Dr. Dobb's Journal March 2004
A lot of the news affecting the future of the computer industry comes from courtrooms now, and lawyers seem to have as much to do with our future as engineers. SCO recently lost a major battle to IBM: The judge ruled that SCO has to disclose precisely what code the company alleges was stolen from it. The case is odd to begin with, since SCO alleges that IBM and others are using its code, but the company has fought to conceal precisely what code its opponents are using.
The open-source community side of this case is presented by Eric Raymond (http://www.opersource.org/sco-vs-ibm.html). SCO's case is a bit harder to find. SCO first claimed that more than 100,000 lines of its code (acquired through a series of purchases including buying UNIX code rights from Bell Labs) are being illegally used. SGI, however, found only about 200 lines of SCO UNIX code in SGI's version of UNIX. SGI rewrote those lines. As near as I can tell, it is to prevent this sort of thing that SCO seeks to conceal precisely which lines of code IBM and others are using illegally.
The result of the SCO lawsuits has been to bring SCO stock from under a dollar to over $15. There is also the complication that Microsoft has paid SCO substantial sums, supposedly in license fees to avoid lawsuits, but speculation is that Microsoft is financing SCO's lawsuits to put a primary hamper on IBM's efforts to develop Linux and other open software systems.
Meanwhile, Microsoft dropped its own bombshell by seeking licensing fees for the FAT disk filesystem. (http://www .microsoft.com/mscorp/ip/tech/fat.asp) The price is $0.25 per unit with a cap of $250,000 total royalties from any manufacturer. Microsoft also seeks to license its ClearType system. These two are different in that the FAT system is ubiquitous. Microsoft claims FAT was developed in 1976 as part of the original Microsoft BASIC. Moreover, Microsoft claims merely to be looking for additional revenue sources, mostly from manufacturers of "keychain" solid-state disk drives and CompactFlash cards who ship these devices preformatted to FAT. Some open-source software people see a more sinister purpose, with the end objective being to end the use of FAT as a filesystem compatible with both Windows and Linux machines.
ClearType, on the other hand, was developed by Microsoft as part of its e-book reader program, then incorporated into Windows XP (and briefly into Windows 2000, where it may remain, but with the simple ways to turn it on removed). ClearType is good stuff, but at least one source—Steve Gibson of Gibson Research (http://www.grc.com/cleartype.htm)—claims that it's relatively simple technology, and the patent is easy to get around. I have no opinion on that. I know of no one who sees a sinister end objective in ClearType licensing.
The late Stefan T. Possony and I once developed a theory of civilization: Societies over the years put more and more of their output into building structure, until progress ceases because there is no room for innovation. Sometimes technology makes a mighty leap and gets away from the regulators: The discovery of the New World, the Industrial Revolution, the Second Industrial Revolution, were all periods of rapid social progress and personal freedom. (Some condemn at least one of those periods as not "free" but subject to the whims of "robber barons"; this is not the place for a discussion of freedom versus equality and social justice.) Eventually, government and regulations and bureaucracy caught up with technology. Wealth creation was less rewarding than controlling its allocation, and those who created the wealth had little to say about what was done with it.
The lawsuits discussed here are interesting because of the effort going into them: The big buzz is that the future of small computers depends as much on lawyers and judges as on engineers and coders. For a while, the computer industry worked the other way, but are we now seeing the signs of increasing structure and decreasing freedom?
The small computer industry is stalled just now. The hardware is much better than the software; in fact, the hardware is so good that it doesn't matter a lot what operating system or software base you choose. Things that were too slow and clunky not long ago run just fine now, and systems developed to run at blazing speed can now pause and add what we used to call "user-friendliness."
Windows has built its dominance on making it easy for users to buy a system and start using it to do useful things. Apple did that, too, but Apple management chose immediate profit over marketshare at several critical times. They sold Apple hardware and held onto their proprietary operating system, so that it cost more to be an Apple user; and Apple was happy to exploit its loyalists. That management is gone and Apple is changing, but some of the legacy remains.
