From owner-freebsd-isp Fri Jul 11 14:12:53 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id OAA03739 for isp-outgoing; Fri, 11 Jul 1997 14:12:53 -0700 (PDT) Received: from etinc.com (et-gw-fr1.etinc.com [204.141.244.98]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id OAA03733 for ; Fri, 11 Jul 1997 14:12:49 -0700 (PDT) Received: from ntws (ntws.etinc.com [204.141.95.142]) by etinc.com (8.8.3/8.6.9) with SMTP id RAA13873; Fri, 11 Jul 1997 17:20:02 -0400 (EDT) Message-Id: <3.0.32.19970711171140.00e09220@etinc.com> X-Sender: dennis@etinc.com X-Mailer: Windows Eudora Pro Version 3.0 (32) Date: Fri, 11 Jul 1997 17:11:43 -0400 To: Robert Shady From: dennis Subject: Re: T1/T3 Upgrade Options? Cc: isp@freebsd.org Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Sender: owner-isp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk At 11:37 AM 7/11/97 -0400, you wrote: >>>> A major factor to consider is that its very difficult to do 86Mbs (T3 is >>>> full duplex) with an addon card on the PCI bus because the sustained >>>> throughput rate is often pretty low. Other bus masters (ethernets, for >>>> example) will futhur reduce the burst capability. While PCI bursts to >>>> 128MB/s, very few PC products have sustained throughput rates over 100Mb/s. >>>> If you have a 100Mbs ethernet card on the same bus (you'd have to) the >>>> number is cut in half. plus bus masters can't be controlled so you have >>>> contention issues. >>> >>> Contention issues I'll agree with for the most part, however.. My >>> calculations are (correct me if I'm wrong): >>> >>> T3 Card = ~12Mbytes/sec [ 90Mbits/sec] >>> Ether = ~25Mbytes/sec [100Mbits/sec] >>> ============= >>> ~37Mbytes/sec >>> >>> 132 Mbytes/sec - PCI bus >>> - 37 Mbytes/sec - Interfaces >>> ================ >>> 95 Mbytes/sec - Left to play with. >> >> Because you have no clue how the PCI bus really works (and that burst >> rates are not sustainable), your Math is totally invalid. >> Remember that ISA is a 64Mb/s medium and if you get 30Mb/s you >> are lucky....the numbers are even more dramaticly worse when you add >> bursts to the formula as with PCI. > >Invalid? I doubt it... In "real-life", how often do *MOST* T3 connections >"sustain" 45Mbit/sec throughput in each direction? this depends on who your selling to. If you're building a card for little weenie ISPs who only have 4Mb/s requirements, then maybe. But I think that big ISPs have pretty busy links, and we're only talking about 2 or 3 large, consecutive packets to potentially have a problem. >In "real-life", how >often do ethernet cards "sustain" 100Mbit/sec throughput in each direction? >I think my figures are right on the money, however.. Let's assume for a >minute that you can only achieve 50% throughput on the PCI bus for whatever >reason.. Assuming that the ethernet and the T3 card were running full tilt >(Very, very unlikely in most situations) then with 1 ethernet and 1 T3 port >you still have ((132Mbytes/2)-37Mbytes) = 29 Mbytes left over just for >lolly-gagging around... "Bursts" are typically only a few words, so the rate you are using is just totally invalid for this computation. 50% is still pretty good (sustained 2 cycle accesses)...and still bursts so only a few words...... > >> Ah, if life was only as easy as your trivial account of the world! > >This sounds a little bit sarcastic Dennis... This would *really* encourage >me to purchase your products. A "Your a dumb-ass, we're not" attitude. Its very sarcastic :-) YOU'RE the one with the "you dont know what you're talking about" attitude, which is OK if you're right, but you're not. Have you designed a PCI card lately? Dennis