[Bug target/46353] [4.6 regression] fma testsuite failures

2010-11-22 Thread jakub at gcc dot gnu.org 
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46353

Jakub Jelinek jakub at gcc dot gnu.org changed:

   What|Removed |Added

 CC||jakub at gcc dot gnu.org

--- Comment #5 from Jakub Jelinek jakub at gcc dot gnu.org 2010-11-22 
10:49:46 UTC ---
I don't see any powerpc/ppc-fma* failures in
http://gcc.gnu.org/ml/gcc-testresults/2010-11/msg01775.html
Has this been resolved?

[Bug target/46353] [4.6 regression] fma testsuite failures

2010-11-22 Thread meissner at gcc dot gnu.org 
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46353

Michael Meissner meissner at gcc dot gnu.org changed:

   What|Removed |Added

 Status|NEW |RESOLVED
 Resolution||FIXED

--- Comment #6 from Michael Meissner meissner at gcc dot gnu.org 2010-11-22 
16:17:44 UTC ---
I believe Richard Henderson's change on November 16th is the change that fixed
this bug.

[Bug target/46353] [4.6 regression] fma testsuite failures

2010-11-15 Thread rguenth at gcc dot gnu.org 
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46353

Richard Guenther rguenth at gcc dot gnu.org changed:

   What|Removed |Added

   Priority|P3  |P1

[Bug target/46353] [4.6 regression] fma testsuite failures

2010-11-10 Thread meissner at gcc dot gnu.org 
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46353

--- Comment #4 from Michael Meissner meissner at gcc dot gnu.org 2010-11-11 
00:19:44 UTC ---
Created attachment 22369
  -- http://gcc.gnu.org/bugzilla/attachment.cgi?id=22369
Patch to fix the FMA instructions.

[Bug target/46353] [4.6 regression] fma testsuite failures

2010-11-08 Thread ubizjak at gmail dot com 
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46353

Uros Bizjak ubizjak at gmail dot com changed:

   What|Removed |Added

 Status|UNCONFIRMED |NEW
   Last reconfirmed||2010.11.08 08:09:05
 Ever Confirmed|0   |1

--- Comment #1 from Uros Bizjak ubizjak at gmail dot com 2010-11-08 08:09:05 
UTC ---
__builtin_fma is now emitted as FMA_EXPR.

BTW: -ffp-contract=off didn't help to prevent this transformation.

[Bug target/46353] [4.6 regression] fma testsuite failures

2010-11-08 Thread ubizjak at gmail dot com 
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46353

Uros Bizjak ubizjak at gmail dot com changed:

   What|Removed |Added

 Status|NEW |UNCONFIRMED
 Ever Confirmed|1   |0

--- Comment #2 from Uros Bizjak ubizjak at gmail dot com 2010-11-08 08:13:40 
UTC ---
(In reply to comment #1)
 __builtin_fma is now emitted as FMA_EXPR.

Oops, this comment was intended for problems with [1]

FAIL: gcc.dg/torture/builtin-math-2.c  -O0  scan-tree-dump-times original fma
 3

[1] http://gcc.gnu.org/ml/gcc-testresults/2010-11/msg00594.html

[Bug target/46353] [4.6 regression] fma testsuite failures

2010-11-08 Thread meissner at gcc dot gnu.org 
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46353

Michael Meissner meissner at gcc dot gnu.org changed:

   What|Removed |Added

 Status|UNCONFIRMED |NEW
   Last reconfirmed|2010-11-08 08:09:05 |2010.11.09 00:34:40
 Ever Confirmed|0   |1

--- Comment #3 from Michael Meissner meissner at gcc dot gnu.org 2010-11-09 
00:34:40 UTC ---
Now that the tree code knows about FMA, the rs6000 backend needs to add
fmav4df4, fmav2df4, etc. named patterns so the vectorizer can properly
vectorize calls to fma, fmas, etc.

[Bug target/46353] [4.6 regression] fma testsuite failures

2010-11-07 Thread rguenth at gcc dot gnu.org 
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46353

Richard Guenther rguenth at gcc dot gnu.org changed:

   What|Removed |Added

   Target Milestone|--- |4.6.0