lamaster@pioneer.arpa (Hugh LaMaster) (09/10/87)
Those with a taste for the obscene may be amused by a little article in the
Sept. 2 Computerworld Extra on DEC. The article is on p53, "Can DEC prove
MIPS claim?" IBM, which doesn't give official MIPS ratings to its computers,
released a brochure on the 9370 "which implied that ""IBM MIPS"" were 2 1/2
times greater than ""DEC MIPS""." DEC apparently claims that the 8700,
compared with the IBM 9370 Model 90, has "net usable MIPS" that are about 42%
greater than IBM's. Of course, it was a little hard to tell how fast a 9370
model 90 was, since at the time of the comparison (March) IBM hadn't delivered
any, though they are available now. Also, the IBM test results were only
published for a microVAX II, according to the article, and then "extrapolated
to other DEC mid-range CPU's". I suppose IBM and DEC are shouting so loudly
at each other in the hope that you won't notice all the other companies.
For those who prefer something a little more substantial than "net usable
MIPS", the latest LINPACK (full precision all Fortran) results show that a
9370-90 comes in at .78 MFLOPS, and an 8700 (VMS) at .97 MFLOPS. For
comparison, a MIPS M/1000 comes in at 1.2 MFLOPS, and a Sun 4/260 at 1.1
MFLOPS. But then, you can't run MVS or VMS on the other machines. Not that
LINPACK is perfect, by any means. But it is one step closer to reality than
the "marketing instructions per second" we see so often.
Hugh LaMaster, m/s 233-9, UUCP {seismo,topaz,lll-crg,ucbvax}!
NASA Ames Research Center ames!pioneer!lamaster
Moffett Field, CA 94035 ARPA lamaster@ames-pioneer.arpa
Phone: (415)694-6117 ARPA lamaster@pioneer.arc.nasa.gov
"IBM will have it soon"
(Disclaimer: "All opinions solely the author's responsibility")