From: Richard Biener <rguenther@suse.de>
To: Roman Zhuykov <zhroma@ispras.ru>
Cc: Segher Boessenkool <segher@kernel.crashing.org>,
"Kewen.Lin" <linkw@linux.ibm.com>,
GCC Patches <gcc-patches@gcc.gnu.org>,
Bill Schmidt <wschmidt@linux.ibm.com>,
"bin.cheng" <bin.cheng@linux.alibaba.com>
Subject: Re: [PATCH 0/4 GCC11] IVOPTs consider step cost for different forms when unrolling
Date: Tue, 11 Feb 2020 13:58:00 -0000 [thread overview]
Message-ID: <nycvar.YFH.7.76.2002111456590.18835@zhemvz.fhfr.qr> (raw)
In-Reply-To: <1ac98132-734e-0ee3-5ea2-7ec256ee92d2@ispras.ru>
[-- Attachment #1: Type: text/plain, Size: 4302 bytes --]
On Tue, 11 Feb 2020, Roman Zhuykov wrote:
> 11.02.2020 11:01, Richard Biener wrote:
> > On Tue, 11 Feb 2020, Segher Boessenkool wrote:
> >
> >> On Tue, Feb 11, 2020 at 08:34:15AM +0100, Richard Biener wrote:
> >>> On Mon, 10 Feb 2020, Segher Boessenkool wrote:
> >>>> Yes, we should decide how often we want to unroll things somewhere before
> >>>> ivopts already, and just use that info here.
> >>>>
> >>>> Or are there advantage to doing it *in* ivopts? It sounds like doing
> >>>> it there is probably expensive, but maybe not, and we need to do similar
> >>>> analysis there anyway.
> >>> Well, if the only benefit of doing the unrolling is that IVs get
> >>> cheaper then yes, IVOPTs should drive it.
> >> We need to know much earlier in the pass pipeline how often a loop will
> >> be unrolled. We don't have to *do* it early.
> >>
> >> If we want to know it before ivopts, then obviously it has to be done
> >> earlier. Otherwise, maybe it is a good idea to do it in ivopts itself.
> >> Or maybe not. It's just an idea :-)
> >>
> >> We know we do not want it *later*, ivopts needs to know this to make
> >> good decisions of its own.
> >>
> >>> But usually unrolling exposes redundancies (catched by predictive
> >>> commoning which drives some unrolling) or it enables better use
> >>> of CPU resources via scheduling (only catched later in RTL).
> >>> For scheduling we have the additional complication that the RTL
> >>> side doesn't have as much of a fancy data dependence analysis
> >>> framework as on the GIMPLE side. So I'd put my bet on trying to
> >>> move something like SMS to GIMPLE and combine it with unrolling
> >>> (IIRC SMS at most interleaves 1 1/2 loop iterations).
> To clarify, without specifying -fmodulo-sched-allow-regmoves it only
> interleaves 2 iterations. With register moves enabled more iterations
> can be considered.
> > SMS on RTL always was quite disappointing...
> Hmm, even when trying to move it just few passes earlier many years ago,
> got another opinion:
> https://gcc.gnu.org/ml/gcc-patches/2011-10/msg01526.html
> Although without such a move we still have annoying issues which RTL
> folks can't solve, see e.q.
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93264#c2
> > It originally came with "data dependence export from GIMPLE to RTL"
> > that never materialized so I'm not surprised ;) It also relies
> > on doloop detection.
> My current attempt to drop doloop dependency is still WIP, hopefully
> I'll create branch in refs/users/ in a month or so. But older (gcc-7
> and earlier) versions are available, see
> https://gcc.gnu.org/ml/gcc-patches/2017-02/msg01647.html
> Doloops are still supported for some kind of backward compatibility, but
> much more loops (which loop-iv can analyze) are considered in new SMS.
> >> Do you expect it will be more useful on Gimple? Moving it there is a
> >> good idea in any case ;-)
> >>
> >> I don't quite see the synergy between SMS and loop unrolling, but maybe
> >> I need to look harder.
> > As said elsewhere I don't believe in actual unrolling doing much good
> > but in removing data dependences in the CPU pipeline. SMS rotates
> > the loop, peeling N iterations (and somehow I think for N > 1 that
> > should better mean unrolling the loop body).
> Yes, this is what theory tells us.
