Re: More about timing

[Moon@MIT-AI]

I have done some investigation of CHAOS:GET-NEXT-PKT.  The results
are a little bizarre.  This function calls 2 other functions,
PROCESS-WAIT and TRANSMIT-STS.  Half the time essentially all of
GET-NEXT-PKT's time is taken up in PROCESS-WAIT.  The other half the
time PROCESS-WAIT takes less time, and TRANSMIT-STS, which isn't
called very often, is very slow (100 ms as compared to 10 ms) and
takes up the rest of the time.  The sum of the two times comes out
rather close to the same.  I suppose this must depend on timing of
stuff happening on the network, whether there are any interrupt
packets, etc.  Anyway it doesn't look like GET-NEXT-PKT can be sped
up, since it looks like its delays are legitimate waiting for the
network rather than computational delays.