|
Week of 29 November 1999 Sunday, 05 December 1999 09:55 A (mostly) daily journal of the trials, tribulations, and random observations of Robert Bruce Thompson, a writer of computer books. |
|
wpoison
|
Monday, 29 November 1999 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Duncan and Malcolm continue to get along pretty well. Duncan occasionally snarls and lunges at Malcolm, but we don't blame him a bit. In fact, we encourage him to do it. Malcolm is still in his fang-it-if-it-moves stage, and Duncan is getting tired of having his face, ears, feet, tail, and other appendages bitten. Most of the time, though, they co-exist peacefully. When Duncan was Malcolm's age, he wanted nothing more than to snuggle up to Kerry to take a nap. Kerry would never let him do that, and we expected that Duncan would feel the same about Malcolm. He doesn't, though. He allows Malcolm to cuddle at nap time. I shot the photo below just as they both woke up from a long nap.
What's really surprising is that he lets Malcolm snuggle up against his rear end. If I were Duncan, I'd be a little more concerned about my dangly bits, especially since Malcolm often celebrates the end of a nap by fanging whatever happens to be closest to him when he wakes up. Now I have to get kiwi in place and running as my main workstation. I've already put the cover back on and moved it to its new home. Now I need to connect cables and start loading software. I have a manual KVM switchbox that I'll use to share the monitor between kiwi and kerby. I won't be able to use it to share a keyboard and mouse, though. The switch box has DB9 serial connectors for the mouse and AT keyboard connectors. Both machines use PS/2 keyboards and mice. I don't have the necessary adapter cables, and I don't feel like making them. Actually, I prefer to have separate keyboards and mice anyway. * * * * *
I figured it had to be something like that. Obviously, since Windows NT 4 Workstation can still boot from and access the volume. I haven't tried installing NT5 first, or if I have I don't remember what happened. Come to think of it, I probably did try that, since I did probably a dozen installs of each of three operating systems, trying to get Windows 98, Windows NT4 Workstation and Windows 2000 Professional to triple-boot. I finally gave up on Windows 98. If you think writing any software book is tough, try a hardware book. I usually do first draft chapters on software books at the rate of one a week. Hardware has turned into more like one a month. There's just no comparison. I'm greatly looking forward to getting these hardware books done and starting in on Windows 2000.
|
|
|
Tuesday, 30 November 1999 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] About a fifth to a quarter of my regular readers have bookmarked my Daynotes home page, and read the current page by clicking on this week's link from the Daynotes Home page. Those readers won't notice any change this morning. The remaining 75% to 80% of my readers have bookmarked thisweek.html, and will have noticed that that page is now a redirector page that automatically displays this named weekly page. It was because of reader requests that I originally started using thisweek.html to provide a bookmarkable page. That method has the advantage of providing one-click access to the current week's page. It does have several drawbacks, though. Probably the biggest of those is that it makes it difficult to link to things I publish on my daily notes page, because this page would be named thisweek.html only until the beginning of next week. There are several other drawbacks, including the fact that search engines index the contents of thisweek.html and present them to searchers. A week or six months from now, someone who locates information in a search engine clicks on the link to thisweek.html and finds that the current thisweek.html page has nothing whatsoever to do with the page they thought they were retrieving. I've tried to get around these problems by maintaining duplicate pages. For example, this page would formerly have been stored on my server as both thisweek.html and 1129RTDN.html. In theory, that provides a static page name that allows someone to link to content on this page without having the link become useless at the end of the week. In practice, no one did that, so it simply meant that I was spending more time keeping two pages synchronized. Brian Bilbrey changed his site yesterday, adding a redirector page that automatically redirects requests to his current weekly notes page to the uniquely named version of that page. That seemed like a good idea to me, so I stole his idea. Henceforth, I will update only the uniquely named weekly page (1129RTDN.html in this case). The page thisweek.html is now a redirector, which I will change only once a week so that it will point to the current weekly page. I've made only two significant changes to Brian's redirector page. First, I increased the timer from 0 seconds to 2 seconds. That avoids the "Hotel California" problem whereby someone who hits the redirector arrives at this page, but cannot use the Back button to get past the redirector page when attempting to "back out". Second, I added the HTML necessary to ensure that thisweek.html is not cached. Because thisweek.html points to a uniquely-named weekly page, users would otherwise have to remember to refresh that