I messed around with some slightly above toy work, implementing expectation-maximization for gaussian mixture models and some other supporting tools for 3D poin

I messed around with some slightly above toy work, implementing expectation-maximization for gaussian mixture models and some other supporting tools for 3D point clouds, and found it really intuitive for someone with a decent amount of FP experience but little GPU experience. Biggest challenge for me was a lack of profiling tools when running OpenCL on an NVIDIA GPU, I got OK performance but had no way to identify hotspots or opportunities for improvement.