Re: Nightly performance regression graphs

> I finally threw together some graphs, see
> http://kegel.com/wine/yagmarkdata/wine-1.1.44-245.html
> Not sure what to make of the results yet.

Cheap thought, expansive solution:

Looking at the graphs and used HW i think this not necessary do what
you expect. Change from 11 -> 10 can be just a normal deviation during
measurement. But change from 100 -> 90 (80...) can show clearly performance
regression.

Unfortunately latest nVidia cards have too high power consumption even
in idle state (although internal logic underclock the GPU/RAM card heavily
when idle). So running 24/7 brings another high cost :-/