linux-mips
[Top] [All Lists]

Re: [PATCH] MIPS: Optimize spinlocks.

To: David Daney <ddaney@caviumnetworks.com>
Subject: Re: [PATCH] MIPS: Optimize spinlocks.
From: Ralf Baechle <ralf@linux-mips.org>
Date: Wed, 24 Feb 2010 16:54:24 +0100
Cc: linux-mips@linux-mips.org
In-reply-to: <20100224155336.GA5130@linux-mips.org>
Original-recipient: rfc822;linux-mips@linux-mips.org
References: <1265311909-1679-1-git-send-email-ddaney@caviumnetworks.com> <20100224155336.GA5130@linux-mips.org>
Sender: linux-mips-bounce@linux-mips.org
User-agent: Mutt/1.5.20 (2009-08-17)
On Wed, Feb 24, 2010 at 04:53:36PM +0100, Ralf Baechle wrote:

> And in your benchmarking patch you wrote:
> 
> >               spin_single   spin_multi
> > base                  106885        247941
> > spinlock_patch  75194               219465
> 
> I did some benchmarking on an IP27 (180MHz, 2 CPU, needs LL/SC workaround):
> 
>               spin_single     spin_multi
> base          229341          3505690
> spinlock_patch        177847          3615326
> 
> So about 22% speedup for spin_single but 3% slowdown for spin_multi.
> 
> Disabling the R10k LL/SC workaround btw. gives another 23% speedup for
> spin_single and marginal 0.3% for spin_multi; the latter may well be
> statistical noise.

And I forgot the most important - patch queued for 2.6.34.

Thanks!

  Ralf

<Prev in Thread] Current Thread [Next in Thread>