public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug tree-optimization/16022] New: Excessive memory usage of vectorizer
@ 2004-06-16 14:47 rguenth at tat dot physik dot uni-tuebingen dot de
2004-06-16 15:04 ` [Bug tree-optimization/16022] " rguenth at tat dot physik dot uni-tuebingen dot de
` (7 more replies)
0 siblings, 8 replies; 9+ messages in thread
From: rguenth at tat dot physik dot uni-tuebingen dot de @ 2004-06-16 14:47 UTC (permalink / raw)
To: gcc-bugs
Trying to compile the tramp3d-v3 testcase
(http://www.tat.physik.uni-tuebingen.de/~rguenth/gcc/tramp3d-v3.cpp.gz)
with vectorization support uses too much memory for my 2GB machine:
~/ix86/gccssa-lno/bin/g++ -O2 -ftree-vectorize -march=pentium4 -c tramp3d-v3.cpp
killed
g++ (GCC) 3.5-tree-ssa-lno 20040615 (merged 20040612)
Note this is lno with the leafify patch, without leafify memory requirements are
down, but nothing is vectorized. Leafify patch is at
http://www.tat.physik.uni-tuebingen.de/~rguenth/gcc/leafify-ssa-2
--
Summary: Excessive memory usage of vectorizer
Product: gcc
Version: lno
Status: UNCONFIRMED
Severity: normal
Priority: P2
Component: tree-optimization
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: rguenth at tat dot physik dot uni-tuebingen dot de
CC: gcc-bugs at gcc dot gnu dot org
GCC build triplet: i686-pc-linux-gnu
GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16022
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Bug tree-optimization/16022] Excessive memory usage of vectorizer
2004-06-16 14:47 [Bug tree-optimization/16022] New: Excessive memory usage of vectorizer rguenth at tat dot physik dot uni-tuebingen dot de
@ 2004-06-16 15:04 ` rguenth at tat dot physik dot uni-tuebingen dot de
2004-06-16 15:36 ` rguenth at tat dot physik dot uni-tuebingen dot de
` (6 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: rguenth at tat dot physik dot uni-tuebingen dot de @ 2004-06-16 15:04 UTC (permalink / raw)
To: gcc-bugs
------- Additional Comments From rguenth at tat dot physik dot uni-tuebingen dot de 2004-06-16 15:04 -------
It looks like -fno-exceptions works around this (first memory peak is down from
1.4Gb to 400Mb). Something is going wrong with exceptions.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16022
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Bug tree-optimization/16022] Excessive memory usage of vectorizer
2004-06-16 14:47 [Bug tree-optimization/16022] New: Excessive memory usage of vectorizer rguenth at tat dot physik dot uni-tuebingen dot de
2004-06-16 15:04 ` [Bug tree-optimization/16022] " rguenth at tat dot physik dot uni-tuebingen dot de
@ 2004-06-16 15:36 ` rguenth at tat dot physik dot uni-tuebingen dot de
2004-06-16 17:50 ` pinskia at gcc dot gnu dot org
` (5 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: rguenth at tat dot physik dot uni-tuebingen dot de @ 2004-06-16 15:36 UTC (permalink / raw)
To: gcc-bugs
------- Additional Comments From rguenth at tat dot physik dot uni-tuebingen dot de 2004-06-16 15:36 -------
Ok, it compiled the testcase with -fno-exceptions and even seems to run
successfully. But the vectorizer didn't do anything interesting. Not one loop
out of 323 was vectorized, mostly due to "complicated pointer arithmetic" or
"can't compute number of iterations". The interesting ones all due to the
latter reason.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16022
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Bug tree-optimization/16022] Excessive memory usage of vectorizer
2004-06-16 14:47 [Bug tree-optimization/16022] New: Excessive memory usage of vectorizer rguenth at tat dot physik dot uni-tuebingen dot de
2004-06-16 15:04 ` [Bug tree-optimization/16022] " rguenth at tat dot physik dot uni-tuebingen dot de
2004-06-16 15:36 ` rguenth at tat dot physik dot uni-tuebingen dot de
@ 2004-06-16 17:50 ` pinskia at gcc dot gnu dot org
2004-06-17 9:09 ` rguenth at tat dot physik dot uni-tuebingen dot de
` (4 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: pinskia at gcc dot gnu dot org @ 2004-06-16 17:50 UTC (permalink / raw)
To: gcc-bugs
------- Additional Comments From pinskia at gcc dot gnu dot org 2004-06-16 17:50 -------
I should note that -fno-excpetions just makes there be a lot less BB's so maybe where there is a huge
amount of BB it takes a huge amount of memory which is reasonable with leafify as every thing is
inlined.