page each week. Eliminating caching on that page means that it's reloaded each time it's retrieved. That shouldn't be a problem, because the page is less than 1 KB, so reloading it takes only a small fraction of a second even on a slow link. Yesterday I was overrun by a horde of lemmings. I moved kiwi to its new location, shutdown kerby, and connected the KVM switch box to the monitor and both machines. When I powered them up, nothing happened. After verifying connections and scratching my head for several minutes, I finally realized that I had the switchbox connected to the wrong video connector on the Matrox Millennium G400 card in kiwi. That card has the dual-head feature, which allows you to connect two monitors, but that feature is supported only under Windows 98. That still didn't explain why I wasn't getting any video out of kerby, though. I powered everything down and connected the cables properly. This time, everything came up normally. I was able to switch between the video on both systems with no problem. Well, one problem. The monitor looked terrible. There was a background shimmering kind of like waves washing across beach sand. My first thought was that I'd gotten one or more of the video cables too close to a power brick or something, so I spent some more time moving cables around and so on. It didn't help. Finally, I unplugged everything and reconnected the monitor cable directly to kerby. The monitor was still shimmering I concluded that either my main monitor had coincidentally developed a problem or that this manual switchbox had damaged it. Pournelle has warned me repeatedly about using cheap manual switchboxes rather than the electronic KVM switches that he uses, but this is the first thing that makes me suspect that he may indeed be right about the hazards of manual switchboxes. I remember years ago that HP voided the warranty on your LaserJet if you connected it to a manual switchbox, so there may be something to that warning. At any rate, I decided that it was time for a new monitor. Fortunately, I didn't need to do much research, having just done so while I was working on the video chapter. There are four monitor manufacturers in the first rank--Hitachi, Mitsubishi, NEC, and Sony. All of the others, including some big names like Nokia, are a step behind the Big Four. One of the biggest names of all, Viewsonic, apparently doesn't actually manufacture monitors. They OEM them from various Pacific Rim sources, which means that we could each buy the same model of Viewsonic and find that the two monitors had actually been made by different manufacturers. How do I know all this? By talking to tech people from the various monitor manufacturers. I ask each of them, "Other than your own company, who makes the best monitors?" Without exception, the Big Four are mentioned as the best manufacturers. When I ask about Nokia, Nanao/Eizo, and other respected makers, the responses are usually similar, "Yes, they make a good monitor, but not quite as good as those from Hitachi, Mitsubishi, NEC, or Sony." When I ask about other popular monitor makers like Samsung or Viewsonic/Optiquest, the responses are similarly similar, "They make decent consumer-grade monitors, but they're not as good as the Japanese Big Four." Also, I'm told that the second- and third-tier brands have greater variability than the Big Four. If you compare two monitors with sequential serial numbers, one may be superb and the other mediocre. That's less a problem with the Big Four. So I knew I wanted a monitor from one of the Big Four. Of those, the one I've been most impressed with is Hitachi. Their models generally have the same or slightly better specs and the same or slightly better price than directly competing models from Mitsubishi, NEC, and Sony. I dithered about 17" versus 19". Barbara, as usual, put her finger on it: "You spend all day working in front of that monitor. Buy the 19" one." So I decided to buy a 19" Hitachi. The last time I looked at the Hitachi web site, they had three 19" models, the 751, 752, and 753. The low-end 751 does everything I need to do, namely 1280X1024 resolution at 85 Hz. The more expensive 752 and 753 would be useful primarily for graphics artists and others who want to run 1600X1200 at high refresh rates. So I'd about decided on the 751 when I visited the Hitachi home page again. I noticed that they have a new model, the 761. It appears to have specs almost identical to those of the 751, but with one difference. The 761 is a short-neck monitor. So I called my contact at Hitachi and told her I wanted to buy a 19" Hitachi, either the 751 or 761. I said, "it appears that the 751 and 761 have nearly identical specs and that the only difference is that the 751 is long-neck and the 761 short-neck. Is that true." She responded, "Well, not exactly." After saying that Hitachi made the absolute best short-neck tubes and that the only other manufacturer of short-neck tubes is Panasonic, which is used by all other monitor manufacturers, and is inferior to the Hitachi short-neck (all of which I believe to be true), she told me that the best short-neck tubes can't compare with even a mid-range long-neck tube as far as display quality. That makes sense, because the deflection angle is 100 degrees for a short-neck versus 90 for a long-neck. When you alter the geometry of a monitor to shorten it, you must make compromises to achieve the tighter bending of the electron beam that the short neck requires. That's why I love talking to techs. They tell the truth. So I ordered a Hitachi SuperScan 751. I got it for the dealer-demo/press price, a little over $400, which is comparable to what I'd have paid for it if I'd bought it from a web site, but I prefer buying directly from Hitachi. * * * * *
