Accelerated Gibbs Sampling on Nvidia GPUs

This project is based on an assignment for UCLA's graduate-level course -- Current Topics in Computer Science: System Design/Architecture: Learning Machines.

Building

Useful commands:

• make -- build project with default configurations
• make NSAMPLES=n KCLASSES=k -- build project and set parameters
• make debug -- build project with -G flag for device (kernel) debugging
• make profile -- build project with --generate-line-info flag to generate line-number info for device code
• make clean -- runs rm -rf on ./obj and ./bin

https://docs.nvidia.com/cuda/curand/host-api-overview.html

https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html

https://docs.nvidia.com/cuda/curand/host-api-overview.html#generator-types

https://scikit-learn.org/stable/modules/mixture.html

https://docs.nvidia.com/cuda/curand/device-api-overview.html#distributions

https://docs.nvidia.com/cuda/curand/device-api-overview.html#poisson-api-example