|
Week of 5/17/99 Sunday, May 23, 1999 10:49 A (mostly) daily journal of the trials, tribulations, and random observations of Robert Bruce Thompson, a writer of computer books. |
|
Monday |
[Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] Monday, May 17, 1999 If you didn't read the updates last weekend, check back to last week. I posted quite a lot of interesting new stuff Saturday and Sunday. I've been testing an Intel Pentium III/550 (actually a QB78ES engineering sample) for about a month now. As you might expect, it's the fastest processor I've ever used. Although this CPU is by far the most overclockable of the Pentium IIIs released to date, I tested it only at its rated speed. I tested it against a Pentium II/450 and a Pentium III/450 using the standard Ziff-Davis benchmarks and found no surprises. The 450 MHz processors provided essentially identical benchmark numbers, and the 550 MHz processor benchmarked about 22% faster than either of them, as might be expected from the relative clock speeds. What really sets the Pentium III apart from the Pentium II and Celeron is the Pentium III's support for Streaming SIMD Extensions (SSE). SSE (formerly Katmai New Instructions, KNI) is essentially a floating-point version of MMX--a set of new instructions that optimize handling of array data structures common to such things as speech processing and graphics. An application must be SSE-aware to benefit from those instructions. As a matter of policy, I do not use vendor-supplied benchmarks. But I had no SSE-aware applications to use in testing, and Intel had supplied an SSE-enabled benchmark suite. So I ran those benchmarks on the three processors. Although I won't report those numbers, I will say that the difference is dramatic. As expected, the PIII/450 and the PIII/550 maintained the same performance relative to each other that you might expect on the basis of simple clock speeds. Both, however, pulled well away from the PII/450, again as expected. Intel is a reputable company, and I have little doubt that the benchmarks they supplied reflect reality. But until SSE-enabled applications appear, the benefit of SSE will remain largely theoretical. Once those applications are available, I suspect the real-world performance benefits of the Pentium III will become obvious. Cruising around the web this morning, I find several reports on the Pentium III/550. Most of them, however, make one or both of the following errors:
Much has been made of the cost of the Pentium III/550. At $744, this is definitely not a mainstream processor. The Pentium III/450 at $268 provides about 82% of the performance at 36% of the cost. Intel is obviously targeting this CPU at the performance market--those for whom even a 20% performance boost or thereabouts is worth a lot. And there are many such people. It makes sense to put a Pentium III/550 on the desk of an executive or engineer who spends a lot of time crunching numbers. That few hundred dollars extra CPU cost is quickly paid for by making that person more efficient. For the rest of us, the real benefit of the Pentium III/550 is the downward price pressure it exerts on the other CPUs that Intel sells. * * * * * This from Tom Syroid [tsyroid@home.com], who has both a new baby and the pieces of a new computer to put together this week:
Sounds like a plan. WinBench should be just fine for stress testing. One minor change I'd suggest: install your CPUs first at 66 MHz FSB to make sure everything works at the rated speed. Then take the steps you mention. Actually, once you see how Windows NT performs on a system with dual 300 MHz CPUs, you may decide not to bother overclocking.
