From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 18893 invoked by alias); 15 Feb 2005 13:14:10 -0000 Mailing-List: contact sid-help@sources.redhat.com; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: sid-owner@sources.redhat.com Received: (qmail 16500 invoked from network); 15 Feb 2005 13:12:59 -0000 Received: from unknown (HELO miranda.se.axis.com) (212.209.10.220) by sourceware.org with SMTP; 15 Feb 2005 13:12:59 -0000 Received: from PCSTARVIK (dh10-84-127-108.se.axis.com [10.84.127.108]) by miranda.se.axis.com (8.12.9/8.12.9/Debian-5local0.1) with ESMTP id j1FECvx8012622 for ; Tue, 15 Feb 2005 15:12:57 +0100 From: "Mikael Starvik" To: Subject: Latency in cache hierarchies Date: Tue, 15 Feb 2005 13:14:00 -0000 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable X-SW-Source: 2005-q1/txt/msg00018.txt.bz2 Lets say you have a system with a L1-cache and a L2-cache. In cache_component::read_line() the latency for the first read is saved as a latency for the complete operation. This looks correct to me because it is the first read that may trigger additional refills down the chain and has the largest latency. In cache_component::write_line() the latency for the last write is returned. This makes me a bit confused. The first write may trigger dirty flushed down the chain and should have the largest latency so why isn't that saved in the same way as in read_line? /Mikael