Sunday, November 5, 2006, 10:40:33 PM, Stefan wrote:
Am Sonntag 05 November 2006 19:46 schrieb Aaron Slunt:
On 11/5/06, ErV ErV2005@yandex.ru wrote:
Dmitrydmitry@codeweavers.com wrote:
Calculating PI is not a representative OS performance test at all, since it doesn't involve actual OS facilities and mostly uses CPU math power only.
Well, I'm not talking about a performance. Since calculating PI is a pure CPU-related task, results must be same or linearly dependand. As I understand, WINE does not modify program, just loads it and resolve references to dlls to correct equivalents. But calculation times surely aren't linearly dependant nor equal. Why?
Victor "ErV" Eremin.
You forget, there are different processes between different OSs that are running, which affects cpu load...
What also affects the performance is the performance of the operating systems process managament and memory management. Linux does a pretty good job with memory management, and this can be seen when windows apps are run with wine.
Benchmarks in the past have shown that cpu-dependent operations are equally fast, memory intensive ones are faster in wine, while graphics(2D gdi) are much slower in wine.
In this case, I think, background processes make the difference. Was the windows box free of malware? Any firewall, virus scanner, quick launch tray applets running?
Well, it isn't possible to make OS states identical, since they are different. Situation was as "identical" as possible. No additional background jobs\applets were running. I had opened browser and established connection in both cases. But connection/browser were inactive/doing nothing... It's difficult to make sure what were doing something and what were not...
I guess the reason of speedup is memory managment. I suppose it run slower on my friends' computer because he uses KDE, and that desktop uses more background jobs|memory...
Thanks for the answer.