[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Looking forward to the transputer


Not exactly answering your question about STBs, but if you want evidence
that the transputer ideas are not dead, this is a news posting I made in
'98, just before WoTUG that year.

And 4Links is not doing too badly just exploiting one aspect of the
transputer legacy.

Best regards

Paul Walker

  ------- Forwarded message follows -------
If you are in any doubt that the transputer's time is yet to come, take
a look at IEEE Computer, January 1998, "Challenges and Trends in
Processor Design, January 1998, pp38-50.

Maybe we need to call it something different, and maybe we want
"transputers" with a wide variety of flovours of processor and link, but
these mainstream industry gurus are all looking forward to something
very much like a transputer.

A few gems are quoted here. Best look at the article, but if you'd like
to see a longer list that has amused some of the people who've seen it,
I'll mail a copy of the longer version to anyone who asks.

Have fun, and hope to see many of you next week in Canterbury


Mark Temblay, Sun Microcomputer Systems:
"We need to run more than one execution thread in parallel" ... "There's
a lot of inherent paralellism in most applications; it's just that we
haven't approached it like that."

Greg Grohoski, Cyrix:
"My personal favorite solution is putting the CPU on a DRAM." ... "The
next step will be getting data from the network or from a disk.
Interconnects like Universal Serial Bus and Firewire will help, but
they're still too slow to keep up with future microprocessor speeds. 

Earl Killian, Mips technologies:
"One thing that is not a performance bottleneck for SGI are bus speeds.
... In their (buses) place, we generally use point-to-point
communication between chips, usually unidirectional. We can actually
communicate at 400MHz over five metres of cable, and can do
significantly better on a PC board over short distances."

Brad Burgess, Motorola:
"There are several problems relating to memory, the foremost of which is
latency. " ... "For large gains, we must find better ways of expressing
and exposing parallelism to the processor."

Robert Colwell, Intel:
"Within the next five years, computers will become primarily
communication devices, as opposed to performing number crunching tasks.
We hadn't noticed this trend until recently... " ... "Antilock brakes
that 'mostly worked' or 'hardly ever crashed wouldn't be acceptable, but
that describes general purpose computing today." ... "In cost sensitive
market segments, as much functionality as is possible will shift to
software because that saves money. ... However the trick is not to
emulate a function in software, but to emulate several functions
simultaneously, especially consodering that many are real-time. ...
Today's most popular operating systems cannot support such functions in
software, so I expect an industry learning experience to accompany this

Paul I Rubinfeld, Digital semiconductor:
"In part, the way software is written today and the language structures
it uses are factors that limit the amount of extractable parallelism." 

(Extracted from IEEE Computer, January 1998, "Challenges and Trends in
Processor Design, January 1998, pp38-50.)
Paul Walker                            Chair of the 1355 Association
paul@xxxxxxxxxxxx    www.4Links.co.uk                   www.1355.org

NB: Please note new address and phone numbers NB:

4Links Limited ---- Boards, chips, IP and consultancy ... for Links
Registered company in England no. 3938960 
P O Box 816, Bletchley Park                  phone  +44 1908 64 2001
Milton Keynes MK3 6ZP, UK                    fax    +44 1908 64 2011