🚀 go-pugleaf

In article <1993Jan22.015827.26653@nas.nasa.gov> fineberg@nas.nasa.gov writes:
 > I don't know too many people that write assembly code, and that is what you
 > need to do to get 35 MFLOPs.  As far as I'm concerned, assembly coded
 > benchnmarks are useless.  And if you can't get more than 60% of peak on an
 > assembly coded matrix multiply, that is bad (when Intel quotes peak speeds
 > on its system it uses the 60MFLOPs number, or 75 for the Paragon).
But as I read it the T800 does not even come near 50% of peak for an
assembly coded matrix multiply.  The article you responded to mentioned
great problems in even getting 25%.
 >                   I don't know what the transputer is capable of, but I would
 > be surprised if it can't do 75-90% of peak for a useless assembly coded 
 > benchmark.
If the peak value is correct a useless assembly coded benchpark should get
much closer to 100% (not really 100% because you always need program startup).
I have reached over 95% on a 25 nsec CDC Cyber 750 (over 38 MFLOPS).  But
of course you could not use the results, they were not even stored anywhere.
And of course the input values were all zero.  That is why useless benchmarks
are just that.  (It is of course still worse if even with a useless benchmark
you can not reach 75% of peak as I recorded in another article.)
-- 
dik t. winter, cwi, kruislaan 413, 1098 sj  amsterdam, nederland
home: bovenover 215, 1025 jn  amsterdam, nederland; e-mail: dik@cwi.nl