From: Samuel Neves <sneves@dei.uc.pt>
To: "Jason A. Donenfeld" <Jason@zx2c4.com>
Cc: Ingo Molnar <mingo@kernel.org>, Jens Axboe <axboe@kernel.dk>,
Matthew Wilcox <willy@infradead.org>,
Al Viro <viro@zeniv.linux.org.uk>, Christoph Hellwig <hch@lst.de>,
linux-fsdevel@vger.kernel.org,
"the arch/x86 maintainers" <x86@kernel.org>
Subject: Re: [RFC] what to do with IOCB_DSYNC?
Date: Fri, 27 May 2022 15:45:02 +0100 [thread overview]
Message-ID: <CAEX_ruHq2VCq_N2oTkYb5mBRZeXHHkzMwY+MLE1y=867jsfe9g@mail.gmail.com> (raw)
In-Reply-To: <CAHmME9q5i971a8gD4ANqQvL79=2nF2MDiDsn+0uA40wGRCyLLg@mail.gmail.com>
[-- Attachment #1: Type: text/plain, Size: 2520 bytes --]
On Fri, May 27, 2022 at 11:15 AM Jason A. Donenfeld <Jason@zx2c4.com> wrote:
>
> Hi Ingo,
>
> On 5/27/22, Ingo Molnar <mingo@kernel.org> wrote:
> >
> > * Jason A. Donenfeld <Jason@zx2c4.com> wrote:
> >
> >> On Mon, May 23, 2022 at 10:03:45AM -0600, Jens Axboe wrote:
> >> > clear_user()
> >> > 32 ~96MB/sec
> >> > 64 195MB/sec
> >> > 128 386MB/sec
> >> > 1k 2.7GB/sec
> >> > 4k 7.8GB/sec
> >> > 16k 14.8GB/sec
> >> >
> >> > copy_from_zero_page()
> >> > 32 ~96MB/sec
> >> > 64 193MB/sec
> >> > 128 383MB/sec
> >> > 1k 2.9GB/sec
> >> > 4k 9.8GB/sec
> >> > 16k 21.8GB/sec
> >>
> >> Just FYI, on x86, Samuel Neves proposed some nice clear_user()
> >> performance improvements that were forgotten about:
> >>
> >> https://lore.kernel.org/lkml/20210523180423.108087-1-sneves@dei.uc.pt/
> >> https://lore.kernel.org/lkml/Yk9yBcj78mpXOOLL@zx2c4.com/
> >>
> >> Hoping somebody picks this up at some point...
> >
> > Those ~2x speedup numbers are indeed looking very nice:
> >
> > | After this patch, on a Skylake CPU, these are the
> > | before/after figures:
> > |
> > | $ dd if=/dev/zero of=/dev/null bs=1024k status=progress
> > | 94402248704 bytes (94 GB, 88 GiB) copied, 6 s, 15.7 GB/s
> > |
> > | $ dd if=/dev/zero of=/dev/null bs=1024k status=progress
> > | 446476320768 bytes (446 GB, 416 GiB) copied, 15 s, 29.8 GB/s
> >
> > Patch fell through the cracks & it doesn't apply anymore:
> >
> > checking file arch/x86/lib/usercopy_64.c
> > Hunk #2 FAILED at 17.
> > 1 out of 2 hunks FAILED
> >
> > Would be nice to re-send it.
>
> I don't think Samuel is going to do that at this point, so I think
> it's probably best if you do it.
For what it's worth, these are the benchmarks I did at the time
comparing the various code paths (generic loop, rep stosd + stosb, rep
stosd + final loop, rep stosb) for sizes from 0 to 4096 (and then to
32k in larger increments) in the following chips:
- Atom C2550 - Avoton
- Xeon E3-1240 v3 - Haswell
- EPYC 7401P - Zen 1
- EPYC 7402P - Zen 2
- Xeon D-1537 - Broadwell
- Core i7-6700HQ - Skylake
- Core i7-3770 - Ivy Bridge
- Xeon Gold 5120 - Skylake-SP
The fields are: bytes, cycles for generic loop, cycles for rep
stosd+stosb, cycles for rep stosd + final loop, cycles for rep stosb.