> > Of course doing "scheduling" on GIMPLE is "interesting" in its own
> > but OTOH our pipeline DFAs are imprecise enough that one could even
> > devise some basic GIMPLE <-> "RTL" mapping to make use of it. But
> > then scheduling without IVs or register pressure in mind is somewhat
> > pointless as well.
> Unfortunately, even with -fmodulo-sched-allow-regmoves it doesn't
> interact much with register pressure.
> > That said - if I had enough time I'd still thing that investigating
> > "scheduling on GIMPLE" as replacement for sched1 is an interesting
> > thing to do.
> Sound good, but IMHO modulo scheduler is not the best choice to be the
> first step implementing such a concept.
True ;) But since the context of this thread is unrolling ...
Not sure how you'd figure the unroll factor to apply if you want
to do unrolling within a classical scheduling framework? Maybe
unroll as much as you can fill slots until the last instruction
of the first iteration retires?
Richard.
next prev parent reply other threads:[~2020-02-11 13:58 UTC|newest]
Thread overview: 45+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-01-16 9:41 Kewen.Lin
2020-01-16 9:43 ` [PATCH 1/4 GCC11] Add middle-end unroll factor estimation Kewen.Lin
2020-01-20 13:12 ` Segher Boessenkool
2020-02-10 6:20 ` [PATCH 1/4 v2 " Kewen.Lin
2020-02-10 23:34 ` Segher Boessenkool
2020-02-11 6:51 ` [PATCH 1/4 v3 " Kewen.Lin
2020-02-11 7:00 ` Segher Boessenkool
2020-02-11 2:15 ` [PATCH 1/4 v2 " Jiufu Guo
2020-02-11 3:04 ` Kewen.Lin
2020-01-16 10:02 ` [PATCH 2/4 GCC11] Add target hook stride_dform_valid_p Kewen.Lin
2020-01-20 10:53 ` Richard Sandiford
2020-01-20 11:47 ` Richard Biener
2020-01-20 13:20 ` Segher Boessenkool
2020-02-25 9:46 ` Kewen.Lin
2020-03-02 11:09 ` Richard Sandiford
2020-03-03 12:26 ` Kewen.Lin
2020-05-13 5:50 ` Kewen.Lin
2020-05-28 2:17 ` Ping^1 [PATCH 2/4 V3] " Kewen.Lin
2020-05-28 10:54 ` Richard Sandiford
2020-01-16 10:06 ` [PATCH 3/4 GCC11] IVOPTs Consider cost_step on different forms during unrolling Kewen.Lin
2020-02-25 9:48 ` [PATCH 3/4 V2 " Kewen.Lin
2020-05-13 5:42 ` [PATCH 3/4 V3 " Kewen.Lin
2020-01-16 10:12 ` [PATCH 4/4 GCC11] rs6000: P9 D-form test cases Kewen.Lin
2020-01-20 13:37 ` Segher Boessenkool
2020-02-10 6:25 ` [PATCH 4/4 v2 " Kewen.Lin
2020-02-10 23:51 ` Segher Boessenkool
2020-01-20 13:03 ` [PATCH 0/4 GCC11] IVOPTs consider step cost for different forms when unrolling Segher Boessenkool
2020-02-10 6:17 ` Kewen.Lin
2020-02-10 21:29 ` Segher Boessenkool
2020-02-11 2:56 ` Kewen.Lin
2020-02-11 7:34 ` Richard Biener
2020-02-11 7:49 ` Segher Boessenkool
2020-02-11 8:01 ` Richard Biener
2020-02-11 12:46 ` Roman Zhuykov
2020-02-11 13:58 ` Richard Biener [this message]
2020-02-11 18:00 ` Segher Boessenkool
2020-02-12 8:07 ` Richard Biener
2020-02-12 21:53 ` Segher Boessenkool
2020-02-11 18:12 ` Segher Boessenkool
2020-02-12 8:13 ` Richard Biener
2020-02-12 10:02 ` Segher Boessenkool
2020-02-12 10:53 ` Richard Biener
2020-02-12 22:05 ` Segher Boessenkool
2020-02-13 7:48 ` Richard Biener
2020-02-13 9:02 ` Segher Boessenkool
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=nycvar.YFH.7.76.2002111456590.18835@zhemvz.fhfr.qr \
--to=rguenther@suse.de \
--cc=bin.cheng@linux.alibaba.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=linkw@linux.ibm.com \
--cc=segher@kernel.crashing.org \
--cc=wschmidt@linux.ibm.com \
--cc=zhroma@ispras.ru \
/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).