Message-ID: <4067383F.1030805@clanhk.org>
Date: Sun, 28 Mar 2004 14:40:31 -0600
From: "J. Ryan Earl" <heretic@clanhk.org>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6b) Gecko/20031205 Thunderbird/0.4
MIME-Version: 1.0
To: Jeff Garzik <jgarzik@pobox.com>
Cc: linux-kernel@vger.kernel.org, linux-ide@vger.kernel.org
Subject: Re: Status of the sata_sil driver for the VT8237
References: <4066342B.4080105@clanhk.org> <406643A8.2050808@pobox.com> <40664AE4.8010003@clanhk.org> <406680AB.8090204@stesmi.com> <4066833B.4080004@clanhk.org> <4067340A.5080709@pobox.com>
In-Reply-To: <4067340A.5080709@pobox.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 953
Lines: 23

Jeff Garzik wrote:

> Do you really plan to get 300MB/sec from two disks?  :)


Yes, for about 50-100ms at a time, and -to- the disks, not from them; 
well, the caches anyway.  On 133MB/s PCI, more like 150-300ms of 
saturation on bursts, and shared with the NICs.  I have very short and 
bursty I/O patterns as is typical in real-time applications.  It's 
easier to guarantee latency when your pipes are always at a small 
fraction of their potential.  100ms higher latency, even for part of a 
second, is quite noticeable in what I serve.

-ryan

PS 2.6 dropped my latencies under load by about 90% over 2.4!  We're 
talking 40-80ms application response time under high load to 8-10ms 
under high load.  <3
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/