Introduction to Graphics Processors and Their Programming Model

GPU Hardware Overview

Updated March 10. We stopped at slide #41.

We will resume from were we left last week and then continue with these slides:

Memory Coalescing, Control Flow, and Shared Memory

No Lecture

Matrix Transpose

Reductions

Example Applications: Histogram and Sparse Arrays

Histogram demonstrates atomic operations and how we can use shared memory to reduce global memory traffic.

Sparse Arrays demonstrates data-dependent optimizations: we use the shared memory as a software managed cache (literally).

Textures

PTX Assembly and Floating Point Considerations

The Fermi Architecture

Introduction to OpenCL