public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug testsuite/48283] New: gcc.dg/graphite/block-[3478].c timeouts
@ 2011-03-25 11:47 ro at gcc dot gnu.org
2011-03-25 13:00 ` [Bug testsuite/48283] " rguenth at gcc dot gnu.org
` (4 more replies)
0 siblings, 5 replies; 6+ messages in thread
From: ro at gcc dot gnu.org @ 2011-03-25 11:47 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48283
Summary: gcc.dg/graphite/block-[3478].c timeouts
Product: gcc
Version: 4.7.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: testsuite
AssignedTo: unassigned@gcc.gnu.org
ReportedBy: ro@gcc.gnu.org
CC: spop@gcc.gnu.org
Host: sparc-sun-solaris2.*, mips-sgi-irix6.5
Target: sparc-sun-solaris2.*, mips-sgi-irix6.5
Build: sparc-sun-solaris2.*, mips-sgi-irix6.5
The for graphite testcases gcc.dg/graphite/block-[3478].c regularly time out on
Solaris/SPARC and IRIX hosts despite already using dg-timeout-factor 4, i.e.
they
take longer than 20 minutes (40 minutes on IRIX where I've doubled the default
to account for the slower CPUs) to complete.
Here are the compile times on onloaded machines:
* i386-pc-solaris2.11, 1.6 GHz Core i7:
real 2:19.73
user 2:16.56
sys 0.10
* sparc-sun-solaris2.10, 1.35 GHz UltraSPARC IV+:
real 5:52.77
user 5:52.22
sys 0.14
* sparc-sun-solaris2.11, 1.2 GHz UltraSPARC-T2:
real 16:23.64
user 16:21.71
sys 0.49
* mips-sgi-irix6.5, 250 MHz MIPS R10000:
real 41:34.05
user 41:21.31
sys 0.78
Both systems are multilibbed, so 4 tests timing out add 160 minutes to the
test time, which I consider way over the top. In light of the policy in
http://gcc.gnu.org/ml/gcc-patches/2011-02/msg00059.html
either consider reducing the runtime of the test to be within the limits
outlined there or flag it as expensive using dg-require-effective-target
run_expensive_tests.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug testsuite/48283] gcc.dg/graphite/block-[3478].c timeouts
2011-03-25 11:47 [Bug testsuite/48283] New: gcc.dg/graphite/block-[3478].c timeouts ro at gcc dot gnu.org
@ 2011-03-25 13:00 ` rguenth at gcc dot gnu.org
2011-03-25 13:02 ` ro at CeBiTec dot Uni-Bielefeld.DE
` (3 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: rguenth at gcc dot gnu.org @ 2011-03-25 13:00 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48283
Richard Guenther <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords| |compile-time-hog
Status|UNCONFIRMED |NEW
Last reconfirmed| |2011.03.25 12:26:55
Ever Confirmed|0 |1
--- Comment #1 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-03-25 12:26:55 UTC ---
Just to clarify - runtime or compile-time issue? I see the testcases
operate on 100^3 arrays - certainly not small for very slow HW.
Compile-time seems to be excessively high for x86_64 machines as well,
Sebastian, why is graphite soooo slow on these very small testcases?
Execution is quick for me. All time is spent in
samples % app name symbol name
1281556 58.0817 libppl.so.9.0.0 /usr/lib64/libppl.so.9.0.0
466710 21.1519 libc-2.9.so /lib64/libc-2.9.so
250459 11.3511 libgmp.so.3.5.0 /usr/lib64/libgmp.so.3.5.0
(note this is a whole-system profile).
I'd say skip graphite tests completely for !run_expensive_tests for now.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug testsuite/48283] gcc.dg/graphite/block-[3478].c timeouts
2011-03-25 11:47 [Bug testsuite/48283] New: gcc.dg/graphite/block-[3478].c timeouts ro at gcc dot gnu.org
2011-03-25 13:00 ` [Bug testsuite/48283] " rguenth at gcc dot gnu.org
@ 2011-03-25 13:02 ` ro at CeBiTec dot Uni-Bielefeld.DE
2011-03-25 14:06 ` ro at gcc dot gnu.org
` (2 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: ro at CeBiTec dot Uni-Bielefeld.DE @ 2011-03-25 13:02 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48283
--- Comment #2 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> 2011-03-25 12:35:03 UTC ---
> --- Comment #1 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-03-25 12:26:55 UTC ---
> Just to clarify - runtime or compile-time issue? I see the testcases
> operate on 100^3 arrays - certainly not small for very slow HW.
Compile time, runtime for block-3.exe is 0.06 s even on the R10k
machine.
> I'd say skip graphite tests completely for !run_expensive_tests for now.
I'd rather tag the four culprits only for the moment to keep some
coverage.
Rainer
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug testsuite/48283] gcc.dg/graphite/block-[3478].c timeouts
2011-03-25 11:47 [Bug testsuite/48283] New: gcc.dg/graphite/block-[3478].c timeouts ro at gcc dot gnu.org
2011-03-25 13:00 ` [Bug testsuite/48283] " rguenth at gcc dot gnu.org
2011-03-25 13:02 ` ro at CeBiTec dot Uni-Bielefeld.DE
@ 2011-03-25 14:06 ` ro at gcc dot gnu.org
2011-05-18 0:01 ` michael.hope at linaro dot org
2013-07-14 5:33 ` spop at gcc dot gnu.org
4 siblings, 0 replies; 6+ messages in thread
From: ro at gcc dot gnu.org @ 2011-03-25 14:06 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48283
--- Comment #3 from Rainer Orth <ro at gcc dot gnu.org> 2011-03-25 13:02:29 UTC ---
Author: ro
Date: Fri Mar 25 13:02:24 2011
New Revision: 171451
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=171451
Log:
PR testsuite/48283
* gcc.dg/graphite/block-3.c: Add dg-require-effective-target
run_expensive_tests.