|
|
Tuesday |
Tuesday, May 18, 1999 The Register this morning reports an interesting story. Apparently, there is some doubt as to whether Windows 2000 will run on Celerons or, more specifically, on CPUs with less than 256KB L2 cache. The Danish computer magazine ComputerDK reports that the Danish Microsoft website reports this requirement, but the U.S. site does not. Although I speak no Danish, I do have enough German that I can usually puzzle out the general sense of things written in Nordic languages, and the Danish site does appear to say that W2K requires a desktop system with at least a 300 MHz processor with 256KB of secondary cache. Well, the thought struck me that I have a Windows 2000 Professional Beta 3 installation CD lying on the desk in front of me, and a Celeron/333 testbed system with an empty partition sitting on the credenza behind me, so I decided to fire it up and see what happened. The first interesting thing was that W2K Setup offered to upgrade the existing installation of Windows 98. I told it I wanted to install a new version rather than upgrade Win98. The installation went normally, with no hangs or long pauses. Setup even detected the LinkSys 100BaseT Ethernet card, which NT4 does not. When Setup completed, I restarted the system and did an exhaustive test by playing a game of FreeCell. Everything appears to work normally, so I conclude that the report about W2K not running on Celerons is incorrect. Perhaps the Danish version requires 256KB of L2 cache, but Beta 3 of the U.S. version does not. One weird thing, though. There are apparently some problems with W2K and DHCP. I first noticed strangeness when I installed W2K on old sherlock (now odin). The installation appeared to complete normally, and odin connected to the network. But when I turned around and looked at the Win98 testbed system on my credenza, it was displaying a DHCP error message saying that there were duplicate machines assigned the same IP address. I went into WINS Manager and killed some entries, and that situation appeared to clear up. Just now, I restarted the testbed system in Win98. It displayed a DHCP error message at startup to tell me that I did not have network connectivity. I went into WINS Manager again, and find that that machine under both W2K and Win98 is assigned 192.168.111.97. The fact that I have "two" machines with the same MAC address is likely causing the problem, so I won't blame that on W2K. * * * * * Which brings me to some comments on W2K itself. Although Microsoft is not completely (or even primarily) responsible for "beta inflation", this "Beta 3" release continues the greek-letter inflation that's been going on for many years. This version is really an Alpha in the original sense of the term. Back when I was involved with software development, an Alpha was an in-house version, with known bugs and missing features. We'd have been embarrassed to allow any outsider to see an Alpha. A Beta was what would now be called a "Release Candidate". A Beta had all features implemented and all known bugs fixed. It was something that we were ready to let the outside world see. We would have a limited number of customers test the Beta in real-world circumstances. Some software companies went further, using a Gamma cycle to make absolutely sure that no bugs remained. Nowadays, what's really an Alpha is called a Beta, and what's really a Beta is called the shipping version. We've somehow gotten to the point where no software is ever really finished and where we all constantly use Betas in production environments. And this so-called Beta 3 of W2K is what I'd call a late Alpha. Right now, to my mind at least, it is very far from being ready for prime time. It crashes frequently, and has numerous weirdities, including a very aggravating random screen flicker. More than anything, Beta 3 reminds me of Windows 9x. It has all the PnP and other stuff that heretofore has been present only in Win9x. It also has the stability problems of Win9x. Obviously, Microsoft can define W2K however they want to. But if they intend to ship a version that includes all the stuff they've promised, I don't see any way they can do it in less than a year. * * * * * Two tragedies in the morning paper. Two children dead. The first, an eleven-year-old girl was killed by her six-year-old brother, who found a shotgun in the weeds, picked it up, pointed it at her, and pulled the trigger. I hope they find the person who left that shotgun there and throw the book at him. The second, a six-year-old boy, was killed by a driver who didn't realize that she was required to stop for a stopped school bus. Barbara and my first reaction was anger that anyone could be so stupid as to do such a thing. Our second reaction was more understanding. The driver was from Costa Rica, works for Sara Lee there, and was visiting a Sara Lee plant here. She told the police that she saw the school bus, but didn't realize that she was required to stop because in Costa Rica there is no such requirement. As I told Barbara, that's analagous to someone from a state where right-turn-on-red is legal making such a turn while visiting a state where right turns are not allowed on red. He's breaking the law, but unintentionally and without malice. Barbara mentioned that on her trip to Nova Scotia last fall, she and her parents were surprised to find that drivers came to a complete stop for pedestrians under any circumstances, and that a visitor from Nova Scotia might easily be run down here because he would expect U.S. drivers to do the same. So it seems to us that, although a little boy is tragically dead, it was indeed simply an accident, with no malice nor even negligence. Just because a very bad thing happens doesn't necessarily mean that someone is at fault. That's no consolation to the parents, of course, but the driver will live with this for the rest of her life.
