Rendering should be done by the GPU so that work can be offloaded from the CPU, and so that less data is transfered between the host and device.
I advise that a CUDA kernel be written that updates a texture which is then updated, so instead of calling a draw pixel function on each particle, the
entire frame is rendered and drawn at once.
Rendering should be done by the GPU so that work can be offloaded from the CPU, and so that less data is transfered between the host and device.
I advise that a CUDA kernel be written that updates a texture which is then updated, so instead of calling a draw pixel function on each particle, the
entire frame is rendered and drawn at once.