Meanwhile, very powerful systems are now cheap, and they will run almost anything you like—including emulations of other chips and systems. For example, my son Frank needed a new machine for his business, which involves video graphics. A couple of years ago, I built him a state-of-the-art system. It's already obsolete. It also has a bad hard drive. Since it has been running continuously in Las Vegas for three years or more, that's perhaps not astonishing. What is astonishing is that Frank, who ought to know better, hadn't been making frequent backups. Fortunately, he had a backup of the most crucial data.
The new machine I built Frank is based on an Intel D865GBF motherboard (http://www.intel.com/products/desktop/motherbd/ index.htm). This board has built-in Intel Extreme graphics, which would be good enough for most of the graphics Frank does, but while I was out at Fry's Electronics (http://www.frys.com/), I found an Asylum nVIDIA GeForce FX 5200 (http://www.nvidia.com/page/fx_desktop.html) with 128-MB video memory for about $70 and that seemed better than the on-board Intel Extreme that I bought it for him. I'm using the on-board sound on the Intel board: None of the sound cards I saw were enough better for the price that I was tempted.
Incidentally, while nVIDIA boards are excellent for 3D graphics, they tend to put rather ugly text on the screen. The on-board Intel Extreme Graphics (http://www.intel.com/products/desktop/chipsets/ graphics/index.htm) has better text, but much worse video processing. Fortunately, the GeForce FX 5200, while having less elegant text than most ATI boards, has text that looks acceptable although not as good as the Intel Extreme. It also has much faster graphics. Since I spend most of my days staring at screens of text, I am perhaps more sensitive to this requirement than most. Also, I recommend setting your system to reasonably high resolution, then using the Georgia font at size 16. That looks gorgeous with an ATI board, decent with Intel Extreme, and not at all bad with the GeForce FX 5200.
The CPU is an Intel 2.4-GHz Pentium 4 with Front Side Bus (FSB) speed of 533. When the 3+ GHz chips come down a bit in price, Frank can upgrade. Until then, this is plenty fast enough for what he'll be doing, and a lot faster than the system it replaces.
The hard drive is a Seagate 80-GB ATA drive. I could have put in one or more serial ATA drives but will leave that for Frank if he wants to upgrade; this one is good enough and was at a good price.
The case is the Antec Sonata, and a story goes with that.
The Antec Sonata case (http://www .antec-inc.com/) is a beautiful "piano" black. It has the True 380-Watt power supply, which is enough for Frank's machine. There's one large fan, with provision for mounting a second. The second one mounts on the drive bay and makes use of some very odd looking rubber mount things: I was confused about them until I looked at how the originally installed big fan is mounted, when the light burst upon me.
The Sonata has well thought out drive bays and mounting schemes with one exception—it is nearly impossible to install an internal drive plus anything that needs to connect to the outside, like a Zip 100, on the same ATA control drive cable. The cables just aren't long enough.
In my case, I put the hard drive on the primary disk controller, with nothing else on the string. It would be simple to add a second ATA hard drive if that's needed. What I can't do is hang the internal Zip on that string, but Zips aren't as important now, given that you can get watchfob "drives" bigger than a Zip that work just fine off the USB ports.
Incidentally, both the hard drive and the DVD drives have jumper positions labeled "master" and "slave," and I set the hard drive and the DVD-ROM drive to "master," with the DVD writers set to "slave." I set them that way because DVD-writer drives tend to be slow in recognizing what kind of disk has been inserted, and experience has shown that making the writer the "slave" doesn't have much effect on the disc creation speed.
Normally, I wouldn't remark on this, but the Director of Purchasing for the City of Los Angeles, responding to a complaint by a city employee forced to work with devices labeled "master" and "slave," has ruled that the city can no longer buy equipment with such derogatory labels. Future vendors must use less inflammatory labels. I live in the City of LA, but I hope I am not in violation of any laws by installing my "master" and "slave" drives.
