From: Jakub Jelinek <jakub@redhat.com>
To: Richard Biener <richard.guenther@gmail.com>
Cc: liuhongt <hongtao.liu@intel.com>, GCC Patches <gcc-patches@gcc.gnu.org>
Subject: Re: [PATCH] Enable auto-vectorization at O2 with very-cheap cost model.
Date: Mon, 6 Sep 2021 11:41:27 +0200 [thread overview]
Message-ID: <20210906094127.GQ920497@tucnak> (raw)
In-Reply-To: <CAFiYyc0NZDnAki3nAi_G6tETHXqMQC3oOVXedyCZbDmb4T4C4w@mail.gmail.com>
On Mon, Sep 06, 2021 at 11:18:47AM +0200, Richard Biener wrote:
> On Mon, Sep 6, 2021 at 10:47 AM liuhongt via Gcc-patches
> <gcc-patches@gcc.gnu.org> wrote:
> >
> > Hi:
> > As discussed in [1], most of (currently unopposed) targets want
> > auto-vectorization at O2, and IMHO now would be a good time to enable O2
> > vectorization for GCC trunk, so it would leave enough time to expose
> > related issues and fix them.
> >
> > Bootstrapped and regtested on x86_64-linux-gnu{-m32,}
> > Ok for trunk?
>
> It changes the cost model used when the user specifices
> -O2 -ftree-vectorize which used 'cheap' before but now sticks to
> 'very-cheap'. I guess adjusting the cost model in process_options
> might be possible when any(?) of the vectorizer flags were set
> explicitly?
process_options would mean it affects only the command line and not
__attribute__((optimize ("O2", "ftree-vectorize")))
etc.
So, shouldn't it be instead done in default_options_optimization, somewhere
among the
if (openacc_mode)
SET_OPTION_IF_UNSET (opts, opts_set, flag_ipa_pta, true);
/* Track fields in field-sensitive alias analysis. */
if (opt2)
SET_OPTION_IF_UNSET (opts, opts_set, param_max_fields_for_field_sensitive,
100);
if (opts->x_optimize_size)
/* We want to crossjump as much as possible. */
SET_OPTION_IF_UNSET (opts, opts_set, param_min_crossjump_insns, 1);
/* Restrict the amount of work combine does at -Og while retaining
most of its useful transforms. */
if (opts->x_optimize_debug)
SET_OPTION_IF_UNSET (opts, opts_set, param_max_combine_insns, 2);
in there?
Like:
/* Use -fvect-cost-model=cheap instead of -fvect-cost-mode=very-cheap
by default with explicit -ftree-{loop,slp}-vectorize. */
if (opts->x_optimize == 2
&& (opts_set->x_ftree_loop_vectorize
|| opts_set->x_ftree_slp_vectorize))
SET_OPTION_IF_UNSET (opts, opts_set, fvect_cost_model_,
VECT_COST_MODEL_CHEAP);
Though, unsure if that will work with -O2 -ftree-vectorize which is an
option without flag with EnabledBy on the other two options.
Also, is:
+ { OPT_LEVELS_2_PLUS, OPT_ftree_loop_vectorize, NULL, 1 },
+ { OPT_LEVELS_2_PLUS, OPT_ftree_slp_vectorize, NULL, 1 },
what we really want, isn't that instead:
+ { OPT_LEVELS_2_PLUS_SPEED_ONLY, OPT_ftree_loop_vectorize, NULL, 1 },
+ { OPT_LEVELS_2_PLUS_SPEED_ONLY, OPT_ftree_slp_vectorize, NULL, 1 },
? I mean, for -Os vectorization even in very-cheap model I'd think it
usually enlarges code size, and for -Og it is seriously harmful for
debugging experience, especially when DWARF <= 5 doesn't have anything that
would help debugging vectorized loops.
Jakub
next prev parent reply other threads:[~2021-09-06 9:41 UTC|newest]
Thread overview: 39+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-09-06 8:46 liuhongt
2021-09-06 8:55 ` Hongtao Liu
2021-09-06 9:18 ` Richard Biener
2021-09-06 9:41 ` Jakub Jelinek [this message]
2021-09-06 10:58 ` Hongtao Liu
2021-09-06 11:01 ` Jakub Jelinek
2021-09-06 11:15 ` Hongtao Liu
2021-09-06 11:15 ` Jakub Jelinek
2021-09-06 12:18 ` Richard Biener
2021-09-06 12:30 ` Jakub Jelinek
2021-09-06 12:43 ` Richard Biener
2021-09-06 11:05 ` Richard Biener
2021-09-06 9:41 ` Hongtao Liu
2021-09-06 16:37 ` Joseph Myers
2021-09-07 2:07 ` Hongtao Liu
2021-09-16 4:33 liuhongt
2021-09-16 8:22 ` Richard Biener
2021-09-16 9:03 ` Hongtao Liu
2021-09-16 12:31 ` Richard Biener
2021-09-17 3:26 ` Hongtao Liu
2021-09-17 7:47 ` Richard Biener
2021-09-17 8:06 ` Hongtao Liu
2021-09-19 20:13 ` Martin Sebor
2021-09-22 1:38 ` Hongtao Liu
2021-09-22 14:21 ` Martin Sebor
2021-09-22 15:03 ` Martin Sebor
2021-09-23 1:48 ` Hongtao Liu
2021-09-23 2:08 ` Hongtao Liu
2021-09-23 6:30 ` Richard Biener
2021-09-23 15:18 ` Martin Sebor
2021-09-24 3:32 ` Hongtao Liu
2021-09-24 14:27 ` Martin Sebor
2021-09-26 3:18 ` liuhongt
2021-09-28 11:18 ` Richard Biener
2021-10-07 15:34 ` Martin Liška
2021-10-07 15:36 ` H.J. Lu
2021-10-08 2:16 ` Hongtao Liu
2021-10-08 10:49 ` Aldy Hernandez
2021-10-08 23:43 ` Martin Sebor
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=20210906094127.GQ920497@tucnak \
--to=jakub@redhat.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=hongtao.liu@intel.com \
--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).