By using this site, you agree to our Privacy Policy and our Terms of Use. Close

I read that PDF file. They're pipelining the rendering through the SPUs, which adds one more frame of lag.

It's a sort of a cheap way of exploiting multicore programming. Not necessarily a method that can be scaled to more cores.

Here is what this means:

1- Ages ago, when GPUs weren't used, a game frame went from the CPU to screen in one step.

2- With GPUs, usually frames go through the CPU to the GPU, while the CPU starts calculating the next frame. Then when the GPU finishes processing the frame, it sends it to the screen. Compared to step 1, this introduces one more frame of lag, without making programming more complicated since the GPU and CPU are working on separate frames. This is called pipelining

3- With multicore, you can go the hard way (using several cores simultaneously to calculate one frame), or you can go the easy way, which is to pipeline things further. So there are actually three stages for each frame (CPU->SPUs->GPU), which adds one more frame of lag.



My Mario Kart Wii friend code: 2707-1866-0957