[rc5] Supercomputers & Bovine, Q&A

Timothy I Mattox tmattox at ecn.purdue.edu
Sun Aug 3 11:50:36 EDT 1997


To answer a few questions that Costin RAIU asked...

A Paragon is a MIMD (Multiple Instruction streams, Multiple Data streams)
machine.  A Cray T3E is also a MIMD, not a vector machine.  A T3E would
not perform well on the RC5 since it is based on the DEC Alpha chip, which
does not have a hardware rotate instruction.

I looked into porting the v1 client to our MasPar MP-1 at Purdue, but after
some "back of the envelope" calculations, it didn't seem worth it.  The MP-1
is a great SIMD machine, but similar to the Alpha chips, it does not have
hardware rotate...  Our 16K processor machine, if it could be fully dedicated
to the effort (which it can't), would probably do from 2M to 6M keys per
second...  which is only equivalent to from 4 to 12 Pentium Pro 200 boxes...
The MP-1 is about 7 years old... amazing what exponential CPU speed growth
has done in 7 years...  Now, if my estimate had been in the 20M keys per
second, I would have been coding on this all last week.

Does anyone actually want a Bovine client for a MasPar?

i.e. Does your company/university have one sitting around with significant
periods of idle time?
I could write one if there was a demand, but I'm not sure how long it would
take (I'm quite busy for the next few weeks).

P.S. - SIMD stands for Single Instruction stream, Multiple Data streams.
-- 
Tim the Enchanter  -  a.k.a. Tim Mattox  -  e-mail: tmattox at ecn.purdue.edu
http://shay.ecn.purdue.edu/~tmattox/
----
To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5' in the body.



More information about the rc5 mailing list