On November 16th I gave a two hour work-shop on GPU Programming as part of the Research Computing Course Week organized by the Univeristy of Oslo.

The workshop consisted of three parts:

1. Introduction to GPU Programming [pdf]
• Abstractions for GPU Programming
• GPU Hardware
2. Beginner and intermediate CUDA [pdf]
• Hello World Example
• CUDA Memories
• Two-pass Gaussian Blur with Shared Memory
3. GPU Algorithm Design and Lifecycle Management [pdf]
• The 13 Dwarves
• Testing and Reporting GPU Algorithms

During the break we also had a discussion on the design of USITs new cluster which will contain some GPU nodes.