Thanks. I'm very glad to hear that. In fact, I reviewed your OnStream ADR 50 SCSI tape drive rather savagely. The problem wasn't the drive itself. The problem was that it used bare pins rather than a shrouded SCSI connector. Those bare pins are connected to a circuit board, which was notched to fit the key on the SCSI cable supplied with the drive. When I aligned the key with the notch and pressed firmly to seat the cable, it mashed the pins flat. After removing the cable and spending the better part of an hour using my needle-nose pliers to straighten the pins, I discovered that the notch in the circuit board had been cut incorrectly. It was off by more than a full pin spacing from where it should have been. Now, of course, it's possible to file down the key on the SCSI cable and connect the cable carefully while keeping all the pins lined up, but I'm sure that's not something that you'd expect an average buyer of your drive to do. By adding a shrouded connector, you remove my last objection to OnStream drives. I will recommend them on that basis. * * * * *
* * * * *
Well, Intel has certainly become dominant in motherboards and chipsets, but it wasn't from any lack of competition. Intel simply makes superb products, despite well-publicized aberrations like their problems with the 820 chipset and RDRAM. I don't think we need worry about monopoly issues, with ALi, SiS, and VIA competing strongly in the chipset market and with any number of motherboard manufacturers cranking out new products every week. As far as your questions, there's enough truth in the accusations to make a good story, but that's about it. There's no grand conspiracy by Intel or anyone else to render products incompatible or obsolete. What we're seeing is simply the march of technology. The problem of newer processors being incompatible with earlier motherboards has several possible causes. First, of course, is physical incompatibility. A socketed processor won't fit a slot and vice versa. But within those constraints, everything is compatible. Any Socket 370 processor, for example, will physically fit any Socket 370 motherboard. There are no arbitrary differences in pinouts between different Socket 370 processors to render one or another unusable with a given motherboard. For example, see the Intel CA810E motherboard, which supports 66 MHz FSB Celeron CPUs, 100 MHz Pentium III CPUs, and 133 MHz Pentium III CPUs. When the 100 MHz FSB Celeron CPUs ship, it will support them as well. Then there's power. Nearly any modern motherboard can detect and switch between the 2.8 V required by the early Klamath-core Pentium II CPUs and the later 2.0 V Deschutes-core Pentium II CPUs. Some motherboards also support using new low-voltage CPUs, but require a BIOS update to do so. The controlling factors are which VRM (voltage regulator module) the motherboard uses and how much current the CPU requires. Some early Intel Seattle SE440BX boards, for example, were produced with VRMs that could only supply adequate current for processors up to and including the Pentium II/400. Using a Pentium II/450 or any Pentium III in those motherboards simply overdrives the VRM and burns it out. Later revisions of the Seattle-1 supported up to the Pentium II/450 or Pentium III/450, but not the Pentium III/500 and higher. The motherboard market is extremely competitive, so criticizing Intel or any other manufacturer for not "over-engineering" their motherboards by using higher-rated VRMs than needed to support currently available processors is really not fair. If anything, Intel (along with EPoX) tends to over-engineer their boards more than most manufacturers. Just look at the size, number, and brand names of the capacitors on Intel or EPoX boards versus those on lesser motherboards. The final issue, of course, is BIOS support. Once again, Intel is among the best in the industry at supplying regular BIOS updates for their boards. So, no, I don't think that there's any grand conspiracy. If anything, these incompatibilities between newer processors and older motherboards are actually a Good Thing. They prove that Intel and other manufacturers are not making the mistake of making backward compatibility a prime issue. That's what stuck us with legacy products like DOS for a decade longer than we should have been. The simple fact is that probably less than 0.1% of systems are ever upgraded by swapping in a faster processor. Designing 100% of motherboards to accommodate that tiny minority simply makes no sense.
|
|
|
Wednesday, 1 December 1999 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Pournelle posted a photo yesterday of the 37 pills he takes each morning, so I figured I could do no less. Here are my two daily pills, which I usually take with dinner. The one on the left is a Centrum From-A-to-Zinc vitamin tablet. The one on the right is a Bayer 325 mg aspirin tablet. The vitamin tablet helps make up for my hideous eating habits. The aspirin will in theory help prevent a heart attack.
