GPU Programming Workshop
Dates: 9:00am-5:00pm August 10 and 11, 2011
Location: Student Computing Center, Room 102B (SCC)
Presenter: Paul A. Navratil, PhD, Visualization Scientist,TACC
This is a hands-on intensive tutorial on GPU programming by a highly experienced and qualified practitioner. Attendants can bring their laptops or use the installed PCs of the SCC. Programming exercises will be carried out on TACC's Longhorn cluster (mostly) and on Eos. Your laptops must have the following software installed: PuTTY, xming, and VNC. A synopsis of the topics to be covered includes:
Day 1 (one-hour lunch break): Connection to Eos and Longhorn GPU Hardware Overview CPU and Device Hardware comparison: NVIDIA, AMD, Intel Host to Device Memory Transfer Kernel Programming and Launch Basics Day 2 (one-hour lunch break) Special device memories: shared, texture and constant memories Kernel Debugging Kernel Profiling Streams CUDA versus OpenCL comparison
Attendance is limited to 35. You must register by e-mailing the SC helpdesk at firstname.lastname@example.org and receive confirmation.
Who can attend? (1) you must have an account on Eos; (2) you must have very good programming skills in C/C++ or Fortran; and, (3) you must be willing to exert considerable effort to acquire GPU programming skills; (4) be on time on both days.
Posted on: 6:04 PM, July 21, 2011