From: Richard Biener <richard.guenther@gmail.com>
To: Alan Lawrence <alan.lawrence@arm.com>
Cc: "gcc-patches@gcc.gnu.org" <gcc-patches@gcc.gnu.org>,
David Edelsohn <dje.gcc@gmail.com>,
Aldy Hernandez <aldyh@redhad.com>,
Steve Ellcey <sellcey@mips.com>,
Eric Christopher <echristo@gmail.com>
Subject: Re: [PATCH 12/14][Vectorizer] Redefine VEC_RSHIFT_EXPR and vec_shr_optab as endianness-neutral
Date: Mon, 22 Sep 2014 10:58:00 -0000 [thread overview]
Message-ID: <CAFiYyc21jou9bztObRdw6J8m6_T5qYFKaO1wr=obbQ3GwNqqTQ@mail.gmail.com> (raw)
In-Reply-To: <541AD353.6020405@arm.com>
On Thu, Sep 18, 2014 at 2:42 PM, Alan Lawrence <alan.lawrence@arm.com> wrote:
> The direction of VEC_RSHIFT_EXPR has been endian-dependent, contrary to the
> general principles of tree. This patch updates fold-const and the vectorizer
> (the only place where such expressions are created), such that
> VEC_RSHIFT_EXPR always shifts towards element 0.
>
> The tree code still maps directly onto the vec_shr_optab, and so this patch
> *will break any bigendian platform defining the vec_shr optab*.
> --> For AArch64_be, patch follows next in series;
> --> For PowerPC, I think patch/rfc 15 should fix, please inspect;
> --> For MIPS, I think patch/rfc 16 should fix, please inspect.
>
> gcc/ChangeLog:
>
> * fold-const.c (const_binop): VEC_RSHIFT_EXPR always shifts towards
> element 0.
>
> * tree-vect-loop.c (vect_create_epilog_for_reduction): always
> extract
> the result of a reduction with vector shifts from element 0.
>
> * tree.def (VEC_RSHIFT_EXPR, VEC_LSHIFT_EXPR): Comment shift
> direction.
>
> * doc/md.texi (vec_shr_m, vec_shl_m): Document shift direction.
>
> Testing Done:
>
> Bootstrap and check-gcc on x86_64-none-linux-gnu; check-gcc on
> aarch64-none-elf.
As said elsewhere I'd like the vectorizer to use VEC_PERM_EXPRs
and the generic vec_perm expansion machinery handle the
case where the permute can be expressed using the vec_shr_optab.
You'd have, for a 1-element shift of V4SI x, VEC_PERM <x, { 0, 0, 0, 0
}, {4, 3, 2, 1 }>
I'd say that if the target says it can handle the constant permute just fine
then use the vec_perm_const expansion path.
Richard.
next prev parent reply other threads:[~2014-09-22 10:58 UTC|newest]
Thread overview: 52+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-09-18 11:41 [PATCH 0/14+2][Vectorizer] Made reductions endianness-neutral, fixes PR/61114 Alan Lawrence
2014-09-18 11:45 ` [PATCH 1/14][AArch64] Temporarily remove aarch64_gimple_fold_builtin code for reduction operations Alan Lawrence
2014-09-24 9:41 ` Marcus Shawcroft
2014-09-18 11:51 ` [PATCH 2/14][Vectorizer] Make REDUC_xxx_EXPR tree codes produce a scalar result Alan Lawrence
2014-09-22 10:34 ` Richard Biener
2014-09-22 13:23 ` Alan Lawrence
2014-09-24 15:02 ` Alan Lawrence
2014-09-24 18:08 ` Segher Boessenkool
2014-09-25 16:07 ` Alan Lawrence
2014-09-18 11:54 ` [PATCH 3/14] Add new optabs for reducing vectors to scalars Alan Lawrence