|
|
Wednesday |
Wednesday, May 19, 1999 I've gotten quite a few messages asking about Jerry Pournelle's site. I don't know exactly what the problem is, but the site was recently moved to a different server, and it seems that the DNS changes have not yet propagated. We don't know at this point how long it's going to take. * * * * * I've given up doing a weekly book recommendation, but the one I'm reading now is worth recommending. It's titled The Professor and the Madman and is about, of all things, the creation of the Oxford English Dictionary. The professor is James Murray, who spent more than 40 years as the editor of the OED, dying before its completion after 70 years of effort. The madman is Dr. W.C. Minor, a delusional American Army Surgeon who was confined to the Broadmoor insane asylum for murdering a man he mistook for an Irish assassin. Via correspondence with Murray, Minor contributed significantly to the OED. This strange but fascinating tale is one of the best books I've read in months. If you have any interest in the English language or the OED, this book is a must-read. You can buy it directly from Amazon.com by clicking here. * * * * * This from Paul Robichaux [paul@robichaux.net]. It's a long one, but I think it's worth the space to post and your time to read:
I've written about this before, and Dean Koontz treated the subject at length in his novel Dark Rivers of the Heart. People who still believe that the federal government is basically good and well-meaning have difficulty believing that this can be true. But it is. Your property can be confiscated at will by government agents. You don't have to be convicted of any crime. In fact, you don't have to be charged with any crime. In fact, a major goal of people who have had their property so confiscated is to get themselves charged with a crime so that they can prove their innocence and have some hope of getting their property back. But, hard though this may be to believe, even if you are charged and subsequently acquitted, the government is under no obligation to return your property, and often doesn't. We are living in a totalitarian state. Believe it. Government is not your friend. * * * * * This from Don Armstrong [darmst@lto.nsw.gov.au]:
I'm not debating the fact that this woman broke the law. What I'm saying is that she had no intent to do so, and should not be punished for it. The old legal dictum you mention is a pragmatic one. Ignorance of the law is not a defense simply because if it were the government would have to prove in each prosecution that the defendant knew and understood the law that he had broken. Proving knowledge of something is a difficult thing to do, and placing that burden on the prosecution would ensure that few crimes were prosecuted. You are correct in the sense that there is absolutely no doubt that this woman could be prosecuted for breaking the law by passing a stopped school bus. But "law" and "justice" are not the same thing. In fact, they seldom intersect nowadays. A principle of common law is that, to be guilty of a such a crime, a person must either (a) take an action that he knows to be forbidden, with malicious intent to do harm to another, or (b) act in reckless disregard of the life or safety of another. It's clear that this woman should not be prosecuted under (a). No one has suggested that she intentionally violated any law, or that she intended to kill the child. As far as (b), the ordinary standard is what action a prudent man would have taken in the circumstances. In Costa Rica, the law requires that drivers who encounter a stopped school bus slow down and proceed with caution. That's exactly what she did, by all accounts. In doing so, she was behaving in accord with the "prudent man" standard by her lights. Your examples are red herrings. If you come to this country and drive on the wrong side of the road, you are doing so in the clear knowledge that that is wrong. The fact that all of the other traffic is driving on the right is sufficient evidence for any prudent man to realize that driving on the left is not a good idea. Had this woman passed a line of stopped cars before striking the child, a case could be made that she had sufficient example from others to indicate that she should have stopped. But that didn't happen. She simply proceeded cautiously as she would have when driving in Costa Rica. As far as your shotgun example, any prudent man realizes that a shotgun is a dangerous weapon and that pulling the trigger will discharge it if it is loaded. Anyone who simply picks up a shotgun, points it at someone and pulls the trigger is guilty, at the least, of reckless disregard of that person's life. But to extend your example further (and this was an actual court case), what happens if the person unloads the magazine but does not realize that doing that does not render the weapon safe. In the case I mentioned, the person had removed the magazine from an automatic pistol, not realizing that he also had to clear the chamber. He pointed the weapon in what appeared to be a safe direction and pulled the trigger. The pistol discharged, and the bullet passed through a wall, killing someone on the other side. Now, it might be argued that someone that ignorant about weapons had no business handling one, but the fact is that that person took what appeared to him to be reasonable precautions. The bad outcome was not due to malice or carelessness on his part. That, incidentally, is why all of my guns are always loaded and everyone around me is made aware of that fact. People don't accidentally shoot other people with "loaded" guns. It's the "unloaded" ones that are horribly dangerous. I didn't realize that I had become one of the world's opinion makers. I thought I just wrote some random daily notes that a few hundred people find interesting enough to read regularly.
