Initial results appear for both but I will re-deploy few boxes tomorrow as a test to confirm all over again. The 5.4 numbers are correct and confirmed as I deployed few servers today. My initial guess was just like yours, maybe immature module causes excessive interrupts as 5.4 is almost 3 years old.
The 5.7 test I need to redo to have apple to apple comparison. Thanks ilya -----Original Message----- From: rhelv5-list-boun...@redhat.com [mailto:rhelv5-list-boun...@redhat.com] On Behalf Of Bryan J Smith Sent: Tuesday, November 08, 2011 6:36 PM To: Red Hat Enterprise Linux 5 (Tikanga) discussion mailing-list Subject: Re: [rhelv5-list] Context Switching and what is normal If support wasn't added for a particular chipset logic or CPU set until an Update after 4 (later than EL5.4), then there may be such issues. Or did you mean you're seeing no difference in EL5.7 from EL5.4? ________________________________ From: "Musayev, Ilya" <imusa...@webmd.net> Sent: Tuesday, November 8, 2011 6:28 PM My check on RHEL 5.7 seems to show the same results for both vendors. I cant believe there ~120 times more interrupts for IBM Intel hardware VS HP AMD hardware, unless there is a bug on CS reporting for AMD or Intel. ________________________________ From:Musayev, Ilya Sent: Tuesday, November 08, 2011 6:20 PM To: ' Red Hat Enterprise Linux 5 (Tikanga) discussion mailing-list' Subject: Context Switching and what is normal I have IBM/Intel 2x10 CPU servers and HP/AMD 2x12 CPU servers running RHEL5.4 stock kernel, soon to go 5.7 latest. What I find weird is that on HP AMD servers with vanilla OS and no apps, my context switch count is very low somewhere in 50 cs/s VS IBM Intel servers with vanilla OS and no apps ranging in 6000 cs/s. Both are idle and not doing much other than typical OS operations. I'm curious why, I did hear in past that AMD CPUs do better with CS due to special instruction sets, but it should not be this drastic. The check is done via sar -w 5 5 and vmstat. Why is this occurring? Is this normal? Does anyone else see this behavior? Thanks ilya _______________________________________________ rhelv5-list mailing list rhelv5-list@redhat.com https://www.redhat.com/mailman/listinfo/rhelv5-list _______________________________________________ rhelv5-list mailing list rhelv5-list@redhat.com https://www.redhat.com/mailman/listinfo/rhelv5-list
