|
Week of 15 May 2000 Friday, 05 July 2002 08:23 A (mostly) daily journal of the trials, tribulations, and random observations of Robert Bruce Thompson, a writer of computer books. |
|
wpoison
Search [tips]
|
Monday, 15 May 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Intel's disastrous problems with the 820 chipset just keep getting worse and worse. My own testing with the CC820 Cape Cod motherboard confirmed other reports about very poor memory performance when using the i820 with SDRAM. It was widely reported a couple of months ago that the i820 with ECC RAM could corrupt data. Now, it appears that the i820 when used with the Memory Translator Hub (MTH) may corrupt data when used with any SDRAM. I actually read about this problem a couple of weeks ago on the Intel site, and wondered why it hadn't hit the news yet. I'd flushed my IE cache, and happened to hit the main Intel web page, where I found a link to a page entitled Memory Translator Hub, with the description "Learn about the Intel® Memory Translator Hub reboot issue." When I read that page and the pages linked to from it, I was horrified. It appeared that Intel was saying that any i820-based board that used the MTH was unreliable and would need to be replaced. I mailed one of my contacts at Intel to ask for clarification, but haven't heard back from him yet. That's unusual for Intel. Their people usually respond quickly. This issue has since hit the news, big time. It appears that Intel is going to have to recall all Intel i820 motherboards boards with the MTH, and replace them with the similar VC820 with RDRAM. That much is clear. What's not clear is exactly how they'll handle recalling and replacing motherboards made by other manufacturers, and just exactly how much and what kind of RDRAM they'll supply. Tom's Hardware posted this article Friday, and The Register posted this article this morning. Apparently, some people are rushing out and buying CC820 motherboards in the expectation that they'll be replaced with VC820 motherboards with a bunch of hideously expensive RDRAM included. The problem, of course, is that (a) there aren't that many VC820 motherboards available (estimates are that as many as 1,000,000 total motherboards are affected), and (b) there is not enough RDRAM on the planet to populate that many motherboards. In addition, there is the question of what speed RDRAM would be supplied. PC800 RDRAM is fast, but it's also hideously expensive and in very short supply. PC700 RDRAM is also extremely expensive, but less so than PC800. There's more of it available, but not enough to go around. Then there's PC600 RDRAM, which is also expensive, albeit less so than PC800 or PC700 RDRAM. There's not enough of it to go around, either. The real problem with PC700 and PC600 RDRAM, though, is that no one in his right mind would want it. PC700 and PC600 RDRAM in a VC820 is, according to Tom's Hardware, slower than SDRAM in an CC820. And I believe him. There are a lot of implications beyond the obvious to this problem. Some of the Taiwanese motherboard makers like ASUS and GigaByte are going to take a major hit to their bottom lines. Nor is it clear how exactly they'll replace motherboards they made with the i820 and MTH. They don't have tens of thousands of i820 RDRAM motherboards lying around, nor do they have warehouses full of RDRAM RIMMs. Perhaps they could ramp up production of VIA Apollo Pro133A motherboards, but it's unclear whether VIA can supply that many chipsets so quickly. Nor is it likely that Intel will want to subsidize replacement of i820 boards with boards using VIA chipsets. Then there's the issue of how much RDRAM Intel will supply, which I haven't seen addressed anywhere. If they replace the motherboard that was sold in a system, the answer is obvious. If Dell sold a system with an i820 and 128 MB of SDRAM, Intel will simply have to replace that 128 MB. But what about motherboards bought bare? If I had a CC820 motherboard and were dishonest, I'd claim that I'd installed 512 MB of SDRAM and wanted the same amount of RDRAM to replace it. How is Intel to know? Will they require that motherboards be returned with the SDRAM still installed? If so, what will they do with all that SDRAM? I think it's more likely that Intel will decide to supply each replacement motherboard with exactly one RDRAM DIMM. If I had to guess, I'd say that DIMM would be 64 MB. The obvious problem there is that 64 MB is clearly inadequate [But it turns out to be 128 MB. See below. RBT] What about an honest buyer who actually did buy a CC820 with 128 MB or 256 MB of SDRAM. Or, for that matter, one who bought a CC820 with no SDRAM and planned to move 256 MB from another system? If Intel enforces a strict limit on how much RDRAM they'll supply, those buyers are screwed. In order to get the amount of memory they'd planned to install, they'll have to pay a very high price for whatever RDRAM Intel doesn't supply. Of course, they won't really, because there simply won't be any RDRAM available for love or money once Intel starts buying every RIMM in sight to populate the replacement boards. It seems to me that there's only one fair solution here. Intel needs to buy back motherboards (and systems) no questions asked, at whatever price the buyer paid. Intel also needs to pay whatever incidental costs buyers incur, whether for shipping, time spent installing and configuring software, on-site service charges, and so on. I doubt that'll happen, because it would probably cost Intel several billion dollars to do it properly, but it's the only way Intel can make things right. Otherwise, the backlash is likely to damage Intel's reputation far more than the infamous Pentium divide bug ever did. I like Intel. They're a superb company, their products are nearly all first rate, and their business practices are as ethical as those of any large company I know of. Intel has some very hard choices facing them right now. It'll be interesting to see how they handle this MTH disaster. [A reader points out that Intel has now posted specifics about how much RDRAM they will supply (128 MB) with VC820 replacement boards for their CC820 boards, although not what type of RDRAM, nor exactly what will happen with non-Intel MTH boards. RBT 5/16]
|
|
wpoison
Search [tips]
|
Tuesday, 16 May 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] I spent all day yesterday working on the Processors chapter for the book Pournelle and I are doing. I have to come up with a less cumbersome way to describe that book. The provisional title that O'Reilly has assigned is PC Hardware: The Definitive Guide. Pournelle calls it The Chaos Manor Guide to Good Enough. For brevity, we refer to PC Hardware in a Nutshell as PCN , PCHIAN, or PC/Nutshell, so I think for convenience I'll just start calling Pournelle's and my book PC/DG. At any rate, the PCN Processors chapter is the basis for the PC/DG Processors chapter, but I'm doing a lot of expansion on it. I spent yesterday compiling detailed tabular data on specific Intel processor models. The so-called S-spec identifies each specific Intel processor. There are, for example, something like 125 individual S-spec versions of the Intel Pentium processor. Getting all of that information, checking it, and pouring it into tables is very hard work. Some of it, of course, is available on the Intel site. But not all, and even that that is is not always easily accessible. (How did you like that sentence? A double-double word.) So I use other sources as well, and sometimes sources conflict with each other. Sorting out those conflicts is very time-consuming. Then there's the fact that different sources present different subsets of the information. One, for example, may list the voltages required by each S-Spec version, but not the acceptable operating temperatures. Other sources may list temperature but not voltage. Tracking down all of the information for each S-Spec, checking it for errors, and getting it into a table is non-trivial. What's particularly frustrating is that this is the type of information that not many readers will need, particularly the information on older processors like the Pentium. But some readers will need it, and when they need it they will really need it. For now, the information is available on the web, so the easy thing to do would be to punt and just include a URL. But the problem with that is that Intel pulls down old information, typically leaving it up for however long the warranty period is on a processor after that processor was last sold. So, for example, on the Pentium, they will leave detailed information posted for three years following the last date the Pentium is officially available for sale and then delete it. A lot of information on older processors has already gone away. But at least detailed information about Intel processors is available, although one has to dig for it. Try finding detailed information about AMD processors. Not easy. So it's back to work for me. * * * * * Tom Syroid mailed me yesterday in regard to a chapter he was reviewing for PC Hardware in a Nutshell. Although most of that message was private, he did bring up something I thought might be of general interest, so I'm extracting that portion of his message and the relevant part of my reply. In his message, he asked:
Yes, you probably could, but there's little point to it. Celerons are now available in two core versions. Mendocino (like you have) is 0.25 micron. That core is good up to perhaps 550 or 600 MHz tops. The fastest version of that currently available is 533 (which is likely to be the fastest version ever made of the Mendocino core). If you install one of the Mendocino-core Celerons, you'd have to run it at 66 MHz FSB, because it won't overclock at 100 MHz FSB. For example, installing a Celeron/500 and running it at 100 MHz would give you a Celeron/750, which wouldn't boot. Conversely, running Celeron/500s at 66 MHz would probably be almost identical in speed to your existing Celeron/450 OC'd system. The 50 MHz you gain in speed would probably be offset by the loss in going from 100 MHz to 66 MHz. You for sure wouldn't be able to tell the difference sitting in front of the system, and benchmarks would probably be close to identical. The Coppermine128-core Celerons are currently available (kind of) at 533A, 566, and 600. The Coppermine core is good for maybe 733 to 800 MHz without extreme measures (like those they use to get the 1 GHz PIII -- running higher voltage than nominal 1.6V, using extreme cooling, etc.) You *might* be able to run Coppermine128-core Celerons in your system, but with the following provisos:
