From: Robert Dewar (dewar_at_gnat.com)
Date: Sun Jan 04 1998 - 20:26:26 EST
<<I would rather rely on Drystones and Whetstones, since they
do more of what I'm looking for. The SI benchmark seems akward, and
more of something a WinDO$ fanatic would use.
>>
Drystone is known to be a completely broken benchmark, and no one should
rely on it for anything (broken means that a compiler can easily subvert
the test, and many compilers do).
Whetstone is a little more reliable if you are measuring floating-point
performance, but gives a completely inadequate reflection of cacehe
effects. The livermore loops are probably much more reliable, or
the new Spec suite.
This archive was generated by hypermail 2.1.3 : Thu Jan 23 2003 - 09:53:52 EST