From: Davide Libenzi <davidel@xmailserver.org>
To: davidm@hpl.hp.com
Cc: Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: performance experiment
Date: Fri, 26 Jul 2002 11:09:31 -0700 (PDT) [thread overview]
Message-ID: <Pine.LNX.4.44.0207261107140.1561-100000@blue1.dev.mcafeelabs.com> (raw)
In-Reply-To: <200207261746.g6QHkjUp005023@napali.hpl.hp.com>
On Fri, 26 Jul 2002, David Mosberger wrote:
> Below is a patch that implements an alternate version of the core-loop
> of do_select(). I'm interested in hearing how the two versions
> (original and new) compare on various architectures. The new loop
> happens to perform better on ia64 and I suspect the same will be true
> for most RISC platforms. It wouldn't surprise me if the new loop
> performed better even on some instances of x86. I suspect on older
> x86s (e.g., 80486), the old loop does better. If someone is running
> Linux on a Transmeta Crusoe chip, I'd be *very* interested in hearing
> how the two loops perform there.
>
> Here is what I'm proposing to do: if a couple of people are willing to
> try out the patch below, I'll collect the results and post a summary.
> To make sure we're comparing apples to apples, I'd like to suggest to
> run LMbench 2 with and without the patch below. Then send me the
> select results from the raw results file. For example, you would run
> lmbench like so:
>
> $ make rerun
>
> Then look at the results file, which is stored in
> results/CONFIG/HOSTNAME.N. For example, on a Pentium III machine
> called "adler", the results of the first run would be stored in
>
> results/i686-pc-linux-gnu/adler.0
>
> I'd prefer if you sent me the complete result files, but if you don't
> want to do that, it should be good enough to mail me the first and
> second line of the file, all the lines starting with "Select", and a
> description of the machine you were testing (CPU type, clock speed,
> chipset, memory size, and compiler version would be ideal). For the
> above example, the lines of interest would be:
i posted a 95% matching patch about one year ago but it fell inside the
Alan drop basket :-) basically Alan requested perf numbers that i did not
have time to supply. glad you did them ...
- Davide
next prev parent reply other threads:[~2002-07-26 18:04 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2002-07-26 17:46 David Mosberger
2002-07-26 18:09 ` Davide Libenzi [this message]
2002-07-26 18:42 ` David Mosberger
2002-07-26 20:49 ` Davide Libenzi
2002-07-27 1:29 ` David Mosberger
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=Pine.LNX.4.44.0207261107140.1561-100000@blue1.dev.mcafeelabs.com \
--to=davidel@xmailserver.org \
--cc=davidm@hpl.hp.com \
--cc=linux-kernel@vger.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
all inboxes | Powered by JetHome®