http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57071
--- Comment #13 from Thomas Koenig ---
Sorry, wrong PR for the comment.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57071
--- Comment #12 from Thomas Koenig ---
Created attachment 30141
--> http://gcc.gnu.org/bugzilla/attachment.cgi?id=30141&action=edit
Another patch that doesn't work...
I tried to follow the advice in comment#9, but I hit a wall
(again).
With th
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57071
Thomas Koenig changed:
What|Removed |Added
Status|ASSIGNED|RESOLVED
Resolution|
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57071
--- Comment #10 from Thomas Koenig 2013-04-30
21:46:11 UTC ---
Author: tkoenig
Date: Tue Apr 30 21:45:13 2013
New Revision: 198476
URL: http://gcc.gnu.org/viewcvs?rev=198476&root=gcc&view=rev
Log:
2013-04-30 Thomas Koenig
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57071
--- Comment #9 from Thomas Koenig 2013-04-28
13:34:08 UTC ---
Author: tkoenig
Date: Sun Apr 28 13:32:59 2013
New Revision: 198369
URL: http://gcc.gnu.org/viewcvs?rev=198369&root=gcc&view=rev
Log:
2013-04-28 Thomas Koenig
PR
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57071
--- Comment #8 from Thomas Koenig 2013-04-27
23:19:13 UTC ---
While special-casing power, it might also make sense
to translate 2**n to ishft(1,n).
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57071
Thomas Koenig changed:
What|Removed |Added
Status|UNCONFIRMED |ASSIGNED
Last reconfirmed|
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57071
--- Comment #6 from Richard Biener 2013-04-26
09:16:41 UTC ---
Note that for the (-1.0)**k case __builtin_powif (-1.0e+0, k) should be
perfectly fine for the middle-end. _gfortran_pow_i4_i4 (-1, k) is of
course unfortunate and should be o
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57071
--- Comment #5 from Tobias Burnus 2013-04-26
07:26:31 UTC ---
(In reply to comment #3)
> 1 - ISHFT(IAND(K,1),1)
For the real version Jakub suspects that (k & 1) ? -1.0 : 1.0 is faster than
the mod/convert to float/subtraction or the
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57071
--- Comment #4 from Joost VandeVondele
2013-04-26 07:12:04 UTC ---
(In reply to comment #3)
> As James Van Buskirk pointed out, the algorithm will fail if k < 0.
note that in the case of k being a loop index, there will be pretty good r
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57071
--- Comment #3 from Tobias Burnus 2013-04-26
07:07:22 UTC ---
As James Van Buskirk pointed out, the algorithm will fail if k < 0. Thus, he
suggests, which gives the expected result:
1 - ISHFT(IAND(K,1),1)
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57071
Joost VandeVondele changed:
What|Removed |Added
CC||Joost.VandeVondele at mat
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57071
Tobias Burnus changed:
What|Removed |Added
CC||burnus at gcc dot gnu.org
--- C
13 matches
Mail list logo