Messages in this thread | | | Date | Wed, 26 Apr 2006 09:46:54 +0900 | From | "Taeyoung Hong" <> | Subject | process affinity problem |
| |
when "root user" runs 4 HPL(High performance Linpack benchmark program which is memory intesive, highly dependent on network latency) processes per node, "top" shows good cpu affinity and good cpu performance.
but when a general user w/o root privilige runs the same benchmark program, "top" shows cpu affinity sometimes broken, furthermore sometimes 2 hpl programs runs the same cpu, after all which gives "poor benchmark result"
I do not use special kernel parameter on boot, and the followings are testbed's spec: system: 4 cpus (2 sockets/ 2 core) opteron 275 blade server memory nodes: 512MB *2 for a node and 2GB*2 for the other node OS: RHEL4 update2 w/ kernel 2.6.9-22.ELsmp network driver: topspin IB implementation
Could you anyone explain about this awkward result? - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |