linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Peter Zijlstra <peter@programming.kicks-ass.net>
To: IWAMOTO Toshihiro <iwamoto@valinux.co.jp>
Cc: Rik van Riel <riel@redhat.com>,
	Marcelo Tosatti <marcelo.tosatti@cyclades.com>,
	linux-mm@kvack.org, Andrew Morton <akpm@osdl.org>,
	Christoph Lameter <christoph@lameter.com>,
	Wu Fengguang <wfg@mail.ustc.edu.cn>,
	Nick Piggin <npiggin@suse.de>, Marijn Meijles <marijn@bitpit.net>
Subject: Re: [PATCH 6/9] clockpro-clockpro.patch
Date: Mon, 06 Feb 2006 11:07:27 +0100	[thread overview]
Message-ID: <1139220448.11539.17.camel@localhost.localdomain> (raw)
In-Reply-To: <20060206093023.E4F227402D@sv1.valinux.co.jp>

> > > Environment: Dell 1850 4GB EM64T CPUx2 HT disabled, x86_64 kernel
> > > Kernel 1: linux-2.6.15-rc5
> > > Kernel 2: linux-2.6.15-rc5 + clockpro patch posted in 2005/12/31
> > > Kernel 3: linux-2.6.15-rc5 + clockpro patch posted in 2005/12/31 +
> > > 	  modification to disable page cache usage from ZONE_DMA
> > > 	  (to rule out possible zone balancing related problem)

> > >  n     Kernel 1    Kernel 2   Kernel 3
> > > ======================================
> > > 100    373600      298720     395818
> > > 200    385639	   272749     272166
> > > 300    371047	   243734     262370
> > > 400    367691	   213974     169714
> > > 500    147130	   126284     103038

> I've built kernels with CONFIG_MM_POLICY_CLOCKPRO enabled and disabled
> and ran the same tests.
> 
>  n   disabled    enabled
> ========================
> 100   366962     392283
> 200   342172	 393087
> 300   339325	 309595
> 400   230029	 283340
> 500   121352	 132298
> 
> The non-clockpro version (left) has got slower in n >= 400, and the
> clockpro version has got faster in n <= 400.  The clockpro version is
> still slower than the "kernel 1" for n=300,400.  There seem to be a
> few percents of measurement variances, so I shouldn't judge something
> from small differences in values.

Hmm, better but still some ways to go. Thanks for the results.

> > esp. the last patch in the series:
> > http://programming.kicks-ass.net/kernel-patches/page-replace/2.6.16-rc1-3/kswapd-writeout-wait.patch
> 
> This change doesn't seem to be related to clockpro.  I wonder if (and
> why) it is significant.

Ah, no patch description there yet. What happens is that most of the new
policies (can) require quite a lot of scanning and are hitting:

 			blk_congestion_wait(WRITE, HZ/10);

way too often without actually having submitted IO.

> I haven't tried but it should be possible to create similar workloads
> by scaling down the arguments to 2read.c.
> 
> BTW, I made some detailed measurments with older kernels.  I'll mail
> the results later.

Yes, I did that, however your results show a greater decrease in
throughput than I was able to get. Will investigate my setup.


Kind regards,

Peter

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

  reply	other threads:[~2006-02-06 10:07 UTC|newest]

