From: Christoph Lameter <christoph@lameter.com>
To: Matthew Wilcox <willy@debian.org>
Cc: Christoph Lameter <clameter@sgi.com>,
William Lee Irwin III <wli@holomorphy.com>,
"David S. Miller" <davem@redhat.com>,
raybry@sgi.com, ak@muc.de, benh@kernel.crashing.org,
manfred@colorfullife.com, linux-ia64@vger.kernel.org,
linux-kernel@vger.kernel.org, vrajesh@umich.edu
Subject: Re: page fault scalability patch v3: use cmpxchg, make rss atomic
Date: Tue, 24 Aug 2004 14:47:39 +0000 [thread overview]
Message-ID: <Pine.LNX.4.58.0408240743180.3853@server.home> (raw)
In-Reply-To: <20040824123416.GL9660@parcelfarce.linux.theplanet.co.uk>
On Tue, 24 Aug 2004, Matthew Wilcox wrote:
> On Mon, Aug 23, 2004 at 10:49:31PM -0700, Christoph Lameter wrote:
> > Unpatched:
> > Gb Rep Threads User System Wall flt/cpu/s fault/wsec
> > 4 3 1 0.157s 11.197s 11.035s 69261.721 69239.940
> > 4 3 2 0.145s 11.445s 6.079s 67849.528 115681.409
> > 4 3 4 0.182s 13.894s 4.027s 55865.834 184108.856
> > 4 3 8 0.196s 24.874s 4.025s 31369.039 184790.767
> >
> > With page fault scalability patch:
> > Gb Rep Threads User System Wall flt/cpu/s fault/wsec
> > 4 3 1 0.176s 11.323s 11.050s 68385.368 68345.055
> > 4 3 2 0.174s 10.716s 5.096s 72205.329 131848.322
> > 4 3 4 0.170s 10.694s 3.040s 72380.552 231128.569
> > 4 3 8 0.177s 14.717s 2.064s 52796.567 297380.041
>
> What kind of variance are you seeing with this benchmark? I'm suspicious
> that your 2 thread case is faster than your single thread case.
What is so suspicious about it? Two CPUs can do the job faster than a
single one. Thats the way it should be and the point of these
patches is to reduce locking so that this can happen in a more efficient
way.
There are some variances in these tests especially if one uses low memory
settings such as 1GB 2GB and to some extend also at 4GB. 16 GB gives quite
stable results but the machine I had available for this only had 8 GB. See
my earlier posts on the subject for other test results.
next prev parent reply other threads:[~2004-08-24 14:47 UTC|newest]
Thread overview: 92+ messages / expand[flat|nested] mbox.gz Atom feed top
2004-08-15 13:50 page fault fastpath: Increasing SMP scalability by introducing pte Christoph Lameter
2004-08-15 20:09 ` page fault fastpath: Increasing SMP scalability by introducing David S. Miller
2004-08-15 22:58 ` Christoph Lameter
2004-08-15 23:58 ` David S. Miller
2004-08-16 0:11 ` Christoph Lameter
2004-08-16 1:56 ` David S. Miller
2004-08-16 3:29 ` Christoph Lameter
2004-08-16 6:59 ` Ray Bryant
2004-08-16 15:18 ` Christoph Lameter
2004-08-16 16:18 ` page fault fastpath: Increasing SMP scalability by introducing pte locks? William Lee Irwin III
2004-08-16 14:39 ` William Lee Irwin III
2004-08-17 15:28 ` page fault fastpath patch v2: fix race conditions, stats for 8,32 Christoph Lameter
2004-08-17 15:37 ` page fault fastpath patch v2: fix race conditions, stats for 8,32 and 512 cpu SMP Christoph Hellwig
2004-08-17 15:51 ` William Lee Irwin III
2004-08-18 17:55 ` page fault fastpath patch v2: fix race conditions, stats for 8,32 Hugh Dickins
2004-08-18 20:20 ` page fault fastpath patch v2: fix race conditions, stats for 8,32 and 512 cpu SMP William Lee Irwin III
2004-08-19 1:19 ` page fault fastpath patch v2: fix race conditions, stats for Christoph Lameter
[not found] ` <B6E8046E1E28D34EB815A11AC8CA3129027B679F@mtv-atc-605e--n.corp.sgi.com>
2004-08-24 4:43 ` page fault scalability patch v3: use cmpxchg, make rss atomic Christoph Lameter
2004-08-24 5:49 ` Christoph Lameter
2004-08-24 12:34 ` Matthew Wilcox
2004-08-24 14:47 ` Christoph Lameter [this message]
[not found] ` <B6E8046E1E28D34EB815A11AC8CA3129027B67A9@mtv-atc-605e--n.corp.sgi.com>
