Comment Re:I wonder when we'll drop the notion... (Score 1) 95
It might be technically possible, but it requires such a big architectural changes that I doubt we will see it anytime soon...
Basically you need to fuse the GPU and the monitor. The VRAM needs to have 2 buses - one for R/W to the GPU (as is now) and one read only (but possibly much faster) going to the panel. So the GPU just renders and puts pixels in the VRAM and the "panel" reads them whenever and changes the corresponding pixel(s). The key difference is that this type of access doesn't have to be sequential - it can be RANDOM as in "Random Access Memory" OR even if it is sequential it will be WAY faster than the pixels of the panel.
And then there is the risk that after someone takes the risk to implement this, the end result looks like shit simply because of how human perception works... who knows.
Basically you need to fuse the GPU and the monitor. The VRAM needs to have 2 buses - one for R/W to the GPU (as is now) and one read only (but possibly much faster) going to the panel. So the GPU just renders and puts pixels in the VRAM and the "panel" reads them whenever and changes the corresponding pixel(s). The key difference is that this type of access doesn't have to be sequential - it can be RANDOM as in "Random Access Memory" OR even if it is sequential it will be WAY faster than the pixels of the panel.
And then there is the risk that after someone takes the risk to implement this, the end result looks like shit simply because of how human perception works... who knows.