From mboxrd@z Thu Jan 1 00:00:00 1970 From: Shane Wegner Subject: Performance with multiple drives Date: Sun, 18 Apr 2004 13:34:08 -0700 Sender: linux-scsi-owner@vger.kernel.org Message-ID: <20040418203408.GA16788@cm.nu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from continuum.cm.nu ([216.113.193.225]:15328 "EHLO continuum.cm.nu") by vger.kernel.org with ESMTP id S264102AbUDRUeK (ORCPT ); Sun, 18 Apr 2004 16:34:10 -0400 Received: from shane by continuum.cm.nu with local (Exim 4.31) id 1BFIzR-0004Rd-Aw for linux-scsi@vger.kernel.org; Sun, 18 Apr 2004 13:34:09 -0700 Content-Disposition: inline List-Id: linux-scsi@vger.kernel.org To: linux-scsi@vger.kernel.org Hello list, I am experiencing a performance hit on my SCSI setup here and am wondering if this is to be expected or whether there is anything I can do about it. In summary, when reading from a single drive, I am getting read performance of approx 73mb/s, when reading from two drives concurrently, regardless of which two, reads drop to 64mb/sec per drive and when reading from all three, read performance is approximately 46mb/sec. The drives are matched Maxtor Atlas 10K IV sca units using a dual channel u160 onboard aic7xxx controler. Two drives are on channel 0 and the third is on channel 1. The drives are sca units so I had to use 80/68 pin converters on all three in case that would make a difference as I couldn't find a scsi backplane which would support splitting the drives across the channels as is necessary due to the 160mb/sec channel limit. The system is a dual Pentium III tualatin setup with 1gb of ram. The board claims to support 64 bit 66mhz pci on the scsi bus and I have no legacy cards on that particular bus so it should be running at 64 bit. However, I don't know of a way to verify that under linux. If anyone has any suggestions on how I can find the bottleneck, I would appreciate it. Best, Shane -- Shane Wegner http://www.cm.nu/~shane/