In short, I wouldn't bother trying to upgrade your system by adding faster Celerons. I'd wait until the 700 and 750 MHz Slot 1 Pentium IIIs drop to about $150 each and do the upgrade then. Overall, I'd guess that installing 750 MHz Coppermines would speed up your system by a factor of between 1.75X and 2X relative to your current Celeron/450 setup. You'd be clocking at about 1.67X faster than your current system, but the larger and more efficient L2 cache on the Coppermines should make up the balance. Also, MPS systems are happier with Pentium IIIs than with Celerons. * * * * *
Actually, bringing up a local mail server isn't contingent on having a fast or full-time link. There's nothing to prevent using a local mail server with a dial-up link, and in fact there are some real advantages to doing so. I haven't really checked out the options, but there are many. I'm not sure that there *is* a mail server that's appropriate for small office. Configuring a mail server is inherently pretty complex, although I'm sure there are commercial products out there that make it easier than, say, trying to bring up sendmail. I'll probably end up installing HP OpenMail on a Linux box. That's free for up to 50 users. * * * * *
Thanks. That part wasn't up when I last looked, but appeared when I did a refresh. The perils of caching pages... And this article in The Register points out why you might not want to go for the refund. Although I don't know why anyone would want a 128 MB stick of PC600 RDRAM...
|
|
wpoison
Search [tips]
|
Wednesday, 17 May 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Another full day yesterday compiling tabular information on Pentium processors. Intel made, get this, more than 250 variants of the P5, including many I'd never heard of. Tracking down accurate information on all of them is non-trivial, to say the least. And that still leaves me with the Pentium Pro, Celeron, Pentium II, and Pentium III left to do. Not to mention processors from other companies. The coverage of those will be sparse compared to Intel processors, simply because other companies don't provide the wealth of information that Intel does, even though getting to that information requires some serious excavation. As an example of the height of something or other, Barbara forwarded me a message from one of her mailing lists.
Apparently, one publisher made the horrible blunder of changing the color of the cover of one of its legal reference books from one edition to the next. None of the law students could find the new edition. I am not making this up. Barbara confirms that she and every other librarian she knows has had exactly the same experience. I know the old saying goes that law students are the ones who weren't smart enough to get into med school, but that's ridiculous. * * * * *
Arrrrghhh. Well, mine at least was unintentional. * * * * *
Thanks. I wasn't aware of any security issues with OpenMail, or at least any more than exist any time one opens a mail server to access from the Internet. As far as non-commercial, I wasn't aware that OpenMail was restricted in that sense. I don't pay much attention to that, because I don't consider TTGNET to be commercial, since it's really just Barbara's and my personal corporation. But you may be right. Maybe I'll run it on one of our "personal" computers rather than on one of the corporate ones. * * * * *
Thanks for the kind words. If it were me, I'd go for the ISA-less version of the board. But that doesn't mean you're truly legacy-free. Just because a board doesn't have any ISA slots doesn't mean that the chipset doesn't have a PCI-to-ISA bridge. If it does (as the 440BX does) you'll still be loading the drivers. But the good news is that that doesn't make much difference. Your real choice here is number and type of slots. Go for the maximum number of PCI slots and don't sweat the ISA issue. * * * * *
No, as far as I'm aware, Intel has not made any changes at all to the Celeron as far as disabling SMP support, although it has long been rumored that they would do so. I understand that they've started putting a label on retail-boxed Celerons that reads something like "Not for use in dual-processor systems", but that's about the extent of it. I suspect any problems that people are encountering running dual Coppermine128-core Celerons are due to the differing pinouts, L2 cache differences, and BIOS support (or lack thereof). I haven't tried running a dual Coppermine128 system, and I may have missed something in the technical documents, but I don't think I did. It may be that this belief arose because someone was able to run a single Coppermine128 Celeron in a dual-capable system, but was unable to get dual processors working. That doesn't necessarily mean that Intel removed SMP capability. There are any number of other things that could account for the problem. If you can point me to an Intel document that says that they've disabled SMP support in the new Celerons, I'd appreciate it.
