From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S272767AbTG1JeT (ORCPT ); Mon, 28 Jul 2003 05:34:19 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S272770AbTG1JeT (ORCPT ); Mon, 28 Jul 2003 05:34:19 -0400 Received: from server.snowfall.se ([213.136.34.4]:33541 "EHLO mail.snowfall.se") by vger.kernel.org with ESMTP id S272767AbTG1JeP (ORCPT ); Mon, 28 Jul 2003 05:34:15 -0400 Date: Mon, 28 Jul 2003 11:49:27 +0200 (CEST) From: Stefan Cars X-X-Sender: stefan@guldivar.globalwire.se To: Jeff Garzik Cc: linux-kernel@vger.kernel.org Subject: Re: ICH5 SATA high interrupt/system load again... In-Reply-To: <3F187DB1.1040309@pobox.com> Message-ID: <20030728114850.F22307@guldivar.globalwire.se> References: <20030718233631.F31074@guldivar.globalwire.se> <3F187DB1.1040309@pobox.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org What also is interesting is that when I configure my kernel to use APIC it hangs during boot just as it found the SATA drives... / Stefan On Fri, 18 Jul 2003, Jeff Garzik wrote: > Stefan Cars wrote: > > Hi! > > > > I've seen the discussion regarding high interrupt / system load on the > > ICH5 SATA and I'm asking what todo about it if I can't put my BIOS into > > "normal" mode. This machine is an Dell Precision 360 and for some stupid > > reason they have for this model removed the possibility in the BIOS to > > change this sort of things (you can't change much really). I'm using > > 2.4.21-ac4. Just to extract a simple tar file brings the system load up > > and the computer is slow... > > > > > > Here is some info: > > tjatte:/import# cat /proc/interrupts > > CPU0 > > 0: 557725 XT-PIC timer > > 1: 102 XT-PIC keyboard > > 2: 0 XT-PIC cascade > > 5: 0 XT-PIC ehci_hcd > > 9: 16409116 XT-PIC libata, usb-uhci, eth0 > > > Hum... interesting. I had seen reports of this before, but they were of > the variety "drivers/ide has high load, libata doesn't". So it seems > intrinsic of the hardware, which is a useful data point. > > Have you tried messing around with interrupt routing in BIOS setup? > Since ATA, USB, and eth0 are all on the same interrupt, changing that > may affect the situation positively. > > Jeff > > >