GPU Computing with Python

1 Credit · Level: Undergraduate

With the emergence of Deep Learning, graphics processing units (GPUs) have become the accelerator of choice for speeding up a variety of software applications. To fully exploit the GPU's potential, however, a programmer typically needs to learn low-level (often C/C++) software libraries, identify parallelism, and understand the GPU's hardware architecture. Fortunately, thanks to recent developments with Python and associated open source libraries, we can now gloss over some of the technical details and write GPU-enabled code via Python while stressing developer productivity and ease of use. GPU Computing with Python is a hands-on, project based course. Students will be exposed to GPU basics, including a brief introduction to the traditional programming model in CUDA, along with best-practices for using GPU-enabled open source software to build and profile high performance applications. While many examples will be tied to the signal processing domain, the goal of this course is for students to GPU-accelerate a problem of choice and to provide supporting evidence including profiler output and trade studies.