I am trying to write an hobby OS with hardware 2D/3D acceleration support.
I don't need code example, just a little bit of theory.
I need some clarification about how the GPU hardware works.
Let say that I send an command to the GPU, to draw a poly to screen. The question is: When the GPU hardware terminate fetching the instruction stream, terminate to read the data from the RAM, then terminate to calculate it all, then, where the stuff are written? Let be a projected shape, where this 2D shape is written - in the video RAM? It is directed to the monitor connector? The specs often talk to finally write to an frame buffer. So is this frame buffer the video ram? There where one write data to draw as in VESA mode? Do I need to enter an special video mode (VESA for example) in order to draw to screen the GPU output?
For example if I want to render through the GPU only one windows content, representing this way only a part of the screen, how achieve it?