--
What |Removed |Added
----------------------------------------------------------------------------
Keywords| |memory-hog
Target Milestone|--- |lno
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16022
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Bug tree-optimization/16022] Excessive memory usage of vectorizer
2004-06-16 14:47 [Bug tree-optimization/16022] New: Excessive memory usage of vectorizer rguenth at tat dot physik dot uni-tuebingen dot de
` (2 preceding siblings ...)
2004-06-16 17:50 ` pinskia at gcc dot gnu dot org
@ 2004-06-17 9:09 ` rguenth at tat dot physik dot uni-tuebingen dot de
2004-06-17 9:53 ` pinskia at gcc dot gnu dot org
` (3 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: rguenth at tat dot physik dot uni-tuebingen dot de @ 2004-06-17 9:09 UTC (permalink / raw)
To: gcc-bugs
------- Additional Comments From rguenth at tat dot physik dot uni-tuebingen dot de 2004-06-17 09:09 -------
Subject: Re: Excessive memory usage of vectorizer
On Wed, 16 Jun 2004, pinskia at gcc dot gnu dot org wrote:
> I should note that -fno-excpetions just makes there be a lot less BB's so maybe where there is a huge
> amount of BB it takes a huge amount of memory which is reasonable with leafify as every thing is
> inlined.
Of course. But I wonder if we cannot clean that up a little and merge
BB's and/or EH regions here. Maybe we simply don't detect nothrow as good
as we should.
Richard.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16022
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Bug tree-optimization/16022] Excessive memory usage of vectorizer
2004-06-16 14:47 [Bug tree-optimization/16022] New: Excessive memory usage of vectorizer rguenth at tat dot physik dot uni-tuebingen dot de
` (3 preceding siblings ...)
2004-06-17 9:09 ` rguenth at tat dot physik dot uni-tuebingen dot de
@ 2004-06-17 9:53 ` pinskia at gcc dot gnu dot org
2004-08-24 5:32 ` pinskia at gcc dot gnu dot org
` (2 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: pinskia at gcc dot gnu dot org @ 2004-06-17 9:53 UTC (permalink / raw)
To: gcc-bugs
--
What |Removed |Added
----------------------------------------------------------------------------
CC| |pinskia at gcc dot gnu dot
| |org
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16022
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Bug tree-optimization/16022] Excessive memory usage of vectorizer
2004-06-16 14:47 [Bug tree-optimization/16022] New: Excessive memory usage of vectorizer rguenth at tat dot physik dot uni-tuebingen dot de
` (4 preceding siblings ...)
2004-06-17 9:53 ` pinskia at gcc dot gnu dot org
@ 2004-08-24 5:32 ` pinskia at gcc dot gnu dot org
2004-08-24 8:41 ` rguenth at tat dot physik dot uni-tuebingen dot de
2004-08-24 14:20 ` pinskia at gcc dot gnu dot org
7 siblings, 0 replies; 9+ messages in thread
From: pinskia at gcc dot gnu dot org @ 2004-08-24 5:32 UTC (permalink / raw)
To: gcc-bugs
------- Additional Comments From pinskia at gcc dot gnu dot org 2004-08-24 05:32 -------
Can you test again, this time on the mainline as the vectorizer has been merged in?
