From: Linus Torvalds <torvalds@linux-foundation.org>
To: Rik van Riel <riel@redhat.com>
Cc: Ingo Molnar <mingo@kernel.org>, "H. Peter Anvin" <hpa@zytor.com>,
Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
Peter Zijlstra <a.p.zijlstra@chello.nl>,
aquini@redhat.com, Andrew Morton <akpm@linux-foundation.org>,
Thomas Gleixner <tglx@linutronix.de>,
Michel Lespinasse <walken@google.com>,
linux-tip-commits@vger.kernel.org,
Steven Rostedt <rostedt@goodmis.org>,
"Vinod, Chegu" <chegu_vinod@hp.com>,
"Low, Jason" <jason.low2@hp.com>
Subject: Re: [tip:core/locking] x86/smp: Move waiting on contended ticket lock out of line
Date: Wed, 27 Feb 2013 09:10:55 -0800 [thread overview]
Message-ID: <CA+55aFzPxJ3rfZvAvrP9LxCXfDU8AAVvhMZnP-OAoe-ycc70aw@mail.gmail.com> (raw)
In-Reply-To: <512E376D.70105@redhat.com>
On Wed, Feb 27, 2013 at 8:42 AM, Rik van Riel <riel@redhat.com> wrote:
>
> To keep the results readable and relevant, I am reporting the
> plateau performance numbers. Comments are given where required.
>
> 3.7.6 vanilla 3.7.6 w/ backoff
>
> all_utime 333000 333000
> alltests 300000-470000 180000-440000 large variability
> compute 528000 528000
> custom 290000-320000 250000-330000 4 fast runs, 1 slow
> dbase 920000 925000
> disk 100000 90000-120000 similar plateau, wild
> swings with patches
> five_sec 140000 140000
> fserver 160000-300000 250000-430000 w/ patch drops off at
> higher number of users
> high_systime 80000-110000 30000-125000 w/ patch mostly 40k-70k,
> wild wings
> long no performance platform, equal performance for both
> new_dbase 960000 96000
> new_fserver 150000-300000 210000-420000 vanilla drops off,
> w/ patches wild swings
> shared 270000-440000 120000-440000 all runs ~equal to
> vanilla up to 1000
> users, one out of 5
> runs slows down past
> 1100 users
> short 120000 190000
Ugh. That really is rather random. "short" and fserver seems to
improve a lot (including the "new" version), the others look like they
are either unchanged or huge regressions.
Is there any way to get profiles for the improved versions vs the
regressed ones? It might well be that we have two different classes of
spinlocks. Maybe we could make the back-off version be *explicit* (ie
not part of the normal "spin_lock()", but you'd use a special
"spin_lock_backoff()" function for it) because it works well for some
cases but not for others?
Hmm? At the very least, it would give us an idea of *which* spinlock
it is that causes the most pain. I think your earlier indications was
that it's the mutex->wait_lock or something?
