From: Nick Piggin <nickpiggin@yahoo.com.au>
To: Paul Jackson <pj@sgi.com>
Cc: Rohit Seth <rohit.seth@intel.com>,
akpm@osdl.org, clameter@engr.sgi.com, torvalds@osdl.org,
linux-mm@kvack.org, linux-kernel@vger.kernel.org,
steiner@sgi.com
Subject: Re: [PATCH]: Free pages from local pcp lists under tight memory conditions
Date: Fri, 02 Dec 2005 11:32:27 +1100 [thread overview]
Message-ID: <438F961B.6060709@yahoo.com.au> (raw)
In-Reply-To: <20051201064446.c87049ad.pj@sgi.com>
Paul Jackson wrote:
> Rohit wrote:
>
>>Can you please comment on the performance delta on the MPI workload
>>because of this change in batch values.
>
>
> I can't -- all I know is what I read in Jack Steiner's posts
> of April 5, 2005, referenced earlier in this thread.
>
It was something fairly large. Basically having a power of 2 batch size
meant that 2 concurrent allocators (presumably setting up the working
area) would alternately pull in power of 2 chunks of memory, which
caused each CPU to only get pages of ~half of its cache's possible
colours.
The fix is not by any means a single value for all workloads, it simply
avoids powers of 2 batch size. Note this will have very little effect
on single threaded allocators and will do nothing for cache colouring
there, however it is important for concurrent allocators.
Nick
--
SUSE Labs, Novell Inc.
Send instant messages to your online friends http://au.messenger.yahoo.com
WARNING: multiple messages have this Message-ID (diff)
From: Nick Piggin <nickpiggin@yahoo.com.au>
To: Paul Jackson <pj@sgi.com>
Cc: Rohit Seth <rohit.seth@intel.com>,
akpm@osdl.org, clameter@engr.sgi.com, torvalds@osdl.org,
linux-mm@kvack.org, linux-kernel@vger.kernel.org,
steiner@sgi.com
Subject: Re: [PATCH]: Free pages from local pcp lists under tight memory conditions
Date: Fri, 02 Dec 2005 11:32:27 +1100 [thread overview]
Message-ID: <438F961B.6060709@yahoo.com.au> (raw)
In-Reply-To: <20051201064446.c87049ad.pj@sgi.com>
Paul Jackson wrote:
> Rohit wrote:
>
>>Can you please comment on the performance delta on the MPI workload
>>because of this change in batch values.
>
>
> I can't -- all I know is what I read in Jack Steiner's posts
> of April 5, 2005, referenced earlier in this thread.
>
It was something fairly large. Basically having a power of 2 batch size
meant that 2 concurrent allocators (presumably setting up the working
area) would alternately pull in power of 2 chunks of memory, which
caused each CPU to only get pages of ~half of its cache's possible
colours.
The fix is not by any means a single value for all workloads, it simply
avoids powers of 2 batch size. Note this will have very little effect
on single threaded allocators and will do nothing for cache colouring
there, however it is important for concurrent allocators.
Nick
--
SUSE Labs, Novell Inc.
Send instant messages to your online friends http://au.messenger.yahoo.com
--
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>
next prev parent reply other threads:[~2005-12-02 0:32 UTC|newest]
Thread overview: 52+ messages / expand[flat|nested] mbox.gz Atom feed top
2005-11-23 0:10 [PATCH]: Free pages from local pcp lists under tight memory conditions Rohit Seth
2005-11-23 0:10 ` Rohit Seth
2005-11-23 5:36 ` Andrew Morton
2005-11-23 5:36 ` Andrew Morton
2005-11-23 5:58 ` Andrew Morton
2005-11-23 5:58 ` Andrew Morton
2005-11-23 18:17 ` Rohit Seth
2005-11-23 18:17 ` Rohit Seth
2005-11-23 6:36 ` Christoph Lameter
2005-11-23 6:36 ` Christoph Lameter
2005-11-23 6:42 ` Christoph Lameter
2005-11-23 6:42 ` Christoph Lameter
2005-11-23 16:35 ` Linus Torvalds
2005-11-23 16:35 ` Linus Torvalds
2005-11-23 17:03 ` Christoph Lameter
2005-11-23 17:03 ` Christoph Lameter
2005-11-23 17:54 ` Rohit Seth
2005-11-23 17:54 ` Rohit Seth
2005-11-23 18:06 ` Mel Gorman
2005-11-23 18:06 ` Mel Gorman
2005-11-23 19:41 ` Rohit Seth
2005-11-23 19:41 ` Rohit Seth
2005-11-24 9:25 ` Mel Gorman
2005-11-24 9:25 ` Mel Gorman
2005-11-23 23:26 ` Rohit Seth
2005-11-23 23:26 ` Rohit Seth
2005-11-23 19:30 ` Christoph Lameter
2005-11-23 19:30 ` Christoph Lameter
2005-11-23 19:46 ` Rohit Seth
2005-11-23 19:46 ` Rohit Seth
2005-11-23 19:55 ` Andrew Morton
2005-11-23 19:55 ` Andrew Morton
2005-11-23 21:00 ` Rohit Seth
2005-11-23 21:00 ` Rohit Seth
2005-11-23 21:25 ` Christoph Lameter
2005-11-23 21:25 ` Christoph Lameter
2005-11-23 22:29 ` Rohit Seth
2005-11-23 22:29 ` Rohit Seth
2005-11-23 21:26 ` Andrew Morton
2005-11-23 21:26 ` Andrew Morton
2005-11-23 21:40 ` Rohit Seth
2005-11-23 21:40 ` Rohit Seth
2005-11-24 3:02 ` Paul Jackson
2005-11-24 3:02 ` Paul Jackson
2005-11-29 23:18 ` Rohit Seth
2005-11-29 23:18 ` Rohit Seth
2005-12-01 14:44 ` Paul Jackson
2005-12-01 14:44 ` Paul Jackson
2005-12-02 0:32 ` Nick Piggin [this message]
2005-12-02 0:32 ` Nick Piggin
2005-11-23 22:01 ` Christoph Lameter
2005-11-23 22:01 ` Christoph Lameter
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=438F961B.6060709@yahoo.com.au \
--to=nickpiggin@yahoo.com.au \
--cc=akpm@osdl.org \
--cc=clameter@engr.sgi.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=pj@sgi.com \
--cc=rohit.seth@intel.com \
--cc=steiner@sgi.com \
--cc=torvalds@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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.