Actually, I usually take a 500 mg enteric-coated aspirin, but all we have at the moment is regular old aspirin tablets. I'm sure that Dr. Jim will question the wisdom of arbitrarily taking aspirin routinely, but I regard routine low-dose aspirin therapy as relatively benign, and 3.0 to 4.7 mg/kg daily is a very small dose. The only real danger is that aspirin irritates the gastric lining, but I always take it with food. Winter may finally be arriving in Winston-Salem. Temperatures last night dropped to about 23F (-5C). As I write this at about 10:00 a.m., our indoor/outdoor thermometer says they've climbed to about 34F (1C). I'm now keeping a parka near the front door for puppy emergencies. I told Barbara last night that I think my Northern-boy immunity to cold has finally worn off. Apparently, there are actual physiological differences between people who live in cold climates and those who live in warmer ones. Kind of a natural anti-freeze that provides some immunity to cold. But if someone accustomed to cold weather moves to a warmer area, that immunity gradually degrades. I remember that when I lived up North 20 years ago, I was reasonably comfortable being outside for a few minutes in a short-sleeve shirt when the temperature was -20F (-29C). Nowadays, if I go outside in short sleeves when the temperature is +20F (-7C) I start shivering after a few minutes. FedEx just showed up with both versions of VMware. I don't have a Linux box available to install it on yet, but I have any number of Windows NT/2000 boxes, so I plan to install VMware for Windows NT/2000 and see what it does. Brian "Brain" Bilbrey posted a detailed narrative, including screen shots, of his experiences installing VMware for Linux, so it's probably just as well that I concentrate on the NT version for now. Well, I couldn't wait, of course. I've just spent the last couple of hours installing and playing with VMware for Windows NT/2000. Installing VMware and configuring a virtual machine for Windows 2000 Professional took about five minutes. Installing W2KP on the virtual machine and playing with it took the rest of that time.
First impressions: VMware is a dynamite product. I can see numerous uses for it. Everything from grabbing difficult screen shots during Setup to creating virtual "sand boxes" for testing applications and operating systems to training system administrators and testing network security. You can download a thirty-day eval version of either the Linux or NT/Windows 2000 versions (or both) here. The single-user electronically-distributed version sells for $299 on either platform ($10 more for a packaged version, which also includes a CD containing a copy of SuSE Linux pre-installed in a Virtual Machine), and is also available for $99/$109 to students and hobbyists (like all of us). I've about given up on multiple-booting different operating systems on one machine. It simply causes too many problems, not least of which is that it sometimes causes Windows Networking to have a virtual mental breakdown. With VMware, you can install numerous operating systems on one physical computer, including MS-DOS 6+, Windows 3.1, Windows 95, Windows 98, Windows NT 4 Server and/or Workstation, Windows 2000 Professional and/or Server; FreeBSD, and various Linices, including Red Hat 5.0 or higher, SuSE 5.3 and higher, and Caldera OpenLinux 1.2 and higher. They're even experimenting with NetWare. If you have any need at all to run multiple operating systems, I strongly suggest that you download the eval version and give it a try. I suspect you'll end up buying it. As usual, mail is backing up. I'm about out of time and energy, so it's short-shrift time.
