public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
From: "rguenth at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug regression/44423] [4.5/4.6] Massive performance regression in SSE code
Date: Sat, 05 Jun 2010 10:47:00 -0000 [thread overview]
Message-ID: <20100605104729.19682.qmail@sourceware.org> (raw)
In-Reply-To: <bug-44423-2736@http.gcc.gnu.org/bugzilla/>
------- Comment #2 from rguenth at gcc dot gnu dot org 2010-06-05 10:47 -------
We have (4.4):
<bb 2>:
va.f[0] = a->r;
va.f[1] = a->g;
va.f[2] = a->b;
va.f[3] = 0.0;
pretmp.40 = va.v;
ivtmp.61 = 0;
<bb 3>:
att.12 = MEM[base: pre1, index: ivtmp.61] * pre2;
tmpatt = {att.12, att.12, att.12, att.12};
tmpatt.66 = __builtin_ia32_mulps (tmpatt, pretmp.40);
D.3720 = __builtin_ia32_addps (tmpatt.66, MEM[base: lpic, index: ivtmp.61,
step: 4]);
MEM[base: lpic, index: ivtmp.61, step: 4] = D.3720;
ivtmp.61 = ivtmp.61 + 4;
if (ivtmp.61 != 80)
goto <bb 3>;
else
goto <bb 4>;
vs. (4.6):
<bb 2>:
va$f$0_3 = a_2(D)->r;
va$f$1_4 = a_2(D)->g;
va$f$2_5 = a_2(D)->b;
<bb 3>:
# ivtmp.31_52 = PHI <ivtmp.31_51(3), 0(2)>
D.4345_11 = MEM[base: pre1_9(D), index: ivtmp.31_52];
att.1_13 = D.4345_11 * pre2_12(D);
tmpatt_37 = {att.1_13, att.1_13, att.1_13, att.1_13};
va.f[0] = va$f$0_3;
va.f[1] = va$f$1_4;
va.f[2] = va$f$2_5;
va.f[3] = 0.0;
D.4347_16 = va.v;
tmpatt_38 = __builtin_ia32_mulps (tmpatt_37, D.4347_16);
D.4350_25 = MEM[base: lpic_20(D), index: ivtmp.31_52, step: 4];
D.4382_39 = __builtin_ia32_addps (tmpatt_38, D.4350_25);
MEM[base: lpic_20(D), index: ivtmp.31_52, step: 4] = D.4382_39;
ivtmp.31_51 = ivtmp.31_52 + 4;
if (ivtmp.31_51 != 80)
goto <bb 3>;
else
goto <bb 4>;
which means that somehow we were not able to hoist loop invariant stuff.
investigating.
--
rguenth at gcc dot gnu dot org changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Ever Confirmed|0 |1
Last reconfirmed|0000-00-00 00:00:00 |2010-06-05 10:47:29
date| |
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44423
next prev parent reply other threads:[~2010-06-05 10:47 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-06-05 10:17 [Bug regression/44423] New: " martin at mpa-garching dot mpg dot de
2010-06-05 10:18 ` [Bug regression/44423] " martin at mpa-garching dot mpg dot de
2010-06-05 10:47 ` rguenth at gcc dot gnu dot org [this message]
2010-06-05 10:56 ` [Bug tree-optimization/44423] [4.5/4.6 Regression] Massive performance regression in SSE code due to SRA rguenth at gcc dot gnu dot org
2010-06-08 13:16 ` jamborm at gcc dot gnu dot org
2010-06-08 13:54 ` martin at mpa-garching dot mpg dot de
2010-06-08 14:03 ` rguenth at gcc dot gnu dot org
2010-06-08 14:29 ` jamborm at gcc dot gnu dot org
2010-06-08 14:51 ` rguenth at gcc dot gnu dot org
2010-06-08 15:00 ` jamborm at gcc dot gnu dot org
2010-06-08 15:11 ` rguenth at gcc dot gnu dot org
2010-06-09 9:03 ` jamborm at gcc dot gnu dot org
2010-06-09 9:06 ` jamborm at gcc dot gnu dot org
2010-06-09 11:21 ` jamborm at gcc dot gnu dot org
2010-06-09 12:06 ` martin at mpa-garching dot mpg dot de
2010-06-14 12:40 ` jamborm at gcc dot gnu dot org
2010-06-14 12:46 ` martin at mpa-garching dot mpg dot de
2010-06-14 12:50 ` jamborm at gcc dot gnu dot org
2010-06-15 9:49 ` jamborm at gcc dot gnu dot org
2010-06-15 10:04 ` jamborm at gcc dot gnu 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=20100605104729.19682.qmail@sourceware.org \
--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).