-=> Quoting John Thompson to Elliott Goodman on 04 Mar 97  07:25:00 <=-
 EG> Sounds pretty good but the system will slow down as each
 EG> program accesses the stack and uses CPU cycles, won't it? The
 EG> advantage has to decrease rapidly, I'd think.
 JT> The system doesn't slow much, but because of the limited
 JT> bandwidth of the network connection the throughput of any single
 JT> task on that connection may eventually suffer.  Because each task
 JT> is separate from the others, when one task is idle (because the
 JT> remote hasn't sent the next packet, for example) the other tasks
 JT> can fill in the void.  As a result, the overall throughput on the
 JT> connection may actually be higher than running a single task at a
 JT> time because less time is wasted waiting for the remote machine to
 JT> send the next packet.
Ah, that makes sense! Thanks, John. I've got to remember that the CPU is
a hell of a lot faster than the modem!
-Elliott
Internet: sysop@miles.tbie.hms.com, elliottg@juno.com, elliottg@qnet.com
Written on Thursday, 6 Mar 1997
... BASIC programmers never die; They GOSUB without RETURN.
--- LoraBBS-OS/2 v2.40pl59+
---------------