About this document
CUDA Memory Management Techniques by Andy Ortiz is a document available to read on EtoBox.
This document discusses CUDA memory types and how to effectively use them in parallel programs. It covers registers, shared memory, and global memory, explaining their scope, lifetime, and importance for memory access efficiency. The document then demonstrates how tiling can improve memory performance by reducing global memory accesses. Tiling divides data into blocks that fit into faster on-chip memory. Threads work cooperatively on a tile, synchronizing to load, compute, and store each tile. This allows a
- Author
- Andy Ortiz
- Language
- EN