Monday, 7 July 2003
10:38 - The monsoons have returned. We had about 5" (12.5 cm) of rain yesterday. The northwestern part of Winston-Salem and Forsyth County got the worst of it, about 2" (5 cm) more than the surrounding area. Even so, that's about a month's worth of rain in a day, which unfortunately is getting to be commonplace around here. Even the ducks are starting to complain.
I'm thinking about planting a stand of teak, mahogany, and other rare rain-forest woods in the back yard. I see that I can order seedlings relatively cheaply, although the catalog warns that these trees require 60 to 100 inches of annual rainfall. That shouldn't be any problem around here.
My "tape drive" problem was indeed caused by AVG AntiVirus being resident while I was backing up. I killed AVG yesterday, and ran a full backup and verify with no problems. I then ran a second full backup to make up for the one I missed last weekend, again with no problems, this to the same tape that had blown up last weekend. I then reloaded AVG AntiVirus and tried running a backup to the tape that had generated errors two weekends ago. It again blew up with a read error during the backup. So I killed AVG and re-ran the backup and compare to the same tape. It completed without problems.
Obviously, AVG is the problem. That's not much of a problem, though. It's a good product. In fact, I prefer it to Norton AntiVirus, and AVG has a version that's free for personal use. I just need to remember to unload AVG before I do a tape backup.
The recent rains and thunderstorms have caused frequent power outages, a few of long duration, but most merely flickers or a few seconds long. That's made me think about something I don't often think about: UPSes. We run our production systems here on three UPSes: my main office desktop system and a couple of Linux boxes are on a Smart Power Systems Sine-Smart 2 kVA unit. My Mandrake Linux border router/gateway and my NT4 Server file server are on a Tripp-Lite 675 VA. Barbara's main office desktop system is on an APC Back-UPS Pro 650.
I don't think much about run-time or battery condition, and that's turned out to be a mistake. When the UPSes were new, I had all kinds of run-time. I could continue working during short power failures, and only think about shutting down after the power had been off for several minutes, which almost never happened. Later, as the batteries aged, they still had plenty of capacity to give me time to walk around and shut down systems gracefully. Until recently, they had at least enough run time to cover short power outages without crashing the systems. No more, except the SPS 2,000 VA system.
A week or two ago, we had a short power blip and Barbara called to me that her system had rebooted. This was really short blip, maybe a tenth or a quarter of a second. That seemed odd, but I put down on my mental to-do list to order her a replacement battery as soon as I had time to do so. Then, several days later, we had another short outage, this one lasting several seconds. I happened to be working on Barbara's system at the time, and it held up with no problem. Very odd.
Yesterday, we had half a dozen very short outages, a fraction of a second each. Every time it happened, my Mandrake Linux router crashed, as did the NT4 Server box that is also connected to the Tripp-Lite 675. The NT4 box at least reboots normally when power returns. After half a minute or so, it's back up and again serving files. The Linux box, on the other hand, doesn't reboot when power returns. Even the slightest power blip takes it down, and it stays down until I come back into my office to push the power switch.
I suppose I could order in some replacement batteries, but I think instead I may just talk to APC, Tripp, SPS, and so on, and get them to send me some of their current products. The ones I'm using now are three or four years old, and I suppose I should look at some current products with an eye to testing them and updating my recommendations.
Tuesday, 8 July 2003
11:10 - Exciting start to the morning. I usually wake up when Barbara gets up or while she's in the shower, but this morning I didn't. She came back to the bedroom to wake me and ask for help. Malcolm had gotten sick during the night, all over the love seat in the den. Yuck. So we unzipped the covers from the cushions, and I put them in the washer before I left to visit my mother. When I returned, the city crew was out front with a jackhammer breaking up a piece of curbing that they'd installed when we had our new main sewer line put in a couple of years ago. That portion of the curbing had sunk 4" (10 cm) or so, along with the gutter, making it a hazard for people walking along the street. Barbara has been complaining for at least a year now, and they finally sent someone out to do something about it.
