🚀 go-pugleaf

In article <24973.9208171506@lion.inmos.co.uk> steved@inmos.co.uk (Stephen Doyle) writes:
>in Article 2466 of inmos.lists.transputer PVR@autoctrl.rug.ac.be
>(Patrick Van Renterghem) writes:
>
>Your statement on Texas is inaccurate. It is a credit to the transputer
>(and perhaps a reflection on Texas' offerings) that there have not
>been design-in losses to Texas. There is still only one real transputer family
>and randomly sticking links on a processor does not make it a contender. An 
>interpretation of the situation could be that even a (previously) absent 
>T9000 is considered better than a real C30/C40.
>
W.J. Dally's (MIT) message driven processor -- the jelly-bean machine
appears to emulate the transputer in several ways: low context switch time
(from multiple register sets), message-passing links (dead-lock free) @
20 Mbytes/sec.  The catch: he has a translate instruction that eases
address translation in a global shared memory structure.  I can't say
that I necessarily support all of the COSMOS software architecture as
a commercially viable PRAM model though.

-- 
Richard Marlon Stein, Internet: rms@well.sf.ca.us
To those who know what is not known;  The Chronicles of Microwave Jim!