jet@flatline.UUCP (It's "Mr. Boyo" to you Dylan) (03/04/90)
This subject is straying, but... In article <8249@netcom.UUCP> hue@netcom.UUCP (Jonathan Hue) writes: >In article <PORTUESI.90Mar2142138@tweezers.esd.sgi.com> portuesi@sgi.com (Michael Portuesi) writes: >They claim 27 to 41 MIPS based upon Dhrystone 1.1, a fairly useless >benchmark. They also say 22.3 to 34.7 Specmarks, which I think is a useful >number. As far as FP goes, they say 7.5 MFLOPs for the slowest boxes, 13 >for the fastest. We (some Univ of Houston math folk) got 5 MFLOPs (on one of the next-to-smallest machines w/ 40Mb of RAM) running some rather large fluid dynamics problems -- lots of conjugate gradients, gaussian elmination, and other things that get you out of cache on a regular basis if you have a moderate-to-large size problem. That was with Fortran code, their C compiler couldn't operate long enough to compile a single file (kept crashing on stray interrupts). The salesgoons assured us that the problem was the software wasn't "the release version" and that in a few months we could see the "release version" of the OS and compiler. Sigh. -- Skate UNIX. J. Eric Townsend -- uunet!sugar!flatline!jet -- jet@flatline.lonestar.org EastEnders mailing list -- eastender@flatline.lonestar.org