From: Andi Kleen <ak@muc.de>
To: Denis Vlasenko <vda@port.imtp.ilyichevsk.odessa.ua>
Cc: "James M." <dart@windeath.2y.net>, linux-kernel@vger.kernel.org
Subject: Re: Obvious one-liner - Use 3DNOW on MK8
Date: 22 Aug 2004 16:47:31 +0200
Date: Sun, 22 Aug 2004 16:47:31 +0200 [thread overview]
Message-ID: <20040822144731.GA85541@muc.de> (raw)
In-Reply-To: <200408221118.45146.vda@port.imtp.ilyichevsk.odessa.ua>
On Sun, Aug 22, 2004 at 11:18:45AM +0300, Denis Vlasenko wrote:
> On Sunday 22 August 2004 04:29, Andi Kleen wrote:
> > "James M." <dart@windeath.2y.net> writes:
> > > Title says it...my Athlon 64 definitely uses 3DNOW. Patch changes
> > > arch/i386/Kconfig and has a 3 line fudge factor(I created it a few
> > > kernels back). Might want to check other arches for the same bug.
> >
> > It it's not a bug, it is a feature. The K8 is better off not using
> > the 3dnow memcpy, which is the only feature this CONFIG controls.
>
> However, 3dnow _copy_page_ is a huge win. I explained why in an emails
On K8? Significant resources were spent on tuning the x86-64
memcpy and memset, and since C stepping K8 rep ; movsl/q are fastest.
Before that an unrolled integer loop was best.
On 32bit the same applies.
Using SSE2 only helps for very large data sets that are
never used in the kernel (several MB). 3dnow wasn't tested, but it is
unlikely to be any better than SSE2.
-Andi
next prev parent reply other threads:[~2004-08-22 14:47 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <2vOfA-7Vg-7@gated-at.bofh.it>
2004-08-22 1:29 ` Obvious one-liner - Use 3DNOW on MK8 Andi Kleen
2004-08-22 5:52 ` James M.
2004-08-22 11:38 ` Marc Ballarin
[not found] ` <200408221118.45146.vda@port.imtp.ilyichevsk.odessa.ua>
2004-08-22 14:47 ` Andi Kleen [this message]
[not found] ` <200408222146.34798.vda@port.imtp.ilyichevsk.odessa.ua>
[not found] ` <20040823195842.GA7952@muc.de>
2004-08-23 22:24 ` Denis Vlasenko
2004-08-23 22:31 ` Denis Vlasenko
2004-08-22 0:14 James M.
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=20040822144731.GA85541@muc.de \
--to=ak@muc.de \
--cc=dart@windeath.2y.net \
--cc=linux-kernel@vger.kernel.org \
--cc=vda@port.imtp.ilyichevsk.odessa.ua \
/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