* gcc.dg/graphite/block-4.c: Likewise.
* gcc.dg/graphite/block-7.c: Likewise.
* gcc.dg/graphite/block-8.c: Likewise.
Modified:
trunk/gcc/testsuite/ChangeLog
trunk/gcc/testsuite/gcc.dg/graphite/block-3.c
trunk/gcc/testsuite/gcc.dg/graphite/block-4.c
trunk/gcc/testsuite/gcc.dg/graphite/block-7.c
trunk/gcc/testsuite/gcc.dg/graphite/block-8.c
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug testsuite/48283] gcc.dg/graphite/block-[3478].c timeouts
2011-03-25 11:47 [Bug testsuite/48283] New: gcc.dg/graphite/block-[3478].c timeouts ro at gcc dot gnu.org
` (2 preceding siblings ...)
2011-03-25 14:06 ` ro at gcc dot gnu.org
@ 2011-05-18 0:01 ` michael.hope at linaro dot org
2013-07-14 5:33 ` spop at gcc dot gnu.org
4 siblings, 0 replies; 6+ messages in thread
From: michael.hope at linaro dot org @ 2011-05-18 0:01 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48283
Michael Hope <michael.hope at linaro dot org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |michael.hope at linaro dot
| |org
--- Comment #4 from Michael Hope <michael.hope at linaro dot org> 2011-05-17 23:36:51 UTC ---
Note that compiling block-4.c on a 1 GHz Cortex-A9 takes 2:30 s on GCC 4.5.2.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug testsuite/48283] gcc.dg/graphite/block-[3478].c timeouts
2011-03-25 11:47 [Bug testsuite/48283] New: gcc.dg/graphite/block-[3478].c timeouts ro at gcc dot gnu.org
` (3 preceding siblings ...)
2011-05-18 0:01 ` michael.hope at linaro dot org
@ 2013-07-14 5:33 ` spop at gcc dot gnu.org
4 siblings, 0 replies; 6+ messages in thread
From: spop at gcc dot gnu.org @ 2013-07-14 5:33 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48283
Sebastian Pop <spop at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |RESOLVED
Resolution|--- |FIXED
--- Comment #5 from Sebastian Pop <spop at gcc dot gnu.org> ---
Works much faster on trunk than what comment#1 reported:
on my ARM laptop exynos5 at 1.6GHz
block-3.c
Graphite loop transforms: 1.61 (71%) usr 0.01 ( 7%) sys 1.68 (57%) wall
61 kB ( 2%) ggc
Graphite data dep analysis: 0.05 ( 2%) usr 0.00 ( 0%) sys 0.05 ( 2%)
wall 0 kB ( 0%) ggc
Graphite code generation: 0.11 ( 5%) usr 0.00 ( 0%) sys 0.11 ( 4%) wall
33 kB ( 1%) ggc
block-4.c
Graphite loop transforms: 1.43 (74%) usr 0.03 (50%) sys 1.82 (78%) wall
39 kB ( 2%) ggc
Graphite data dep analysis: 0.38 (20%) usr 0.00 ( 0%) sys 0.35 (15%)
wall 0 kB ( 0%) ggc
block-5.c
Graphite loop transforms: 0.52 (67%) usr 0.00 ( 0%) sys 0.51 (61%) wall
30 kB ( 1%) ggc
Graphite data dep analysis: 0.02 ( 3%) usr 0.00 ( 0%) sys 0.02 ( 2%)
wall 0 kB ( 0%) ggc
Graphite code generation: 0.02 ( 3%) usr 0.00 ( 0%) sys 0.03 ( 4%) wall
17 kB ( 1%) ggc
block-6.c
Graphite loop transforms: 0.22 (67%) usr 0.00 ( 0%) sys 0.21 (55%) wall
19 kB ( 1%) ggc
block-7.c
Graphite loop transforms: 2.34 (75%) usr 0.01 (11%) sys 2.41 (72%) wall
49 kB ( 2%) ggc
Graphite data dep analysis: 0.43 (14%) usr 0.01 (11%) sys 0.46 (14%)
wall 0 kB ( 0%) ggc
Graphite code generation: 0.14 ( 4%) usr 0.01 (11%) sys 0.15 ( 5%) wall
45 kB ( 2%) ggc
block-8.c
Graphite loop transforms: 2.34 (73%) usr 0.01 (17%) sys 2.37 (72%) wall
59 kB ( 2%) ggc
Graphite data dep analysis: 0.42 (13%) usr 0.00 ( 0%) sys 0.42 (13%)
wall 0 kB ( 0%) ggc
Graphite code generation: 0.13 ( 4%) usr 0.01 (17%) sys 0.13 ( 4%) wall
45 kB ( 2%) ggc
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2013-07-14 5:33 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2011-03-25 11:47 [Bug testsuite/48283] New: gcc.dg/graphite/block-[3478].c timeouts ro at gcc dot gnu.org
2011-03-25 13:00 ` [Bug testsuite/48283] " rguenth at gcc dot gnu.org
2011-03-25 13:02 ` ro at CeBiTec dot Uni-Bielefeld.DE
2011-03-25 14:06 ` ro at gcc dot gnu.org
2011-05-18 0:01 ` michael.hope at linaro dot org
2013-07-14 5:33 ` spop at gcc dot gnu.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).