There are aligned (to cacheline) and unaligned (cacheline+1)
destination buffer numbers, as that seems to be relevant for rep stos
performance.
Make of that what you will; as Jason said, I'm not particularly
interested in reviving this.
Samuel.
> Jason
[-- Attachment #2: numbers.tar.gz --]
[-- Type: application/gzip, Size: 247755 bytes --]
next prev parent reply other threads:[~2022-05-27 14:51 UTC|newest]
Thread overview: 60+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-06-21 0:46 [RFC] what to do with IOCB_DSYNC? Al Viro
2021-06-21 13:59 ` Christoph Hellwig
2021-06-21 14:03 ` Matthew Wilcox
2021-06-21 14:09 ` Christoph Hellwig
2021-06-21 14:16 ` Matthew Wilcox
2021-06-21 14:22 ` Christoph Hellwig
2021-06-21 14:32 ` Al Viro
2021-06-21 14:35 ` Christoph Hellwig
2021-06-21 15:22 ` Jens Axboe
2022-05-21 17:48 ` Al Viro
2022-05-21 19:03 ` Jens Axboe
2022-05-21 22:14 ` Jens Axboe
2022-05-22 7:45 ` Christoph Hellwig
2022-05-22 10:23 ` Matthew Wilcox
2022-05-22 10:36 ` Al Viro
2022-05-22 11:15 ` Matthew Wilcox
2022-05-22 11:45 ` Christoph Hellwig
2022-05-22 12:39 ` Jens Axboe
2022-05-22 12:48 ` Al Viro
2022-05-22 13:02 ` Jens Axboe
2022-05-22 13:07 ` Al Viro
2022-05-22 13:09 ` Jens Axboe
2022-05-22 18:06 ` Jens Axboe
2022-05-22 18:25 ` Al Viro
2022-05-22 18:29 ` Jens Axboe
2022-05-22 18:39 ` Al Viro
2022-05-22 18:48 ` Jens Axboe
2022-05-22 19:04 ` Al Viro
2022-05-22 20:03 ` Jens Axboe
2022-05-23 0:42 ` Al Viro
2022-05-23 1:22 ` Jens Axboe
2022-05-23 1:28 ` Jens Axboe
2022-05-23 1:50 ` Jens Axboe
2022-05-23 2:43 ` Jens Axboe
2022-05-23 14:22 ` Al Viro
2022-05-23 14:34 ` Jens Axboe
2022-05-23 14:47 ` Al Viro
2022-05-23 15:12 ` Jens Axboe
2022-05-23 15:44 ` Jens Axboe
2022-05-23 15:49 ` Matthew Wilcox
2022-05-23 15:55 ` Jens Axboe
2022-05-23 16:03 ` Jens Axboe
2022-05-26 14:46 ` Jason A. Donenfeld
2022-05-27 10:09 ` Ingo Molnar
2022-05-27 10:15 ` Jason A. Donenfeld
2022-05-27 14:45 ` Samuel Neves [this message]
2022-05-27 10:25 ` Ingo Molnar
2022-05-27 10:36 ` Borislav Petkov
2022-05-28 20:54 ` Sedat Dilek
2022-05-28 20:38 ` Sedat Dilek
2022-05-28 20:39 ` Sedat Dilek
2022-05-23 16:15 ` Al Viro
2022-05-25 14:34 ` Matthew Wilcox
2022-05-26 23:19 ` Al Viro
2022-05-27 14:51 ` Jens Axboe
2022-05-22 12:21 ` Al Viro
2022-05-22 7:43 ` Christoph Hellwig
2022-05-22 12:41 ` Al Viro
2022-05-22 12:51 ` Christoph Hellwig
2021-06-21 14:22 ` Al Viro
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='CAEX_ruHq2VCq_N2oTkYb5mBRZeXHHkzMwY+MLE1y=867jsfe9g@mail.gmail.com' \
--to=sneves@dei.uc.pt \
--cc=Jason@zx2c4.com \
--cc=axboe@kernel.dk \
--cc=hch@lst.de \
--cc=linux-fsdevel@vger.kernel.org \
--cc=mingo@kernel.org \
--cc=viro@zeniv.linux.org.uk \
--cc=willy@infradead.org \
--cc=x86@kernel.org \
/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 NNTP newsgroup(s).