Slow processor on portmasters? (fwd)

Brian 'MegaZone' Bikowicz ((no email))
Wed, 1 Nov 1995 13:50:15 -0800 (PST)

Once upon a time Kevin M Lynn shaped the electrons to say...
> He told me that PM2e's have a rather slow processor in them and
>that when going with a fully stocked PM2e you cannot have all 30 ports
>set to 115200 or you'll overrun the processor and some people will start
>getting loss of carrier because of this..

WRONG WRONG WRONG

He doesn't know what he is talking about. Besides, we don't control carrier
in the first place, the modem does that. No way is load related to our
cutting someone off or not answering, the main processor doesn't deal with
that anyway, the serial controller chips do.

In fact, we have 1 386 processor on a PM-2e-30. Another well known TS maker
users 2 386 chips on a 32 port model.

We have higher throughput.

There is much more to it than chip speed or the number of processors. Our
hardware and software is just more efficient then they are.

An answer I just gave on comp.dcom.servers:

We can handle all ports set at 115200 with no loss via buffering. Depending
on overhead we can pass 100,000-150,000 bytes at a sustained rate from the
ether to the serial ports. (PPP takes more overhead that SLIP for example.)
1-1.5 Megabits, not 6.9 Megabits. Sorry if I wasn't clear.

In other words, we can handle roughly a T1 of throughput on a PM. And even
at full load you don't loose any data, nor will it effect connections. If
load gets that heavy, we buffer and use flow control to regulate it.

-MZ

--
Livingston Enterprises Technical Support
Phone: 800-458-9966      FAX: 510-426-8951
support@livingston.com  <http://www.livingston.com/> 
6920 Koll Center Parkway  #220, Pleasanton, CA 94566