Subject: Re: scheduler woes on MPACPI kernel
To: Johnny Billquist <bqt@Update.UU.SE>
From: Rick Kelly <>
List: current-users
Date: 01/18/2005 15:45:20
Johnny Billquist said:

>Hmm. I haven't given much thought on how much impact shared cache and TLB 
>can have. Is it really that bad?

From what I've read on comp.arch, there is no OS that is a win on hyper-
threaded processors. This makes dual P4 Xeon systems rather unattractive
unless the scheduler is intelligent enough to run processes on the true
physical processors.

On plain old SMP boxes, applications like Apache 2.x can be a win when
compiled with threading and the env variable "PTHREAD_CONCURRENCY=2" or
whatever is set in the processes environment.
Rick Kelly