Re: [DotGNU]Could someone test this on all platforms ?

[Rhys Weatherley]

On Saturday 08 February 2003 12:50 am, Paolo Molaro wrote:

> But 1024 is still too little: you're trying to measure the size of an
> egg with a feet and, after crashing the egg, all you can say is that it
> was somewhere from 0 to 2 feet big ;-)

I agree with this.  1024 was an insufficient number of runs to smooth out JIT 
artifacts and to overcome clock resolution problems.

Could we have a moratorium on "my engine is faster than yours" unless 
well-understood benchmarks like the CaffeineMark/PNetMark are used?  It is 
too easy to measure the wrong thing and draw the wrong conclusions.

Benchmarking is hard.  And irrelevant - what matters is performance on real 
applications, not compute-bound spin loops.

Cheers,

Rhys.