Thread overview: 66+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2005-12-30 22:40 [PATCH] vm: page-replace and clockpro Peter Zijlstra
2005-12-30 22:40 ` [PATCH 01/14] page-replace-single-batch-insert.patch Peter Zijlstra
2005-12-31  7:03   ` Marcelo Tosatti
2005-12-31  9:43     ` Peter Zijlstra
2005-12-31 14:44     ` Rik van Riel
2005-12-31 22:19       ` Marcelo Tosatti
2005-12-30 22:40 ` [PATCH 02/14] page-replace-try_pageout.patch Peter Zijlstra
2005-12-30 22:40 ` [PATCH 03/14] page-replace-remove-sc-from-refill.patch Peter Zijlstra
2005-12-30 22:40 ` [PATCH 04/14] page-replace-activate_page.patch Peter Zijlstra
2005-12-30 22:41 ` [PATCH 05/14] page-replace-remove-loop.patch Peter Zijlstra
2005-12-30 22:41 ` [PATCH 06/14] page-replace-move-macros.patch Peter Zijlstra
2005-12-30 22:41 ` [PATCH 07/14] page-replace-move-isolate_lru_pages.patch Peter Zijlstra
2005-12-30 22:41 ` [PATCH 08/14] page-replace-candidates.patch Peter Zijlstra
2005-12-30 22:41 ` [PATCH 09/14] page-replace-reinsert.patch Peter Zijlstra
2005-12-30 22:41 ` [PATCH 10/14] page-replace-remove-mm_inline.patch Peter Zijlstra
2005-12-30 22:42 ` [PATCH 11/14] page-replace-move-refill.patch Peter Zijlstra
2005-12-30 22:42 ` [PATCH 12/14] page-replace-rotate.patch Peter Zijlstra
2005-12-30 22:42 ` [PATCH 13/14] page-replace-init.patch Peter Zijlstra
2005-12-30 22:42 ` [PATCH 14/14] page-replace-kswapd-incmin.patch Peter Zijlstra
2005-12-31  1:15   ` Marcelo Tosatti
2005-12-31  9:40     ` Peter Zijlstra
2005-12-30 22:42 ` [PATCH 1/9] clockpro-nonresident.patch Peter Zijlstra
2005-12-31  1:13   ` Marcelo Tosatti
2005-12-31  9:54     ` Peter Zijlstra
2005-12-31 14:53       ` Rik van Riel
2005-12-31 22:20         ` Marcelo Tosatti
2005-12-30 22:42 ` [PATCH 2/9] clockpro-nonresident-del.patch Peter Zijlstra
2005-12-30 22:43 ` [PATCH 3/9] clockpro-PG_test.patch Peter Zijlstra
2005-12-30 22:43 ` [PATCH 4/9] clockpro-use-once.patch Peter Zijlstra
2005-12-30 22:43 ` [PATCH 5/9] clockpro-ignore_token.patch Peter Zijlstra
2005-12-30 22:43 ` [PATCH 6/9] clockpro-clockpro.patch Peter Zijlstra
2005-12-31  0:24   ` Marcelo Tosatti
2005-12-31  1:22     ` Rik van Riel
2005-12-31  3:27       ` Marcelo Tosatti
2005-12-31  5:24         ` Rik van Riel
2005-12-31 10:57           ` Peter Zijlstra
2005-12-31 10:48     ` Peter Zijlstra
2005-12-31 22:12       ` Marcelo Tosatti
2006-01-03 19:30         ` Christoph Lameter
2005-12-31 11:29     ` Peter Zijlstra
2006-01-05  9:47       ` IWAMOTO Toshihiro
2006-01-05 13:32         ` Rik van Riel
2006-01-06  9:01           ` IWAMOTO Toshihiro
2006-01-24  6:30             ` IWAMOTO Toshihiro
2006-01-24  7:25               ` IWAMOTO Toshihiro
2006-01-25  8:00                 ` Peter Zijlstra
2006-02-03  9:25                 ` Peter Zijlstra
2006-02-06  9:30                   ` IWAMOTO Toshihiro
2006-02-06 10:07                     ` Peter Zijlstra [this message]
2006-02-08 10:05                   ` IWAMOTO Toshihiro
2006-02-08 20:00                     ` Peter Zijlstra
2006-02-09  6:57                       ` Peter Zijlstra
2006-02-09  7:22                       ` IWAMOTO Toshihiro
2006-02-09 10:07                         ` IWAMOTO Toshihiro
2006-02-09 15:23                       ` Rik van Riel
2006-02-08  9:53                 ` IWAMOTO Toshihiro
2005-12-31 22:40   ` Marcelo Tosatti
2006-01-01 10:37     ` Peter Zijlstra
2006-01-03 12:21       ` Marcelo Tosatti
2006-02-14  7:29   ` IWAMOTO Toshihiro
2006-02-15  6:35     ` Peter Zijlstra
2006-02-16  6:25       ` IWAMOTO Toshihiro
2005-12-30 22:43 ` [PATCH 7/9] clockpro-remove-old.patch Peter Zijlstra
2005-12-30 22:43 ` [PATCH 8/9] clockpro-rename_PG_active.patch Peter Zijlstra
2005-12-30 22:44 ` [PATCH 9/9] clockpro-clockpro-stats.patch Peter Zijlstra
2005-12-31 18:59 ` [PATCH 10/9] clockpro-document.patch Peter Zijlstra

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=1139220448.11539.17.camel@localhost.localdomain \
    --to=peter@programming.kicks-ass.net \
    --cc=akpm@osdl.org \
    --cc=christoph@lameter.com \
    --cc=iwamoto@valinux.co.jp \
    --cc=linux-mm@kvack.org \
    --cc=marcelo.tosatti@cyclades.com \
    --cc=marijn@bitpit.net \
    --cc=npiggin@suse.de \
    --cc=riel@redhat.com \
    --cc=wfg@mail.ustc.edu.cn \
    /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