Path: utzoo!censor!geac!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!caen!uflorida!gatech!udel!wuccrc!maria!jps From: jps@maria.wustl.edu (James Sterbenz) Newsgroups: comp.arch Subject: Host--network interface architecture (was: Re: The Future of Buses...) Message-ID: <2600@olympus.wustl.edu> Date: 17 Dec 90 19:04:59 GMT References: <36734@cup.portal.com> <1178@shakti.ncst.ernet.in> Sender: usenet@dworkin.wustl.edu Organization: Washington University, St Louis MO Lines: 40 Nntp-Posting-Host: maria In article <1178@shakti.ncst.ernet.in> shri@ncst.ernet.in (H.Shrikumar) writes: >#2 How would High Speed Networking affect design ? > If one watches the high-speed protocol people, such as people at >Uwash, St. Louis, MO, they are worrying about how to handle the guzzling I/O >rates of Gigabit networking. And one of the problems is that conventional I/O mechanisms just aren't suited to this kind of bandwidth. Communicaitons looks a lot more like memory than I/O at these speeds. >And most solutions tend to split memory >into several units, each with a high speed interface (sort of like the >video-RAM chips, only these are multi-chip boards). And the most >general backplane interconnect is a cross-point like one. this would >alter the picture considerably, since the each board has lots of memory >AND I/O, the backplane then carries only a small fraction of the load. There is a limit on how much you can push through a bus. Very high performace systems require a more sophisticated interconnect than a simple bus. Mainframes and high performace multiprocessors have and will continue to use interconnects with higher connectivity, such as crossbars [good but O(n^2)] and binary routing fabrics [Banyan, etc. -- O(log n)]. High data rate communication just makes the bus saturate all the faster, and has implication on interference between the local processor-memory traffic and external communicaton. We've got a set of papers out on the Axon host--network interface architecture; e-mail to me if interested. -- James Sterbenz Computer and Communications Research Center Washington University in St. Louis +1 314 726 4203 INTERNET: jps@wucs1.wustl.edu 128.252.123.12 UUCP: wucs1!jps@uunet.uu.net