Something else for a change / Parallel Computer for APL - in case you didn't know 
Author Message
 Something else for a change / Parallel Computer for APL - in case you didn't know

Hi All,

whilst you are all talking about APL, J and other things, I'd just like to
make something public concerning APL and a friend of mine - which
probably many of you are not aware of. Maybe this might interest some of you
and prompt discussions ...

In 1987-89, Jrgen Sauermann built a parallel computer in the Uni
Saarbrcken in Germany for the purpose of his doctorate. He received his
doctorate with a note of 1+ ( which is the highest ) and the machine worked
nicely. ( Professor Paul / Uni Saarbrcken )

The machine did not have an operating system as such - APL was loaded
directly at boot up and controlled the machine itself. Also, an APL
interpreter was written by some 10+ students working under Jrgen for the
machine.

The machine spec. was something like the following:

- 64 mother-boards, each with a Motorola 68020 and 1 MB of static memory and
10MB harddisk directly attached
- 32-bit VME Bus as data-communications transporter
- MIRAGE OS in EPROM on each motherboard ( only to control hardware and
requests from the APL environment )
- a mainframe was attached for - if I remember correctly  - for
preprocessing APL instructions for use within the parallel environment.
   ( this was of course a research project - not an attempt to build a
"machine for the market" )

Bill Dowling - the author of MIRAGE, which is a small and very fast
real-time prepemptive multitasking OS based on
68xxx Processors was also co-operated closely on this project with Jrgen as
did MicroAPL in London !
( Jrgen has a very high opinion of Bill Dowling and MIRAGE - as I also have
! )

The German Bundespost used this computer on the Hannover Fair to demonstrate
the speed of their fibre-glas data highway
way back in ca. 1989. Apparently even mainframes of the day had problems
handling the high bit-rates of this new technology.

The machine performance - if I remember correctly - was something like 2
Gigaflops "constant" .
I think the Cray at the time managed this only on burst-mode - but may be
wrong !
Also, I'm not sure what the MHz rate was on the 68020's but probably not
more than 40 MHz.

The whole project was done on a shoe-string budget of ca. 250,000 DM. At the
same time, another similar
project was being done in Berlin ( Suprenum ) which cost ca. 1000 mio. DEM
I'm told .

What has happened to the two projects now, I have no idea !

By the way, Jrgen is something of a genius - he built a motherboard for me
with the following spec.
in 1990 using a board design product called "Tango" - from Australia.

- PC-board dimensions
- VME Bus socket
- ISA Bus sockets
- SCSI on board
- 68040 Processor ( we could then put the board into a PC casing, but run
Motorola rather than Intel )
- 32 MB memory

... within 10 weeks from start to finish !

3 prototypes were built and worked immediatey.
The boards were then in service for some 8 years without any problems.
There were 2500 drill-holes in the board using multi-layer techniques.

He also built an 8-port VME card in half-EuroVME format in 2 days !!!

Jrgen is now working in research for the top mobile communications
companies,
I think Ericcson - where his talents are being put to good use I'm sure !



Mon, 24 Mar 2003 03:00:00 GMT  
 
 [ 1 post ] 

 Relevant Pages 

1. Why didn't variable change?

2. Upper case / Lower case I'm a lost case

3. If dynamic typing didn't exist you'd have to invent it :-)

4. Can't set IDT from C (hope I didn't send this before)

5. (meta) 'Sorry if my cancellations didn't work

6. CFP: ACPC'96 - parallel DBs and parallel I/O

7. CFP: 3rd Int'l ACPC Conf: parallel DBS and parallel I/O

8. CFP: 3rd Int'l ACPC Conf: parallel DBS and parallel I/O

9. I didn't know the Beetle relied on Smalltalk

10. Here are 3 ways to get window on top that didn't work for me

11. key didn't get primed

12. I DIDN'T BELIEVE UNTIL I TRIED..

 

 
Powered by phpBB® Forum Software