Small Intro - Thanks for PB v4 (bughunting version added)

Jan Vooijs · Post by **Jan Vooijs** » Thu Feb 09, 2006 3:05 pm

va!n wrote:
Jan Vooijs wrote: Note the " WaitWindowEvent( 2)" and removal of "; ClearScreen(0)" and the "Delay(1)". But the program stops if the mouse is NO longer on the window (side effect)... This version runs about 30% and still NO stutter!!! I even can play a DVD while this litle demo version runs...

Hope this helps....
What should WaitWindowEvent(2) do??? This command dont use/need any arguments! So i dont understand the "2"... In the original effect code i dont used any ClearScreen(), because the complete screen will be redrawed each frame, so i didnt used ClearScreen() .. Delay(1) was used because i dont wanted 100% CPU ussage...

In PB4 it uses a parameter to indicate a "waint untill" time almost the sam e in the delay() function but goes really back to windows. But using the WaitWindow() function the usage drops from 100% to around 30% meaning the delay() uses 70% of the time?!?! I suspect the delay() is "in house" so cpnsuming all the proccessor cycles... Will research more, later...

chris_b · Post by **chris_b** » Thu Feb 09, 2006 3:16 pm

va!n wrote:yes... possible TimeGetTime_ is the result of not running smooth??

are you calling TimeBeginPeriod_(1) (I think that's the correct function?) before your main loop? otherwise TimeGetTime_() may not return the time with millisecond accuracy.

Nik · Post by **Nik** » Thu Feb 09, 2006 4:27 pm

Maybe the problems on the AMD X2 are coming from the problems they have with measuring time when speedstep is active, you could test without....

THCM · Post by **THCM** » Thu Feb 09, 2006 5:05 pm

Since I have overclocked my X2 I don't have Cool'N'Quiet activated.

djes · Post by **djes** » Thu Feb 09, 2006 6:30 pm

Maybe the new flipbuffer argument (2) correct this. I think this is this function that have to wait for synchro (1) without taking extra cycles, and not some tricks made by ourself.

I've tried flipbuffers(3), and it seems that it is timer based with synchro activated (and it looks like a triple buffering). However, it's a bit slower than just flipbuffers(1) when sync is activated in gfx card drivers.