|
|
wpoison
Search [tips]
|
Thursday, 18 May 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Barbara just left on a trip to Knoxville, for a meeting with her Southern Searchers group. She'll be back late tomorrow evening, so I get to dog sit while she's gone. Given Malcolm's current behavior, I'm not likely to get much work done. I sometimes get mail from morons, which I usually just trash after reading the first few words. But this is the first time that I've gotten mail from multiple morons at the same organization, in this case flatplanet.org. They're upset about my response to J. H. Ricketson's message back on February 8, in which I commented that it's stupid for webmasters to create sites that are non-functional for people who browse with ActiveX, Java, JavaScript, and persistent cookies disabled. They whine about my supposed security concerns about using Flash, which I didn't mention, and take me to task for using Outlook (whatever that has to do with anything) and for using Internet Explorer rather than Netscape. According to them, I don't know anything about anything, and they think it's unfortunate that I should be allowed to have a web page to publish my clearly ignorant opinions. So, for the first time, I've added an entire domain to my kill file. Lots of mail about dual Copppermine128 Celerons. Here's a selection: * * * * *
Well, obviously, Intel has not previously disabled the ability of Celerons to run in a dual configuration, since there are many people doing just that. What we were discussing was whether or not Intel had truly disabled the ability of the Coppermine128-core Celerons to run in SMP mode. The only way to disable SMP absolutely is at the core level, which I don't believe they've done. Intel could obviously discourage casual use of the Celeron in dual-CPU mode by fiddling pin connections, but unless there is an actual signal (as opposed to just a voltage) on the pin(s) in question, it's easy enough for a slocket manufacturer to spoof that pin. And I suspect that's just what will happen. * * * * *
I suspect you're right, although I haven't even tried to run dual Mendocino-core Celerons, let alone dual Coppermine128-core Celerons. But I'd be surprised if those companies advertising support for dual Coppermine128 Celerons were blowing smoke. I don't think dual Celeron support is going to be a problem. * * * * *
I seem to recall that it's pin AN15 that's been at issue since the earliest days of the dual Celeron hack. I don't recall the details, and don't have time to look them up, but I suspect that AN15 carries only a steady voltage rather than a signal. If that's the case, it's easy enough for a slocket to spoof the correct state for AN15 simply by asserting whatever voltage is necessary. And I'm out of time.
|
|
wpoison
Search [tips]
|
Friday, 19 May 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] I was hoping that with Barbara away on a business trip the dogs would let me sleep in a bit later, but it was not to be. Duncan (not Malcolm) snouted me at 0548 this morning. I spoke harshly to him and tried to go back to sleep. A couple minutes later, he snouted me again. Okay, I figured, perhaps he really does need to go. So I got up, put on my sweats and slippers and took them out front. Duncan sniffed around for a few minutes, did a 2-second squirt on a bush, and then headed for the front door, expecting his usual morning treat. Some days I like these dogs better than I do other days. In the last month or so, Malcolm has started growling at Barbara and me when we pet him. Not always, but sometimes, and we can always tell when he's in one of his moods by his body language. He's also attacking Duncan, who doesn't fight back. I don't think Malcolm is going to turn out to be an aggressive dog, though. This is the first time we've had a male puppy growing up with an intact male dog around. The start of the growling corresponds with when he started lifting his leg, so I think he's just going through puberty. Apparently, all teen-age boys are obnoxious, regardless of species. This morning, he growled particularly loudly at me, so I grabbed and squeezed his snout (a dog thing, showing dominance). He growled all the louder, so I grabbed his collar, making sure to get some fur in my grip, and gave him a good shake (a mother-dog thing, showing she is not amused), dragged him off the couch and across the den floor, and tossed him into his crate. Like any teenage boy, Malcolm needs his own room, where he can go to be by himself and be angry at the world. Think twice before you apply the Outlook Email Security Update (ILOVEYOU fixes). Among other things, that update sometimes doesn't "take", which requires you to uninstall and re-install the entire Office 2000 suite. There's a list of the downsides to applying the fixes here. And speaking of fixes, the recent revelations that Clippy, the disgusting little animated help sprite, is actually a general-purpose backdoor prompted me to do something. As usual, my first thought was to use a blunt instrument, so that's what I did. Windows 98 allows you to remove Windows Scripting Host from the Windows Setup page of Control Panel - Add/Remove Programs, but NT4 offers no such option. Two files in the the \Winnt\System32 folder, wscript.exe and cscript.exe, are used to run scripts. I don't have any desire