NCSA Webinar - Introduction to CUDA on Lincoln

February 12, 2009
1:00 PM - 4:30 PM Central Time

Instructor: Galen Arnold, NCSA Consulting and Training

NCSA is hosting a webinar on February 12, 2009 about using CUDA on the TeraGrid's only GPU resource, NCSA's Lincoln Cluster. This webinar will be delivered via the Elluminate web conferencing software. Interaction between the instructor and participants through audio connection and synchronous chat will be supported and encouraged.

The webinar will provide an introduction to using the CUDA software environment on Lincoln. Some simple CUDA programs will be run with code walk-throughs explaining the CUDA basics; debugging techniques will be demonstrated using gdb with CUDA in device emulation mode; and a CUDA program will be combined with MPI to show how both of these high performance computing (HPC) tools can be used together. Hands-on participant activities will not be supported during the session but participants are welcome to download the example programs to use for practice after the workshop.

Registration

Participation in this workshop is FREE but registration is limited to 40 participants. You do not need to have a TeraGrid allocation to participate but preference will be given to applicants who have one. Once your application has been accepted, instructions will be sent to you describing how to access the webinar.

Application Form

Prerequisites

This workshop is not meant to serve as a thorough introduction to CUDA and GPU computing. If these concepts are new to you, we recommend reviewing the following resources prior to attending:

NVIDIA's CUDA Programming Guide

Scalable Parallel Programming with CUDA on Manycore GPUs (Stanford)

NVIDIA on GPGPU programming with CUDA from SC08

John Stone on NAMD, VMD, and CUDA from SC08

Note that we are only recommending these resources for their content. The speakers shown in the videos will not be presenting at this workshop.