Overview

This course addresses the need for effectively utilizing emerging multi-core (e.g., Quad-core processors of AMD and Intel) and many-core (e.g., NVIDIA Graphics Processing Unit (GPU)) computer processors.  

The course will cover the architectures of multi- and many-core processors, system view of threads, multi-thread programming  and message passing programming.  In addition, the course will cover the popular GPU programmings: NVIDIA Compute Unified Device Architecture (CUDA) and the emerging industrial standard, Open Computing Language (OpenCL) for the heterogeneous platforms consisting of CPUs, GPUs, and other processors.

Selected case studies will cover major scientific and engineering disciplines. 

Structure of Course

This 3 day workshop will combine lecture with hands-on exercises using UMBC’s BlueGrit supercomputer.

Prerequisites

Students should have experience with the C, C++ or Fortran programming language.

Audience

Various scientific and engineering personnel seeking to develop new or reengineer existing single-processor codes and gain the significant performance improvement through parallelism in multi- and many-core processors.

Contact Information

E-mail Heith Hart or call (443) 692-6599 if you are interested in running this course for your organization.