BBS: Inland Empire Archive
Date: 02-25-93 (18:36) Number: 219
From: DAVE CLEARY Refer#: NONE
To: ZACK JONES Recvd: NO
Subj: Fossil Conf: (2) Quik_Bas
In a message of <23 Feb 93 14:30:30>, Zack Jones (1:387/641) writes:
>Could you take the time to explain this to us. I haven't done any
>programming
>involving communications - but would be interested to hear about the
>limits
>a FOSSIL imposes.
A Fossil works on the principle of interrupt calls. While this would be fine
on the receiving end, it slows things down miserably on the
transmitting side. Todays 386 and 486 micros are probably
running in protected mode. If you are running QEMM or
386Max or even EMM386, your machine is in protected mode.
When you issue an interrupt call from a real mode program
(DOS), the up issues an exception and the OS (or EMS
manager) intercepts the call and has to do some magic to
make it work. If you are just running a memory manager,
then the overhead on the Fossil isn't that much because you
only have one virtual machine. However, a much more
interrupt intensive application like QB's floating point
emulation will show you a major decrease in throughput.
Under Windows and OS/2, which have multiple virtual
machines, the overhead is quite nasty limiting you to
transmitting at around 550 CPS. DSZ is able to achieve a
rate of over 2900 CPS under Windows for Workgroups just by
accessing the port directly. I always wondered why DOS
services were interrupt based while new OS' are call based.
It has to be the overhead.
Later Dave
--- msged 1.99S ZTC
* Origin: The BASIC Code Cache (203)426-5958 (fidonet.org 1:141/777.1)