No, actually. I was referring to the Daynotes home page (listed in the left column) rather than the main index page for the site. * * * * *
I wasn't knocking Eizo monitors, really. I'm sure they make a very good product. In fact, when I started asking questions, I assumed that Nokia and Nanao/Eizo would fall in the top group. But no one mentioned them when I queried them as to the best competing models. There is, of course, greater variability in monitors from sample to sample than with any other computer component. Shipping also plays a role. A monitor that leaves the factory as a superb specimen may be only mediocre by the time the shipping company drops it at your front door. * * * * *
Arghh. Screwed by Windows NT and/or FrontPage again! You can access that file using lower-case "rtdn". Microsoft's inexcusably casual treatment of case has once again caused a problem. What's worse, is that FP won't let me rename the file to 0817RTDN.html from 0817rtdn.html, because it regards them as the same file name! I'm going to have to go back, rename each problem file to some arbitrary name and then re-rename it back to the correct file name, but using upper-case letters. And even then, there's no guarantee that it'll work, or if it does work that it will stay the way it is. FP arbitrarily converts the case of file names (along with changing their date/timestamp) for no reason whatsoever. I hate Microsoft. The obvious answer is that if you want to use FP, you should choose a web host that runs Windows NT Server rather than the case-sensitive UNIX. I hate Microsoft. I'll get to the problem files when I can, but it's likely to be a while. Until then, if you get such an error, changing the case in the browser should allow you to retrieve the file. I hate Microsoft. * * * * *
Thanks! I wasn't aware of that. The last time I spoke with my contact at Matrox, she told me that NT support for dual-head wasn't implemented yet. * * * * *
I don't know. Perhaps one of my readers will. But I suspect Adaptec is your best bet. * * * * *
Well, yes, but again the question is "so what?" The point is that current motherboards accommodate a broad range of processors. The fact that old motherboards don't accept some new processors is certainly nothing out of the ordinary. When Intel started shipping split-voltage P55C Pentium/MMX CPUs, no one thought it unusual that they needed a new motherboard, both for split-voltage support and MMX support. It's the same situation here. * * * * *
No, you're correct that that link is an artifact. I have more and more of those as the web site gets older. That one's fixed now, anyway.
|
|
|
Thursday, 2 December 1999 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Today is Barbara's birthday. She turns twenty-twenty-five today, and doesn't look a day older than twenty-eleven. I couldn't figure out what to get her. She'd asked me to buy her a copy of Family Tree Maker (or something like that) back in about March. I'd tried to do it several times, but every time I tried the vendors were out of stock. Apparently, these programs are updated frequently. At any rate, I finally got serious about it last week and did some reading on the web. PC Magazine reviewed four such programs last April, three of which were from different divisions of the same company. There was FT Maker, FT Creator, and Ultimate FT. Reading the PC Magazine review, it looked as though Ultimate FT was the most powerful. I checked with Insight. They had it in stock, so I ordered it. I just hope it does what she wants, because I'm clueless about genealogy (or is that "geneology"?). We were supposed to go to a nice restaurant for dinner tonight. I've come down with some sort of bug, so it's not clear whether we'll go or not. Barbara insists not, and says it's no big deal, but she does think birthdays and holidays are special, so we'll see. * * * * *
Thanks very much for the offer, but I'm in the same position you are. I have half a dozen cable sets myself. Incidentally, all KVM boxes I've seen use those MM cables. You're correct that it'd make more sense to use MF cables, which could also be used as normal extension cables, but perhaps there's a reason why they do it that way. If I were you, I'd put those cables somewhere where you can find them, because computers have a way of proliferating. You may find in a year or two that you suddenly need to hook up a third (or fourth) machine to run Linux or something. Thanks again. * * * * *
I must admit that I was surprised to learn that children's aspirin is still available, what with all the to-do about Reyes Syndrome a few years back. I've never been a big fan of acetaminophen or the other NSAIDs, but I figured that a parent who gave a kid aspirin nowadays would be charged with child abuse or something. I'll speak to Barbara about upping my multivitamin dose to two per day. Thanks. * * * * *
Thanks, but I already have it. Being a Real Man, I don't hesitate to download 70 MB files over my 31.2 dial-up connection. If it takes eight hours, so what? It's all in the background anyway. * * * * *
Excellent point, and one I should have mentioned. Inexpensive monitors simply don't last as long as good monitors. I've seen that myself around here. I have a Princeton that died about a month after its warranty expired, and a couple of Mag Innovision monitors whose picture quality started to degrade severely after little more than a year. I use them for servers and test bed systems. I've talked to various people in the monitor industry about this problem, and the consensus is that inexpensive monitors are likely to degrade severely after as little as a year of sustained use. The more expensive monitors use better grade components, have better quality control, and are likely to last for three years or more of full-time use (more if they aren't used 8 hours/day). Powering down your monitor when it won't be used for an extended period extends its life, as does using a screen-blanking screen saver (versus one that displays changing images).
