From: "謝昇達(Shen-Ta Hsieh)" <ibmibmibm.tw@gmail.com>
To: Paul Zimmermann <Paul.Zimmermann@inria.fr>
Cc: libc-alpha@sourceware.org
Subject: Re: [PATCH v6 2/3] x86_64: roundeven with sse4.1 support
Date: Sun, 23 May 2021 08:41:50 +0800 [thread overview]
Message-ID: <CA+fY3Ldipj7UR_FD6pvd8E+2A=sQmdLTiYrk0gPc6P861SddZg@mail.gmail.com> (raw)
In-Reply-To: <mwbl93mku6.fsf@tomate.loria.fr>
Dear Paul,
Paul Zimmermann <Paul.Zimmermann@inria.fr> 於 2021年5月22日 週六 下午12:52寫道：
>
> Dear Shen-Ta,
>
> > Here is a benchmark result on my AMD Ryzen 9 3900X system:
> >
> > * benchmark result before this commit
> > | | roundeven | roundevenf |
> > |------------|--------------|--------------|
> > | duration | 3.77659e+09 | 3.77504e+09 |
> > | iterations | 3.97043e+08 | 4.36752e+08 |
> > | max | 83.714 | 58.861 |
> > | min | 7.144 | 6.27 |
> > | mean | 9.51179 | 8.64345 |
> >
> > * benchmark result after this commit
> > | | roundeven | roundevenf |
> > |------------|--------------|--------------|
> > | duration | 3.76913e+09 | 3.76923e+09 |
> > | iterations | 5.55921e+08 | 5.64822e+08 |
> > | max | 211.698 | 439.09 |
> > | min | 6.498 | 6.422 |
> > | mean | 6.77998 | 6.6733 |
>
> I wonder why the max times have increased by a factor 2.5 and 7.5.
> In my experiments I noticed that the "mean" time was quite stable,
> while the "max" time could vary a lot between different runs. Thus
> I usually run 5 times "make bench" and keep the smallest times:
>
> $ ./testrun.sh benchtests/bench-roundeven
>
> You can also try make USE_RDTSCP=1 bench (cf benchtests/README).
>
> And it would be nice to have figures on another hardware (for example Intel).
>
> Best regards,
> Paul Zimmermann
>
I've tried to run another benchmark following your advises, and here
is the result:
# AMD Ryzen 9 3900X 12-Core Processor
* benchmark result before this commit
| | roundeven | roundevenf |
|------------|--------------|--------------|
| duration | 3.75587e+09 | 3.75114e+09 |
| iterations | 3.93053e+08 | 4.35402e+08 |
| max | 52.592 | 58.71 |
| min | 7.98 | 7.22 |
| mean | 9.55563 | 8.61535 |
* benchmark result after this commit
| | roundeven | roundevenf |
|------------|---------------|--------------|
| duration | 3.73815e+09 | 3.73738e+09 |
| iterations | 5.82692e+08 | 5.91498e+08 |
| max | 56.468 | 51.642 |
| min | 6.27 | 6.156 |
| mean | 6.41532 | 6.3185 |
# Intel(R) Pentium(R) CPU D1508 @ 2.20GHz
* benchmark result before this commit
| | roundeven | roundevenf |
|------------|--------------|--------------|
| duration | 2.18208e+09 | 2.18258e+09 |
| iterations | 2.39932e+08 | 2.46924e+08 |
| max | 96.378 | 98.035 |
| min | 6.776 | 5.94 |
| mean | 9.09456 | 8.83907 |
* benchmark result after this commit
| | roundeven | roundevenf |
|------------|--------------|--------------|
| duration | 2.17415e+09 | 2.17005e+09 |
| iterations | 3.56193e+08 | 4.09824e+08 |
| max | 51.693 | 97.192 |
| min | 5.926 | 5.093 |
| mean | 6.10385 | 5.29507 |
I ran each tests about 20 times, and pick a stable result out.
next prev parent reply other threads:[~2021-05-23 0:42 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-05-22 0:22 [PATCH v6 1/3] math: redirect roundeven function Shen-Ta Hsieh
2021-05-22 0:22 ` [PATCH v6 2/3] x86_64: roundeven with sse4.1 support Shen-Ta Hsieh
2021-05-22 4:52 ` Paul Zimmermann
2021-05-23 0:41 ` 謝昇達(Shen-Ta Hsieh) [this message]
2021-05-23 16:14 ` Paul Zimmermann
2021-05-23 17:10 ` H.J. Lu
2021-05-23 17:12 ` Noah Goldstein
2021-05-22 0:22 ` [PATCH v6 3/3] Use GCC builtins for roundeven functions if desired Shen-Ta Hsieh
2021-05-23 19:03 ` H.J. Lu
2021-05-24 1:22 ` 謝昇達(Shen-Ta Hsieh)
2021-05-23 18:54 ` [PATCH v6 1/3] math: redirect roundeven function H.J. Lu
2021-05-24 1:39 ` 謝昇達(Shen-Ta Hsieh)
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='CA+fY3Ldipj7UR_FD6pvd8E+2A=sQmdLTiYrk0gPc6P861SddZg@mail.gmail.com' \
--to=ibmibmibm.tw@gmail.com \
--cc=Paul.Zimmermann@inria.fr \
--cc=libc-alpha@sourceware.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 read-only IMAP folder(s) and NNTP newsgroup(s).