Essentially, GPU cache could be a linear texture that is both CPU and GPU visible (no coherency needed). WR already knows which parts are used for what, so it can manage updates easily.
This could be an optimization - need to implement and benchmark.
Essentially, GPU cache could be a linear texture that is both CPU and GPU visible (no coherency needed). WR already knows which parts are used for what, so it can manage updates easily.
This could be an optimization - need to implement and benchmark.