|
|
Thursday |
Thursday, May 20, 1999 Once more, the First Amendment is under attack. This time, it's a bunch of concerned idiots who want laws passed to prevent bomb-making information from being disseminated on the Internet. Laying aside the constitutional objections, consider the practical ones. First, the Internet is global, while legislation is local. No law passed by any legislative body can eliminate this information from the Internet. All it can do is drive one more nail into the coffin of the First Amendment. Then there's the small matter that the Internet is just one of many sources of such information. I can visit any decent library and find all the information I need to construct a bomb. I can order any number of books from any number of publishers that provide step-by-step instructions. Even the U.S. government disseminates such information. I can order books from the Government Printing Office that describe in detail how to construct a bomb from readily available materials. They're called Field Manuals, and the military produces them in great abundance. For that matter, I don't even have to order these manuals (placing my name on a list somewhere). I can simply visit any of the many Depository Libraries (those that maintain copies of government documents) and read the information at my leisure. These demands simply reflect the trend to blame anyone except the person who takes the action. Shootings are the fault of the gun manufacturer or of the gun itself. Bombings are the fault of the Internet for making the how-to information too easily available. How much longer can it be before drunk driving is the fault of the automobile? The fault, folks, lies with the person who commits the crime. Period. * * * * * The Register reports that Intel has officially declared the Pentium II dead. It will no longer be available after the end of this year. Of course, we’ve been saying that for some months now. We're now recommending the Celerons for folks who are conscious of bang-for-the-buck; the Pentium III/450 for those who believe that SSE will be well supported and is worth the extra cost; and the Pentium III/500 and /550 for those willing to spend a lot more for the absolute best performance available. We don't currently recommend any AMD offering. * * * * * This from Bo Leuf [bo@leuf.org]:
I agree with all of your points. And I'll go further: "responsibility" applies to actions, not to words or thoughts. I remember reading when I was very young about a man who had been arrested for "inciting to riot." It seemed to me then, and it still seems to me, that this man did nothing to deserve being arrested and imprisoned. He was simply exercising his First Amendment rights to free speech. Although a riot did in fact ensue, this man did not participate in it. Those who actually participated should have been (and in fact were) charged with various crimes, including battery, destruction of property, and so on. Actions, in other words. I think it was Rene Descartes who said, "I may disagree with what you say, but I will defend to the death your right to say it." It seems clear to me that the framers of the Constitution took that position, and intended the First Amendment to prohibit the government from restricting speech in any way whatsoever. The First Amendment has been under constant attack since it was written, and now has so many holes that it has become ineffective. I think the death of the First Amendment really began with Oliver Wendell Holmes red herring about "crying fire in a crowded theatre." It's been chipped away at in the name of "National Security" and restricting pornography. People have been charged with crimes for uttering racial epithets. The whole purpose of the First Amendment was to prevent the government from restricting what someone says because they don't like what he is saying. They now have that power, and the First Amendment is effectively dead.
