>On Sat, 17 May 2003, Todd Walton wrote:
>Also, what's the status of NIO?  Do we know yet the effect it will have on
>load?  Also, we have yet to see what insights performance profiling will
>give.

One thing that I know for sure is that nio will definitely make profiling
performance easier. Any time that is spent performing taskswitches makes the
collected performance data contain values further from reality. In my
opinion, as long as you have significantly more threads running, or waiting
blocked, than you have processors you will always suffer performancewise
unless you are actually FORCED to have that many threads to get acceptable
performance.. *cough*non-nio*cough*. (I deliberatly ignore talking about
other things than performance that threads might be good for..
programatically and otherwise they can be very useful..)

That and the fact that fred usually stands for 80-90% of all  threads
running on my one-CPU machine makes me vote for a nio-as-fast-as-possible
soloution.

/N

_______________________________________________
devl mailing list
devl at freenetproject.org
http://hawk.freenetproject.org:8080/cgi-bin/mailman/listinfo/devl

Reply via email to