CS 6230 - Advanced Topics in Parallel Computing

Fall. 3 credits. Student option grading.

Students should be familiar with C or a related language and be familiar with general computer architecture and memory hierarchy. Prior experience in computer architecture at the CS3410 level and in parallel programming at the CS5220 level will be useful, though not strictly required.

G. Guidi.

The course will be divided into modules. The course will start with an overview of parallel machines and parallel programming. The course then will cover parallel computing topics in machine learning and deep learning combinatorial scientific computing, heterogeneous parallel programming and architectures, and high-performance domain-specific languages.

Outcome 1: Describe data parallelism and model parallelism in parallel machine learning, identify such parallelism modes in published work, and implement such parallelism modes yourself.

Outcome 2: Explain, design, and apply combinatorial techniques, especially in the context of graph analysis challenges, and identify combinatorial approaches in published work.

Outcome 3: Recognize and describe various heterogeneous parallel computer architectures and their communication characteristics and performance; and explain approaches in published work.