|
|
|
Friday, 3 December 1999 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Barbara had the CD changer crammed full of Saturnalia CDs yesterday afternoon. It was playing along happily, finished a track, and crashed. The first thing I checked for was a Windows Inside logo, but that didn't seem to be the problem. The thing was still lit up, but pushing any of the buttons did nothing. Not a good sign. I suppose everything has to die sometime, but that CD changer was bought in 1994 and has probably had less than 200 hours total use in the intervening five years. It's a JVC, and I'm becoming less and less happy with them. I wouldn't have minded the CD changer dying so much, but it has five of Barbara's CDs in a cartridge, and it won't let go of them. I put it on the kitchen table and pulled the cover. There's an immensely complex physical mechanism that moves the CD from either the individual tray or the 5-CD cartridge back to the actual CD player mechanism at the rear of the CD changer (to the left in the photo.) That mechanism has taken hold of that cartridge and won't let go of it. I was hoping that there would be an obvious motor fuse or manual release visible, but I can't find anything. So, the choice is either to carefully disassemble the mechanism or to use brute force. I'll probably use the latter.
I'm used to working on PCs, which have all for years used surface-mount technology for mounting components, so I was rather surprised to see how primitive this five-year old JVC CD player really is. The circuit board is neatly done and well-dressed, but the component density is lower than anything I've seen in a computer in twenty years or more, and it uses mostly discrete components.
Well, we've temporarily solved the problem, anyway. I had a spare JVC CD player in my office, which we used to replace the failed unit. What's really weird is that as soon as we put that one in place and fired it up, Barbara observed that it sounded a lot better. I told her that was ridiculous, and that even inexpensive CD players sound just as good as the most expensive ones. I mean, when you have flat frequency response from 2 Hz to 20+ kHz and distortion of about 0.00%, what difference can there be? Unfortunately, if I'm honest, I'd have to say that I think it sounds better, too. The sound seems richer somehow. Of course, that may be because I'd never cleaned the changer. It was coming due for its five-year cleaning, though. Malcolm is turning into a back-sleeper, just like Duncan. Here's one of him near the front door. I continue to be amazed by how well the Olympus D400-Z focuses in low light. When I took this photo, it was so dark in the foyer that I had a hard time telling through the optical viewfinder whether or not all of Malcolm was in the frame. The Olympus took it in stride, though. It focused and exposed properly. A very nice little camera.
Brian Bilbrey reports problems viewing PNG images in Netscape. He can see the thumbnail PNGs fine, but he's unable to see the full-size PNG image when he clicks on the thumbnail. That's actually pretty common behavior. Many browsers display in-line images (those embedded with an image tag in an HTML page) just fine, but are unable to display directly-linked PNG files. There's a great deal more detail about this problem here. The problem is, I don't have a great deal of choice. I won't use GIFs because of Unisys's ridiculous demand for licensing fees from web-site operators. I tried converting the PNGs to JPEGs, but a small PNG turns into a huge JPEG, by a factor of at least five and sometimes ten or more. I could create a separate HTML page for each large image and use an image tag within that page, but I don't really have time to do that. My guess is that PNGs will ultimately triumph, so the real answer is to start using a browser that supports them properly. Supposedly, Opera and IE5 do so, with minor exceptions. In fact, I've seen exactly the same misbehavior that Brian describes in some (but not all) installations of IE5 and IE5.01. We did end up going out to dinner last night, and very nice it was. I'd list what we had, but I don't know the French names for any of it and it doesn't sound nearly as impressive in English. * * * * *
* * * * *
I hadn't thought about the iron issue, but that makes sense. In line with Dr. Jim's advice, I turned to Barbara last night and said, "Dr. Jim says I should take two vitamins instead of just one." She responded, "Nonsense. One is plenty." So I'm taking one. * * * * *
Pournelle does indeed have two doctorates, and Thomas Pabst is an MD. I suppose next you'll be questioning my two doctorates (in Chemistry and Mathematics), but I bought those fair and square. * * * * *
My current view of AMD is basically unchanged. They have a long, hard row to hoe if they're to make any significant progress against Intel. As I've said repeatedly, technology is not the issue. AMD has frequently produced processors with very good technology. I question their ability to execute, which has historically been their problem. That said, they're doing well at getting Athlon's in increasingly faster versions to market. Even more important, although largely overlooked, is the fact that they're starting to ship their 0.18 micron versions. That's good in two respects. First, the Achilles Heel of the Athlon has been its huge appetite for power, which in turn made it difficult to design decent Athlon motherboards and resulted in high heat production. Second, going to 0.18 micron from 0.25 micron means they can fit many more processors on a wafer, which greatly reduces their production costs. It seems to me that AMD's main weakness right now is that they are getting killed by Intel in the mainstream market. The Celeron is low-margin, granted, but Intel sells them by the million. It's all well and good to claim pride of place in the high-margin 733/750 MHz market, but the fact is that relatively few of these processors are sold. AMD needs something to compete with the Celeron, and the K6* Socket 7 processors aren't it. AMD needs to introduce a downscale Slot A version of the Athlon that they can sell in huge volumes if they really want to compete with Intel on an equal footing. The trouble is, I don't think AMD has the fab capacity to do that. * * * * *
Thanks for the kind words. I think it's too early to tell about this new technology. As you say, there have been a lot of good technologies that never went anywhere. With any new technology like this, you basically go from science to engineering to production to marketing. There are a lot of pitfalls along the way, and people are naturally optimistic when they estimate how long something will take and how easy it will be to overcome difficulties. My guess is that it will be at least two or three years, if ever, before this technology becomes widely available. Even then, bad marketing might kill it. I wouldn't hold my breath.
