mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Erich Focht <efocht@ess.nec.de>
To: Michael Hohnbaum <hohnbaum@us.ibm.com>
Cc: "Martin J. Bligh" <Martin.Bligh@us.ibm.com>,
	Robert Love <rml@tech9.net>, Anton Blanchard <anton@samba.org>,
	Ingo Molnar <mingo@elte.hu>,
	Stephen Hemminger <shemminger@osdl.org>,
	linux-kernel <linux-kernel@vger.kernel.org>
Subject: Re: NUMA scheduler BK tree
Date: Fri, 8 Nov 2002 17:57:22 +0100	[thread overview]
Message-ID: <200211081757.22876.efocht@ess.nec.de> (raw)
In-Reply-To: <1036712788.3178.15.camel@dyn9-47-17-164.beaverton.ibm.com>

Michael,

thanks for the results, they look really good!

There's a problem with your script for Schedbench: the column labels
are permuted! Instead of
>                              Elapsed   TotalUser    TotalSys     AvgUser
you should have
>                              AvgUser   Elapsed      TotalUser    TotalSys

The numbers make more sense like that ;-)
Could you please send me your script for schedbench?

BTW: the bk repository has now two distinct changesets, one for the
"core" the other one for the initial balancing. Larry McVoy showed me
how to do things right with the names assigned to the changesets.

Thanks!

Regards,
Erich


On Friday 08 November 2002 00:46, Michael Hohnbaum wrote:
> On Wed, 2002-11-06 at 08:34, Erich Focht wrote:
> > in order to make it easier to keep up with the main Linux tree I've
> > set up a bitkeeper repository with our NUMA scheduler at
> >        bk://numa-ef.bkbits.net/numa-sched
> > (Web view:  http://numa-ef.bkbits.net/)
...
> Tested this on a 4 node NUMAQ.  Worked fine.  Results:
>
> $reportbench stock46 sched46
> Kernbench:
>                              Elapsed        User      System         CPU
>                  stock46      20.66s    194.062s      53.39s     1197.4%
>                  sched46     19.988s    191.302s     50.692s     1210.4%
>
> Schedbench 4:
>                              Elapsed   TotalUser    TotalSys     AvgUser
>                  stock46       27.27       40.64      109.13        0.85
>                  sched46       23.10       41.32       92.42        0.76
>
> Schedbench 8:
>                              Elapsed   TotalUser    TotalSys     AvgUser
>                  stock46       39.18       55.12      313.56        1.68
>                  sched46       34.45       51.24      275.63        2.28
>
> Schedbench 16:
>                              Elapsed   TotalUser    TotalSys     AvgUser
>                  stock46       56.39       72.44      902.45        5.12
>                  sched46       56.73       71.31      907.88        4.19
>
> Schedbench 32:
>                              Elapsed   TotalUser    TotalSys     AvgUser
>                  stock46       90.47      203.28     2895.41       10.39
>                  sched46       60.95      143.21     1950.72       10.31
>
> Schedbench 64:
>                              Elapsed   TotalUser    TotalSys     AvgUser
>                  stock46      105.00      439.04     6720.90       25.02
>                  sched46       59.07      262.98     3781.06       19.59
>
> The schedbench runs were ran once each.  Kernbench is the average of
> 5 runs.
>
>           Michael


  reply	other threads:[~2002-11-08 16:51 UTC|newest]

Thread overview: 29+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2002-11-06 16:34 Erich Focht
2002-11-06 18:10 ` Michael Hohnbaum
2002-11-07 23:05   ` Erich Focht
2002-11-07 23:46 ` Michael Hohnbaum
2002-11-08 16:57   ` Erich Focht [this message]
2002-11-11 15:13 ` [PATCH 2.5.47] NUMA scheduler (1/2) Erich Focht
2002-11-11 15:14   ` [PATCH 2.5.47] NUMA scheduler (2/2) Erich Focht
2002-11-12  0:24   ` [PATCH 2.5.47] NUMA scheduler (1/2) Michael Hohnbaum
2002-11-18 19:40 ` NUMA scheduler BK tree Martin J. Bligh
2002-11-19 16:26   ` [PATCH 2.5.48] NUMA scheduler (1/2) Erich Focht
2002-11-19 16:27     ` [PATCH 2.5.48] NUMA scheduler (2/2) Erich Focht
2002-12-02 15:29     ` [PATCH 2.5.50] NUMA scheduler (1/2) Erich Focht
2002-12-02 15:30       ` [PATCH 2.5.50] NUMA scheduler (2/2) Erich Focht
2002-12-06 17:39       ` [PATCH 2.5.50] NUMA scheduler (1/2) Michael Hohnbaum
2002-12-18 16:21       ` [PATCH 2.5.52] " Erich Focht
2002-12-18 16:23         ` [PATCH 2.5.52] NUMA scheduler (2/2) Erich Focht
2002-12-20 14:49         ` [PATCH 2.5.52] NUMA scheduler: cputimes stats Erich Focht
2002-12-20 15:17         ` [PATCH 2.5.52] NUMA scheduler (1/2) Christoph Hellwig
2002-12-20 17:44           ` Erich Focht
2002-12-31 13:29         ` [PATCH 2.5.53] NUMA scheduler (1/3) Erich Focht
2002-12-31 13:29           ` [PATCH 2.5.53] NUMA scheduler (2/3) Erich Focht
2002-12-31 13:30           ` [PATCH 2.5.53] NUMA scheduler (3/3) Erich Focht
2003-01-04  1:58           ` [PATCH 2.5.53] NUMA scheduler (1/3) Michael Hohnbaum
2003-01-05  5:35             ` Martin J. Bligh
2003-01-06  3:58               ` Michael Hohnbaum
2003-01-06  6:07                 ` Martin J. Bligh
2003-01-07  2:23                   ` Michael Hohnbaum
2003-01-07 11:27                     ` Erich Focht
2003-01-07 23:35                       ` Michael Hohnbaum

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=200211081757.22876.efocht@ess.nec.de \
    --to=efocht@ess.nec.de \
    --cc=Martin.Bligh@us.ibm.com \
    --cc=anton@samba.org \
    --cc=hohnbaum@us.ibm.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=rml@tech9.net \
    --cc=shemminger@osdl.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®