Date: Mon, 10 Oct 2005 23:21:43 +0800
From: WU Fengguang <wfg@mail.ustc.edu.cn>
To: Yingchao Zhou <yc_zhou@ncic.ac.cn>
Cc: linux-kernel <linux-kernel@vger.kernel.org>
Subject: Re: [RFC] use radix_tree for non-resident page tracking
Message-ID: <20051010152141.GA6714@mail.ustc.edu.cn>
Mail-Followup-To: WU Fengguang <wfg@mail.ustc.edu.cn>,
	Yingchao Zhou <yc_zhou@ncic.ac.cn>,
	linux-kernel <linux-kernel@vger.kernel.org>
References: <20051010134345.C9E7CFB046@gatekeeper.ncic.ac.cn>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20051010134345.C9E7CFB046@gatekeeper.ncic.ac.cn>
User-Agent: Mutt/1.5.9i
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1103
Lines: 24

On Mon, Oct 10, 2005 at 09:37:47PM +0800, Yingchao Zhou wrote:
> The paper "The Performance Impact of Kernel Prefetching On Buffer 
> Cache Replacement Algorithms" appeared in SIGMETRICS05 gives some 
> interesting relative results of replacement algorithm when taking 
Thanks for the information, it should help my work on read-ahead.
I noticed similar problem when testing read-ahead behavior for fs-on-loop.
The basic conclusion is read-ahead in the block device level can
complement file level read-ahead, for the latter does not do inter-file
read-ahead.  But it has to be limited/conservative, because the risk of
cache miss is much higher.
> into account of prefetching. How about the situation in CLOCK-Pro
> project?
Sorry, I'm new to CLOCK-Pro, too...

-- 
WU Fengguang
Dept. of Automation
University of Science and Technology of China
Hefei, Anhui
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/