2004-08-26 15:20 ` page fault scalability patch v4: reduce page_table_lock use, atomic Christoph Lameter
[not found] ` <B6E8046E1E28D34EB815A11AC8CA3129027B67B4@mtv-atc-605e--n.corp.sgi.com>
2004-08-27 23:20 ` page fault scalability patch final : i386 tested, x86_64 support Christoph Lameter
2004-08-27 23:36 ` page fault scalability patch final : i386 tested, x86_64 support added Andi Kleen
2004-08-27 23:43 ` page fault scalability patch final : i386 tested, x86_64 David S. Miller
2004-08-28 0:19 ` page fault scalability patch final : i386 tested, x86_64 support Christoph Lameter
2004-08-28 0:23 ` page fault scalability patch final : i386 tested, x86_64 David S. Miller
2004-08-28 0:36 ` Andrew Morton
2004-08-28 0:40 ` David S. Miller
2004-08-28 1:05 ` page fault scalability patch final : i386 tested, x86_64 support added Andi Kleen
2004-08-28 1:11 ` page fault scalability patch final : i386 tested, x86_64 David S. Miller
2004-08-28 1:17 ` page fault scalability patch final : i386 tested, x86_64 support added Andi Kleen
2004-08-28 1:02 ` Andi Kleen
2004-08-28 1:39 ` page fault scalability patch final : i386 tested, x86_64 Andrew Morton
2004-08-28 2:08 ` Paul Mackerras
2004-08-28 3:32 ` page fault scalability patch final : i386 tested, x86_64 support Christoph Lameter
2004-08-28 3:42 ` page fault scalability patch final : i386 tested, x86_64 Andrew Morton
2004-08-28 4:24 ` page fault scalability patch final : i386 tested, x86_64 support Christoph Lameter
2004-08-28 5:39 ` page fault scalability patch final : i386 tested, x86_64 Andrew Morton
2004-08-28 5:58 ` page fault scalability patch final : i386 tested, x86_64 support Christoph Lameter
2004-08-28 6:03 ` page fault scalability patch final : i386 tested, x86_64 support added William Lee Irwin III
2004-08-28 6:06 ` page fault scalability patch final : i386 tested, x86_64 Andrew Morton
2004-08-30 17:02 ` page fault scalability patch final : i386 tested, x86_64 support added Herbert Poetzl
2004-08-30 17:05 ` Andi Kleen
2004-08-28 13:19 ` page fault scalability patch final : i386 tested, x86_64 Andi Kleen
2004-08-28 15:48 ` page fault scalability patch final : i386 tested, x86_64 support added Matt Mackall
2004-09-01 4:13 ` page fault scalability patch final : i386 tested, x86_64 Benjamin Herrenschmidt
2004-09-02 21:26 ` page fault scalability patch final : i386 tested, x86_64 support added Andi Kleen
2004-09-02 21:55 ` page fault scalability patch final : i386 tested, x86_64 David S. Miller
2004-08-28 21:41 ` page fault scalability patch final : i386 tested, x86_64 support added Daniel Phillips
2004-09-01 4:24 ` page fault scalability patch final : i386 tested, x86_64 Benjamin Herrenschmidt
2004-09-01 5:22 ` David S. Miller
2004-09-01 16:43 ` page fault scalability patch final : i386 tested, x86_64 support Christoph Lameter
2004-09-01 23:09 ` page fault scalability patch final : i386 tested, x86_64 Benjamin Herrenschmidt
[not found] ` <Pine.LNX.4.58.0409012140440.23186@schroedinger.engr.sgi.com>
[not found] ` <20040901215741.3538bbf4.davem@davemloft.net>
2004-09-02 5:18 ` page fault scalability patch final : i386 tested, x86_64 support added William Lee Irwin III
2004-09-09 15:38 ` page fault scalability patch: V7 (+fallback for atomic page table Christoph Lameter
2004-09-02 16:24 ` page fault scalability patch final : i386 tested, x86_64 support Christoph Lameter
2004-09-02 20:10 ` page fault scalability patch final : i386 tested, x86_64 David S. Miller
2004-09-02 21:02 ` page fault scalability patch final : i386 tested, x86_64 support Christoph Lameter
2004-09-02 21:07 ` page fault scalability patch final : i386 tested, x86_64 David S. Miller
