Quantifying NUMA and Contention Effects in Multi-GPU Systems

Conference · Sat Jan 01 00:00:00 EST 2011

DOI:https://doi.org/10.1145/1964179.1964194· OSTI ID:1042806

Spafford, Kyle L ^[1]; Meredith, Jeremy S ^[1]; Vetter, Jeffrey S ^[1]

---

1. ORNL

As system architects strive for increased density and power efficiency, the traditional compute node is being augmented with an increasing number of graphics processing units (GPUs). The integration of multiple GPUs per node introduces complex performance phenomena including non-uniform memory access (NUMA) and contention for shared system resources. Utilizing the Keeneland system, this paper quantifies these effects and presents some guidance on programming strategies to maximize performance in multi-GPU environments.

• Research Organization: Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States)
• Sponsoring Organization: USDOE Office of Science (SC)
• DOE Contract Number: AC05-00OR22725
• OSTI ID: 1042806
• Resource Relation: Conference: The Fourth Workshop on General Purpose Computation on Graphics Processors (GPGPU-4)., Newport Beach, CA, USA, 20110305, 20110306
• Country of Publication: United States
• Language: English