Daynotes Journal Week of 12 May 2003 Latest Update : Sunday, 18 May 2003 09:23 -0400

Monday, 12 May 2003

[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week]
[Daynotes Journal Forums] [HardwareGuys.com Forums] [TechnoMayhem.com Forums]

10:26 - I have most of the components for Barbara's new computer on the kitchen table. Here's what I'm using:

You can duplicate this system for not much more than $800, which to me is simply incredible. Make no mistake. This is a very high performance desktop system. These are all top-notch components. Sure, I could have used a bit faster processor, but the performance boost would be indistinguishable except running benchmarks, and the price would have been a lot higher. Similarly, for another $125 or so I could have substituted a Plextor PX-504A DVD+R/RW drive for the Plextor combination DVD-ROM/CD-RW drive. There wasn't any point to that, because 20X CD writing is plenty fast for Barbara's occasional use, and the PlexCombo reads DVD-ROM discs. A DVD writer would make sense as an inexpensive backup device, but Barbara's system is backed up across the network.

You could sit this system down beside the fastest single-processor desktop systems available, and it wouldn't embarrass itself. I won't give this system to Barbara immediately, because I want to run some benchmarks on it, although I'm pretty sure already how they'll come out. I used the 2.8 GHz Pentium 4 because I've already tested it in an Intel D850EMV2 motherboard with 512 MB and 1 GB of PC1066 RDRAM as well as in an Intel D875PBZ motherboard with 512 MB and 1 GB of PC2700 and PC3200 DDR-SDRAM. I consider those two systems to have essentially identical performance, although one or the other wins one or another benchmark by a few percent. I expect the Intel D8-<I-can't-talk-about-this-yet> motherboard to be a small step behind the others. We'll see what happens.

In fact, this configuration would also be a good basis for an inexpensive system. I could simply substitute a sub-$100 Celeron processor, halve the memory, use a lower capacity Barracuda ATA drive, and substitute an inexpensive DVD-ROM drive or CD burner for the PlexCombo drive. You'd end up with a reasonably fast system for around $500, not counting monitor, keyboard, mouse, and software. It's no wonder that system makers are going nuts trying and failing to sell more expensive systems, when even basic systems are so incredibly fast compared to what was available not all that long ago.

I do a spam-check periodically, just to see what percentage of my incoming messages are spam, and how many of them are being caught by SpamAssassin. This morning, I had 271 new messages from overnight, of which 97 were spam, or 35.8%. Of those 97 spams, SpamAssassin caught 92, or 94.8%.

On an average day, I get between 200 and 300 spams. With a 95% catch ratio, between 10 and 15 of those end up in my inbox rather than in my trash folder. That doesn't really bother me, because I can delete them with a single click each. What does bother me is that I've been forced by the volume of spam to assume that SpamAssassin has zero false positives. On the heaviest day I've counted, I got nearly 600 spams. I don't have the time or the patience to wade through that anything near that number, so nowadays I simply empty my trash without even looking at it. That means that I sometimes delete real messages without realizing it. That angers me, but there's not much I can do about it. If you send me email and don't get a reply, chances are good that I never saw the message. I hate that.

And here's an interesting comment from Dr. Mark Huth on what I'd written about Linux needing to grab serious market share quickly.

-------- Original Message --------
Subject: re your linux desktop
Date: Fri, 9 May 2003 11:35:45 -0700
From: Mark Huth
To: Robert Bruce Thompson

Bob,

Problem for linux is even bigger than the desktop in my view.

We are about to purchase 4 million dollars worth of cathlab equipment, image storage devices, and databases. I've asked each vendor from whom we are making our purchases (GE, Siemens, Heartlab, and Witt) if they can provide or are planning to provide a linux solution. Not one has any plans to EVER support linux in their medical applications. EVER. Now 4 million isn't a huge amount of money in the real world...although paying for it makes all of us very anxious, but if you say that every hospital/cardiologist office in the US does this every 5-7 years, it starts to add up. I keep asking myself if these huge corporations are populated with ignorant boobs who don't understand the benefits of open source, is it just momentum, or is the Microsoft development environment such that development is just easier for the vendors.

It will be time to replace my electronic medical record system next year. We are currently using an AIX based system, but need to migrate to a more modern system (our current system uses a cobol based database). I set forth to find a modern system and would love to have it be open source based/with linux as a backend. I've looked at all of the systems listed in sourceforge for medicine and have yet to find one that actually works, is compliant with all the reporting agencies, and is supported. I have looked at a dozen systems running in the Microsoft environment and all of them work.. I actually discussed helping one opensource group get funding to do a project with the ACC, they couldn't put together a cogent presentation and we went over to the dark side to do our project.

Linux on the desktop would help, but linux in the tools we use would go a long way to getting it on my desktop. I'm not sure how linux cracks the medical market....the vendors don't seem to see the merits. Would we use it if it were available? Sure, but we aren't in the business of supporting development of applications by part timers. Lets have someone do what all these other vendors who have made inroads in the medical market have done. Get funding, form a corporation, develop a product, and sell it to me.

I recognize that this kind of comment is regarded as a sign of ignorance among linux users. Bring me products.

11:53 - I seem to have lost 5.8 GHz worth of Pentium 4 processors. Never mind. I just found 3.0 GHz worth, sitting in an Intel D875PBZ motherboard. But I know I have a Pentium 4/2.8G here because I've used it before. It's probably sitting in a motherboard somewhere, buried under other stuff. Or, worse, I may have put it back in the little tiny cardboard box it originally came in. Perhaps I should have Barbara design me a check-in/check-out system just like her library.

David Magda mentioned that I hadn't said much about the forthcoming total Lunar eclipse. He also included a good link. The eclipse occurs during the early morning hours of Friday, 16 May, but that's UT. Here on the east coast, totality occurs at 23:14 local time on Thursday, 15 May. Weather permitting, our astronomy club will be out in force at SciWorks to give the public a close-up view of the event.

The full eclipse is visible from the eastern half of North America, all of Central America except the western half of Mexico, and all of South America. For a detailed map of eclipse visibility, see this page. Even those areas shown in shades will be able to see some or most of the eclipse. For example, the westernmost parts of the continental US will be able to see parts of totality, although totality begins before moonrise in those areas. Similarly, western Europe will be able to parts of totality, although totality ends after moonset in those areas. For more information about viewing and photographing the eclipse, see Sky & Telescope magazine.

13:56 - I received the following warning from Roland Dobbins just as I was about to send out my own warning. I've gotten four of these things so far today. They appear to originate with a domain registered in Belize. This message does not come from Microsoft. Clicking on it will do who knows what damage to your system.

-------- Original Message --------
Subject: Windows update email scam
Date: Mon, 12 May 2003 10:44:19 -0700
From: Roland Dobbins
To: Jerry Pournelle, Robert Thompson, Tom Syroid, Brian Bilbrey, Greg Lincoln

This is bogus, and probably an attempt to trojanize Windows boxes:

-----

From: <windowsupdate@windowsupdatenow.com> Sent: Monday, May 12, 2003 4:50 AM Subject: Windows Update Notification

WINDOWS SECURITY WARNING!!

A VIRUS HAS BEEN DETECTED ON YOUR COMPUTER. IN ORDER FOR YOUR COMPUTER NOT TO CRASH YOU WILL NEED TO GO TO:

HTTP://WWW.WINDOWSUPDATENOW.COM

AND IT WILL AUTOMATICALLY UPDATE YOUR COMPUTERS SECURITY PATCHES.

SIMPLY TYPE IN WWW.WINDOWSUPDATENOW.COM INTO YOUR BROWSER. OTHERWISE YOU WILL KEEP RECEIVING THIS SECURITY ALERT EMAIL EVERY DAY.