I had to park out on the street in front of a neighbor's house. When I got inside, the dogs were going nuts. Not that I blamed them. The jackhammer sounded like a heavy machinegun. I could feel the vibrations sitting in my office. Fortunately, they finished up before I'd been home long, put up an orange cone, and took off. The guy told me that the concrete crew would be along sometime to pour a new curb and gutter. My bet is that the orange cone lasts at most a day before the neighborhood punks carry it off and hang it on a street sign or something. Too bad the city crew wasn't allowed to booby-trap it.
I'm still working hard on the HardwareGuys.com web site. The board I'm recommending for the Performance Intel System is the Intel D865PERL. I just noticed that Tom's Hardware has a new article up, comparing 24 865/875P motherboards. He has really nasty things to say about the Intel D865PERL and D875PBZ based on performance.
As usual, Tom misses the point. Although he does give a nod in passing to the stability of the Intel motherboards, he damns them for poor performance. Let's take a look at an example I chose at random. In the SiSoft SANDRA 2003 CPU bench test, the fastest board (a Gigabyte 865 board) has Dhrystone/Whetstone scores of 6219 and 10043. The Intel D865PERL is the second slowest board in that chart, and has scores of 5797 and 9504. The Intel scores are 6.8% and 5.4% slower, respectively.
Huge difference, right? Except that that difference would be impossible to detect when actually using the systems, probably even if they were set up side-by-side. A basic principle of benchmarking is that an imperceptible difference is no difference at all. Giving such heavy weight to imperceptible speed differences is like evaluating parachutes based solely on which one gets you to the ground quickest. After all, the parachute that fails to open gets you there faster than any of the others.
As is so often the case, Tom does his readers a disservice by focusing attention on meaningless benchmarks. That's probably because benchmarks, as useless as they are, look impressive and are easy to run. And it's easy to set up the Intel board as a straw man, because most of the other boards Tom tested are being overclocked. Not by Tom, but by the board manufacturer. Those manufacturers are pushing the 865 chipset to use memory timings that Intel says the 865 will not support reliably. In effect, the makers have tried to make 875P boards using less expensive 865 chipsets, thereby risking users' data in return for a meaningless small increase in performance. Intel has issued an official warning against this practice, but the Taiwanese motherboard makers persist in doing it. And Tom Pabst, in my opinion, should know better than to let them get away with it.
Writing about the Intel D865PERL, Tom damns it as the slowest motherboard in the group, and then says "There are, at any rate, no doubts when it comes to stability." Well, duh. Could it be that Intel has produced the fastest board possible that also meets their standards for stability?
There's a reason why Intel sells motherboards by the millions to OEM system makers. Those makers realize that minor speed differences are meaningless, but stability is a huge issue. A board that is 5% faster means nothing to their users. But a board that is even 5% less stable (and that's nowhere near the actual stability differences) costs those OEMs real money in terms of support calls and unhappy customers. Intel motherboards have very high build quality, and are as rock solid as any motherboard can be. That's why OEMs use them. And that's why you should, too.
Wednesday, 9 July 2003
9:01 - Here's one that made it through SpamAssassin. It had a score of only 0.2, and it wouldn't have had a score that high except for the "gappy subject line". Looks like SpamAssassin needs to refine its tests to keep up with the spammers.
But this does bring up an interesting question. It's well known that Canadian ethical drug prices are much, much lower than those in the US. Basically, the government of Canada negotiates with the US drug companies for much better prices than US consumers pay. So why should US consumers pay US prices when Canada is right next door? I have no idea of the legalities involved, but it seems to me that if my wife has a prescription for, say, Claritin, she should be able to fax or scan & email that prescription to a Canadian pharmacist and have the drugs shipped here. Shipping costs should be fairly minor, especially if she gets her doctor to prescribe her a three- or six-month supply at one time.
