Hi Alex, Identified another significant issue with 'igb'. Bottom half CPU consumption is excessive, even with DCA enabled. This server is now dropping data where awhile back it was not, possibly due to the move from 2.6.27.1 to 2.6.27.7. I might try it without DCA to see if it works better that way.
Here is the same 5 second 'top' interval on the E5430/82575 system and the 2354/82571 system processing exactly the same load. I attribute much of the E5430's weakness to the high latency memory architecture, nevertheless the SI number still looks bad. Another difference is that the 82575 system is running kernel.org 2.6.27.7 where the 82571 system is running RHEL 5 2.6.18-92.1.18.el5, and I've noticed that the kernel.org kernel seem to report high CPU consumption for the same loads than the RH kernel. Is there any chance that 82575 with multiple queues will ever work with a RHEL kernel? I'm about ready to give up on this project and wait six months or a year for things to mature. David top - 15:58:07 up 1 day, 17:21, 7 users, load average: 14.59, 12.05, 12.55 Tasks: 139 total, 2 running, 137 sleeping, 0 stopped, 0 zombie Cpu(s): 39.5%us, 36.6%sy, 0.0%ni, 12.7%id, 0.0%wa, 0.7%hi, 10.5%si, 0.0%st Mem: 16472476k total, 7230208k used, 9242268k free, 87552k buffers Swap: 2031608k total, 120k used, 2031488k free, 6388308k cached top - 15:58:08 up 5 days, 14:22, 7 users, load average: 4.31, 4.50, 4.55 Tasks: 147 total, 1 running, 146 sleeping, 0 stopped, 0 zombie Cpu(s): 35.6%us, 9.0%sy, 0.0%ni, 50.5%id, 1.3%wa, 0.5%hi, 2.9%si, 0.0%st Mem: 15599048k total, 15547520k used, 51528k free, 27780k buffers Swap: 2097144k total, 248k used, 2096896k free, 14579004k cached ------------------------------------------------------------------------------ SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada. The future of the web can't happen without you. Join us at MIX09 to help pave the way to the Next Web now. Learn more and register at http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/ _______________________________________________ E1000-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/e1000-devel
