Re: [Jikesrvm-core] Performance tests

[Peter D. <pe...@re...>]

On 1/27/07, Steve Blackburn <Ste...@an...> wrote:
> Any other thoughts?

Decreasing the variability of the scores would obviously be the first
thing to do if possible and not to difficult. Another thing we could
do is track the variance of the test overtime and for score result
determine if the variance is statistically significant.

-- 
Cheers,

Peter Donald