to run any kind of script, so I deleted them. Batch files work just fine for anything I want to automate, and I've never heard of a batch-file virus. Speaking of ILOVEYOU, there's a very nasty variant loose in the wild. The Register reports that this version varies the subject line by choosing a filename from the MRU list and prepending FW: to that filename. Also, the virus apparently is designed to mutate by adding random lines of code to itself, which increases its size as it propagates, putting an even heavier load on mail systems. Only Microsoft could write a programming language that permits random lines of code to be interspersed in a working program without causing execution errors. The really annoying thing about this ongoing mess is that very few of these security holes that seem to pop up nearly every day are a result of actual bugs in the programs themselves. Most of them are instead a direct result of poorly-chosen default configuration settings, and there's no excuse for that. Simply by changing a few of the default configuration settings, Microsoft could make Outlook a reasonably secure mail client (or Internet Explorer a reasonably secure browser). That ILOVEYOU and similar viruses have been able to proliferate so readily is a direct result of Microsoft's preference for convenience over safety. They need to start distributing their software in a form that installs securely by default, requiring users who want to enable dangerous options like scripting to do so explicitly. Had Microsoft done that in the first place, no one would have ever heard of Melissa or ILOVEYOU. Probably 99% of Outlook and IE users simply use the product with the default security settings. Had those default settings been secure, Melissa and ILOVEYOU simply would not have been able to propagate. Without minimizing the responsibility of those who actually create such viruses, it must be said that Microsoft is at the very least an accomplice.
|
|
wpoison
Search [tips]
|
Saturday, 20 May 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] The good news is that I'm to be interviewed by the Wall Street Journal on Monday. Not bad. So far this year, The Times of London and the Wall Street Journal. If I keep this up, I may end up becoming a pundit. The bad news is that the interview is for an article on the history of dishwashers. Apparently, I'm a world-renowned authority on dishwasher history. As evidence of that, during a web search the author of the article found this material from my journal of nearly two years ago:
Actually, what the author liked was my "Dishwasher Darwinism" construct. Who knows? Perhaps the term will catch on and I'll become rich and famous. Probably not, though. I invented another term that is now in widespread use, but I've never gotten credit for it. That was the term "coaster" as applied to a ruined CD-R disc. I remember the exact moment I coined the term. My friend John Mikol had gotten his hands on one of the very first CD-burners, which back then cost $20,000. Even worse, the blanks were something like $25 each. He fired up the burner and promptly ruined his first blank. I said, "Jesus, John. You just made a $25 coaster." Alas, I neglected to trademark the term, and so have not benefited financially from my inventiveness. Had I only thought to trademark that term and license it for, say, $0.01 per use, I could have retired a wealthy man by now. So, not wanting to make the same mistake twice, I hereby claim a trademark on the term Dishwasher Darwinism™. Take that, phrase pirates everywhere. * * * * *
Dunno. I have two or three Win9X boxes around here, but I don't use them for anything serious, so I don't pay much attention to Win9X. Perhaps one of my readers will know the answer. * * * * *
That's certainly odd behavior, but I don't think HP is at fault. If I'd encountered that problem, I'd be more upset with Microsoft. Given that Windows 98 is really what amounts to Windows 95.1, it seems to me that Microsoft's upgrade process butchered the system configuration. There's no reason I know of that Win98 should not have simply used the printer drivers already in place for Win95. If there is some technical reason that the older drivers wouldn't work under Windows 98, the Win98 install should pop up a dialog "We see that you have an HP PhotoSmart P-1000 printer installed. Unfortunately, the current driver won't work with Windows 98 ..." and going on to, at the least, explain what the user needs to do. Better still, it would provide a link to the HP page with the new drivers. Clicking on that link would uninstall your old drivers, saving any custom configuration information you'd entered, and then download and install the drivers for you and configure them to your preferences. I agree that HP should have sent email to all registered users of that printer model describing the problem and the workaround. I can't agree, however, that HP should have sent a CD to all users. After all, only a tiny percentage of Win95 machines are ever upgraded to Win98. For that tiny percentage, posting the updated drivers on the web site seems a reasonable solution. I don't see that HP "goofed" in any respect. If any mistake was made, it was made by Microsoft. And, of course, any time one upgrades his operating system, he should expect some glitches. That's just a fact of life.
