material about GPU architecutre and scheduling

Where can I find the material, which address on the GPU architecture and scheduling in detail.


scheduling is a very hidden part of the whole CUDA framework. Only thing known is what is mentioned about half-warps in the programming guide, and how ones try to occupy the SMs to avoid being hit memory latency.

Check out the slides in UIUC course…7/Syllabus.html

The best place to get detailed info is right from NVIDIA.

First, The CUDA proramming guide has an very good discussion on the architecture. Have you read it, front to back?

Second, you can read the articles that NVIDIA has published on the architecture.

  • J. Nickolls et al. “Scalable Programming with CUDA” ACM Queue, vol. 6 no. 2 Mar./Apr. 2008 pp 40-53

  • E. Lindholm et al. “NVIDIA Tesla: A Unified Graphics and Computing Architecture,” IEEE Micro, vol. 28 no. 2, Mar.Apr. 2008, pp 39-55