I have two machines, Master and slave. It can distribute the render as expected and I can set the slave to either Cuda or CPU. For the master it does not seem to matter whether I set it to Cuda or CPU it will only render on the cpu? Also, the combined render is much slower then an individual render ?