From: Ramana Radhakrishnan <ramana.radhakrishnan@linaro.org>
To: Jakub Jelinek <jakub@redhat.com>
Cc: Richard Guenther <richard.guenther@gmail.com>,
Marc Glisse <marc.glisse@inria.fr>,
gcc-patches@gcc.gnu.org
Subject: Re: combine permutations in gimple
Date: Wed, 15 Aug 2012 12:21:00 -0000 [thread overview]
Message-ID: <CACUk7=W1_gsCTaWXBjqK1=B_qz9ziwBueY5-EFwaTTd3dBuy7w@mail.gmail.com> (raw)
In-Reply-To: <20120815120703.GB1999@tucnak.redhat.com>
On 15 August 2012 13:07, Jakub Jelinek <jakub@redhat.com> wrote:
> On Wed, Aug 15, 2012 at 01:46:05PM +0200, Richard Guenther wrote:
>> Well, we're waiting for someone to break the tie ... I'd go with the original
>> patch, improving the backends where necessary.
>
> E.g. i?86/x86_64 with just plain -msse2 has only very small subset of
> constant shuffles (and no variable shuffle), so by doing the transformation
> you could end up with a scalarized shuffle instead of two constant vector
> shuffles. Expecting the backend to figure it out and doing the two constant
> vector shuffles in every case is not realistic, i386/x86_64 has way too many
> different shuffling instructions (and worse different ISA levels have
> different subsets of them) and while a lot of effort has been spent on
> it already by Richard, me, Marc and others, we are nowhere close to having
> optimal sequence in many cases for various modes and ISA levels. Doing a
> brute force might be too expensive.
>
Neon has atleast 7-8 such forms. Brute force computing this would be expensive.
I don't know what happens on other vector targets - surely other SIMD
targets in GCC will also have the same problem.
Ramana
> Jakub
next prev parent reply other threads:[~2012-08-15 12:21 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-08-10 21:40 Marc Glisse
2012-08-10 21:48 ` Marc Glisse
2012-08-11 12:25 ` Marc Glisse
2012-08-13 12:23 ` Richard Guenther
2012-08-13 13:04 ` Marc Glisse
2012-08-13 13:10 ` Richard Guenther
2012-08-13 13:29 ` Marc Glisse
2012-08-13 15:43 ` Marc Glisse
2012-08-15 11:22 ` Marc Glisse
2012-08-15 11:46 ` Richard Guenther
2012-08-15 12:09 ` Jakub Jelinek
2012-08-15 12:15 ` Richard Guenther
2012-08-15 12:34 ` Jakub Jelinek
2012-08-15 12:37 ` Richard Guenther
2012-08-15 12:59 ` Jakub Jelinek
2012-08-15 13:09 ` Richard Guenther
2012-08-15 12:21 ` Ramana Radhakrishnan [this message]
2012-08-18 10:00 ` Marc Glisse
2012-08-20 11:35 ` Richard Guenther
2012-08-13 13:12 ` Ramana Radhakrishnan
2012-08-13 13:13 ` Richard Guenther
2012-08-13 13:21 ` Marc Glisse
2012-08-13 14:05 ` Ramana Radhakrishnan
2012-08-13 14:29 ` Marc Glisse
2012-08-15 11:57 ` Ramana Radhakrishnan
2012-08-15 12:06 ` Richard Guenther
2012-08-17 3:03 ` Hans-Peter Nilsson
2012-08-13 13:27 ` Jakub Jelinek
2012-08-13 13:45 ` Marc Glisse
2012-08-13 13:59 ` Jakub Jelinek
2012-08-13 14:12 ` Ramana Radhakrishnan
2012-08-13 14:43 ` Richard Guenther
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='CACUk7=W1_gsCTaWXBjqK1=B_qz9ziwBueY5-EFwaTTd3dBuy7w@mail.gmail.com' \
--to=ramana.radhakrishnan@linaro.org \
--cc=gcc-patches@gcc.gnu.org \
--cc=jakub@redhat.com \
--cc=marc.glisse@inria.fr \
--cc=richard.guenther@gmail.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).