|
|
wpoison
Search [tips]
|
[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Oh, ye of little faith. Several people have sent mail expressing disbelief that (a) the Wall Street Journal is doing an article on dishwashers, and/or (b) that they would choose to interview me even if they were doing such an article. I won't name names, so as not to embarrass the guilty. But herewith the original message and my reply:
You have my permission to quote, in whole or in part, what I wrote in my Daynotes Journal for Saturday, September 5, 1998 regarding Dishwasher Darwinism. I would appreciate you attributing it to me, Robert Bruce Thompson, and (if possible) including the URL for my current weekly journal page. You may also be interested to know that I have found other rather unconventional uses for the dishwasher. One of them, which I describe in the book I am just now completing (PC Hardware in a Nutshell, O'Reilly & Associates, 2000) is for cleaning computer keyboards, mice, and similar peripherals. The dishwasher does a wonderful job of getting a grungy input device clean. I've been doing this for 20 years now, and have never had any problems (except the time that I was in too much of a hurry to let it dry completely). I have also used the dishwasher to clean particularly grungy CDs, again without problems. I'll be more than happy to speak with you at your convenience. I work at home as a freelance writer, and am generally in the office and available from about 8:00 a.m until 5:00 p.m. Eastern. This is a home office, so I'm also generally available at other times. My number is [telephone number deleted. RBT]. You hit what sounds like an answering machine (to drive off telemarketers) but is really an automated attendant. Press 3 to ring my office. Today is an exception to my normal schedule. My wife, who also works at home, is out of town on a business trip, and I have a previous commitment that I must leave for immediately. If your need is urgent, please call me after 6:00 p.m. tonight. Otherwise, I'll be around Monday. If it's any consolation, my wife and my mother (both of whom can usually tell when I'm making up something outrageous) fell for this one. They both swore I was making it up. Which I'm not. Heh, heh, heh. * * * * * CNET just trashed the Plextor 8/4/32A ATAPI CD burner. CNET's main complaint seems to be that the Plextor is slow, yielding only "19K/sec", whatever that means. The drive they liked, a Ricoh model, supposedly yielded "92K/sec". They don't document their testing methodology, though, so these numbers are meaningless. They go on to state that the Plextor required more than 20 minutes to burn a 311 MB data CD, while the Ricoh required "just over 8 minutes". I haven't seen the Ricoh, so I can't comment on the respective speeds of the two drives. However, I do have a Plextor 8/4/32A, and I can say that in my testing it has proven to be a very fast drive, coming very close to meeting its advertised 8X speed when burning CD-Rs. It takes me much less than 20 minutes to burn a full 650 MB data CD on the Plextor, so the only thing I can assume is that whoever tested the drive for CNET had it configured for 4X instead of 8X. In the User Opinions section, the Plextor has 75 responses, which are 95% positive and 5% negative. The Ricoh has only four responses, two positive and two negative. In the CNET summaries, under "The Bad" for the Plextor, they list "more expensive and slower than some competitors", but under "The Good" for the Ricoh, they say, "inexpensive". That's pretty odd, considering what they show for price ranges for each drive. The lowest price they show for the Plextor is $187 and the lowest they show for the Ricoh is $10 more at $197. So how does that make the Plextor "more expensive" and the Ricoh "inexpensive". Perhaps it's because the highest price shown for the Plextor is $369, whereas the highest for the Ricoh is $245. But the Ricoh price range is based on 22 merchants, all of whom are US-based, while the Plextor price range uses 40 merchants, including one Canadian merchant who charges $369 (Canadian). If you exclude that one Canadian merchant, the highest price quoted by the 39 US merchants is $259. Give me a break. Apparently Plextor didn't buy enough ads on CNET. And that's why I don't trust advertising-supported sources for PC information.
[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] |