Drug prices here in the US are a real rip-off, and in particular because typical insurance plans have very high deductibles. For example, one of the allergy drugs Barbara takes has a $40 deductible for a month's supply. She can get the same drug over-the-counter, if I recall correctly, but then she has to pay for it all herself, and the cost is--you guessed it--$40 a month.
So, if Barbara wanted to order in her drugs from Canada, how would she go about it? I have no idea. Suggestions appreciated. Please post them over on the messageboard.
10:24 - I was reading Michael Connelly's Lost Light yesterday, and I think I've stumbled across a very subtle clue. It came when I read this paragraph, describing a female FBI agent who had disappeared.
That immediately struck me as very odd. Off the top of my head, I thought that a man in superb condition, such as an active-duty SEAL, might have 6% body fat. A woman in equally superb physical condition might have 12% body fat. As it happened we were visiting our friends Paul and Mary last night. Mary runs marathons regularly. She stays in shape by running fifty miles or more a week--the equivalent of two marathons--and is literally nearly Olympic-class for her age group. I asked Mary what percentage body fat a female marathon runner would have. She told me something in the 10% to 15% range. When I asked about the possibility of a woman with 4% body fat, Mary replied, "Only if she's dead."
So, assuming that the 4% figure is correct, the rest must be mis-direction. Rather than a female FBI agent in superb condition, this character must actually be a male, anorexic, transvestite zombie. I haven't finished the book yet, but I can't wait to see how Connelly ties it all together.
Our washing machine died yesterday. I was doing a small load of mom's clothes, and when I went down to put them in the dryer I found the washer was still full of soapy water. I jiggled things around a bit, hoping that it had just gotten jammed (which it's done before), but no joy. I pulled the clothes out and put them in a bucket, hoping I'd find something tangled in the agitator or something. Nope. I tried resetting the dial to the part where it empties and spins, but it just sat there buzzing at me.
My first thought was to fix it, but it's probably not worth it. It's definitely not worth a service call, because it's 23 years old. I hate it when our appliances die premature deaths, but it looks like the best alternative is to head for Sears and buy a new one. I see that they have a sale on some models right now. The huge capacity ones go for anything from $270 to upwards of $1,000. I think we'll end up with a Kenmore that's a step or two above the basic model. I hope they can get one delivered in the next couple of days.
13:45 - I've been getting some interesting private mail about buying drugs from Canada. Here's an example, from someone who asked that I keep him anonymous.
10:19 - If you have any interest in desktop Linux, go read this document. It's long, but it's well worth the time to read it. I think they nailed it. Of course, the views expressed in this document are remarkably similar to my own, which I'm sure prejudices me in its favor. Oh, I might take issue with some minor points and some estimates of timeframes, but overall I think they have it right.
One paragraph did strike me as funny, though:
I think they meant "leery/leeriness" or "wary/wariness" rather than "weary/weariness". There is a word to describe the combination of two real words into a new hybrid word that conveys the essence of both words, but I forget that term. This isn't that, precisely, because the two words they've combined are synonyms and because the resulting word is a real word.
I don't know why I can never remember that term. I have a similar problem with the term that is almost but not quite a synonym for acronym. Although it's often misused for any set of initials, an acronym is properly a set of initials that can be pronounced normally as a word. For example, SMART (Self-Monitoring And Repair Technology) is an acronym, while AGP is not. There is a term for those not-quite-acronyms, but I can never remember what it is.
Barbara and I headed down to Sears yesterday afternoon to buy a new washing machine. Washing machine makers have apparently standardized industry-wide on the following terms to describe capacities:
There's something deranged about perverting the language to the degree that "extra-large" means the smallest available.
At any rate, we ended up with a Super Capacity Plus Kenmore model that's five models up from the bottom. There were 34 Kenmore models available, ranging in price from $260 to $1,000. The one we ended up with was $400, which included delivery. Sears charges $50 to deliver, and waives that fee for any model that sells for $399 or more, so I'm not sure why anyone would buy the $350 model. But the important thing is that they deliver it tomorrow, so I'll be able to do laundry again.
