Re: ULE 2.0

From: Jeff Roberson <jroberson_at_chesapeake.net>
Date: Fri, 5 Jan 2007 17:39:16 -0800 (PST)
On Fri, 5 Jan 2007, Jeff Roberson wrote:

> On Fri, 5 Jan 2007, David Xu wrote:
>
>> 
>>> Anyway, thanks for the pointer Xu.  I may hack this up and compare it to 
>>> ULE's current balancing.
>> 
>> Yesterday, I have tested super-smack benchmark on 2-cpu machine, ULE
>> decreased performance about 40%, this might be a regression though.
>
> I just fixed a regression in the load balancer.  On my 8way opteron ULE is 
> now 22% faster than 4BSD for select-key.smack with 32 threads.  This is for 
> both KSE and libthr.  libthr is of course significantly faster overall.  I 
> believe I can improve this even more by borrowing some tricks from solaris.

An initial implementation of what I felt were the most significant parts 
of the solaris algorithm pushed it up to 50% faster than 4BSD.  I probably 
won't have this ready to commit for a few days however.

Thanks,
Jeff.

>
> Jeff
>
>> 
>>> 
>>> Cheers,
>>> Jeff
>>> 
>> 
>> Regards,
>> David Xu
>> 
>> _______________________________________________
>> freebsd-current_at_freebsd.org mailing list
>> http://lists.freebsd.org/mailman/listinfo/freebsd-current
>> To unsubscribe, send any mail to "freebsd-current-unsubscribe_at_freebsd.org"
>> 
> _______________________________________________
> freebsd-current_at_freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-current
> To unsubscribe, send any mail to "freebsd-current-unsubscribe_at_freebsd.org"
>
Received on Sat Jan 06 2007 - 00:40:42 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:04 UTC