2004-09-18 23:23 ` page fault scalability patch V8: [0/7] Description Christoph Lameter
[not found] ` <B6E8046E1E28D34EB815A11AC8CA312902CD3243@mtv-atc-605e--n.corp.sgi.com>
2004-09-18 23:24 ` page fault scalability patch V8: [1/7] make mm->rss atomic Christoph Lameter
2004-09-18 23:26 ` page fault scalability patch V8: [2/7] avoid page_table_lock in Christoph Lameter
2004-09-19 9:04 ` page fault scalability patch V8: [2/7] avoid page_table_lock in handle_mm_fault Christoph Hellwig
2004-09-18 23:27 ` page fault scalability patch V8: [3/7] atomic pte operations for Christoph Lameter
2004-09-18 23:28 ` page fault scalability patch V8: [4/7] universally available cmpxchg Christoph Lameter
[not found] ` <200409191430.37444.vda@port.imtp.ilyichevsk.odessa.ua>
2004-09-19 12:11 ` page fault scalability patch V8: [4/7] universally available cmpxchg on i386 Andi Kleen
2004-09-20 15:45 ` page fault scalability patch V8: [4/7] universally available Christoph Lameter
[not found] ` <200409202043.00580.vda@port.imtp.ilyichevsk.odessa.ua>
2004-09-20 20:49 ` Christoph Lameter
2004-09-20 20:57 ` page fault scalability patch V8: [4/7] universally available cmpxchg on i386 Andi Kleen
[not found] ` <200409211841.25507.vda@port.imtp.ilyichevsk.odessa.ua>
2004-09-21 15:45 ` Andi Kleen
[not found] ` <200409212306.38800.vda@port.imtp.ilyichevsk.odessa.ua>
2004-09-21 20:14 ` Andi Kleen
2004-09-23 7:17 ` page fault scalability patch V8: [4/7] universally available Andy Lutomirski
2004-09-23 9:03 ` page fault scalability patch V8: [4/7] universally available cmpxchg on i386 Andi Kleen
[not found] ` <Pine.LNX.4.58.0409271204180.31769@schroedinger.engr.sgi.com>
2004-10-15 19:02 ` page fault scalability patch V10: [0/7] overview Christoph Lameter
2004-10-15 19:03 ` page fault scalability patch V10: [1/7] make rss atomic Christoph Lameter
2004-10-15 19:04 ` page fault scalability patch V10: [2/7] defer/omit taking page_table_lock Christoph Lameter
2004-10-15 20:00 ` Marcelo Tosatti
2004-10-18 15:59 ` page fault scalability patch V10: [2/7] defer/omit taking Christoph Lameter
2004-10-19 5:25 ` [revised] " Christoph Lameter
2004-10-15 19:05 ` page fault scalability patch V10: [3/7] IA64 atomic pte operations Christoph Lameter
2004-10-15 19:06 ` page fault scalability patch V10: [4/7] cmpxchg for 386 and 486 Christoph Lameter
2004-10-15 19:06 ` page fault scalability patch V10: [5/7] i386 atomic pte operations Christoph Lameter
2004-10-15 19:07 ` page fault scalability patch V10: [6/7] x86_64 " Christoph Lameter
2004-10-15 19:08 ` page fault scalability patch V10: [7/7] s/390 " Christoph Lameter
2004-09-18 23:29 ` page fault scalability patch V8: [5/7] atomic pte operations for Christoph Lameter
2004-09-18 23:30 ` page fault scalability patch V8: [6/7] " Christoph Lameter
2004-09-18 23:31 ` page fault scalability patch V8: [7/7] " Christoph Lameter
[not found] ` <200409191435.09445.vda@port.imtp.ilyichevsk.odessa.ua>
2004-09-20 15:44 ` page fault scalability patch V8: [7/7] atomic pte operations Christoph Lameter
2004-08-15 22:38 ` page fault fastpath: Increasing SMP scalability by introducing Benjamin Herrenschmidt
2004-08-16 17:28 ` Christoph Lameter
2004-08-17 8:01 ` Benjamin Herrenschmidt
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=Pine.LNX.4.58.0408240743180.3853@server.home \
--to=christoph@lameter.com \
--cc=ak@muc.de \
--cc=benh@kernel.crashing.org \
--cc=clameter@sgi.com \
--cc=davem@redhat.com \
--cc=linux-ia64@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=manfred@colorfullife.com \
--cc=raybry@sgi.com \
--cc=vrajesh@umich.edu \
--cc=willy@debian.org \
--cc=wli@holomorphy.com \
/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