As usual, Barbara and I descended on Sears like a SWAT team. Others may enjoy shopping and comparing, but we walked into the store, headed for the major appliances section, told them what we wanted, paid, and got out. It was 18 minutes from the time we left our truck in the parking lot until we returned. I think I beat my record from last year when we bought a dryer by a minute or two. I absolutely despise going to the mall.
13:07 - Regarding Intel, good news and bad news.
Bad news first. Contradicting earlier reports, this AnandTech article says current 875P and 865-series Socket 478 motherboards will not be compatible with the Prescott-core processor, due to be released later this year (and probably to be called the Pentium 5). The processor itself will initially ship in Socket 478, but apparently the voltage regulator modules on current motherboards won't support the Prescott-core processors. I haven't even bothered to ask my Intel contacts whether this is true. Intel's policy is to refuse comment on such things, and I've learned that there's no point even trying.
But there is some good news. This ExtremeTech article reports that the new version of the Intel Application Accelerator software (download here) adds support for RAID 1 for those motherboards that use the 875P or an 865-series chipset with the ICH5R southbridge. The previous version of the software supported only RAID 0 (striping), which provides faster performance but less data security than a single drive.
RAID 1 (mirroring) replicates data to two drives. If one of those drives fails, all data can be recovered from the other drive. RAID 1 is typically about as fast as a single drive or a bit slower for writes, but nearly as fast as RAID 0 for reads. Since most people read a lot more than they write, a RAID 1 array offers much of the performance of RAID 0, but your data is much better protected. With large, fast ATA drives selling for very little money, a lot of people may find it worthwhile to install RAID 1 even on personal systems. (Note that RAID 1 protects you only against a drive failure; it is not a substitute for backing up).
16:02 - Thanks to Alan, one of my subscribers, who cleared up the question that had been bugging me. The word I was looking for was "initialism".
9:18 - This is the funniest web page I have ever read in my life. It's called thingsmygirlfriendandihavearguedabout.com, and it's a series of snapshots from the life of Mil, a British guy, and Margret, his German girlfriend. It's a mistake to read it while drinking your favorite beverage. Trust me on this. I sprayed my Coke out through my nose, which is even more unpleasant than it sounds. The page is very long, but it's well worth the time to read. Actually, he could have had a much shorter web page if only he'd named the domain thingsmygirlfriendandihavenotarguedabout.com.
Disclaimer: If you are male, as most of my readers are, and particularly if you are married and have been for some time, you will find yourself nodding and thinking, "He got that right...". This guy's girlfriend, whom he's been with for 14 years and with whom he's had two children, is even less prone to using Earth Logic than most women. Two small examples, chosen more or less at random:
If you are a woman, you will probably hate the page, at least at first. Barbara's first reaction was that the guy was ridiculing his girlfriend, but that's not the case at all. It's more like a long, loving, prose version of Billy Joel's She's Always a Woman. But much, much funnier. Dave Barry had better watch out. This guy is funnier than he is.
In checking the Internet, I found that this guy is a phenomenon in England. His web page turned into a book deal, and eventually into a job as a newspaper columnist. The book is fiction, but is clearly based on Mil and Margret. I may pick up a copy.
10:39 - I have more women readers than I thought, apparently. A couple didn't like the site, but most were good humored about it and thought the site was funny. More than a couple took exception to my crack about "Earth Logic". What can I say? It's something every man knows to be true, but few are brave (or foolish) enough to speak aloud about it.
Here is an example of male versus female logic. Barbara has a pinched nerve on her right side, which was causing her a great deal of pain until she went to the doctor and got some muscle relaxants and other stuff. Exerting force with her right arm can aggravate the problem. So, the other night, we went over to visit my mother, bringing a take-out dinner along. As we were getting in Barbara's Trooper to leave, I reached down to release the parking brake, which is between the seats.
