| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Wed, 15 Jun 2011 22:32:17 +0200
From: Peter Zijlstra <a.p.zijlstra@...llo.nl>
To: Tim Chen <tim.c.chen@...ux.intel.com>
Cc: Andi Kleen <ak@...ux.intel.com>, Shaohua Li <shaohua.li@...el.com>,
Andrew Morton <akpm@...ux-foundation.org>,
Linus Torvalds <torvalds@...ux-foundation.org>,
Hugh Dickins <hughd@...gle.com>,
KOSAKI Motohiro <kosaki.motohiro@...fujitsu.com>,
Benjamin Herrenschmidt <benh@...nel.crashing.org>,
David Miller <davem@...emloft.net>,
Martin Schwidefsky <schwidefsky@...ibm.com>,
Russell King <rmk@....linux.org.uk>,
Paul Mundt <lethal@...ux-sh.org>,
Jeff Dike <jdike@...toit.com>,
Richard Weinberger <richard@....at>,
"Luck, Tony" <tony.luck@...el.com>,
KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>,
Mel Gorman <mel@....ul.ie>, Nick Piggin <npiggin@...nel.dk>,
Namhyung Kim <namhyung@...il.com>,
"Shi, Alex" <alex.shi@...el.com>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"linux-mm@...ck.org" <linux-mm@...ck.org>,
"Rafael J. Wysocki" <rjw@...k.pl>
Subject: Re: REGRESSION: Performance regressions from switching
anon_vma->lock to mutex
On Wed, 2011-06-15 at 11:43 -0700, Tim Chen wrote:
> Wonder if you can provide the profile on your run so I can compare with
> what I got on 4 sockets?
Sure, so this is on an Westmere-EP (2 sockets, 6 cores/socket, 2
threads/core), what I did was:
perf record -r2 -gf make bench
perf report > foo.txt
bzip2 -9 foo.txt
Both files are about 0.5M, the tip one has the sirq-rcu patch and linus'
patch applied (could do one without if wanted).
http://programming.kicks-ass.net/sekrit/tip.txt.bz2
http://programming.kicks-ass.net/sekrit/39.txt.bz2
However, looking at them, the weird thing is, they're both dominated by
(taken from 39.txt):
7.44% exim [kernel.kallsyms] [k] format_decode
|
--- format_decode
|
|--93.07%-- vsnprintf
| |
| |--98.83%-- seq_printf
| | show_stat
| | seq_read
| | proc_reg_read
| | vfs_read
| | sys_read
| | system_call
| | __GI___libc_read
| | |
| | |--99.47%-- (nil)
| | --0.53%-- [...]
| |
| --1.17%-- snprintf
| proc_flush_task
| release_task
| wait_consider_task
| do_wait
| sys_wait4
| system_call
| |
| |--93.15%-- __libc_wait
| |
| --6.85%-- __waitpid
|
|--6.84%-- seq_printf
| show_stat
| seq_read
| proc_reg_read
| vfs_read
| sys_read
| system_call
| __GI___libc_read
| |
| |--99.56%-- (nil)
| --0.44%-- [...]
--0.10%-- [...]
I've no idea why its doing that, I've had massive trouble getting this
MOSBENCH crap working in the first place since its all in python, but
what I basically done was rip out everything !exim in config.py and put
cores = [24]. In hosts.py I too ripped out everything !exim, cleared out
the clients list and made 'tom' my localhost (removing that perflock
thing).
After that things more or less ran, I saw exim, and its giving me those
msgs/sec/core numbers like:
# perf record -r2 -gfo 39.perf.data make bench
python config.py
Starting results in: results/20110615-221914
*** Starting configuration 1/1 (benchmark-exim) ***
Starting Host.host-westmere...
sending westmere: /./
del.ing westmere: out/log/EximLoad.trial-2.host-westmere
del.ing westmere: out/log/EximLoad.trial-1.host-westmere
del.ing westmere: out/log/EximLoad.trial-0.host-westmere
del.ing westmere: out/log/
del.ing westmere: out/EximDaemon.host-westmere.configure
del.ing westmere: out/
sending westmere: /home/root/
sending westmere: /home/root/test/mosbench/
Starting Host.host-westmere... done
Starting HostInfo.host-westmere...
Starting HostInfo.host-westmere... done
Starting FileSystem.host-westmere.fstype-tmpfs-separate...
Starting FileSystem.host-westmere.fstype-tmpfs-separate... done
Starting SetCPUs.host-westmere...
FATAL: Module oprofile not found.
FATAL: Module oprofile not found.
Kernel doesn't support oprofile
CPUs 0-23 are online
CPUs 0-23 are online
Starting SetCPUs.host-westmere... done
Starting EximDaemon.host-westmere...
Starting EximDaemon.host-westmere... done
Waiting on EximLoad.trial-0.host-westmere...
[EximLoad.trial-0.host-westmere] => 86983 messages (15.0032 secs, 241.568 messages/sec/core)
Waiting on EximLoad.trial-0.host-westmere... done
Waiting on EximLoad.trial-1.host-westmere...
[EximLoad.trial-1.host-westmere] => 86770 messages (15.004 secs, 240.964 messages/sec/core)
Waiting on EximLoad.trial-1.host-westmere... done
Waiting on EximLoad.trial-2.host-westmere...
[EximLoad.trial-2.host-westmere] => 86987 messages (15.0035 secs, 241.574 messages/sec/core)
Waiting on EximLoad.trial-2.host-westmere... done
Stopping EximDaemon.host-westmere...
Stopping EximDaemon.host-westmere... done
Stopping HostInfo.host-westmere...
Stopping HostInfo.host-westmere... done
Stopping Host.host-westmere...
copying westmere: ./
copying westmere: EximDaemon.host-westmere.configure
copying westmere: log/
copying westmere: log/EximLoad.trial-0.host-westmere
copying westmere: log/EximLoad.trial-1.host-westmere
copying westmere: log/EximLoad.trial-2.host-westmere
Stopping Host.host-westmere... done
Stopping ResultPath...
Results in: results/20110615-221914/benchmark-exim
Stopping ResultPath... done
All results in: results/20110615-221914
[ perf record: Woken up 3774 times to write data ]
[ perf record: Captured and wrote 979.494 MB 39.perf.data (~42794760 samples) ]
CPUs 0-23 are online
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists