https://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
Dominique d'Humieres changed:
What|Removed |Added
Status|NEW |RESOLVED
Resolution|---
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
William J. Schmidt changed:
What|Removed |Added
CC||bergner at gcc dot gnu.org
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
William J. Schmidt changed:
What|Removed |Added
CC||wschmidt at gcc dot gnu.org
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #42 from Dominique d'Humieres
2012-11-16 14:42:33 UTC ---
> Fatigue now gets all inlining with -O3 -fwhole-program, with -O3 it gets only
> half of inlining because jump functions are not able to track array
> descriptors
> in
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #41 from Jan Hubicka 2012-11-15
02:28:26 UTC ---
mgrid regression is now PR55334
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #40 from Jan Hubicka 2012-11-14 23:54:44
UTC ---
mgrid do not seem to be sensitive to --param min-inline-speedup, so it seems
independent regression of this change.
No idea what goes wrong.
Honza
mgrid do not seem to be sensitive to --param min-inline-speedup, so it seems
independent regression of this change.
No idea what goes wrong.
Honza
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #39 from Jan Hubicka 2012-11-14
23:22:40 UTC ---
Hmm, indeed. Good catch. I will look into it.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
Igor Zamyatin changed:
What|Removed |Added
CC||izamyatin at gmail dot com
---
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #37 from Jan Hubicka 2012-11-12
12:16:18 UTC ---
Fatigue now gets all inlining with -O3 -fwhole-program, with -O3 it gets only
half of inlining because jump functions are not able to track array descriptors
in both calls to gen
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #36 from Jan Hubicka 2012-11-11
18:14:40 UTC ---
Author: hubicka
Date: Sun Nov 11 18:14:35 2012
New Revision: 193406
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=193406
Log:
PR middle-end/48636
* ipa-in
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #35 from Jan Hubicka 2012-11-08
16:46:28 UTC ---
Author: hubicka
Date: Thu Nov 8 16:46:18 2012
New Revision: 193331
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=193331
Log:
PR middle-end/48636
* ipa-inli
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #34 from Jan Hubicka 2012-11-07
11:17:28 UTC ---
Created attachment 28629
--> http://gcc.gnu.org/bugzilla/attachment.cgi?id=28629
Array index hint
This patch should help to inline when array descriptors become known, such
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #33 from Jan Hubicka 2012-11-07
09:34:25 UTC ---
Created attachment 28628
--> http://gcc.gnu.org/bugzilla/attachment.cgi?id=28628
Final patch (I hope)
This is version of path I will commit today or tomorrow (depending on w
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #32 from vincenzo Innocente
2012-10-28 11:27:22 UTC ---
In a small test (that I will eventually publish here) the new patch at -O2
looks superior to 4.7.2 at O3.
I would like to build a test with multiple source files where lto matte
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #31 from Jan Hubicka 2012-10-28
10:11:13 UTC ---
Concerning vincenzo's request about 4.7 version, it won't work - it depends on
improvements of inline metric and ipa-prop we made for 4.8
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #30 from Jan Hubicka 2012-10-28
10:08:23 UTC ---
Created attachment 28543
--> http://gcc.gnu.org/bugzilla/attachment.cgi?id=28543
Updated patch
Hi,
this is updated patch I am testing. It fixes the big speedup test and als
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #29 from Thomas Koenig 2012-10-20
12:10:49 UTC ---
Another approach (not for the benchmarks) would be to
make inlining tunable by the user, e.g. support
!GCC$ ATTRIBUTES always_inline :: procedure_name
See PR 41209.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #28 from Dominique d'Humieres
2012-10-20 11:22:16 UTC ---
If I understand correctly the patch, the default value for
max-inline-min-speedup is 20. This could be over-agressive: for fatigue.f90 the
threshold is between 94 (fast)
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #27 from Jan Hubicka 2012-10-20
10:34:58 UTC ---
Thank you for testing. It seems that the patch works well for small benchmarks,
I will look into lapack/test_fpu slowdown.
There is problem that it really causes inacceptable gro
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #26 from vincenzo Innocente
2012-10-19 08:45:03 UTC ---
I'm interested to test the patch on our large application currently compiled
with 4.7.2.
would it be possible to get the same patch against gcc-4_7-branch?
