Date: Wed, 02 Sep 2009 08:59:22 -0400 From: "Derek (freebsd lists)" <482254ac@razorfever.net> To: "James R. Van Artsdalen" <james-freebsd-current@jrv.org> Cc: FreeBSD Current <freebsd-current@freebsd.org> Subject: Re: siis/atacam/ata/gmirror 8.0-BETA3 disk performance Message-ID: <4A9E6C2A.7090305@razorfever.net> In-Reply-To: <4A9E6759.7090700@jrv.org> References: <4A9E5F34.2090700@razorfever.net> <4A9E6759.7090700@jrv.org>
next in thread | previous in thread | raw e-mail | index | archive | help
James R. Van Artsdalen wrote: > Derek (freebsd lists) wrote: >> I get about 40MB/s read, and 30MB/s write. See attached >> bench-*.txt files. > > That's too slow. As I posted on July 31 I got sustained read rates of > 875 MB/s across ten disks, through two 3124 cards and four port multipliers. > Right, so directly to a single disks I see 40/30MB/s read/write respectively. "Benchmarks" on the Internet peg the i/o to be around 100/80MB/s average read on these disks. > The system I tested was a dual processor Opteron 244 (1.8 GHz) running > amd64. Perhaps more important, my disk controllers were on two > different PCI-X buses with ample bandwidth (100 MHz 64-bit and 133 MHz > 64-bit). From the dmesg, this is a dual 1.4Ghz P3, and the controller is standard 32-bit/33Mhz PCI. Mind you, I think even 80MB/s should not saturate the PCI bus, if there's nothing else going on. > Is your disk controller on a PCI bus, and is there any other PCI traffic > at the time? The only traffic would be a little ssh traffic on fxp0, but otherwise, I can't think of any other traffic that would be occurring. - Derek
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4A9E6C2A.7090305>