https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
Martin Liška changed:
What|Removed |Added
CC||marxin at gcc dot gnu.org
--- Comment #55
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #54 from rsandifo at gcc dot gnu.org
---
Author: rsandifo
Date: Thu Apr 4 07:35:34 2019
New Revision: 270142
URL: https://gcc.gnu.org/viewcvs?rev=270142&root=gcc&view=rev
Log:
DF usage in loop-invariant.c (PR46590)
- df_live is al
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #53 from Richard Biener ---
After both are fixed there's still non-linearity from loop-invariant, this
time in the DF RD problem.
Doubling the testcase goes from 9.1s to 25.5s, doubling again to 82.2s
with RD going from
df reachin
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #52 from Richard Biener ---
Author: rguenth
Date: Mon Apr 1 11:36:25 2019
New Revision: 270055
URL: https://gcc.gnu.org/viewcvs?rev=270055&root=gcc&view=rev
Log:
2019-04-01 Richard Biener
PR tree-optimization/46590
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #51 from Richard Biener ---
Alternative patch, the first one doesn't seem to work (probably
all_blocks is sometimes NULL, will debug).
Index: gcc/df.h
===
--- gcc/df.h
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
Richard Biener changed:
What|Removed |Added
CC||ebotcazou at gcc dot gnu.org,
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #49 from Richard Biener ---
I have a fix for the copy-headers issue. Afterwards:
tree copy headers : 0.03 ( 0%) 0.01 ( 3%) 0.03 ( 0%)
4763 kB ( 3%)
with the remaining worst offenders being
df live&i
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #48 from Thomas Koenig ---
The test case from comment#5 and comment#6 has regressed for M7/8/9:
$ time gfortran-4.8 -O1 gener-4.f90
real0m11.509s
user0m11.356s
sys 0m0.148s
$ time gfortran-7 -O1 gener-4.f90
real
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #46 from Jeffrey A. Law ---
Author: law
Date: Thu Dec 10 19:40:23 2015
New Revision: 231542
URL: https://gcc.gnu.org/viewcvs?rev=231542&root=gcc&view=rev
Log:
PR tree-optimization/61515
PR tree-optimization/46590
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
Jeffrey A. Law changed:
What|Removed |Added
CC||law at redhat dot com
--- Comment #47 f
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
Bug 46590 depends on bug 61515, which changed state.
Bug 61515 Summary: [4.9 Regression] Extremely long compile time for generated
code
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61515
What|Removed |Added
---
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
Richard Biener changed:
What|Removed |Added
CC||law at gcc dot gnu.org
--- Comment #45
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #44 from Richard Biener ---
Author: rguenth
Date: Fri Jan 17 14:49:18 2014
New Revision: 206714
URL: http://gcc.gnu.org/viewcvs?rev=206714&root=gcc&view=rev
Log:
2014-01-17 Richard Biener
PR tree-optimization/46590
* opts.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #43 from Richard Biener ---
Author: rguenth
Date: Fri Jan 17 14:40:11 2014
New Revision: 206709
URL: http://gcc.gnu.org/viewcvs?rev=206709&root=gcc&view=rev
Log:
2014-01-17 Richard Biener
PR tree-optimization/46590
* vec.h
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #42 from Richard Biener ---
Another interesting bit - -O0 takes 17s (and also ~1GB memory) but -Og is
not much faster than -O1 (116s).
> ./f951 -quiet -Og t2.f90 -ftime-report
Execution times (seconds)
df reaching defs: 62
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #41 from Richard Biener ---
After I will have committed the patch the original testcase will use about 1GB
of memory (regardless of optimization level) on x86_64, 122s at -O1 and 161s
at -Ofast with the following main contributors:
>
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #40 from Richard Biener ---
On the full testcase tree LIM uses too much memory (I didn't merge some of
the patches that only benefit this kind of testcase ... bah). Without LIM
we use around 1GB of memory at -O2 and
df reaching defs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #39 from Richard Biener ---
Author: rguenth
Date: Thu Jan 16 13:48:51 2014
New Revision: 206663
URL: http://gcc.gnu.org/viewcvs?rev=206663&root=gcc&view=rev
Log:
2014-01-16 Richard Biener
PR rtl-optimization/46590
* lcm.c
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
Richard Biener changed:
What|Removed |Added
Blocks||47344
Target Milestone|4.6.4
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #6 from Thomas Koenig 2010-11-21
19:48:41 UTC ---
Without -fno-ivopts:
i...@linux-fd1f:~/Krempel/Dep-c> time
~/libexec/gcc/x86_64-unknown-linux-gnu/4.6.0/f951 -ftime-report -O2
gener-4.f90
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #7 from Thomas Koenig 2010-11-21
21:07:08 UTC ---
Created attachment 22478
--> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22478
Test case from comment #5 and comment #6
Here is a test case that is a little shorter.
A bit hard t
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #4 from Richard Guenther 2010-11-21
18:56:58 UTC ---
Please strip down the testcase and provide -ftime-report output (I suppose
you hit IVOPTs slowness, so try -fno-ivopts).
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #5 from Thomas Koenig 2010-11-21
19:39:23 UTC ---
With -fno-ivopts, the alias statement walking is the clear culprit
(a somewhat shorter test case):
i...@linux-fd1f:~/Krempel/Dep-c> time
~/libexec/gcc/x86_64-unknown-linux-gnu/4.6.0/f
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
Thomas Koenig changed:
What|Removed |Added
Keywords||memory-hog
--- Comment #3 from Thomas Koe
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #2 from Dominique d'Humieres 2010-11-21
16:59:39 UTC ---
On my laptop the test in comment #1 takes ~4mn to compile and at -O1 I had to
interrupt the compilation after ~20mn during which all the time was spent in
memory pagination!-(wi
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #1 from Thomas Koenig 2010-11-21
13:22:50 UTC ---
Created attachment 22475
--> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22475
gzipped test case
Test case.
26 matches
Mail list logo
