lkml.org 
[lkml]   [1999]   [Jun]   [9]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectResults of schedule() switching test with my patches ...
Date
Hi guys,

I've done a patch to 2.3.5 that will improve schedule() switching speed
from 40 up to 70 %.
This patch use a set of "slot-of-goodness" in which schedule() search
the best task to run.
This results in a hi schedule() speed since the best task can be found
even in only 1 test while the previous one scan the entire runqueue list.
With this mail I include the patch and the little program
( great THX to Willy Terreau for his contribute ) used to test it.
The test program is very simple, it creates 450 threads that do nothing
but increment a counter and call syscall(SYS_sched_yield).
My results are:

2.3.5 Test 1 = 6169
2.3.5 Test 2 = 6087

Patch Test 1 = 8811
Patch Test 2 = 10325

Average gain = + 56 % !!!!

I'll be happy if this patch will be tested by other Linux users ( better if
in high-loaded
environmets ) to get other results.

Good work to everybody, cheers,
Davide.

--
"Debian, the Freedom in Freedom."


[unhandled content-type:application/x-gzip][unhandled content-type:application/x-gzip]
\
 
 \ /
  Last update: 2005-03-22 13:52    [W:0.017 / U:0.400 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site