--- Additional Comments From pcarlini at suse dot de 2005-02-02 16:48
---
This specific issue is fixed for 4.0.0 and 3.4.4.
--
What|Removed |Added
Status|ASSIGNE
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-02-02
16:47 ---
Subject: Bug 19642
CVSROOT:/cvs/gcc
Module name:gcc
Branch: gcc-3_4-branch
Changes by: [EMAIL PROTECTED] 2005-02-02 16:46:40
Modified files:
libstdc++-v3 : Change
--
What|Removed |Added
Target Milestone|--- |3.4.4
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19642
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-01-30
14:10 ---
Subject: Bug 19642
CVSROOT:/cvs/gcc
Module name:gcc
Changes by: [EMAIL PROTECTED] 2005-01-30 14:09:59
Modified files:
libstdc++-v3 : ChangeLog
libstdc++-v3/confi
--- Additional Comments From pcarlini at suse dot de 2005-01-28 09:58
---
Thanks. We are now reasonably confident that the patch is safe, besides
providing a good performance improvement. Most probably we shall apply
it soon, I'll take care of the next steps. Thanks again.
--
--- Additional Comments From joerg dot richter at pdv-fs dot de 2005-01-28
07:54 ---
(In reply to comment #26)
> Excellent. Can you please check whether on AIX this is ok ("de_DE" is just
> an example, any actually installed named locale should do):
> [code]
> In short, we should make su
--- Additional Comments From pcarlini at suse dot de 2005-01-27 16:56
---
> There's known problems with setlocale not being "stable" on AIX.
Thanks for the pointer. I can find only some rather vague hints, around,
however... "(un)stable"?!? Or 'just' plain slow?!? ;) ;)
> You're right,
--- Additional Comments From bkoz at gcc dot gnu dot org 2005-01-27 16:38
---
Paolo, just a note. This issue has been raised before on the mailing list. (Or
bugzilla, sorry so vague.) If you search stable/AIX/setlocale you might find
something.
There's known problems with setlocale not
--- Additional Comments From pcarlini at suse dot de 2005-01-27 15:59
---
Excellent. Can you please check whether on AIX this is ok ("de_DE" is just
an example, any actually installed named locale should do):
#include
#include
#include
int main()
{
char* save = setlocale(LC_ALL, "
--- Additional Comments From joerg dot richter at pdv-fs dot de 2005-01-27
15:05 ---
With patch draft_19642_3 and count=100
$ time t > /dev/null
real 0,64
user 0,60
sys0,01
$ time t cout > /dev/null
real 1,14
user 1,12
sys0,00
Now its only 2 times slower than sprin
--- Additional Comments From pcarlini at suse dot de 2005-01-27 12:57
---
Eh! The runtimes are indeed **very** interesting, thanks! And, also, it looks
like I was wrong in the last comment: in case LC_ALL is set, this is
automatically
seen in setlocale(LC_NUMERIC, 0), therefore your sug
--- Additional Comments From joerg dot richter at pdv-fs dot de 2005-01-27
12:54 ---
I think this runtimes of the different setlocale() calls might be interesting:
(This is for 100 calls)
category locale time
LC_ALL00.70s
LC_NUMERIC00.29s
LC_ALL "C"
--- Additional Comments From pcarlini at suse dot de 2005-01-27 12:46
---
Thanks! For the correctness issue, notice that LC_ALL overrides any LC_*,
therefore checking only LC_NUMERIC seems weaker, still, probably effective in
most circumstances...
--
http://gcc.gnu.org/bugzilla/show_
--- Additional Comments From joerg dot richter at pdv-fs dot de 2005-01-27
12:39 ---
A quick test shows, that std::setlocale(LC_ALL, NULL) returns "C C C C C C".
But std::setlocale(LC_NUMERIC, NULL) returns "C"
I think it is enough to test LC_NUMERIC.
I'll try to test a modified patch.
--- Additional Comments From pcarlini at suse dot de 2005-01-27 12:31
---
Notice, however, that as-is, the patch is not perfect from the correctness point
of view: we should also check LC_NUMERIC...
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19642
--- Additional Comments From pcarlini at suse dot de 2005-01-27 12:25
---
Can you test the attached? Should help, and we can apply it in 3_4-branch, maybe
in mainline too depending on the res of 17140.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19642
--- Additional Comments From pcarlini at suse dot de 2005-01-27 11:11
---
Hi everyone. No, switching by hand to "C" locale cannot help, because, given
the current generic locale model, the setlocale calls are issued *anyway*.
