https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98348
--- Comment #14 from Hongtao.liu ---
*** Bug 96891 has been marked as a duplicate of this bug. ***
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98348
--- Comment #13 from Hongtao.liu ---
Created attachment 49832
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49832&action=edit
gcc11-pr98348_v3.patch
1. Use REAL_VALUE_TO_TARGET_SINGLE/DOUBLE in the
"float_vector_all_ones_operands" predica
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98348
--- Comment #12 from Hongtao.liu ---
(In reply to Jakub Jelinek from comment #11)
> I'm not sure about the knot changes, isn't that too risky at least at this
> point?
> I mean, can't we instead just match what knot emits?
>
As indicated in
http
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98348
--- Comment #11 from Jakub Jelinek ---
I'm not sure about the knot changes, isn't that too risky at least at this
point?
I mean, can't we instead just match what knot emits?
As for the new predicate, I think we should check CONST_DOUBLE_AS_FLOAT
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98348
--- Comment #10 from Hongtao.liu ---
Created attachment 49819
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49819&action=edit
Incremental to gcc11-pr99348.patch
Update patch.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98348
--- Comment #9 from Hongtao.liu ---
(In reply to Jakub Jelinek from comment #8)
> Created attachment 49806 [details]
> gcc11-pr98348.patch
>
> So, if we go for GCC11 the way of pre-reload define_insn_and_split, this is
> some incremental unteste
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98348
--- Comment #8 from Jakub Jelinek ---
Created attachment 49806
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49806&action=edit
gcc11-pr98348.patch
So, if we go for GCC11 the way of pre-reload define_insn_and_split, this is
some incrementa
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98348
--- Comment #7 from Jakub Jelinek ---
Seems the adjustment for last UNSPEC_PCMP operand between all_ones, const0 and
const0, all_ones is GEN_INT (INTVAL (operands[3]) ^ 4).
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98348
--- Comment #6 from Jakub Jelinek ---
In the light of the recent discussions I've been wondering about doing it as
combine splitters only, like roughly:
--- sse.md.jj 2020-12-03 10:04:35.862093285 +0100
+++ sse.md 2020-12-19 11:00:14.27214
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98348
--- Comment #5 from Hongtao.liu ---
Created attachment 49804
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49804&action=edit
This patch can fix this issue
I'm testing the patch.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98348
--- Comment #4 from Hongtao.liu ---
(In reply to Jakub Jelinek from comment #3)
> Started with r10-5250-g8b905e9b0c09530c0f660563540257f3d181c2ac
> Perhaps peephole2s or something similar to catch that?
A patch(with peephole2) is posted at
https
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98348
Jakub Jelinek changed:
What|Removed |Added
CC||jakub at gcc dot gnu.org
--- Comment #3
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98348
H.J. Lu changed:
What|Removed |Added
CC||crazylht at gmail dot com,
|
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98348
--- Comment #1 from Daniel Han-Chen ---
I also just noticed that in GCC 10, an extra movdqa is done, which is also not
necessary.
14 matches
Mail list logo