[Top]

Tuesday, 13 May 2003

[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week]
[Daynotes Journal Forums] [HardwareGuys.com Forums] [TechnoMayhem.com Forums]

10:48 - Barbara went in to work early this morning so that she could leave early to play golf with her dad. I was awakened at 0719 by Malcolm slurping my face. I didn't even hear the garage door when Barbara left.

Barbara's new system is still sitting on the kitchen table. I still haven't found the Pentium 4/2.8G processor that I want to use in it. I'm also missing the 120 GB Seagate Barracuda ATA V P-ATA hard drive that I planned to use in it. Several readers have suggested that Malcolm is again accumulating parts for a PC for himself, and that may be the explanation. I do have four 120 GB Seagate Barracuda Serial ATA V hard drives in the work room, so I may just use one of those. I forgot to look, and I wonder if the Antec TruePower 380W power supply in the Sonata case has S-ATA power connectors. Most of the S-ATA motherboards I get have one or two S-ATA data cables with them, but few of them come with S-ATA power adapters. I can already see it coming. I'll go to the computer store and buy some S-ATA data cables and power adapter cables, planning to keep a few spares for later use. I'll store them somewhere that I can't forget them, and then when I need one I won't be able to find it, so I'll go to the computer store, buy the ones I need as well as some spares, store the spares ...

Spam-check: This morning, I had 232 new messages from overnight, of which 100 were spam, or 43.1%. Of those 100 spams, SpamAssassin caught 96, or 96.0%. Of those 100 spams, there was one false positive, which was a search report from FreeFind. I don't really count that as a false positive, because it in fact looks like spam and SpamAssassin had assigned it a spam level that was just barely above the threshold value needed to categorize it as spam.

Although I haven't done any counts during the workday, my impression is that the percentage of spam is lower then. Not because there are fewer spam messages--those guys send their garbage 24 hours a day--but because there are more "real" messages. If I had to guess, I'd say my percentage of spam overall is probably something like 25%. That's ridiculously high, but not nearly as high as some organizations claim. Most of them, of course, have a financial interest in exaggerating the amount of spam. The flipside to that, though, is that Greg Lincoln and Brian Bilbrey, who run rocket, the server my sites are hosted on, use numerous blackhole lists and other resources to kill spam before I even see it. I have no idea how much spam they catch that way, but if 25% of the email traffic I see is spam, it wouldn't surprise me if the real number is 50% or higher overall.

I think the only solution to spam is vigilantism. Laws cannot operate across jurisdictions, and spammers take advantage of that. The legal solution, a uniform law across all jurisdictions, is worse than the problem it would be intended to solve. So the only solution I see, short of a complete overhaul of the email system, is for volunteers to make it literally dangerous to be a spammer. Various organizations such as spamhaus.org publish lists of major spammers, including their real names and physical addresses.

It is a truism of law enforcement that it's possible to murder anyone and get away with it if you have no discernable connection to the victim. That's why serial killers are so hard to catch. If a vigilante started hunting down and killing spammers, his chance of being caught would be very small. Where would the cops even start looking? The victim pissed off a hundred million people all over the world every day. There's no point to interviewing the spammer's family, friends, enemies, competitors, and so on. Nor would the cops be too inclined to waste a lot of time investigating such a crime. Cops get spam, too. They'd rapidly categorize that murder as an AVA (asshole-versus-asshole) killing, and give it very low priority.

Does anyone have detailed information on the windowsupdatenow.com scam I mentioned yesterday? I posted a warning about it to one of the non-computer mailing lists I'm on. Last night, I got an anguished message from one of the list members, telling me she'd clicked on the link and updated her system. She wanted to know what to do now, and I had no idea what to tell her. (Taking her disk down to bare metal and re-installing everything is not a realistic option). I checked the Symantec and McAfee web sites, and neither had any mention of windowsupdatenow.com. Does anyone have any idea what this site does to a system and how to clean up an infected system?

