OSDev.org

note that there are open-source drivers that can do hardware-accelerated blitting for many cards (including nvidia's). That'd mean (unless i'm mistaken) that you could handoff the RAM-to-VRAM copy (in a single PCI/DRAM cycle) to the card itself rather than "fill cpu cache with offscreen bytes, send bytes over PCI bus" of a movsb ...

First thanks for all your in puts, i must admit that up to now i have all ways concentrated on getting the code to work and never got round to any optimising, which is a poor showing, for someone who is suppose to be making a OS for speed :-[ .
I will use the info learnt from this topic to good effected and let you know if i find any little surprises.
Thanks again.

Edit: @Brendan, 34 FPS .

UPDATE: I wrote some code to implemented the:

For a constructive addition, you could set the MTRR of the video memory to write-combining if you haven't done so already.

As Brendan and Candy suggested.
Its made a big difference, the FPS as gone from 34 FPS to 54 FPS , that really is a big difference, thanks again for some good advice.

Is that a "small" piece of code? Would you mind sharing that?

No problem, give me a day or to to implement checks, using "cpuid" as you need to make shore mtrr's is supported, as i did not implement checks, as i new that my PC supported it, but this was temp to see if it made much difference, now i know it does i will do the checks code.
I will send you a PM, when its done and post it on my web site.

PS: With no checks, about 16 line of asm code.

Thanks!