Lance ..

 "L Anyway, the problem is this - with node 1 only 
 "L running, CPU usage is around
 "L 3%, when the second node's started, it goes up to 99%.

 "L Any ideas why this might be the case ?

 "L Even with Tame running, it takes *both* VDM's up to 
 "L around 20% each (which
 "L shouldn't happen).

I'm not suggesting an answer, just have a further question.  Long ago and
far away in the WIN-95/98 world, it was said in the OS/2 circles that the
prime reason for this sort of thing and also the inability to run more than
one communications related DOS application at a time was this; WIN at that
time was still a process-oriented issue, with respect to task switching. 
OS/2 on the other hand, was a thread-oriented issue, with respect to task
switching. Thus, although the exact problem you describe wasn't the issue
for work we were doing, but something else maybe related might be.

We found that you simply could kiss goodbye to more than one instance of
running some of my older DOS executables on a WIN95/98 box.  The whole box
slowed to a crawl when the second application was started on a given box.
In contrast, I've had as many as half a dozen DOS-VDM's running in OS/2
with my programs happy as a clam with each other, and with multiple comm
port I/O at the same time.

Your description of your problem makes me wonder.  Is WIN-XP still
suffering from this same lack of thread-oriented task pre-emption, and
thus, is still not a true thread pre-emptive multi-tasker?  If the answer
to that question is, "Yes!", then maybe that is a start toward an
answer for you.  But it will take people who really know WIN-XP to get you
the full answer, even if the above is a partial one.

Just curious here, too.


--> Sleep well; OS/2's still awake! ;)

Mike {at} 1:117/3001

--- Maximus/2 3.01