|
|
Friday |
Friday, May 21, 1999 I've been spending a lot of time on the phone and exchanging email with Pournelle the last couple of days, trying to help him do something about his website and mail problems. As things stand now, his site is again accessible, although he's still limited to updating only his major pages. If you'd given up on his site because of all the problems, try it again now. Hopefully, things will be back to normal in the next few days. Well, I need to get my network backup going, and I have a bunch to do on the books today... * * * * * This from Paul Robichaux [paul@robichaux.net]:
Duh. I knowed that. Sometimes my fingers don't type what I tell them to. Let's see, was Descartes the "Cogito, ergo sum" guy? For some reason I always have trouble keeping straight all you great writers with French names. I think I once credited your Registry book to Dumas pere.
* * * * * This from Francisco Garcia Maceda [maceda@pobox.com]:
Freedom of speech is either absolute or does not exist. You say that "just as most freedoms it should not be infinite". But if it is not infinite, then it is not free. Obviously, if limitations are placed on speech, someone or something has the power to determine what is and is not acceptable speech. As the Romans observed, "Qui custodiet ipsos custodes." There is a big difference between someone exercising his right to free speech and someone giving orders to subordinates. In the case I mentioned, the man was speaking before a crowd. He had no power to compel action on the part of any of them, so whether or not he directly advocated a riot is immaterial. Each member of that crowd made his own decision. In the cases you mention, the leaders are in a position to compel behavior from subordinates. * * * * * This from Don Armstrong [darmst@yahoo.com.au]:
I think the problem is that you, like most people, associate the legal concepts "legal" and "illegal" with the moral concepts "right" and "wrong", respectively. That's a facile answer, but the unfortunate fact is that there are many things that are illegal that are not morally wrong, just as there are many things that are completely legal but are morally wrong. In fact, there are many times when the morally correct action is very much illegal. I prefer to apply Immanuel Kant's Categorical Imperative. I can't remember a time when I had anything but contempt for the law. Laws are made by men. Unfortunately, in the United States at least, the men who make our laws are the least suited to do so properly. Politicians in this country are rightly perceived by the majority of the people as being the scum of the earth. I have no interest in any of the laws these people pass. I obey their laws only to the extent that (a) a particular law happens to co-incide with my perception of correct behavior, and (b) that the product of the penalty for breaking it and the likelihood of being caught while doing so exceeds the inconvenience of obeying it. And I'm by no means alone. Everyone in this country breaks numerous laws every day, without necessarily even being aware that they are doing so. The unfortunate result of having too many laws, and of having them made by stupid, evil, corrupt lawmakers, is that people develop contempt for all laws. Reasonable laws are obeyed by the vast majority of people, but unreasonable laws simply dilute the value of the reasonable laws. When you pass a law setting a 55 MPH speed limit on a road whose free-market speed limit is 75 MPH, you simply cause people who instinctively understand that that's a stupid law to become lawbreakers. When you make it illegal to buy a toilet that actually works, you simply cause otherwise law-abiding people to become lawbreakers simply to get a functional toilet. That woman behaved in a reasonable manner which had an unexpectedly bad outcome. To say that she should be punished or that she is somehow more responsible for the outcome simply because some idiot passed a law that she was not aware of is simply not on.
