From: Sven-Thorsten Dietrich <sdietrich@novell.com>
To: "Peter W. Morreale" <pmorreale@novell.com>
Cc: paulmck@linux.vnet.ibm.com,
"Bill Huey (hui)" <bill.huey@gmail.com>, Andi Kleen <ak@suse.de>,
Gregory Haskins <ghaskins@novell.com>,
mingo@elte.hu, a.p.zijlstra@chello.nl, tglx@linutronix.de,
rostedt@goodmis.org, linux-rt-users@vger.kernel.org,
linux-kernel@vger.kernel.org, kevin@hilman.org,
cminyard@mvista.com, dsingleton@mvista.com, dwalker@mvista.com,
npiggin@suse.de, dsaxena@plexity.net, gregkh@suse.de,
mkohari@novell.com
Subject: Re: [PATCH [RT] 08/14] add a loop counter based timeout mechanism
Date: Fri, 22 Feb 2008 23:36:30 -0800 [thread overview]
Message-ID: <1203752190.4772.192.camel@sven.thebigcorporation.com> (raw)
In-Reply-To: <1203712566.4147.56.camel@hermosa.morrealenet>
On Fri, 2008-02-22 at 13:36 -0700, Peter W. Morreale wrote:
> On Fri, 2008-02-22 at 11:55 -0800, Sven-Thorsten Dietrich wrote:
> >
> > In high-contention, short-hold time situations, it may even make sense
> > to have multiple CPUs with multiple waiters spinning, depending on
> > hold-time vs. time to put a waiter to sleep and wake them up.
> >
> > The wake-up side could also walk ahead on the queue, and bring up
> > spinners from sleeping, so that they are all ready to go when the lock
> > flips green for them.
> >
>
> I did try an attempt at this one time. The logic was merely if the
> pending owner was running, wakeup the next waiter. The results were
> terrible for the benchmarks used, as compared to the current
> implementation.
Yup, but you cut the CONTEXT where I said:
"for very large SMP systems"
Specifically, what I mean, is an SMP system, where I have enough CPUs to
do this:
let (t_Tcs) be the time to lock, transition and unlock an un-contended
critical section (i.e. the one that I am the pending waiter for).
let (t_W) be the time to wake up a sleeping task.
and let (t_W > t_Tcs)
Then, "for very large SMP systems"
if
S = (t_W / t_Tcs),
then S designates the number of tasks to transition a critical section
before the first sleeper would wake up.
and the number of CPUs > S.
The time for an arbitrary number of tasks N > S which are all competing
for lock L, to transition a critical section (T_N_cs), approaches:
T_N_cs = (N * t_W)
if you have only 1 task spinning.
but if you can have
N tasks spinning, (T_N_cs) approaches:
T_N_cs = (N * t_Tcs)
and with the premise, that t_W > t_Tcs, you should see a dramatic
throughput improvement when running PREEMPT_RT on VERY LARGE SMP
systems.
I want to disclaim, that the math above is very much simplified, but I
hope its sufficient to demonstrate the concept.
I have to acknowledge Ingo's comments, that this is all suspect until
proven to make a positive difference in "non-marketing" workloads.
I personally *think* we are past that already, and the adaptive concept
can and will be extended and scaled as M-socket and N-core based SMP
proliferates into to larger grid-based systems. But there is plenty more
to do to prove it.
(someone send me a 1024 CPU box and a wind-powered-generator)
Sven
>
> What this meant was that virtually every unlock performed a wakeup, if
> not for the new pending owner, than the next-in-line waiter.
>
> My impression at the time was that the contention for the rq lock is
> significant, regardless of even if the task being woken up was already
> running.
>
> I can generate numbers if that helps.
