On Thu, Mar 11, 2004 at 08:29:30PM +0100, Brad Knowles wrote:
> This is a transition period. 4BSD took a long time to get where
> it is today, and it's going to take a while to get ULE to the point
> where it is always better than 4BSD. But we have to make the
> cut-over somewhere, and we have to get people using it more widely.
> If we don't, then ULE will never get to the point where it could be
> as good as 4BSD, much less better.
Who is actively working on ULE, getting feedback from users,
improving ULE, etc.? I asked some questions about the
late tool used to evaluate ULE, and at first didn't get any responses,
but was finally directed to look at: http://www.chesapeake.net/~jroberson/late.tgz
I also mentioned on freebsd-arch that the following tool developed
at University of Utah might be useful for evaluating scheduler performance
on FreeBSD:
http://www.cs.utah.edu/~regehr/hourglass/
I know a lot of researchers who are interested in operating system
schedulers (especially for real-time systems), so leveraging off of their
work couldn't hurt, and might lead to a better ULE implementation.
--
Craig Rodrigues
http://crodrigues.org
rodrigc@crodrigues.org