moskowit@paul.rutgers.edu (Len Moskowitz) (07/07/90)
We are in the process of selecting a processor to be used to implement a small (64 processors or less) parallel architecture. Our uniprocessor baseline is a 68030 running at 20 megaHertz and we would like to better its performance by 20 to 100 times. We will run a parallel production system, perhaps OPS5 or CLIPS. Parallelism will be at the rule or condition element level. This implies that we will not be doing heavy number crunching (mostly integers, no arrays, no vectors). Logical and numerical comparisons will be ubiquitous; hash tables and interprocessor communications too. We have a space problem so compactness is a plus. So far we've identified the transputers, the 88000, and the i860. We are interested in comments and recommendations about the suitability of these and other processors for our task. Len Moskowitz moskowitz@bendix.com moskowit@paul.rutgers.edu
chan@dg.dg.com (Allen Chan) (07/09/90)
the i860 does not have hardware that keeps caches coherent in a multiprocessor system. the m88k does have snoop logic that does this. in my opinion, you should not even try to build a parallel/multi processor system with the i860. disclaimer: "i speak only for myself"
markv@gauss.Princeton.EDU (Mark VandeWettering) (07/11/90)
In article <609@dg.dg.com> uunet!dg!chan (Allen Chan) writes: >the i860 does not have hardware that keeps caches coherent >in a multiprocessor system. the m88k does have snoop logic >that does this. in my opinion, you should not even try to >build a parallel/multi processor system with the i860. > >disclaimer: "i speak only for myself" Well, I almost agree, but then again, not all parallel machines have to share memory, in which case your cache strategy is kind of moot. The i860 is not a real elegant chip. As one who has immersed himself in the bowels of i860 assembly language, I can say with great conviction that it isn't the most fun thing I have ever done. Because compilers are not able to use the chip effectively yet, the chip itself is an incredible time sink. But, for all its foibles, it does go really fast for carefully coded programs. Like REALLY fast. You have an fft you want done, hey, the i860 will blast it out at a screaming 40 megaflops, double precision. The 88K, MIPS R3000, or whathaveyou just ain't that fast. Someday (hopefully soon) real compilers will be available that will make this chip really fun and fast to use. Till then, it may not be the easiest, but it certainly gives an enormous performance for relatively little cost. Mark
rbn@umd5.umd.edu (Ron Natalie) (07/18/90)
While the lack of cache coherency makes multiprocessor work on the i860 interesting, it by no means disqualifies it from multiprocessor work. I am working now on a multiprocessor i860 system. Of course, Intel has realiazed this draw back and is fixing this in the follow on chip. I think the other advantages of the i860 make it more interesting the the 88000 anyway. You have to realize that having the chips doing cache coherency accross the intercpu bus has it's own design issues as well. -Ron
feustel@well.sf.ca.us (David Alan Feustel) (07/21/90)
When will the followon chip to the 860 be announced?
--
Phone: (work) 219-482-9631; MCI mail: DFEUSTEL
E-mail: feustel@well.sf.ca.us {ucbvax,apple,hplabs,pacbell}!well!feustel
USMAIL: Dave Feustel, 1930 Curdes Ave, Fort Wayne, IN 46805-2710