>
> -PWM
>
>
>
next prev parent reply other threads:[~2008-02-23 7:37 UTC|newest]
Thread overview: 59+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-02-21 15:26 [PATCH [RT] 00/14] RFC - adaptive real-time locks Gregory Haskins
2008-02-21 15:26 ` [PATCH [RT] 01/14] spinlocks: fix preemption feature when PREEMPT_RT is enabled Gregory Haskins
2008-02-21 15:26 ` [PATCH [RT] 02/14] spinlock: make preemptible-waiter feature a specific config option Gregory Haskins
2008-02-22 19:09 ` Pavel Machek
2008-02-21 15:26 ` [PATCH [RT] 03/14] x86: FIFO ticket spinlocks Gregory Haskins
2008-02-21 15:26 ` [PATCH [RT] 04/14] disable PREEMPT_SPINLOCK_WAITERS when x86 ticket/fifo spins are in use Gregory Haskins
2008-02-21 15:26 ` [PATCH [RT] 05/14] rearrange rt_spin_lock sleep Gregory Haskins
2008-02-22 13:29 ` Gregory Haskins
2008-02-22 13:35 ` Steven Rostedt
2008-02-22 13:40 ` Peter Zijlstra
2008-02-22 13:35 ` Ingo Molnar
2008-02-22 13:43 ` Steven Rostedt
2008-02-22 13:46 ` Steven Rostedt
2008-02-21 15:26 ` [PATCH [RT] 06/14] optimize rt lock wakeup Gregory Haskins
2008-02-21 15:27 ` [PATCH [RT] 07/14] adaptive real-time lock support Gregory Haskins
2008-02-22 19:14 ` Pavel Machek
2008-02-21 15:27 ` [PATCH [RT] 08/14] add a loop counter based timeout mechanism Gregory Haskins
2008-02-21 16:41 ` Andi Kleen
2008-02-21 17:02 ` Gregory Haskins
2008-02-21 17:04 ` Peter W. Morreale
2008-02-21 17:06 ` Sven-Thorsten Dietrich
2008-02-22 19:08 ` Paul E. McKenney
2008-02-22 19:19 ` Bill Huey (hui)
2008-02-22 19:21 ` Bill Huey (hui)
2008-02-22 19:43 ` Paul E. McKenney
2008-02-22 19:55 ` Sven-Thorsten Dietrich
2008-02-22 20:23 ` Paul E. McKenney
2008-02-22 22:03 ` Gregory Haskins
2008-02-23 12:31 ` Andi Kleen
2008-02-23 16:32 ` Paul E. McKenney
2008-02-25 23:52 ` Sven-Thorsten Dietrich
2008-02-22 20:36 ` Peter W. Morreale
2008-02-23 7:36 ` Sven-Thorsten Dietrich [this message]
2008-02-22 20:15 ` Peter W. Morreale
2008-02-21 15:27 ` [PATCH [RT] 09/14] adaptive mutexes Gregory Haskins
2008-02-21 15:27 ` [PATCH [RT] 10/14] adjust pi_lock usage in wakeup Gregory Haskins
2008-02-21 16:48 ` Steven Rostedt
2008-02-21 17:09 ` Peter W. Morreale
2008-02-21 15:27 ` [PATCH [RT] 11/14] optimize the !printk fastpath through the lock acquisition Gregory Haskins
2008-02-21 16:36 ` Andi Kleen
2008-02-21 16:47 ` Gregory Haskins
2008-02-22 19:18 ` Pavel Machek
2008-02-22 22:20 ` Gregory Haskins
2008-02-23 0:43 ` Bill Huey (hui)
2008-02-25 5:20 ` Gregory Haskins
2008-02-25 6:21 ` Bill Huey (hui)
2008-02-25 9:02 ` Bill Huey (hui)
2008-02-21 15:27 ` [PATCH [RT] 12/14] remove the extra call to try_to_take_lock Gregory Haskins
2008-02-21 15:27 ` [PATCH [RT] 13/14] allow rt-mutex lock-stealing to include lateral priority Gregory Haskins
2008-02-21 15:27 ` [PATCH [RT] 14/14] sysctl for runtime-control of lateral mutex stealing Gregory Haskins
2008-02-21 16:05 ` [PATCH [RT] 00/14] RFC - adaptive real-time locks Gregory Haskins
2008-02-21 21:24 ` Ingo Molnar
2008-02-21 21:33 ` Bill Huey (hui)
[not found] ` <20080221214219.GA27209@elte.hu>
2008-02-21 21:56 ` Gregory Haskins
2008-02-21 22:53 ` Bill Huey (hui)
2008-02-21 21:40 ` Gregory Haskins
2008-02-21 22:12 ` Peter W. Morreale
2008-02-21 22:42 ` Peter W. Morreale
2008-02-23 8:03 ` Andrew Morton
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=1203752190.4772.192.camel@sven.thebigcorporation.com \
--to=sdietrich@novell.com \
--cc=a.p.zijlstra@chello.nl \
--cc=ak@suse.de \
--cc=bill.huey@gmail.com \
--cc=cminyard@mvista.com \
--cc=dsaxena@plexity.net \
--cc=dsingleton@mvista.com \
--cc=dwalker@mvista.com \
--cc=ghaskins@novell.com \
--cc=gregkh@suse.de \
--cc=kevin@hilman.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-rt-users@vger.kernel.org \
--cc=mingo@elte.hu \
--cc=mkohari@novell.com \
--cc=npiggin@suse.de \
--cc=paulmck@linux.vnet.ibm.com \
--cc=pmorreale@novell.com \
--cc=rostedt@goodmis.org \
--cc=tglx@linutronix.de \
/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
all inboxes | Powered by JetHome®