|
|
|
Saturday, 4 December 1999 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Barbara is off this morning visiting a prospective adoptive home for one of their rescue Border Collies, leaving me with demon puppy. Actually, he started turning really demonic several days ago. He was also breaking house training. He had been almost perfectly house trained, but had started having accidents. We noticed that he was whipping around biting at his tail and tail head. That, of course, is not unusual behavior for a puppy, but we were suspicious because Malcolm is Duncan's half brother, and Duncan is subject to impacted anal glands. If you've ever read the Herriot stories, Tricki Wu suffered from impacted anal glands, which Mr. Pumphrey called "flop-bott". The symptoms of impacted anal glands are when the dog chews its back and tail-head and/or drags its butt along the ground. Malcolm was doing both. So we took Malcolm into the bathroom for what we fondly call a "butt squeeze." Sure enough, Malcolm has it too, and badly. I got to hold the sharp end while Barbara was expressing Malcolm's anal glands, and he bit me. I don't really blame him. I'd probably have bitten him had the positions been reversed. Barbara made an emergency appointment with the vet, where Malcolm got a cortisone injection and a course of antibiotics for his impacted and infected anal glands. They're gradually clearing up, although he's still having problems. I think he needs another butt squeeze, but Barbara wants to wait until the swelling goes down some more. When we do it this time, I'll wear heavy gloves. Unfortunately, this is likely to be an ongoing problem, although it can sometimes be minimized by proper feeding. Barbara feeds Duncan a high-fibre premium dog food and sprinkles bran flakes over the food before serving it, which seems to have minimized the problem for Duncan. Fortunately, the problem is disgusting but not dangerous. When we discovered this problem with Duncan, our vet (and friend) Sue Stephens told us that the anal glands could be removed surgically, but she didn't recommend it. There's about a 2% chance that the surgery will render the animal incontinent. Many people apparently take that chance because they are unwilling to express the anal glands manually. We'd rather do something disgusting periodically rather than risk the health of our dogs. * * * * *
Well, again, I don't see the problem. I simply assumed that the processor layout required making that change, which I'll continue to assume until someone proves otherwise. And I'd expect that that would already have been done, given that that information is publicly available and there are plenty of people who don't like Intel. It doesn't much matter anyway, as few earlier motherboards support VRM 8.4. Once again, I ask you what would be the point of Intel making that change arbitrarily? The only possible answer I can think of is to force new motherboard sales to people who want to upgrade processors. Why would Intel want to make it hard to buy a high-margin processor in order to sell a low-margin motherboard, especially since about 99.9% of the people who would consider making such an upgrade are using non-Intel motherboards and would be almost certain to replace them with other non-Intel motherboards? * * * * *
Ah, I see. When Dr. Jim referred to "Children's Aspirin" he meant what used to be called that, but is now called "Adult Low Strength" in order to allow people to continue to give aspirin to their children while preventing being sued. I hate lawyers. * * * * *
Hmmm. Perhaps, although I was under the impression that specifications were based on the Line Out signal, and I've always considered that advertised differences in DA converters and so on were simply marketing hype. I still think the fact that I'd never cleaned the thing in five years might have something to do with it, but you may be right. Incidentally, that URL for the Philips 5-disc changer has the all-time record for the longest URL I've ever posted. I had to embed it to prevent it from totally screwing up the display of the page. * * * * *
I know a lot of people who have a Ph.D. and they seem to be as level-headed and to have as much common sense as anyone else I know. Usually more. I think this "yeah, they may be smart but they have no common sense" attitude is usually sour grapes from people who aren't bright. Just as the argument that the SATs (or GMATs or MCATs) are meaningless or biased is always made by people who (probably coincidentally) didn't do well on them. Bright people are usually bright in all aspects of living, excepting the so-called idiot savant, and I've never known one of those.
