[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <45B5703E.5010400@redhat.com>
Date: Mon, 22 Jan 2007 21:17:34 -0500
From: Rik van Riel <riel@...hat.com>
To: Christoph Lameter <clameter@....com>
CC: Balbir Singh <balbir@...ibm.com>,
Andrea Arcangeli <andrea@...e.de>,
Niki Hammler <mailinglists@...aq.net>,
linux-kernel@...r.kernel.org,
Vaidyanathan Srinivasan <svaidy@...ux.vnet.ibm.com>
Subject: Re: Why active list and inactive list?
Christoph Lameter wrote:
> On Mon, 22 Jan 2007, Rik van Riel wrote:
>
>> It would be really nice if we came up with a page replacement
>> algorithm that did not need many extra heuristics to make it
>> work...
>
> I guess the "clock" type algorithms are the most promising in that
> area. What happened to all those advanced page replacement endeavors?
> What is the most promising of those? You seem to have done a lot of work
> on those.
CLOCK-Pro seems the most promising algorithm, because it can
act well both as a first level cache (operating system running
applications) and as a second level cache (operating system
running as a file server), because it tracks both recency and
frequency well.
However, there are a few unanswered questions on clock-pro.
The big one is how we are to do some background aging in a
clock-pro system, so referenced bits don't just pile up when
the VM has enough memory - otherwise we might not know the
right pages to evict when a new process starts up and starts
allocating lots of memory.
At least we've solved the problems of keeping track of the
recently evicted pages in a cheap way, and balancing the
pressure/hotness of different caches against each other.
--
Politics is the struggle between those who want to make their country
the best in the world, and those who believe it already is. Each group
calls the other unpatriotic.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists