Abstract : The ever-growing level of parallelism within the multi-core and multi-processor nodes in clusters leads to the generalization of distributed memory banks and busses with non-uniform access costs. These NUMA effects have been mostly studied in the context of threads scheduling and are known to have an influence on high-performance networking in clusters. We present an evaluation of their impact on communication performance in multi-Opteron machines. NUMA effects exhibit a strong and asymmetric impact on high-bandwidth communications while the impact on latency remains low. We then describe the implementation of an automatic NUMA-aware placement strategy which achieves as good communication performance as a careful manual placement, and thus ensures performance portability by gathering hardware topology information and placing communicating tasks accordingly.
https://hal.inria.fr/inria-00175747 Contributor : Brice GoglinConnect in order to contact the contributor Submitted on : Monday, October 1, 2007 - 10:12:17 AM Last modification on : Monday, December 20, 2021 - 4:50:11 PM Long-term archiving on: : Monday, September 24, 2012 - 12:55:56 PM
Stéphanie Moreaud, Brice Goglin. Impact of NUMA Effects on High-Speed Networking with Multi-Opteron Machines. PDCS, Nov 2007, Cambridge, United States. ⟨inria-00175747⟩