The secondary controller gets the DVD-ROM drive and a Sony DRU-510A DVD +- R CD drive. That's two drives to make copying CDs (mostly) and DVDs easier. I normally prefer Plextor for CD or DVD writer drives, but I've used those Sony drives in several systems without problems. The advantage of the Sony is that it can write to any DVD format you like, so you don't need to take a stand in the DVD+RW versus DVD-RW wars. You pay a bit of a premium for that, but it's not much, and in my judgment is worth it if you're in the kind of enterprise that might encounter all of those formats.
The Sony drive was on sale at Fry's; if it hadn't been I probably would have gotten it anyway to avoid choosing a DVD format. It's really convenient to be able to work with any format, and the Sony software is quite elegant and complete.
Of course, some would say the DVD format wars are over, and the only reason to burn a DVD-R (instead of a DVD+R) disc is to use up an old supply of blanks. Others claim that DVD-R is the proper format for use with standalone DVD players, and the wars aren't over at all. This machine works in an environment where the ability to read/write both formats can be important.
I can recommend the Sony DVD RW drives for internals, although I usually prefer Plextor. Plextor drives and Nero Burning ROM software have been the standard at Chaos Manor for years. However, I've used Sony in other machines and it works, and I didn't have immediate access to a Plextor PX-708A, which writes to all formats. Plextor is also making external drives now, and all Plextor products are good. I can also recommend Backpack external DVD writers.
I can build a new Intel system from scratch in an Antec Sonata case in about two hours with time out to answer phone calls. The drive bays are well laid out and the mounting arrangements make it all easy to do. Antec has thought things out and builds quality stuff.
One way I speed things up is to have a 5-mm nut driver. The little brass hex things that hold the motherboard up off the floor of the case are 5 mm, and it helps a lot if you tighten them down good with a nut driver before you put on the motherboard.
This installation took a bit longer than usual, and there are instructive lessons in the story.
The Antec Sonata case comes with a bewildering number of wires from the front panel—ways to install USB 2.0, FireWire, and your speakers. There's also a power connection for the beautiful blue lights it will turn on if you like that sort of thing.
The Intel D865GBF board comes with on-board sound, and a set of pins that let you send that sound out to the front panel if you connect them properly. Alas, connecting them properly is a pain.
First, Antec has provided a keyed plug that you should be able to connect to the pin/jumper block on the Intel board, but Antec has got it wrong. That block won't physically connect there. Fortunately, Antec also provides seven leads out of that block, each marked with what it is, for connecting one pin at a time to the Front Panel output block on the D865 motherboard. Unfortunately, there are eight active pins on that board: one of them, AUD-VDC, is missing from the Sonata front panel wire harness.
I found that out by going to the Intel web site and getting the D865 documents. It's all there and laid out well, including the designation of AUD-VDC for what you connect to pin 4 of the block. I have no idea of what AUD-VDC stands for.
The Intel D865 board also comes with two jumpers on that block. Since I had intended to connect the front panel wires up—I had done that once before with a D865 board—I pulled the jumpers off. Alas, I did not record where they were before I pulled them. I mean, I could always go to the Intel web site and find out, right? But it was a big mistake.
Anyway, I brought up the machine and installed Windows XP Pro. This didn't take long. So far there was neither sound nor Ethernet, but once XP was installed, I could run the Intel D865 Installation disk. That installed the chip set information, and drivers for the on-board features including Ethernet and SoundMax, as well as neat little items like Intel Active Monitor to look after cooling and fan speeds.
The Ethernet came up fine. The sound didn't. Instead, the program kept asking me what I had plugged into the front panel sound output jack. When I said "nothing," it asked what I had put into the rear panel sound jack, and then said I wouldn't hear any sound because SoundMax had detected that I had headphones connected to the front panel headphones output jack.
Only, the front panel wasn't connected at all. I hadn't got around to doing that yet. Now what?
