From mboxrd@z Thu Jan 1 00:00:00 1970 From: cnielsen@pobox.com To: 9fans@cse.psu.edu Subject: Re: [9fans] small boxed PC and dual CPU file server? Message-ID: <20020719202807.GW42736@cassie.foobarbaz.net> References: <724ddada05d9bb9643865ae2030bd664@plan9.bell-labs.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.1i Date: Fri, 19 Jul 2002 13:28:07 -0700 Topicbox-Message-UUID: d31c89c2-eaca-11e9-9e20-41e7f4b1d025 On Fri, Jul 19, 2002 at 11:54:49AM -0500, Doc Shipley wrote: > > Are you sure you're saturating the ethernet adapter and not the PCI > bus? A full 64-bit, 33MHz PCI bus gives something on the order of 275 > MB, or 2200 Mb. That's _total_ available bandwidth. Even in a totally > unloaded setup, I can't see getting a gigabit through it without > knocking the box on its nose. > Never mind the bus overhead when you start moving bits from disk > instead of the, um, ether. 64-bit, 66MHz PCI should be able to handle it, in theory. you get 4224Mbps for the PCI bus, so it's less likely you're going to saturate the bus. at that point it becomes a game of chase the bottleneck. from my limited testing with hardware similar to what jmk currently has, i see about what he does. part of that may be due to the 620Ts having only 1M SRAM onboard. iirc, the SysKonnect cards have 2M SRAM and perform better as a result. i have an SK-9821 kicking around. i just need to sit down and write a p9 driver for it. :-) -- Christopher Nielsen - Metal-wielding pyro techie "Those who are willing to trade freedom for security deserve neither freedom nor security." --Benjamin Franklin