Re: [patch] sched: fix improper load balance across sched domain

2007-10-17 Thread Ken Chen
On 10/16/07, Siddha, Suresh B <[EMAIL PROTECTED]> wrote: > On Tue, Oct 16, 2007 at 12:07:06PM -0700, Ken Chen wrote: > > We recently discovered a nasty performance bug in the kernel CPU load > > balancer where we were hit by 50% performance regression. > > > > When tasks are assigned to a subset of

Re: [patch] sched: fix improper load balance across sched domain

2007-10-17 Thread Ingo Molnar
* Ken Chen <[EMAIL PROTECTED]> wrote: > We recently discovered a nasty performance bug in the kernel CPU load > balancer where we were hit by 50% performance regression. > > When tasks are assigned to a subset of CPUs that span across > sched_domains (either ccNUMA node or the new multi-core d

Re: [patch] sched: fix improper load balance across sched domain

2007-10-16 Thread Siddha, Suresh B
On Tue, Oct 16, 2007 at 12:07:06PM -0700, Ken Chen wrote: > We recently discovered a nasty performance bug in the kernel CPU load > balancer where we were hit by 50% performance regression. > > When tasks are assigned to a subset of CPUs that span across > sched_domains (either ccNUMA node or the

[patch] sched: fix improper load balance across sched domain

2007-10-16 Thread Ken Chen
We recently discovered a nasty performance bug in the kernel CPU load balancer where we were hit by 50% performance regression. When tasks are assigned to a subset of CPUs that span across sched_domains (either ccNUMA node or the new multi-core domain) via cpu affinity, kernel fails to perform pro