thanks
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #25 from Dominique d'Humieres
2012-10-17 14:05:51 UTC ---
> I also see many failures for the gcc.dg/tree-ssa/slsr-* tests: slsr-2.c to
> slsr-11.c, slsr-14.c to slsr-20.c, slsr-24.c, and slsr-25.c, and for
> gfortran.dg/vect/ve
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #24 from Dominique d'Humieres
2012-10-17 13:13:24 UTC ---
Summary for the polyhedron tests (pb05):
(a) revision 192449 unpatched
(b) revision 192516 with patch in comment #20
options: -fprotect-parens -Ofast -funroll-loops -
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
Jakub Jelinek changed:
What|Removed |Added
CC||jakub at gcc dot gnu.org
--- Co
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #22 from Dominique d'Humieres
2012-10-16 20:58:58 UTC ---
With the patch I see a ~10% slowdown in the Test4 - Lapack 2 (1001x1001) of
test_fpu.f90 compared to revision 192449
[macbook] lin/test% time /opt/gcc/gcc4.8c/bin/gfor
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #21 from Dominique d'Humieres
2012-10-16 17:57:52 UTC ---
Before the patch in comment #20, I get
-rwxr-xr-x 1 dominiq staff 73336 Oct 16 19:19 a.out*
[macbook] lin/test% time gfc -fprotect-parens -Ofast -funroll-loops
-ftree
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #20 from Jan Hubicka 2012-10-16
16:38:27 UTC ---
Created attachment 28456
--> http://gcc.gnu.org/bugzilla/attachment.cgi?id=28456
Path I am considering
Hi,
I am considering to enable inlining when inline-analysis says tha
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #19 from Jan Hubicka 2012-09-12
21:51:21 UTC ---
Author: hubicka
Date: Wed Sep 12 21:51:14 2012
New Revision: 191232
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=191232
Log:
PR fortran/48636
* gcc.dg/ipa/inlinehint-
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #18 from Jan Hubicka 2012-08-21
08:14:33 UTC ---
With loop_iterations hint, we should now hint the bar function of testcase in
comment #4, but we don't because the value is used conditionally:
# iftmp.11_3 = PHI <_12(3), 1(2)>
a.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #17 from Jan Hubicka 2012-08-21
06:54:09 UTC ---
Author: hubicka
Date: Tue Aug 21 06:54:01 2012
New Revision: 190556
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=190556
Log:
PR fortran/48636
* ipa-inline.c (want_inli
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #16 from Martin Jambor 2012-08-11
10:50:29 UTC ---
Author: jamborm
Date: Sat Aug 11 10:50:24 2012
New Revision: 190313
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=190313
Log:
2012-08-11 Martin Jambor
PR fortran/4863
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #15 from Martin Jambor 2012-07-03
17:43:35 UTC ---
Hi,
(In reply to comment #12)
> Hi,
> I discussed some of the issues today with Martin. For the array descriptor
> testcase, we really want ipa-cp to be propagate the constant array
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #14 from Jan Hubicka 2011-06-04
18:06:01 UTC ---
Yeah, the fnspec issue is something we ought to solve. ipa-cp should be
effective on fortran so it should not disable itself there ;)
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #13 from Tobias Burnus 2011-05-04
17:30:31 UTC ---
(In reply to comment #12)
> Perhaps frontend could help us here since the descriptors are probably
> constant after they are initialized, or is there way to change existing
> descript
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #12 from Jan Hubicka 2011-05-04
16:09:19 UTC ---
Hi,
I discussed some of the issues today with Martin. For the array descriptor
testcase, we really want ipa-cp to be propagate the constant array bounds
instead of making Inliner to bl
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #11 from Janne Blomqvist 2011-04-20
18:14:20 UTC ---
(In reply to comment #9)
> > But do we actually do this? I did some tests a while ago, and IIRC for
> > assumed
> > shape dummy arguments the procedure always calculates new bounds
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #10 from Thomas Koenig 2011-04-20
16:40:46 UTC ---
(In reply to comment #6)
> Not strictly related to inlining, but in the new descriptor we'll have a field
> specifying whether the array is simply contiguous, so it might make sense
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #9 from Tobias Burnus 2011-04-20
15:39:47 UTC ---
> But do we actually do this? I did some tests a while ago, and IIRC for assumed
> shape dummy arguments the procedure always calculates new bounds such that
> they
> start from 1. Th
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #8 from Janne Blomqvist 2011-04-20 13:09:51
UTC ---
(In reply to comment #7)
> (In reply to comment #6)
> > > Here is some sample code (extreme, I admit) which profits a lot from
> > > inlining:
> > >
> > > - Strides are known to be
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #7 from Tobias Burnus 2011-04-20
12:29:02 UTC ---
(In reply to comment #6)
> > Here is some sample code (extreme, I admit) which profits a lot from
> > inlining:
> >
> > - Strides are known to be one when inlining (a common case, but
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
Janne Blomqvist changed:
What|Removed |Added
CC||jb at gcc dot gnu.org
--- Comment #6 fr
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #5 from Dominique d'Humieres 2011-04-17
14:12:30 UTC ---
I have investigated why test_fpu is slower with --param
max-inline-insns-auto=400 (11.18s) compared to -finline-limit=600 (10.84s) in
the timings of comment #2. This is due to t
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #4 from Thomas Koenig 2011-04-17
13:32:11 UTC ---
(In reply to comment #3)
> The second item is interesting - it would be cool if backend was able to work
> out that the code is supposed to simplify after inlining. Either by itself o
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #3 from Jan Hubicka 2011-04-17
10:44:23 UTC ---
I am slowly starting to look into fortran issues now. For years it was
non-issue since we had the non-one-decl-per-function problem. This is finally
solved
One additional problem is th
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #2 from Dominique d'Humieres 2011-04-17
10:23:03 UTC ---
As shown by the following results it seems that --param max-inline-insns-auto=*
is the way to go.
Date & Time : 17 Apr 2011 11:22:05
Test Name : pbharness
Compile Com
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
Richard Guenther changed:
What|Removed |Added
Status|UNCONFIRMED |NEW
Last reconfirmed|
46 matches
Mail list logo