|
|
|
Sunday, 5 December 1999 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Important Notice pair Networks is moving to a new data center tomorrow. This site, along with others hosted at pair Networks (including www.jerrypournelle.com, www.wakeolda.com, www.robichaux.net, www.tomshardware.com, and many others) will be down during the move. pair Networks says that the move will commence around 7:00 a.m. EST and should be 50% completed by noon, at which time they will cut over network connectivity to the new site. They are moving servers in small groups, which means that some servers may be inaccessible for as little as an hour, while others (the first and last servers moved) may be inaccessible for as much as five hours. An average server will be down for 1/4 of the total move time, or about 2.5 hours. That's assuming that everything goes as planned, which it may of course not. This also means that mail will be down. If you send mail to any ttgnet.com address while the server is down, you will likely receive a "transient delivery failure" warning message. Your SMTP server will continue attempting to deliver the message, so there is no need to resend it. The message will be delivered once the ttgnet.com server is again accessible. If you urgently need to send email to us during the outage, please address it to thompsrb@bellsouth.net. Please note that this address should be used only for sending urgent mail while the server is down. We apologize for the inconvenience. When I walked into the hall bathroom last night and flipped on the light, here's what I saw:
Malcolm discovers toilet paper * * * * *
You have my sympathy. BellSouth says that my service is "unlimited" and I take them at their word. I'm normally connected for at least 10 hours straight. They have an autodisconnect after 12 hours, but even that's not really a problem. As soon as they drop the connection, WinGate re-dials it. The only time I worry about their 12 hour disconnect timer is if I have a long download to do. In that case, I simply drop the connection manually and then re-dial it, which resets the timer to 12 hours. * * * * *
I doubt the change will have any effect at all on the VIA/Cyrix Socket 370 CPUs. Every new Socket 370 board I know of has those two pins jumpered (which allows using either PPGA Celeron CPUs or FC/PGA Pentium III CPUs in the same motherboard). Why would any motherboard vendor do otherwise? So the VIA/Cyrix CPUs should work fine, assuming proper BIOS support. As far as PNGs, my problem is that IE won't touch them at all unless they're referenced by an in-line image tag in an HTML page. When I click on one of my thumbnails, IE displays the download dialog, so the full-size PNG never makes it to the cache. * * * * *
Thanks. Like you, I believe that taking Vitamin C and other antioxidants and free radical inhibitors is a good idea. I've asked Barbara to get me some 1 g Vitamin C tablets, some garlic, and the other stuff you mention. * * * * *
Standardized tests do indeed correlate strongly with future performance. When I started the MBA program at Wake Forest University's Babcock School in 1983, they did not require the GMAT or any other standardized test for admission. They were in the process of getting the program accredited, and one of the requirements of the accrediting agency was that students be required to take the GMAT. So all of us ended up having to take the GMAT between the first and second years. The faculty tried everything they could think of to prove that GMAT scores were not predictive of success in the program. What they found was that GMAT scores correlated strongly with future success in the program, whereas nothing else they attempted to correlate came even close. Others have found that the same is true for the MCATs in medical school, the LSATs in law school, and so on. People who do well on the MCATs do well in medical school and subsequently do well practicing medicine. People who have lower MCAT scores but are admitted to med school on a quota typically graduate near the bottom of their class and become mediocre physicians. Obviously, there are exceptions, but the correlation is very strong. As far as Ph.D.'s, I think the question is "in what and from where?" Like other things, the Ph.D. has been diluted. But if I encounter someone with a Ph.D. in Chemistry from CalTech or MIT, I'm prepared to believe that the guy knows his stuff. If his Ph.D. is from Podunk U, I won't assume the same. * * * * *
[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] |
