[Discuss-gnuradio] Eric, will give you a call this week.

[David Beberman]

Will try to reach you this week.

Your concept of shifting pieces back and forth between processors sounds
interesting.  Not sure how you accomplish that, or are planning to.
Sounded like you wanted it to happen dynamically.  Should be interesting to
read.   I'm guessing that it looks a lot different then the previous 
architecture for
threading. 

On realtime kernel issues, Timesys makes claims that their approach to 
kernel preemption is superior to 2.6.  I'll have to read their docs 
again, but I think the claim is that with SMP, their approach is 
superior because they don't lock up all processes on a processor when in 
a critical section.  They don't use spinlocks.  The also have their own 
wrapper for device drivers to prioritize interrupt handlers.  Only 
looked through their docs so far, last week.  Can't tell what is really 
there, and have yet to look at 2.6. 

I will look through the sections you recommended this week as well.

David