I just noticed something. I was previously doing a blit from a system
surface to a video surface and then doing a stretchblit from there to the
primary surface. It worked nicely. Then I tried to do a stretchblit
directly from the system surface to the primary surface. Performance was
poor and the result didn't have filtering. I assume blits from system
memory are done by software. Why isn't the hardware used here?

I have a G400 so I think it should be able to do an ILOAD from anywhere in
the system memory. What would be even nicer is allocate the system
surfaces from AGP memory (or have an AGP surface type) so the card could
also blit to system memory. What do you think?

-- 
Ville Syrj�l�
[EMAIL PROTECTED]
http://www.sci.fi/~syrjala/


-- 
Info:  To unsubscribe send a mail to [EMAIL PROTECTED] with 
"unsubscribe directfb-dev" as subject.

Reply via email to