2012-02-20 Keeneland Workshop at GT


On February 20-21, 2011, the Keeneland project and the Georgia Tech NVIDIA CUDA Center of Excellence are scheduled to present a two day tutorial on GPU heterogeneous processing for computational science. At the tutorial, we will provide attendees with temporary access to the Keeneland Initial Delivery system to port applications and learn how to use tools. Be sure to bring your laptop. If you have a code you would like to try on Keeneland, bring it. Instructors from the Keeneland project will lead the lectures and help you during the hands-on sessions.

For those who could not travel to Atlanta to take advantage of the hands-on training, we webcast and recorded presentations.  Video from portions of the tutorial will posted.  Check here for links to the videos and for copies of the presentation materials.

Presentations, videos, and examples

Revised Agenda

Monday, February 20
Loading the player ...
  • Welcome, Introduction, and Administrative Details
  • Scalable Heterogeneous Processing Overview
  • HANDS-ON: Getting started on Keeneland
  • Introduction to CUDA and OpenCL
  • Software Development on Scalable GPU Architectures
  • Libraries Overview – NVIDIA & MAGMA
  • HANDS-ON: building and running CUDA/OpenCL codes on Keeneland
  • Compiler Directives: OpenACC, PGI, CAPS
  • Kernel Walkthroughs and SHOC Overview
Tuesday, February 21
  • HANDS-ON: Compilers and Libraries
  • Integrated performance tools for Scalable GPU Computing
  • HANDS-ON: Performance tools
  • Application Case Studies: Pulling it all together
  • Correctness tools
  • HANDS-ON: Correctness tools
  • Advanced optimization techniques for SHCs
  • HANDS-ON: Bring your own code, or apply optimization techniques to example codes