MM performance benchmark

MM performance benchmark

[Randy Appleton]

[-- Attachment #1.1: Type: text/plain, Size: 799 bytes --]

Hi!

I'm a professor of computer science at Northern Michigan University.
Three students and myself
have been benchmarking the Linux kernel.  At
http://euclid.nmu.edu/~benchmark you will see
graphs describing performance for mmap() and page faults.

Both graphs show a huge improvement between 2.2 and 2.3.  We see a 100x
performance
gain.  My questions is ... Why has performance improved so much?  What
changed between
2.2 and 2.3 to account for a 100x performance improvement?

-Much Thanks
-Randy

--
========================================================================
|| Randy Appleton, Professor of Computer Science at Northern Michigan ||
|| University.  And a big fan of Linux!                               ||
================= mailto:randy@euclid.nmu.edu ==========================



[-- Attachment #1.2: Type: text/html, Size: 1193 bytes --]

[-- Attachment #2: Card for Randy Appleton --]
[-- Type: text/x-vcard, Size: 193 bytes --]

begin:vcard 
n:Appleton;Randy
x-mozilla-html:TRUE
org:Northern Michigan University
adr:;;;;;;
version:2.1
email;internet:rappleto@nmu.edu
title:Dr.
x-mozilla-cpt:;0
fn:Randy Appleton
end:vcard

Re: MM performance benchmark

[Eric W. Biederman]

Randy Appleton <rappleto@nmu.edu> writes:

> Hi!
> 
> I'm a professor of computer science at Northern Michigan University.
> Three students and myself
> have been benchmarking the Linux kernel.  At
> http://euclid.nmu.edu/~benchmark you will see
> graphs describing performance for mmap() and page faults.
> 
> Both graphs show a huge improvement between 2.2 and 2.3.  We see a 100x
> performance
> gain.  My questions is ... Why has performance improved so much?  What
> changed between
> 2.2 and 2.3 to account for a 100x performance improvement?

The page cache usage was rewritten to be threaded, and more
importantly to not double buffer data between the page cache and the
buffer cache.  

Eric
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/