My first reaction was to do something you can't do—call Robert Bruce Thompson who, with Barbara Thompson, writes the O'Reilly PC Hardware in a Nutshell I often recommend. He didn't immediately see what the problem was, but wondered if the sound block needed jumpers; at which point I remembered removing two. Aha.
Next step was to look through the Intel online documentation for that board. There's a section on jumper settings. That told me I would need to install a jumper connecting pins 1 and 2. That's microphone and ground, and seemed odd, but the Intel text warned me not to do anything else lest I damage the board. Pins 1 and 2, and no others.
I recalled there had been two jumpers, and regretted exceedingly that I hadn't made a note of where they were installed when I got the board, but it was a bit late for that. Besides, here was the authority, Intel itself, telling me pins 1 and 2. So I did that, and up on the screen came the question, "What have you plugged into the front panel?" and then the warning that the system has detected headphones plugged into the front panel so there will be no speaker sound.
By now, I expect you have figured this out. I went back to the Intel pin layout documentation and determined that pins 5 and 6 are AUD-FtPanel-R and AUD-Ret-R. This sounded like the place the jumper ought to be. I also recalled that the two jumpers I had removed were together, and that would work because the next jumper pair was pin 7 and the missing 8 (the key), so it would look like two jumpers in place although only one was working.
I jumpered 5 and 6 and the system worked just fine after that. Note that a headphone jack has a switch that opens if you insert a headphone plug; that's the equivalent of the jumper across those pins.
The moral of this story is clear. Before you remove jumpers from a system, write down what you did. Also, while system documentation is usually correct, it's not always so. Logic helps.
There are two approaches to new machines—build to the state of the art, or build at the best price/performance point. The 2.54-GHz and 533-FSB D865GBF-based system is an example of the latter, but it has great upgrade potential. The FSB will go to 800 MHz and I have another D865GBF with a 3+ GHz CPU at that FSB speed. In fact, for almost any purpose other than accumulating benchmarks, there's not a lot of difference between those two systems. As I said, the hardware has outstripped all the software except games, and even there you have to go to extremes before you can make use of the hardware easily available.
Build to the sweet spot. It's unlikely you will ever notice that your system could have been even faster, and look at the money you'll save.
We're in an era of computational plenty. Enjoy it.
Last month, I recommended Neal Stephenson's Quicksilver before I had finished reading it. I am obliged to tell you that half way into the book, everything changes, and instead of following however deviously a plot line, you are subjected to endless pages designed largely to show you the results of Stephenson's research on Central Europe in the late 17th Century. If you like that sort of thing, fine. But of plot there is none, and while the characters are interesting at first, long periods in which they don't do much, and act out of character, makes them less so. The book reminds me of Ewell Gibbons on eating a pine tree: Many parts are edible.
The book of the month is Thomas Powers, Intelligence Wars: American Secret History from Hitler to Al-Queda (New York Times Review Books, 2002; ISBN 1590170237). Powers is not nonpartisan, but his is probably the most objective account in the unclassified press. His biography of Richard Helms was based on interviews with Helms, and Helms himself vetted at least two of Powers's books on intelligence. The intelligence game is shadowy, and no history of that period can be complete without some reference to my old mentor, the late Stefan T. Possony (who doesn't appear in this book at all); but this book is well worth reading for its broad scope and sometimes it gets into depths that surprised me. It's certainly a better book than most of the conspiratorial volumes.
The computer book of the month is Mitch Tulloch's Windows Server 2003 In A Nutshell, O'Reilly & Associates, 2003; ISBN 0596004044). For Windows 2000 Server you will want Roger Jennings's Using Windows 2000 Server (Que, 2000; ISBN 0789721228), which is complete and invaluable in my efforts to get my Macintosh and Windows Active Directory Network in communication. I don't believe there is anything similar for Windows 2003 Server, and the O'Reilly book, while more reference than introductory, is the best I've seen on 2003 Server so far.
DDJ