11:33 - The end of an era. AMD today releases the Athlon XP 3200+, which is the grand finale for the Athlon product line. With the Barton core and a 400 MHz FSB, the Athlon XP 3200+ is a very fast processor, roughly as fast as an Intel Pentium 4/3.0G. But AMD is devoting all of their development resources to the Athlon 64, so the sole remaining role for the Athlon XP is to serve as a stopgap processor until AMD can begin shipping the much-delayed Athlon 64. AMD plans to continue making and selling the Athlon XP for the next two years or so, although it will gradually fall further and further behind the mainstream as Intel continues to introduce faster Pentium 4s and AMD ramps up Athlon 64 production. But during that period, the Athlon will continue to provide a low-cost, high-performance alternative. Although the Athlon XP 3200+ is currently a $400+ processor, it won't be all that long before we'll see it positioned as an entry-level processor, selling for well under $100.

We should all be thankful for AMD and the Athlon. Without the Athlon, Intel would not have been forced to improve their processors and cut their prices nearly as fast as they've done. As a credible competitor, the Athlon forced development further and faster than anyone would have believed possible just a few years ago. Nowadays, most people can be perfectly happy with a sub-$100 processor. That's true only because AMD forced Intel's hand. The competition between Intel and AMD has benefited all of us.

13:25 - And, boy, has Tom's Hardware trashed the Athlon XP 3200+. According to them, the Athlon XP 3200+ should really have been designated the 2800+. They detail some of the odd things that AMD suggested for benchmarking, including testing with obsolete game engines, and even substituting a new DLL for the one that was supplied with a benchmark program. Tom's Hardware's review is an exception. Most of the reviews on the web wax enthusiastic about the Athlon XP 3200+. But I think the Tom's Hardware review is closer to reality. AMD has been inflating their PR ratings noticeably over the last year or so. AnandTech has commented on it, as have others. (AnandTech so far has no review of the Athlon XP 3200+ posted, which is surprising. It'll be interesting to see what Anand has to say about it.)

Of course, AMD claims that their PR ratings are not intended to be compared to Pentium 4 clock speeds. Perhaps so, but that's how everyone takes them, and I'm sure AMD is aware of that. In my own testing, the AMD PR ratings are inflated relative to the Pentium 4. The fastest AMD Athlon XP I have is a 2600+ with the 333 MHz FSB. Overall, it about matches the Pentium 4/2.4, and is slower in most benchmarks than the Pentium 4/2.53. What AMD calls the Athlon XP 2600+, I'd have called the Athlon XP 2300+ or 2400+. But then I'm not trying to sell Athlons (or Pentium 4s, come to that.)

[Top]

Wednesday, 14 May 2003

[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week]
[Daynotes Journal Forums] [HardwareGuys.com Forums] [TechnoMayhem.com Forums]

11:04 - Barbara wants her new system, so yesterday she announced that she was going to help me find the Intel Pentium 4/2.8G processor. Success! The processor had cunningly concealed itself in a small box on my workroom table, misleadingly labeled "Intel Pentium 4/2.8G". Now the only problem is that I don't have a heatsink/fan unit for it. I'd forgotten that when I built my new den system using the Pentium 4/2.53G I'd used the HSF from the Pentium 4/2.8G. I have a couple of HSF units on the way in from Dynatron Corporation.

Dynatron is not particularly well known. When Tom's Hardware or AnandTech does a big CPU cooler comparo, they almost never include Dynatron units. That's a shame. Dynatron makes a full range of CPU coolers. They're inexpensive, effective, and reasonably quiet. I confess that I'd never heard of Dynatron until last year, when Damon Muzny of AMD sent me an early engineering sample of the Athlon XP 2600+ processor. He didn't have an HSF to send along with it. I thought that wouldn't be a problem, because I had a stack of HSFs on the shelf.

Alas, when I checked, I found that none of the coolers I had was rated for the Athlon 2600+. Not only that, but at the time none of the big-name HSF makers had a unit available that was rated for the 2600+. I asked Damon what to do, and he recommended a Dynatron unit. He said they used them internally, and what's good enough for AMD is good enough for me. I contacted Dynatron for some samples, and have since used them on many processors. The standard Dynatron units are more effective than the stock HSF units. Also, unlike some aftermarket HSFs, the Dynatrons don't sound like a jet taking off.