|
|
Saturday |
Saturday, May 22, 1999 I did my standard network backup yesterday, to the Seagate Travan TR4 drive installed in my resource server. As usual, the drive spent almost no time streaming, and was constantly backing up and restarting. The throughput when backing up the local drives is about 38 MB/min, but on network drives it drops to 15 or 20 MB/min. Although I can't imagine that anything but the network itself or Windows NT is at fault, I decided as an experiment to try a backup with compression turned off, on the theory that perhaps the slow (Pentium/133) processor in the resource server was having problems compressing data in a timely manner. No difference with compression on or off, so that theory gets pitched. But at least it was worth doing the experiment. I suppose I really should go out and buy a 100BaseT hub. All my network cards and cabling are 100BaseT-capable, so swapping in the new hub should be the job of a few minutes. * * * * * This from Chuck Waggoner [waggoner@gis.net]:
Thanks for the kind words. I try, but I suspect we're on a downward slope that will not end short of a crash. As far as the phone license, I got it back in the middle 1970's. I never worked in broadcasting per se, but I was a ham operator, and did get involved in running the college radio station. As far as the book, things are progressing slower than we'd like, but they are progressing. This is the first hardware book I've written. I was used to doing first drafts on software books at about a chapter per week. With hardware, I have to spend more time on the phone with vendors and building and testing stuff than I can writing. Also, doing two similar books, I thought there would be a great deal more synergy than there turns out to be. I hope PC Hardware in a Nutshell will be complete this fall and the book Pournelle and I are doing not too long thereafter. * * * * * This from Bo Leuf [bo@leuf.com]:
No comment... * * * * * This from Robin Gould [rgould@ihcc.org]:
Yes, I shouild have thought of that. That's one of the downsides to the fact that I work mostly with very new equipment constantly. PnP didn't have much to do with your problem. It's possible (although unlikely) that you simply had a bad PCI slot. What's more likely is that the card was a bus master and the original slot did not support bus mastering. Nowadays, all of the PCI slots on most motherboards support bus mastering. On older motherboards, that was often not the case. My guess is that yours is such a motherboard, and that you simply had a bus master network card installed in a slot that didn't support bus mastering.
|
|
Sunday |
Sunday, May 23, 1999 The usual house cleaning, laundry, etc. today. Not much to write about. * * * * * Tomorrow, I'm going to start a Current Topics page, where I'll ask for your help. The Current Topics home page will link to pages on various subjects. Unlike this page, those pages will not be chronologically-oriented. I'll put them up and add material to them as it arrives. I may write an intro for a given page and then throw it open for contributions by readers. The pages will remain as resources organized by topic for convenient reference. I'll use these pages, among other things, to learn more about things I'm writing about. For example, some time back we had a great on-going discussion about CD-R and CD-RW drives, IDE versus SCSI interfaces, incompatibility of various media types, etc.. That material is still accessible via the Search function, but it would be better organized in one place. This idea is similar to what Pournelle does on his site with some topics, but I intend to make this page more structured and formal. Also, for now at least, I plan to restrict these pages to discussion of computer-oriented topics rather than politics and so on. I hope all of you will choose to participate both by reading and by contributing to the pages on topics you know something about. I'll maintain a permanent name for each page, so they will be bookmarkable. * * * * * This from Francisco Garcia Maceda [fremen@pobox.com]:
A "right" that is not absolute and unbounded is not a right at all. It is a privilege, which may be modified or withdrawn at any time by the issuing authority. Bo Leuf observed several days ago that Europeans have difficulty understanding the way we look at things, just as we have difficult understanding how Europeans look at things. In Europe, the rights of individuals have always been subordinated to the rights of states. The US was created on exactly the opposite principle, although our government has been doing its best for at least 150 years to take those rights away from us. In essence, Europeans have no rights and never have had. Certainly, some states, notably Britain, have been better historically at formalizing respect for individual rights (e.g. the Magna Carta), but even Britain puts the state above individuals, as D Notices and the fact that their constitution is unwritten illustrates. As far as the power of speech, I was not debating that. I was simply saying that advocating an action and performing that action are two entirely different things. For example, I abhor everything the Ku Klux Klan, Nazi Party, and similar organizations stand for, but I will vigorously defend their right to speak freely and to advocate violence against blacks, Jews, Catholics, and others they consider worthy of their hatred. Despicable though I may consider their opinions to be, they have a right to express them. If they attempt to put those beliefs into action by lynching a black person or burning a synagogue, that's another matter. If they try to do that, I'll happily shoot the sons of bitches.
|