netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Stephen Hemminger <shemminger@linux-foundation.org>
To: Andi Kleen <andi@firstfloor.org>
Cc: Jan Engelhardt <jengelh@linux01.gwdg.de>,
	linux-kernel@vger.kernel.org, netdev@vger.kernel.org
Subject: Re: [RFC] div64_64 support
Date: Mon, 5 Mar 2007 15:57:14 -0800	[thread overview]
Message-ID: <20070305155714.3abe1b5e@freekitty> (raw)
In-Reply-To: <p73slcnawh3.fsf@bingen.suse.de>

On 03 Mar 2007 03:31:52 +0100
Andi Kleen <andi@firstfloor.org> wrote:

> Stephen Hemminger <shemminger@linux-foundation.org> writes:
> 
> > Here is another way to handle the 64 bit divide case.
> > It allows full 64 bit divide by adding the support routine
> > GCC needs.
> 
> Not supplying that was intentional by Linus so that people
> think twice (or more often) before they using such expensive
> operations. A plain / looks too innocent.
> 
> Is it really needed by CUBIC anyways?  It uses it for getting
> the cubic root, but the algorithm recommended by Hacker's Delight
> (great book) doesn't use any divisions at all. Probably better 
> to use a better algorithm without divisions.
> 

I tried the code from Hacker's Delight.
It is cool, but performance is CPU (and data) dependent:

Average # of usecs per operation:

		Hacker		Newton
Pentium 3	68.6	<	90.4
T2050		98.6	>	92.0
U1400		450	>	415
Xeon		70 	<	90
Xeon (newer)	71	<	78

EM64T		21.8	<	24.6
AMD64		23.4	<	32.0

It might be worth the change for code size reduction though.


-- 
Stephen Hemminger <shemminger@linux-foundation.org>

  reply	other threads:[~2007-03-05 23:57 UTC|newest]

Thread overview: 57+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-02-24  1:05 [RFC] div64_64 support Stephen Hemminger
2007-02-24 16:19 ` Sami Farin
2007-02-26 20:09 ` Jan Engelhardt
2007-02-26 21:28   ` Stephen Hemminger
2007-02-27  1:20     ` H. Peter Anvin
2007-02-27  3:45       ` Segher Boessenkool
2007-02-26 22:31   ` Stephen Hemminger
2007-02-26 23:02     ` Jan Engelhardt
2007-02-26 23:44       ` Stephen Hemminger
2007-02-27  0:05         ` Jan Engelhardt
2007-02-27  0:07           ` Stephen Hemminger
2007-02-27  0:14             ` Jan Engelhardt
2007-02-27  6:21     ` Dan Williams
2007-03-03  2:31     ` Andi Kleen
2007-03-05 23:57       ` Stephen Hemminger [this message]
2007-03-06  0:25         ` David Miller
2007-03-06 13:36           ` Andi Kleen
2007-03-06 14:04           ` [RFC] div64_64 support II Andi Kleen
2007-03-06 17:43             ` Dagfinn Ilmari Mannsåker
2007-03-06 18:25               ` David Miller
2007-03-06 18:48             ` H. Peter Anvin
2007-03-06 13:34         ` [RFC] div64_64 support Andi Kleen
2007-03-06 14:19           ` Eric Dumazet
2007-03-06 14:45             ` Andi Kleen
2007-03-06 15:10               ` Roland Kuhn
2007-03-06 18:29                 ` Stephen Hemminger
2007-03-06 19:48                   ` Andi Kleen
2007-03-06 20:04                     ` Stephen Hemminger
2007-03-06 21:53                   ` Sami Farin
2007-03-06 22:24                     ` Sami Farin
2007-03-07 16:11                       ` Chuck Ebbert
2007-03-07 18:32                         ` Sami Farin
2007-03-08 18:23                       ` asm volatile [Was: [RFC] div64_64 support] Sami Farin
2007-03-08 22:01                         ` asm volatile David Miller
2007-03-06 21:58                   ` [RFC] div64_64 support David Miller
2007-03-06 22:47                     ` [PATCH] tcp_cubic: faster cube root Stephen Hemminger
2007-03-06 22:58                       ` cube root benchmark code Stephen Hemminger
2007-03-07  6:08                         ` Update to " Willy Tarreau
2007-03-08  1:07                           ` [PATCH] tcp_cubic: use 32 bit math Stephen Hemminger
2007-03-08  2:55                             ` David Miller
2007-03-08  3:10                               ` Stephen Hemminger
2007-03-08  3:51                                 ` David Miller
2007-03-10 11:48                                   ` Willy Tarreau
2007-03-12 21:11                                     ` Stephen Hemminger
2007-03-13 20:50                                       ` Willy Tarreau
2007-03-21 18:54                                         ` Stephen Hemminger
2007-03-21 19:15                                           ` Willy Tarreau
2007-03-21 19:58                                             ` Stephen Hemminger
2007-03-21 20:15                                             ` [PATCH 1/2] div64_64 optimization Stephen Hemminger
2007-03-21 20:17                                               ` [PATCH 2/2] tcp: cubic optimization Stephen Hemminger
2007-03-22 19:11                                                 ` David Miller
2007-03-22 19:11                                               ` [PATCH 1/2] div64_64 optimization David Miller
2007-03-08  4:16                                 ` [PATCH] tcp_cubic: use 32 bit math Willy Tarreau
2007-03-07  4:20                       ` [PATCH] tcp_cubic: faster cube root David Miller
2007-03-07 12:12                         ` Andi Kleen
2007-03-07 19:33                           ` David Miller
2007-03-06 18:50               ` [RFC] div64_64 support H. Peter Anvin

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=20070305155714.3abe1b5e@freekitty \
    --to=shemminger@linux-foundation.org \
    --cc=andi@firstfloor.org \
    --cc=jengelh@linux01.gwdg.de \
    --cc=linux-kernel@vger.kernel.org \
    --cc=netdev@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;
as well as URLs for NNTP newsgroup(s).