From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S266200AbUHHTlY (ORCPT ); Sun, 8 Aug 2004 15:41:24 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S266201AbUHHTlX (ORCPT ); Sun, 8 Aug 2004 15:41:23 -0400 Received: from shawidc-mo1.cg.shawcable.net ([24.71.223.10]:62152 "EHLO pd4mo3so.prod.shaw.ca") by vger.kernel.org with ESMTP id S266200AbUHHTlW (ORCPT ); Sun, 8 Aug 2004 15:41:22 -0400 Date: Sun, 08 Aug 2004 13:32:10 -0600 From: Robert Hancock Subject: Re: ide-cs using 100% CPU To: linux-kernel Message-id: <000f01c47d7e$669727f0$6401a8c0@northbrook> MIME-version: 1.0 X-MIMEOLE: Produced By Microsoft MimeOLE V6.00.2800.1441 X-Mailer: Microsoft Outlook Express 6.00.2800.1437 Content-type: text/plain; charset=iso-8859-1 Content-transfer-encoding: 7bit X-Priority: 3 X-MSMail-priority: Normal References: Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org ----- Original Message ----- From: "Hamie" > Robert Hancock wrote: > > >It isn't that the CPU is doing so much work, it's mostly waiting. However > > > > > > That was my point... While waiting, shouldn't the CPU be off doing > something else? Like giving X some attention... > > >with this type of PIO access, the CPU must do all the reads/writes from the > >buffer and while doing this the CPU is blocked and cannot do anything else. > > > > > > > > Or is the CF requirements such that it's spending it's time doing the > actual reads & writes from the buffer, and it's the hardware inserting > wait-states when it's being accessed? > That's basically what's happening, when the CPU does a write or a read of some data from the buffer, it has to wait for that to go all the way across the bus and to/from the card, meanwhile no other useful work can be done while it is waiting. As Alan Cox mentioned, having a hyperthreaded CPU helps tremendously in such cases, since the other "half" of the CPU can be doing useful work while the first half is blocked.