--
What |Removed |Added
----------------------------------------------------------------------------
Target Milestone|lno |---
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16022
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Bug tree-optimization/16022] Excessive memory usage of vectorizer
2004-06-16 14:47 [Bug tree-optimization/16022] New: Excessive memory usage of vectorizer rguenth at tat dot physik dot uni-tuebingen dot de
` (5 preceding siblings ...)
2004-08-24 5:32 ` pinskia at gcc dot gnu dot org
@ 2004-08-24 8:41 ` rguenth at tat dot physik dot uni-tuebingen dot de
2004-08-24 14:20 ` pinskia at gcc dot gnu dot org
7 siblings, 0 replies; 9+ messages in thread
From: rguenth at tat dot physik dot uni-tuebingen dot de @ 2004-08-24 8:41 UTC (permalink / raw)
To: gcc-bugs
------- Additional Comments From rguenth at tat dot physik dot uni-tuebingen dot de 2004-08-24 08:40 -------
Subject: Re: Excessive memory usage of vectorizer
On 24 Aug 2004, pinskia at gcc dot gnu dot org wrote:
> Can you test again, this time on the mainline as the vectorizer has been merged in?
With the parts of the vectorizer merged there is not a difference
between using (with leafification turned off)
-O2 -funroll-loops -ftime-report -march=pentium4 -mfpmath=sse
vs.
-O2 -funroll-loops -ftime-report -march=pentium4 -mfpmath=sse
-ftree-vectorize
we're using about 450MB for this (Machine with 2GB ram).
with leafification turned on (i.e. for some loops generate an
awful lot of BBs) its
550MB vs. 552MB.
Numbers do improve a little (470MB) if using -fno-exceptions, but
nowhere like the problem was on lno.
So on mainline this doesn't seem to be a problem.
Btw. on lno the problem seems to be fixed as well.
Thanks,
Richard.
--
Richard Guenther <richard dot guenther at uni-tuebingen dot de>
WWW: http://www.tat.physik.uni-tuebingen.de/~rguenth/
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16022
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Bug tree-optimization/16022] Excessive memory usage of vectorizer
2004-06-16 14:47 [Bug tree-optimization/16022] New: Excessive memory usage of vectorizer rguenth at tat dot physik dot uni-tuebingen dot de
` (6 preceding siblings ...)
2004-08-24 8:41 ` rguenth at tat dot physik dot uni-tuebingen dot de
@ 2004-08-24 14:20 ` pinskia at gcc dot gnu dot org
7 siblings, 0 replies; 9+ messages in thread
From: pinskia at gcc dot gnu dot org @ 2004-08-24 14:20 UTC (permalink / raw)
To: gcc-bugs
------- Additional Comments From pinskia at gcc dot gnu dot org 2004-08-24 14:20 -------
So closing as fixed.
--
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |RESOLVED
Resolution| |FIXED
Target Milestone|--- |lno
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16022
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2004-08-24 14:20 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2004-06-16 14:47 [Bug tree-optimization/16022] New: Excessive memory usage of vectorizer rguenth at tat dot physik dot uni-tuebingen dot de
2004-06-16 15:04 ` [Bug tree-optimization/16022] " rguenth at tat dot physik dot uni-tuebingen dot de
2004-06-16 15:36 ` rguenth at tat dot physik dot uni-tuebingen dot de
2004-06-16 17:50 ` pinskia at gcc dot gnu dot org
2004-06-17 9:09 ` rguenth at tat dot physik dot uni-tuebingen dot de
2004-06-17 9:53 ` pinskia at gcc dot gnu dot org
2004-08-24 5:32 ` pinskia at gcc dot gnu dot org
2004-08-24 8:41 ` rguenth at tat dot physik dot uni-tuebingen dot de
2004-08-24 14:20 ` pinskia at gcc dot gnu dot org
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).