Spam-check: This morning, I had 252 new messages from overnight, of which 93 were spam, or 36.9%. Of those 93 spams, SpamAssassin caught 87, or 93.5%.

Mozilla crashed on me again last night. This time it was on the new den system. I have a nasty habit of opening many, many browser instances, and at some point Mozilla decides it's just not going to take it any more and crashes. That's annoying, because it takes down not just one instance but Mozilla itself. In the process, it loses all of the browser instances I had opened, many of which I hadn't bookmarked. Last night, I had Mozilla Mail open and clicked on the "Get Msgs" button. Mozilla started to retrieve my mail and then crashed horribly.

I exited Mozilla completely and restarted it, as the error message suggested. The only problem with doing that is that a Mozilla crash apparently screws up TCP/IP connectivity on Windows 2000. The normal symptom is that TCP/IP connectivity works, but at a snail's pace. A web page that ordinarily loads almost instantly over the cable modem may take literally 30 seconds to load. If I try to POP mail, it may take a full minute or more between clicking on the Get Msgs button and when the mail actually starts to download. Even after it's started, downloading mail takes forever. Nor is the problem limited to Mozilla components. If I fire up Opera or IE, response is still very slow. The only solution I've found is to reboot the system.

The first time this happened to me, I assumed it was a problem with the Linux gateway/router or the cable modem. I shut down the Linux box and the cable modem and restarted them. When they came back up, I checked response on my office desktop system, and everything looked normal. Convinced the problem had been with the Linux box or the cable modem, I went back into the den only to find that my den system was still running at about 5% of normal Internet performance. This has since happened to me on several different systems, all of which exhibited the same symptoms. Killing Mozilla on the taskbar doesn't work, nor does firing up Task Manager and killing the Mozilla process. The only cure I've found is rebooting the affected system.

So be warned. A Mozilla crash somehow "poisons" TCP/IP, at least on Windows 2000. I'd planned to migrate away from the Mozilla suite to the standalone components once they're a bit closer to being finalized. Perhaps I should do that now instead of waiting.

14:39 - I speculated the other day about how many spams the RBLs on rocket catch before they even make it to my mailbox. Greg Lincoln tells me that rocket receives a total of about 2700 messages a day, of which 1200 to 1500 are blackholed before we even see them. As Greg says, something must be done about spam.

Pournelle called a little while ago to talk about building another project system for his column. Like me, he has an Intel D8<we-can't-talk-about-it-yet> motherboard, and he was thinking about using it to build a kick-ass system. I told Jerry that he's built so many high-end systems lately that perhaps he should give it a rest and build an inexpensive system instead. As much as we all enjoy reading about his experiences with D850E-based motherboards with two gigs of RDRAM or D875 boards with 3.0 GHz Pentium 4s, I thought his readers might enjoy a column where Jerry tries to put together a great system on a small budget. If he does that, we'll read about it in his June column. It'd be interesting to see how an inexpensive system compares in performance to one of his high-end ones. It'll be slower, of course. But the real question is, how much slower?

[Top]

Thursday, 15 May 2003

[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week]
[Daynotes Journal Forums] [HardwareGuys.com Forums] [TechnoMayhem.com Forums]

[Top]

Friday, 16 May 2003

[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week]
[Daynotes Journal Forums] [HardwareGuys.com Forums] [TechnoMayhem.com Forums]

[Top]

Saturday, 17 May 2003

[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week]
[Daynotes Journal Forums] [HardwareGuys.com Forums] [TechnoMayhem.com Forums]

[Top]

Sunday, 18 May 2003

[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week]
[Daynotes Journal Forums] [HardwareGuys.com Forums] [TechnoMayhem.com Forums]

[Top]

Copyright © 1998, 1999, 2000, 2001, 2002, 2003, 2004 by Robert Bruce Thompson. All Rights Reserved.