jeffr_tech ([info]jeffr_tech) wrote,
@ 2007-06-26 18:24:00
Previous Entry  Add to memories!  Tell a Friend  Next Entry
So I updated the image for the second time today to include Ingo's cfs scheduler. This kernel is from the rpm on his website. I double checked that it was not using tcmalloc at the time and switching back to a 2.6.21 kernel returned to the expected perf.

Basically, it has the same performance as the FreeBSD 4BSD scheduler now. Which is to say the peak is terrible but it has virtually no dropoff and performs better under load than the default 2.6.21 scheduler.

(Read 3 comments) - (Post a new comment)

my numbers are somewhat different
(Anonymous)
2007-06-27 02:39 am UTC (link)
Jeff, my numbers are somewhat different:
First row is number of threads and the second row is transactions / second.

* 2.6.21 *
#1 527
#2 1051
#4 870
#8 684
#16 665
#32 661

* 2.6.21 + cfsv18 *
#1 550
#2 1070
#4 1040
#8 996
#16 934
#32 878

This is on a Core2Duo 1,8Ghz and 1GB Ram. Your machine is much more powerful than mine, but I outperform you by a magnitude on 2 threads and I'm somewhat on par at 4 threads. How is this possible!? Compared to mainline 2.6.21 your numbers are considerably worse. By looking at the graph it seems that your and my 2.6.21 numbers are about the same (~1000).

(Reply to this) (Thread)
Re: my numbers are somewhat different
[info]jeffr_tech
2007-06-27 08:12 pm UTC (link)
Your individual processors are faster than mine so the times for low numbers of threads are better. With fewer processors there is also less contention so you don't see the poor behavior.

(Reply to this) (Parent)

(Read 3 comments) - (Post a new comment)

Create an Account
Forgot your login or password?
Login w/ OpenID
English • Español • Deutsch • Русский…