From mboxrd@z Thu Jan  1 00:00:00 1970
From: Peter Zijlstra <peterz@infradead.org>
Subject: Re: [PATCH -v8][RFC] mutex: implement adaptive spinning
Date: Mon, 12 Jan 2009 17:50:15 +0100
Message-ID: <1231779015.4371.138.camel@laptop>
References: <1231774622.4371.96.camel@laptop>
	 <alpine.LFD.2.00.0901120802430.6528@localhost.localdomain>
	 <alpine.LFD.2.00.0901120818070.6528@localhost.localdomain>
	 <1231778757.22806.24.camel@think.oraclecorp.com>
Mime-Version: 1.0
Content-Type: text/plain
Content-Transfer-Encoding: 7bit
Cc: Linus Torvalds <torvalds@linux-foundation.org>,
	Ingo Molnar <mingo@elte.hu>,
	"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>,
	Gregory Haskins <ghaskins@novell.com>,
	Matthew Wilcox <matthew@wil.cx>,
	Andi Kleen <andi@firstfloor.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	linux-fsdevel <linux-fsdevel@vger.kernel.org>,
	linux-btrfs <linux-btrfs@vger.kernel.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	Nick Piggin <npiggin@suse.de>,
	Peter Morreale <pmorreale@novell.com>,
	Sven Dietrich <SDietrich@novell.com>,
	Dmitry Adamushko <dmitry.adamushko@gmail.com>
To: Chris Mason <chris.mason@oracle.com>
Return-path: <linux-btrfs-owner@vger.kernel.org>
In-Reply-To: <1231778757.22806.24.camel@think.oraclecorp.com>
Sender: linux-btrfs-owner@vger.kernel.org
List-Id: linux-fsdevel.vger.kernel.org

On Mon, 2009-01-12 at 11:45 -0500, Chris Mason wrote:
> On Mon, 2009-01-12 at 08:20 -0800, Linus Torvalds wrote:
> > 
> > On Mon, 12 Jan 2009, Linus Torvalds wrote:
> > > 
> > > You made it back into the locked version.
> > 
> > Btw, even if you probably had some reason for this, one thing to note is 
> > that I think Chris' performance testing showed that the version using a 
> > lock was inferior to his local btrfs hack, while the unlocked version 
> > actually beat his hack.
> > 
> 
> The spinning hack was faster than everything before v7 (we'll call it
> the Linus-fix), and the v7 code was much faster than my spin.
> 
> This is somewhere in between, with slightly better fairness than v7.
> 
>              spin         v7         v8
> dbench 50    580MB/s      789MB/s    421MB/s
> file creates 152 file/s   162 file/s 195 file/s
> file stat    3.8s total   2.3s total 5.3s total
> 
> (the file stat run is total run time, so lower is better.  The other
> numbers are files or MB per second, so higher is better)
> 
> For the file create run, v8 had much lower system time than v7,
> averaging 1s of sys time per proc instead of 1.6s.

Right, how about the spread in completion time, because that is the only
reason I tried this fairness stuff, because you reported massive
differences there.