2014-09-22 10:40 ` Richard Biener
2014-09-22 13:26 ` Alan Lawrence
2014-09-22 13:38 ` Richard Biener
2014-09-25 14:33 ` [PATCH/RFC v2 " Alan Lawrence
2014-09-25 15:31 ` Richard Biener
2014-09-25 16:12 ` Alan Lawrence
2014-09-25 19:20 ` Segher Boessenkool
2014-09-18 11:59 ` [PATCH 4/14][AArch64] Use new reduc_plus_scal optabs, inc. for __builtins Alan Lawrence
2014-09-24 9:44 ` Marcus Shawcroft
2014-09-18 12:02 ` [PATCH 5/14][AArch64] Use new reduc_[us](min|max)_scal optabs, inc. for builtins Alan Lawrence
2014-09-24 9:47 ` Marcus Shawcroft
2014-09-18 12:05 ` [PATCH 6/14][AArch64] Restore gimple_folding of reduction intrinsics Alan Lawrence
2014-09-24 9:48 ` Marcus Shawcroft
2014-09-18 12:19 ` [PATCH 7/14][Testsuite] Add tests of reductions using whole-vector-shifts (multiplication) Alan Lawrence
2014-09-22 10:41 ` Richard Biener
2014-09-18 12:25 ` [PATCH 8/14][Testsuite] Add tests of reductions using whole-vector-shifts (ior) Alan Lawrence
2014-09-22 10:42 ` Richard Biener
2014-09-18 12:27 ` [PATCH 9/14] Enforce whole-vector-shifts to always be by a whole number of elements Alan Lawrence
2014-09-22 10:50 ` Richard Biener
2014-09-18 12:34 ` [PATCH 10/14][AArch64] Implement vec_shr optab Alan Lawrence
2014-09-18 12:35 ` [PATCH 11/14] Remove VEC_LSHIFT_EXPR and vec_shl_optab Alan Lawrence
2014-09-22 10:52 ` Richard Biener
2014-10-27 18:45 ` Alan Lawrence
2014-10-27 20:24 ` Richard Biener
2014-09-18 12:43 ` [PATCH 12/14][Vectorizer] Redefine VEC_RSHIFT_EXPR and vec_shr_optab as endianness-neutral Alan Lawrence
2014-09-18 13:12 ` David Edelsohn
2014-09-22 13:27 ` Bill Schmidt
2014-09-22 10:58 ` Richard Biener [this message]
2014-09-18 12:45 ` [PATCH 13/14][AArch64_be] Fix vec_shr pattern to correctly implement endianness-neutral optab Alan Lawrence
2014-09-22 10:52 ` Richard Biener
2014-09-18 12:48 ` [PATCH 14/14][Vectorizer] Tidy up vect_create_epilog / use_scalar_result Alan Lawrence
2014-09-22 10:53 ` Richard Biener
2014-11-14 17:29 ` PUSHED: " Alan Lawrence
2014-09-18 12:58 ` [PATCH/RFC 15 / 14+2][RS6000] Remove vec_shl and (hopefully) fix vec_shr Alan Lawrence
2014-09-23 12:50 ` David Edelsohn
2014-09-18 13:02 ` [PATCH 16 / 14+2][MIPS] " Alan Lawrence
2014-09-22 11:21 ` [PATCH 0/14+2][Vectorizer] Made reductions endianness-neutral, fixes PR/61114 Richard Biener
2014-09-22 11:26 ` Richard Biener
2014-10-06 17:31 ` Alan Lawrence
[not found] ` <5432D1A5.6080208@arm.com>
2014-10-07 7:45 ` Richard Biener
2014-10-07 7:46 ` Richard Biener
[not found] ` <5436C138.50208@arm.com>
2014-10-09 17:13 ` Alan Lawrence
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CAFiYyc21jou9bztObRdw6J8m6_T5qYFKaO1wr=obbQ3GwNqqTQ@mail.gmail.com' \
--to=richard.guenther@gmail.com \
--cc=alan.lawrence@arm.com \
--cc=aldyh@redhad.com \
--cc=dje.gcc@gmail.com \
--cc=echristo@gmail.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=sellcey@mips.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).