What I’m trying to do using openGL is :
Divide a scene to 4 (2x2).
Each time read a quarter of the scene to the back-buffer,
which is multi-sampled (16).
Then copy the pixels to a dummy render target (same size as back-buffer, but no multi-sampling).
Then copy the contents of the dummy render target to a temp render target, to the appropriate quarter of the temp.
For example - the first quarter of the scene will be rendered to BB, then copied to DUMMY (same size), then copied to the
top-left quarter of the TEMP (resized by 4 times).
Then, when I finished going over all 4 quarters - I copy back
from the TEMP to the BACK BUFFER, and swap to the screen.
This will SUPER-SAMPLE for me.
The problem is - on Windows XP this runs ok (10 ms per frame).
On Windows Vista - this takes 60 ms per frame !
I saw that Vista can use CPU memory if it’s memory gets full,
but I’ve watched the process with gDebugger, and it doesn’t
look like I use more than 100 MB at any given time …
So why is this performance issue on Vista ?
And can this be fixed ?
Thank you all in advance…