Linus
next prev parent reply other threads:[~2013-02-27 17:10 UTC|newest]
Thread overview: 54+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-02-06 20:03 [PATCH -v5 0/5] x86,smp: make ticket spinlock proportional backoff w/ auto tuning Rik van Riel
2013-02-06 20:04 ` [PATCH -v5 1/5] x86,smp: move waiting on contended ticket lock out of line Rik van Riel
2013-02-13 12:06 ` [tip:core/locking] x86/smp: Move " tip-bot for Rik van Riel
2013-02-13 16:20 ` Linus Torvalds
2013-02-13 18:30 ` Linus Torvalds
2013-02-14 0:54 ` H. Peter Anvin
2013-02-14 1:31 ` Linus Torvalds
2013-02-14 1:56 ` H. Peter Anvin
2013-02-14 10:50 ` Ingo Molnar
2013-02-14 16:10 ` Linus Torvalds
2013-02-15 15:57 ` Ingo Molnar
2013-02-15 6:48 ` Benjamin Herrenschmidt
2013-02-13 19:08 ` Rik van Riel
2013-02-13 19:36 ` Linus Torvalds
2013-02-13 22:21 ` Rik van Riel
2013-02-13 22:40 ` Linus Torvalds
2013-02-13 23:41 ` Rik van Riel
2013-02-14 1:21 ` Linus Torvalds
2013-02-14 1:46 ` Linus Torvalds
2013-02-14 10:43 ` Ingo Molnar
2013-02-27 16:42 ` Rik van Riel
2013-02-27 17:10 ` Linus Torvalds [this message]
2013-02-27 19:53 ` Rik van Riel
2013-02-27 20:18 ` Linus Torvalds
2013-02-27 21:55 ` Rik van Riel
[not found] ` <CA+55aFwa0EjGG2NUDYVLVBmXJa2k81YiuNO2yggk=GLRQxhhUQ@mail.gmail.com>
2013-02-28 2:58 ` Rik van Riel
2013-02-28 3:19 ` Linus Torvalds
2013-02-28 4:06 ` Davidlohr Bueso
2013-02-28 4:49 ` Linus Torvalds
2013-02-28 15:13 ` Rik van Riel
2013-02-28 18:22 ` Linus Torvalds
2013-02-28 20:26 ` Linus Torvalds
2013-02-28 21:14 ` Rik van Riel
2013-02-28 21:58 ` Linus Torvalds
2013-02-28 22:38 ` Rik van Riel
2013-02-28 23:09 ` Linus Torvalds
2013-03-01 6:42 ` Rik van Riel
2013-03-01 18:18 ` Davidlohr Bueso
2013-03-01 18:50 ` Rik van Riel
2013-03-01 18:52 ` Linus Torvalds
2013-02-06 20:04 ` [PATCH -v5 2/5] x86,smp: proportional backoff for ticket spinlocks Rik van Riel
2013-02-13 12:07 ` [tip:core/locking] x86/smp: Implement " tip-bot for Rik van Riel
2013-02-06 20:05 ` [PATCH -v5 3/5] x86,smp: auto tune spinlock backoff delay factor Rik van Riel
2013-02-13 12:08 ` [tip:core/locking] x86/smp: Auto " tip-bot for Rik van Riel
2013-02-06 20:06 ` [PATCH -v5 4/5] x86,smp: keep spinlock delay values per hashed spinlock address Rik van Riel
2013-02-13 12:09 ` [tip:core/locking] x86/smp: Keep " tip-bot for Eric Dumazet
2013-02-06 20:07 ` [PATCH -v5 5/5] x86,smp: limit spinlock delay on virtual machines Rik van Riel
2013-02-07 11:11 ` Ingo Molnar
2013-02-07 21:24 ` [PATCH fix " Rik van Riel
2013-02-13 12:10 ` [tip:core/locking] x86/smp: Limit " tip-bot for Rik van Riel
2013-02-07 11:25 ` [PATCH -v5 5/5] x86,smp: limit " Stefano Stabellini
2013-02-07 11:59 ` Raghavendra K T
2013-02-07 13:28 ` Rik van Riel
2013-02-06 20:08 ` [PATCH -v5 6/5] x86,smp: add debugging code to track spinlock delay value Rik van Riel
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=CA+55aFzPxJ3rfZvAvrP9LxCXfDU8AAVvhMZnP-OAoe-ycc70aw@mail.gmail.com \
--to=torvalds@linux-foundation.org \
--cc=a.p.zijlstra@chello.nl \
--cc=akpm@linux-foundation.org \
--cc=aquini@redhat.com \
--cc=chegu_vinod@hp.com \
--cc=hpa@zytor.com \
--cc=jason.low2@hp.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-tip-commits@vger.kernel.org \
--cc=mingo@kernel.org \
--cc=riel@redhat.com \
--cc=rostedt@goodmis.org \
--cc=tglx@linutronix.de \
--cc=walken@google.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;
as well as URLs for NNTP newsgroup(s).