linuxer4fun

I think that pursuing option 4) might be interesting because - It doesn't take too much work to retrieve the displaybuffer (No matter how many vertexbuffers you use, you will always end up having a displaybuffer in adressable memory) - It would allow to write apps that are future proof (NVidia has already made some advances in this field) The step you would have to take would be 1) Remove the output to a frame (for efficiency) 2) Create a set of sockets that send and listen to browser clients 3) Calculate the framebuffers and send them over the sockets 4) listen, recalculate etc...