old modems & a packet switcher

From: Ethan Dicks <erd_at_infinet.com>
Date: Mon Aug 3 01:26:50 1998

tony writes:
> > Does anyone on this list have any classic networking stuff beyond
> > ethernet/token thing cards?
>
> Does anyone else remember the 'York Box'. This was an interface between
> either an async RS232 port or a DR11-like parallel port and an X25 line.
> It was built from standard DEC modules (SBC21 + 64Kbyte Q-bus RAM card +
> DRV11 + DPV11) in a BA11-V cabinet. They were used on the UK JANET
> network to link VAXen (on the DR11 port) and just about anything else (on
> the RS232 port -

I don't remember the York Box specifically, but I remember JANET from my
DECUS days. The York sounds just like an X.25 version of the HASPBOX with
more modern cards (the HASPBOX dates from 1977 to 1982). It was an 11/03
(KD-11A?) or 11/23 (KDF-11) or 11/04, depending on exact vintage, with RAM
and a DPV11 or similar (COM5025 UART), again, depending on vintage, all
running a custom monolithic I/O front-end program that was downloaded once
per phone call, and handled retransmission of flawed packets and routing to
a local printer port, all without distubing the host PDP-11. It was a great
boon the RSTS adminstrators to offload print jobs from the IBM mainframe.
The pre-RSTS v8 print spooler ate gobs of CPU (later versions did as well,
I suspect, but by then, there were more cycles to go around). The application
software on the PDP-11 host saw a perfect data stream, blocked to 512 bytes
per block, DMAed into memory, ready to chew on. Very efficient.

A funny side note, when HASPBOX was replaced by COMBOARD(R), a 68000-based,
single-board descendent, the serial I/O board often had more horsepower than
the PDP-11 or VAX it went into (an 8Mhz 68K is rated at a nominal .6 MIPS,
which compares favorably to a VAX 11/750 at .6 VUPs) It could be termed an
I/O coprocessor at that level.

Second funny side note: the original HASPBOX code, in PDP-11 MACRO, was
converted to 68K assembler by TECO macros. the macros did over 90% of
the work, humans did a quick desk check and fired it up. It was an
amazingly clean migration, and a testement to how much the Motorola
designers borrowed from DEC in terms of processor philosophy. Both
CPUs rank up there in my top 5 to program (the 6502 and 1802 all competing
for number one). 80x6 assembler sucks, in case anyone cares to hear.

-ethan
Received on Mon Aug 03 1998 - 01:26:50 BST

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:30:43 BST