http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #30 from Marc Glisse glisse at gcc dot gnu.org 2012-05-14
20:19:39 UTC ---
Author: glisse
Date: Mon May 14 20:19:30 2012
New Revision: 187479
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=187479
Log:
2012-05-14 Marc Glisse
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #28 from Marc Glisse marc.glisse at normalesup dot org 2012-04-11
16:48:47 UTC ---
A difficulty I hadn't foreseen is that the code that canonicalizes permutations
(and in particular checks if one of the operands is unused) is in
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #29 from Marc Glisse marc.glisse at normalesup dot org 2012-04-11
20:35:00 UTC ---
Created attachment 27136
-- http://gcc.gnu.org/bugzilla/attachment.cgi?id=27136
V4DF generic shuffle
A patch (independent from the others)
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #27 from Marc Glisse marc.glisse at normalesup dot org 2012-04-09
16:50:47 UTC ---
Notes to self (or other):
- Intel's SDE makes it possible to test without appropriate hardware;
- for V4DF shuffles, there seems to be a very simple
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #25 from Marc Glisse marc.glisse at normalesup dot org 2012-03-31
09:37:51 UTC ---
The test for AVX2 in expand_vec_perm_interleave2 might be too strict. For the
V4DF shuffle 4,0,2,6, removing that check lets the compiler generate a
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
Marc Glisse marc.glisse at normalesup dot org changed:
What|Removed |Added
Attachment #26979|0 |1
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
Richard Henderson rth at gcc dot gnu.org changed:
What|Removed |Added
Status|UNCONFIRMED |NEW
Last
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #24 from Marc Glisse marc.glisse at normalesup dot org 2012-03-29
14:19:11 UTC ---
(In reply to comment #23)
(In reply to comment #18)
+ if (!d-testing_p)
+dsecond.target = gen_reg_rtx (dsecond.vmode);
+ dfirst.op1 =
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
Jakub Jelinek jakub at gcc dot gnu.org changed:
What|Removed |Added
Status|NEW |UNCONFIRMED
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #20 from Jakub Jelinek jakub at gcc dot gnu.org 2012-03-27
17:10:00 UTC ---
Thanks for working on this.
I don't like much the calls to ix86_expand_vec_perm_const_1, if you are looking
for exactly two insn permutations, then really
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #21 from Marc Glisse marc.glisse at normalesup dot org 2012-03-27
18:21:39 UTC ---
(In reply to comment #20)
I don't like much the calls to ix86_expand_vec_perm_const_1, if you are
looking
for exactly two insn permutations,
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #22 from Marc Glisse marc.glisse at normalesup dot org 2012-03-27
20:57:16 UTC ---
(In reply to comment #20)
Lastly for each routine it is desirable to think whether it might be useful
for
other vector modes (likely 32-byte only)
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
Marc Glisse marc.glisse at normalesup dot org changed:
What|Removed |Added
Attachment #26938|0 |1
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
Jakub Jelinek jakub at gcc dot gnu.org changed:
What|Removed |Added
Status|UNCONFIRMED |NEW
Last
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #13 from Jakub Jelinek jakub at gcc dot gnu.org 2012-03-20
16:26:08 UTC ---
Author: jakub
Date: Tue Mar 20 16:25:54 2012
New Revision: 185577
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=185577
Log:
PR target/52607
*
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #14 from Jakub Jelinek jakub at gcc dot gnu.org 2012-03-20
16:51:50 UTC ---
Author: jakub
Date: Tue Mar 20 16:51:41 2012
New Revision: 185579
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=185579
Log:
PR target/52607
*
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #15 from Marc Glisse marc.glisse at normalesup dot org 2012-03-20
19:00:32 UTC ---
If I am not mistaken, the V8SF shuffle 22022246 is doable by a vperm2f128 that
takes 01234567 to 01230123, followed by a vshufps (mask 138 maybe). Was
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #16 from Marc Glisse marc.glisse at normalesup dot org 2012-03-20
19:05:22 UTC ---
(In reply to comment #15)
If I am not mistaken, the V8SF shuffle 22022246 is doable by a vperm2f128 that
takes 01234567 to 01230123, followed by a
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
Marc Glisse marc.glisse at normalesup dot org changed:
What|Removed |Added
Attachment #26912|0 |1
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #7 from Jakub Jelinek jakub at gcc dot gnu.org 2012-03-19
11:13:52 UTC ---
Created attachment 26915
-- http://gcc.gnu.org/bugzilla/attachment.cgi?id=26915
gcc48-pr52607.patch
AVX2 changes.
This improves:
1) for {x, x, x, x}
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
Jakub Jelinek jakub at gcc dot gnu.org changed:
What|Removed |Added
CC||jakub at gcc dot
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #9 from Marc Glisse marc.glisse at normalesup dot org 2012-03-19
18:29:50 UTC ---
(In reply to comment #8)
I'm not very keen on having too many different routines, the more generic they
are, the better.
Agreed, that was one of my
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #10 from Jakub Jelinek jakub at gcc dot gnu.org 2012-03-19
19:18:33 UTC ---
Created attachment 26925
-- http://gcc.gnu.org/bugzilla/attachment.cgi?id=26925
gcc48-vselect.patch
Patch to decrease memory cost of
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #11 from Richard Henderson rth at gcc dot gnu.org 2012-03-19
19:36:03 UTC ---
The vselect patch looks pretty good.
+ if (icode = 0 !testing_p)
+x = copy_rtx (PATTERN (vselect_insn));
...
+ if (!testing_p)
+emit_insn (x);
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
Marc Glisse marc.glisse at normalesup dot org changed:
What|Removed |Added
Attachment #26909|0 |1
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #6 from Marc Glisse marc.glisse at normalesup dot org 2012-03-18
18:58:44 UTC ---
Created attachment 26912
-- http://gcc.gnu.org/bugzilla/attachment.cgi?id=26912
generic shuffle of a single v8sf
An additional function (I should
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #2 from Marc Glisse marc.glisse at normalesup dot org 2012-03-17
19:20:36 UTC ---
Created attachment 26908
-- http://gcc.gnu.org/bugzilla/attachment.cgi?id=26908
copy-paste patch for 0213 and 1302
This seems to handle 0213 and 1302
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #3 from Marc Glisse marc.glisse at normalesup dot org 2012-03-17
19:55:18 UTC ---
Uh. I feel silly, but it looks like vshufpd could replace vpermilpd+vblendpd in
many cases, including the original 1230 from PR52568...
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
Marc Glisse marc.glisse at normalesup dot org changed:
What|Removed |Added
Attachment #26908|0 |1
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52607
--- Comment #1 from Marc Glisse marc.glisse at normalesup dot org 2012-03-17
01:05:57 UTC ---
Note that {1,2,0,3} seems harder, I need one extra vpermilpd. Actually, it
looks like every v4df shuffle can be realized as a vblendpd of a vpermilpd
30 matches
Mail list logo