Re: Solaris real-time thread capabilities

From: Greg Menke (gregm-xyzpdq_at_toadmail.com)
Date: 09/07/05 

Date: 06 Sep 2005 20:24:15 -0400

capnwhit@yahoo.com writes:

> Greg Menke wrote:
> > Have you run empty loops using those various techniques to see
> > if you can achieve the timing you're looking for?
> >
> Hello Greg, All of my tests so far have been using empty loops! I can't
> get the timing right even on an empty loop... that is why I didn't
> bother explaining the size constraints for our jitter buffers, our
> packages, or the rest of our app.
>
> > there wasn't even the results of prstat showing how busy
> > the cpu was while executing your job.
> >
> The reason I didn't post this is because it was a non-issue (1 or
> 2%)... like I said the test loop is empty, just calling nanosleep to
> try and achieve the timing... When using the spin loop technique, of
> course, the CPU usage went up, but I only tried said technique when
> using a processor set, and only when my test program was the only LWP
> assigned to the target processor. Therefore the system still had one
> entire processor in which to schedule other LWPs.

Ahh thats the kind of info that we need. I'll sniff around as well.

Gregm