Date: Tue, 7 Aug 2001 04:44:19 -0700 From: "Ted Mittelstaedt" <tedm@toybox.placo.com> To: "j mckitrick" <jcm@freebsd-uk.eu.org>, "Terry Lambert" <tlambert2@mindspring.com> Cc: "Wes Peters" <wes@softweyr.com>, <freebsd-advocacy@FreeBSD.ORG> Subject: RE: time to step up to the SMP plate? Message-ID: <002601c11f36$4b0b2080$1401a8c0@tedm.placo.com> In-Reply-To: <20010807121133.A73889@dogma.freebsd-uk.eu.org>
next in thread | previous in thread | raw e-mail | index | archive | help
One of the things about Open Source development is that the laws of diminishing returns set in sooner than in commercial software. This gives rise to some interesting permutations in the software. For example, everyone criticizes sysinstall and would love a GUI installer - but nobody can possibly justify the effort of throwing out a program that you basically use ONCE then toss aside. By contrast, you take commercial software like Windows where they have totally overengineered the installers to the point that the installer makes so many decisions for you that you can end up with a system totally crunched because the installer decided to put THIS over THERE ontop of THAT. I think if you look at the parts of any Open Source program where development has slowed to a crawl, you will find some feature lacking that maybe 5% of the users want - and the other 95% don't give a rat's ass for. Of course that 5% is very vocal about their unhappiness! Even this discussion over SMP. Throughout the years there have been numberous flirtations with SMP on the desktop market. I can remember a dual 486/66 for example that I loaded NT 3.1 on, made by Vtech if you can belive it. (yes, the kid's toy manufacturer) The problem is that as Terry pointed out, SMP is not easy. It turns out that due to the leaps and bounds growth in desktop hardware ability, that in just about all of the installations out there, for the same money, often less, you can replace your existing uniprocessor hardware with the next generation uniprocessor hardware and get the power growth that you need as it would be to completely redesign the kernel from scratch to take advantage of SMP to the utmost. To give you an example, Compaq came out with rack mounted dual Pentium 200Mhz servers not too long ago - within a year the uniprocessor Pentium 400's had outperformed the dual 200's in most benchmarks because the applications that people were running on them were not multithreaded, and the bottleneck was the disk I/O which was greatly speeded up in the later machines. So, while adding SMP to FreeBSD was kind of a "gee whiz, we proved we can do it" kind of thing, you can see why most of the attention has been focused elsewhere. SMP is still useful on FreeBSD for some things and it espically helps those who happen to have a dual-slotted motherboard and are interested in just tossing a couple hundred dollars into something that will make the system go a little bit faster without doing a forklift upgrade. This isn't to say that we will never see SMP regularly used - indeed we are already seeing it. Most users don't, of course, recognize it. But the fact is that the CPU in a graphics card or a smart peripheral is a form of multiprocessing - it's not SMP but it is distributed processing. As far as the slow pace of development goes - well at the current time FreeBSD is useful for quite a lot of things. I daresay that it's useful for ALL of the things that MOST of the people using it want to do with it. The folks that are still complaining about the slow pace of development on it may have legitimate beefs - but they are doing things that are more esoteric than the average user. Ted Mittelstaedt tedm@toybox.placo.com Author of: The FreeBSD Corporate Networker's Guide Book website: http://www.freebsd-corp-net-guide.com >-----Original Message----- >From: j mckitrick [mailto:jcm@freebsd-uk.eu.org] >Sent: Tuesday, August 07, 2001 4:12 AM >To: Terry Lambert >Cc: Wes Peters; Ted Mittelstaedt; freebsd-advocacy@FreeBSD.ORG >Subject: Re: time to step up to the SMP plate? > > >| FreeBSD is setting itself up to be similarly limited; Linux >| already is hitting its head on the same issue. > >Is it possible the real limit here is just the practical limit of open >source development? When almost everyone is a volunteer working in their >spare time, not only is management and design difficult, development can >slow to a crawl for a myriad reasons. > >I've seen more than one comment lately where this is becoming a concern. > > >jm >-- >"Investigators have discovered the cause of the TWA 800 explosion >was a frayed wire. The wire became frayed when it was struck >by a missile." > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-advocacy" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?002601c11f36$4b0b2080$1401a8c0>