View Single Post
Old 2002-09-06, 14:20   #4
Xyzzy's Avatar
Aug 2002

24·3·173 Posts

I think the benchmark should run a bit longer than it does... It is so short right now that any activity on the box will make the result invalid... If it ran each FFT size for maybe one to five minutes it would average out all these irregularities... What I do now is run the benchmark a few times and take the best one...
Xyzzy is offline   Reply With Quote