A slight digression: Barbara was driving, which she nearly always does when we're together. Probably something to do with that night at a mystery convention a couple of years ago when I drove because Barbara had had a couple glasses of wine. On the way back to the hotel, we came to a major intersection, which was under construction and very poorly lit. I am on a ten-lane road, counting turn lanes, intersecting another ten-lane road. I was trying to turn left from the leftmost left turn lane of the three turn lanes present into the intersecting ten-lane road. Okay, I admit it. I turned into oncoming traffic, of which there wasn't much, instead of into the proper lane. There was one of those raised concrete lane dividers between the wrong-way lanes I was in and the ones I wanted to be in. Being a man, I recognized my mistake immediately and set out to correct it. I was driving a 4X4, so I simply drove over the lane divider, at speed. I didn't hit anything. There was never any danger that I was going to hit anything. It was an honest mistake. So sue me.
At any rate, back to the story. Barbara turns on the ignition and reaches down to pull up on the parking brake to release it. I say, "Let me do that. You might hurt yourself." Barbara says, "You can't do it for me at work tomorrow, so what's the point?" And here we have the difference between male logic and female logic encapsulated. The man thinks, "At least I can do that for her this time, even if I can't always do it for her. Each time she does it she risks aggravating her injury, so the more often I can do it for her the better." The woman thinks, "He can't do it for me every time, so there's no point to him doing it for me this time."
I confess that I thought this was just an aberration in Barbara's thinking, but the other night when we were visiting our friends Paul and Mary, the subject came up. Paul, being a smart guy (a Ph.D. organic chemistry professor), kept his mouth shut. Mary, a smart woman and also a Ph.D. organic chemist, instantly understood Barbara's rationale. I still don't understand it, and I think it's safe to say that no man alive does either. This is what I mean by Earth Logic. Men are from Earth. Women are from somewhere else. Where, I'm not entirely sure. I'm convinced that we are actually completely separate species that just happen to be able to interbreed.
Note that I am not criticizing or belittling women here. As long ago as my college years, I'd understood that smart women come to correct conclusions and decisions by a completely different process than men use. Men use slow, plodding logic. Step A to Step B to Step C, and so on, eventually ending up at Result Z. Women leap straight to result Z with nothing apparently in between. I don't understand it. I've never understood it. But it is the way women's minds work. I suppose that's what's meant by intuition. An unconscious chained association of facts that occurs at such a high speed that women aren't even aware of the intermediate steps. And if women aren't aware of them, men have no prayer of ever understanding.
Perhaps we men actually are Neanderthals. I mean, think about it. Neanderthals are known for being robust, plodding, and having poor (or nonexistent) communication skills. That means women are the true Homo sapiens. H. sapiens is known for being gracile, fast, and excelling in communications skills.
Some interesting stuff going on on Pournelle's back-channel mailing list over the weekend. One of his servers, Creon, died with an awful sound (with apologies to Deep Purple). When he opened the box, he found that the heatsink/fan retainer bracket had broken and the HSF unit had literally fallen off the processor. Reasonably enough, Jerry was getting ready to go out and buy a replacement processor, on the assumption that the original processor had burnt itself to a crisp. I told him the processor was almost certainly fine, because Intel Pentium III processors have good thermal protection. Jerry ended up standing there holding the heatsink/fan down on top of the processor while he fired up the system. It came up fine and ran normally. Now all Jerry needs is either a new HSF unit or some way to clamp the existing one to the processor with the proper amount of force.
During the exchange of emails, Jerry mentioned covering this incident in his column. I sent the following reply, which I think is worth re-posting here, if only to remind people of thermal issues.
You can also mention the following in the column with regard to overheating:
Copyright © 1998, 1999, 2000, 2001, 2002, 2003, 2004 by Robert Bruce Thompson. All Rights Reserved.