CS324:High Performance Computing
AY 2020-21
Division 5

∙        Data Structures

∙        Computer Organization & Architectures

The course High Performance Computing covers basics of Parallel Programming Paradigms at hardware and software level. The course also touch upon the tools used to parallelize the serial code and the techniques to do the same.

Introduction to High Performance Computing:

Evolution of HPC:

Hardware evolution: Superscalar architectures &  Multi-core architecture, Limitations of Memory, Dichotomy of Parallel Computing Platforms.

Software Evolution: Concept of Serial program, Concept of parallelism & parallel program, basic introduction to OpenMP

Significance of HPC, Performance Metrics for Parallel Systems – SpeedUp, Execution time, total parallel overhead, response time, efficiency and cost.

Applications of HPC: Real-time applications, Case study.

Designing Parallel Algorithms

Preliminaries,  Decomposition Techniques,  Characteristics of Tasks and Interactions, Mapping Techniques for Load Balancing, parallel computing model.

Performance Analysis of Parallel Programming:

The effect of Granularity on Performance, Scalability of Parallel Systems, and Interconnection Networks*: Bus, Ring, Star, Crossbar, Minimum execution time and minimum cost

CUDA programming:

The Age of Parallel Processing, the Rise of GPU Computing, A Brief History of GPUs, Early GPU.

CUDA: CUDA Architecture, Using the CUDA Architecture, Applications of CUDA, Writing a basic CUDA program