https://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
Steven Bosscher changed:
What|Removed |Added
Keywords||missed-optimization
Status
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
Joost VandeVondele changed:
What|Removed |Added
CC||Joost.VandeVondele at mat dot
ethz
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
vincenzo Innocente changed:
What|Removed |Added
Known to fail||4.9.1
--- Comment #27 from vincenzo
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #26 from vincenzo Innocente
2011-11-30 09:13:08 UTC ---
using a minimal configuration like
./configure --prefix=/something --enable-languages=c,c++ --disable-multilib
--disable-bootstrap --enable-lto -enable-gold=yes -disable-libitm
I
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #25 from vincenzo Innocente
2011-11-29 15:24:02 UTC ---
Thanks Jakub for the new patch.
Using it I get this: any Idea how to fix/avoid?
home/data/newsoft/gcc-trunk/host-x86_64-unknown-linux-gnu/gcc/xgcc
-B/home/data/newsoft/gcc-trunk
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #24 from Jakub Jelinek 2011-11-29
14:46:59 UTC ---
Created attachment 25945
--> http://gcc.gnu.org/bugzilla/attachment.cgi?id=25945
gcc47-pr50374.patch
This updated patch applies (had to remove various parts from the older patch
th
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #23 from vincenzo Innocente
2011-11-29 10:10:06 UTC ---
a working patch could be useful to check performance and quality of the
generated ASM w.r.t. manually crafted soutions
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #22 from Jakub Jelinek 2011-11-29
09:58:40 UTC ---
I gave up on it, because it still needed lots of changes in the generic
vectorizer. Guess I can update the patch so that it at least applies, but
there is more work on it.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #21 from vincenzo Innocente
2011-11-29 09:54:49 UTC ---
My understanding is that we cannot count to see a "min/max location pattern"
vectorization in 4.7, isn't it?
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #20 from Jakub Jelinek 2011-09-22
15:52:11 UTC ---
The reason why many of the loops in *-12.c aren't vectorized is that for
idxtype other than unsigned int which is the type of i there is a cast from i
to the type of pos.
OT, one big
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
Jakub Jelinek changed:
What|Removed |Added
Attachment #25333|0 |1
is obsolete|
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #18 from Ira Rosen 2011-09-22 07:51:35 UTC
---
(In reply to comment #16)
> and -3.c fails with an ICE in the vectorizer, Ira,
> could you look at that?
--- tree-vect-stmts.c 2011-09-22 09:48:34.0 +0200
+++ tree-vect-stmts
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #17 from Jakub Jelinek 2011-09-21
16:57:08 UTC ---
Seems my reduction patterns just give some extreme's index, but not necessarily
the first or last extreme's index. It was just narrowing down the index vector
together with the compa
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
Jakub Jelinek changed:
What|Removed |Added
Attachment #25331|0 |1
is obsolete|
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #15 from Ira Rosen 2011-09-21 12:20:47 UTC
---
(In reply to comment #14)
> Another thing is that this really ought to work even with -ftree-pre, having a
> vectorization that requires users to disable PRE would be weird. I believe
>
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
Jakub Jelinek changed:
What|Removed |Added
Attachment #25320|0 |1
is obsolete|
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #13 from Jakub Jelinek 2011-09-20
14:01:55 UTC ---
Ira, thanks for your help, I'll continue working on this now.
To vincenzo, this patch isn't finished, so it is premature to test it.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #12 from vincenzo Innocente
2011-09-20 13:46:16 UTC ---
I'm getting these errors
../.././gcc/optabs.c: In function 'optab_d* optab_for_tree_code(tree_code,
const_tree, optab_subtype)':
../.././gcc/optabs.c:470:9: error: cannot conver
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
Ira Rosen changed:
What|Removed |Added
Attachment #25323|0 |1
is obsolete|
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
Ira Rosen changed:
What|Removed |Added
Attachment #25322|0 |1
is obsolete|
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #9 from vincenzo Innocente
2011-09-20 12:05:01 UTC ---
does not compile to me
../.././gcc/tree-vect-loop.c: In function 'vect_is_simple_reduction_1':
../.././gcc/tree-vect-loop.c:2237:35: warning: suggest parentheses around '&&'
with
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #8 from Ira Rosen 2011-09-20 11:47:00 UTC
---
Created attachment 25323
--> http://gcc.gnu.org/bugzilla/attachment.cgi?id=25323
complete patch including my fix
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #7 from Ira Rosen 2011-09-20 11:45:53 UTC
---
Created attachment 25322
--> http://gcc.gnu.org/bugzilla/attachment.cgi?id=25322
fix
Here is the fix (it's a diff relative to your patch).
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #6 from Ira Rosen 2011-09-20 08:27:08 UTC
---
Thanks for working on this!
It looks like the problem is with the way the stmts are marked. We don't insert
pattern stmts now, so the things are more tricky. I'll try to fix this.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #5 from Jakub Jelinek 2011-09-19
15:59:06 UTC ---
Created attachment 25320
--> http://gcc.gnu.org/bugzilla/attachment.cgi?id=25320
gcc47-pr50374-wip.patch
Here is the ported patch with rejects hopefully resolved and a few bugfixes
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #4 from Jakub Jelinek 2011-09-19
12:51:59 UTC ---
I've looked briefly at the
http://gcc.gnu.org/ml/gcc-patches/2010-08/msg00631.html
patch, it seems the UNSPEC_REDUC* is unnecessary there (only used in the
expanders ending with DONE,
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
--- Comment #3 from vincenzo Innocente
2011-09-13 08:45:40 UTC ---
with
gcc version 4.7.0 20110910 (experimental) (GCC)
int lmin(float const * __restrict__ c, int N) {
int k=0;
for (int i=1; i!=N; ++i)
k = (c[k] > c[i]) ? i : k;
re
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
Jakub Jelinek changed:
What|Removed |Added
CC||jakub at gcc dot gnu.org
--- Comment #2 f
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50374
Ira Rosen changed:
What|Removed |Added
CC||irar at il dot ibm.com
--- Comment #1 from Ir
29 matches
Mail list logo
