Video card vs computer memory

if you read from a surface it should be in system memory otherwise it should be in vram.

if you blit to a system buffer, then all buffers that you blit to the system buffer should also be in the system buffer.

the only time you blit a system buffer to vram is for final blits to the back buffer.

you should never need to blit more then a screen worth of pixels to the vram back buffer from a system buffer.

if you have a system buffer as a back buffer, you should stil have two buffers in vram for "proper: double buffer using flip(). this allows eaiser way to sync to the refresh rate if desired. though you could blit directly to the primary if you really want to.

Copying system memory into the screen video memory backbuffer, is quite expensive on the cpu cycles.
I have made a program which does nothing but drawing the FPS into the system memory buffer, then blits it using bltfst into the screen backbuffer.
On a 800*600*32 the FPS was at top 40.
A program doing the same, only dont have the system memory and does the drawing directly into the buffer had a FPS of 75.
My questions are: is there a way to make that system to screen memory blt even faster?

That''s why people try to store their stuff completely in vram, vram->vram blits are very fast, and have been for the last 5 years at least.

------------
- outRider -