Indeed,
we could envisage improving on this...
--
http://
--- Additional Comments From joerg dot richter at pdv-fs dot de 2005-01-27
08:08 ---
std::cout.imbue( std::locale( "C" ) );
nor
std::cout.imbue( std::locale::classic() );
nor
export LANG=C
does change anything
Joerg
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19642
--- Additional Comments From dje at gcc dot gnu dot org 2005-01-26 19:31
---
AIX does not yet implement uselocale(). I am not aware of any faster
interface. Does "C" locale help?
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19642
--
What|Removed |Added
Status|WAITING |NEW
Ever Confirmed||1
Last reconfirmed|-00-00 00:00:00 |2005-01-
--- Additional Comments From pcarlini at suse dot de 2005-01-26 17:34
---
FWIW, the issue doesn't affect linux even if configured --enable-clocale=generic
(thus using setlocale). Therefore, on *some* targets setlocale is slow and
causes
it, not everywhere.
--
http://gcc.gnu.org/bugz
--- Additional Comments From pcarlini at suse dot de 2005-01-26 16:17
---
> If it helps you, here the first lines of prof. But in practice some
> backtraces
> help me a lot more than a a [g]prof output. Its very fast in finding big
> bottlenecks, like setlocale in this PR.
Ok, thank
--- Additional Comments From joerg dot richter at pdv-fs dot de 2005-01-26
16:00 ---
If it helps you, here the first lines of prof. But in practice some backtraces
help me a lot more than a a [g]prof output. Its very fast in finding big
bottlenecks, like setlocale in this PR.
count =
--- Additional Comments From pcarlini at suse dot de 2005-01-26 15:45
---
> Doesn't a function that is on 99% of all call stacks automatically consume
> 99%
> of the runtime? Ok, the time can also be spend in the childs like here in
> __issetuid, but this is an implementation detail o
--- Additional Comments From joerg dot richter at pdv-fs dot de 2005-01-26
15:41 ---
> ;) No, I mean *relative* percentages: setlocal in your profile is on top,
higher
> than __convert_from_v and everything else. I'm asking: is consuming 1%, 10%,
or
> 99% of the total runtime?
Doesn't
--- Additional Comments From pcarlini at suse dot de 2005-01-26 15:37
---
For reference, on linux, a typical profile (count = 1000) begins as
follows, and:
1- The various snprintf are on top.
2- uselocale (the thread safe equivalent of setlocale) is way below
Therefore, we can easily
--- Additional Comments From pcarlini at suse dot de 2005-01-26 15:22
---
> Ok, lets say I took 100 backtraces and got 99 times this one. I hope this is
> enough percentage.
;) No, I mean *relative* percentages: setlocal in your profile is on top, higher
than __convert_from_v and every
--- Additional Comments From joerg dot richter at pdv-fs dot de 2005-01-26
15:18 ---
Ok, lets say I took 100 backtraces and got 99 times this one. I hope this is
enough percentage.
_GLIBCXX_USE_C99 is defined
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19642
--- Additional Comments From pcarlini at suse dot de 2005-01-26 15:10
---
Honestly, this doesn't help much, the percentages are important: suppose, for
some reason, setlocale is very slow on AIX... Is _GLIBCXX_USE_C99 defined, on
AIX? I'm asking because internally __convert_from_v uses s
--- Additional Comments From joerg dot richter at pdv-fs dot de 2005-01-26
15:00 ---
I took some backtraces during a longer run of the program and most of the time
it looked like this: (I removed some templates to make it readable)
__issetuid
load_locale
setlocale
int std::__convert_fr
--
What|Removed |Added
Status|UNCONFIRMED |WAITING
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19642
--- Additional Comments From pcarlini at suse dot de 2005-01-26 14:47
---
Wait a minute... Ok, we have got 17140, but here, if the issue is confirmed,
we are talking about 2 orders of magnitude!?!?!
Indeed, on my P4-2400 (gcc3.4.3, actually, similar bevavior for mainline)
I cannot confi
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-01-26
14:34 ---
*** This bug has been marked as a duplicate of 17140 ***
--
What|Removed |Added
--
What|Removed |Added
Component|c++ |libstdc++
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19642
34 matches
Mail list logo
