dot net
2023-01-06 17:29 ` pinskia at gcc dot gnu.org
2023-01-06 17:31 ` pinskia at gcc dot gnu.org
2023-01-06 19:29 ` tkoenig at gcc dot gnu.org
2023-01-06 22:14 ` tkoenig at gcc dot gnu.org
2023-01-07 11:14 ` tkoenig at gcc dot gnu.org
2023-01-07 12:14 ` tkoenig at gcc dot gnu.org
2008-01-04 18:56 [Bug c/34678] New: Optimization generates incorrect code with -frounding-math option merkert at comcast dot net
2008-01-05 12:08 ` [Bug middle-end/34678] Optimization generates incorrect code with -frounding-math option (#pragma STDC FENV_ACCESS not implemented) rguenth at gcc dot gnu dot org
2008-01-05 16:48 ` merkert at comcast dot net
2008-01-05 21:00 ` joseph at codesourcery dot com
2008-01-05 21:29 ` rguenth at gcc dot gnu dot org
2008-01-06 16:21 ` joseph at codesourcery dot com
2008-01-06 17:19 ` rguenth at gcc dot gnu dot org
2008-01-06 18:03 ` joseph at codesourcery dot com
2008-10-15 15:44 ` rguenth at gcc dot gnu dot org
2008-10-15 21:31 ` vincent at vinc17 dot org
2008-10-15 22:15 ` rguenth at gcc dot gnu dot org
2008-10-15 22:35 ` vincent at vinc17 dot org
2008-10-16 9:46 ` rguenth at gcc dot gnu dot org
2008-10-16 11:58 ` merkert at comcast dot net
2008-10-16 14:21 ` vincent at vinc17 dot org
2008-10-16 16:40 ` joseph at codesourcery dot com
2008-10-16 17:41 ` vincent at vinc17 dot org
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=bug-34678-4-SeVOgDsgJl@http.gcc.gnu.org/bugzilla/ \
--to=gcc-bugzilla@gcc.gnu.org \
--cc=gcc-bugs@gcc.gnu.org \
/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).