|
Week of 7 February 2000 Friday, 05 July 2002 08:17 A (mostly) daily journal of the trials, tribulations, and random observations of Robert Bruce Thompson, a writer of computer books. |
|
wpoison
Search [tips] |
Monday, 7 February 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] I'm debating what machine to test this ATI All-In-Wonder 128 in. Obviously, I need a machine with an AGP slot, and as I thought about that I realized that those are surprisingly rare around here.
And so on. I can just see me going to Barbara and saying, "Believe it or not, dear, I need another computer." Yeah, right. * * * * *
Yeah, they sent me a thank-you note, too. The difference was, I then sent them a message something like, "Well, thanks for the thank-you note, but don't you think I deserve a t-shirt or coffee cup for finding such an egregious error and notifying you of it?" I'm interested in the ATI AIW-128 also. I want to be fair to it, which means I want to run it with different processors, everything from a Celeron/400 up to whatever the fastest Pentium III I can get my hands on is. As far as disconnecting service, it's my understanding that the LEC can't do that. The long-distance carrier can refuse further service, but as long as you pay your local phone bill, the local telco shouldn't be able to cut off service. I think what I'm going to do is discontinue long-distance service entirely and just dial 10-10-220 when I want to make a long-distance call. I haven't done that so far because I didn't want to make life harder for my mother, but I suspect she can deal with dialing 10-10-220. That still doesn't address the problem of 800 numbers being invisibly redirected, but that practice is so clearly fraudulent that I'm not sure why it's even permitted. * * * * *
Have you ever considered starting your own daily journal page? * * * * *
Yes, Microsoft-speak is downright Orwellian at times. I kind of understand why Microsoft goes through all the hoops. DMA is, of course, safe on the vast majority of newer systems out there and dramatically increases performance. In theory, the ATA interface and devices report whether or not they are DMA-capable. If you enable DMA checking, Windows queries the interface and devices. If both the interface and all devices on a channel report that they are DMA-capable, Windows uses DMA for that channel. The problem is that some supposedly DMA-capable ATA interfaces and some supposedly DMA capable ATA devices are in fact not, but lie and say they are. What may result is system lockups and lost data. I'd be very cautious about enabling DMA on older hardware, but any hardware two years old or less should be fine.
|
|
Search [tips] |
Tuesday, 8 February 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Dr Keyboard's Diary has a new name and a new home. You can read about life at Chateau Keyboard here. Chris Ward-Johnson, aka Dr. Keyboard, maintains a daily diary of his adventures as he speeds across Europe in his BMW Seān, eating and drinking his way through an incredible array of foods and wines with foreign sounding names. Oh, yeah. He also writes about computers when he's not too busy waxing lyric about his latest meal. If you've not read Chris's page, give it a try. I think you'll enjoy it. I certainly do. Barbara is off to play golf this morning with her father. We've been loosening the restrictions on Malcolm lately. We've taken down most of the baby gates, although we still have Malcolm blocked from the library. First editions and book-iverous puppies do not mix. My office is now freely accessible. I'd gotten very tired of having to step over a three foot high baby gate, and had nearly fallen a couple of times while doing it. In general, Malcolm is doing pretty well with his new-found freedom. I sometimes see Malcolm trotting out of my office with a floppy disk or something in his mouth, but so far he's been pretty good. Still, my heart sank when I came out of my office this morning just after Barbara had left to play golf. The hall bathroom door was standing wide open. I looked in, and sure enough the toilet paper roll was missing. Dreading what I knew I'd find, I looked in the den. Sure enough, Malcolm had shredded an entire new roll of toilet paper. Bits of it covered the sofa, the love seat, both ottomans, two of three end tables, and the entire floor. I put Malcolm in his crate while I cleaned up the mess. After taking him outside to give him a chance, I brought him back in and went back to work. Within a minute or less, I heard shredding noises. I ran out to the den to find Malcolm chewing on my latest InfoWorld. I explained to him that crimes committed while on parole deserve no sympathy at sentencing time, and tossed him back in the crate. He'll stay there until Barbara gets back from playing golf. I have lost all respect for PC Magazine. This has been an on-going process. At first, I merely disagreed with much of what they had to say, but didn't question their objectivity. Then, several years ago, I began to question their objectivity, as it seemed more and more that they were becoming simply shills for whoever was willing to buy a lot of ads. As of late, it's apparent that PC Magazine is no longer relevant. The straw that broke the camel's back was the Windows 2000 issue that arrived yesterday. In it, PC Magazine basically confirms my earlier observations that Windows 2000 Professional is slower than NT 4 Workstation. In Business Winstone under various memory and filesystem configurations, Windows NT4 Workstation wins three of four tests, and loses by small margin to W2KP in the fourth. But, according to PC Magazine, W2KP is faster despite these results. Then PC Magazine trumpets that Windows 2000 is as much as 24% faster than W2K in the Content Creation Winstone but that is on systems with IDE hard disks with NT4W using PIO mode and W2KP using DMA mode. As anyone who knows anything about NT4 knows, DMA dramatically increases performance over PIO, particularly in benchmarks as disk intensive as these. Of course W2KP is faster using DMA mode than NT4W is using PIO mode. Duh. Any idiot would know that. PC Magazine points out that they used PIO mode on NT4W and DMA mode on W2K, but they make it sound like NT4W supports only PIO mode. That is so misleading that it makes their objectivity suspect, to say the least. And it makes me wonder what their results on Business Winstone would have been had they compared DMA apples to DMA apples. I'd bet money that NT4W would have blown away W2KP completely. PC magazines in general are a dying breed, and PC Magazine is no exception. All of them are trying desperately to re-invent themselves, but the fundamental problem is that knowledgeable users no longer turn to magazines for PC information. They look to the web. PC Computing, another ZD publication, has recently completed a work-over that turns it into what amounts to a general business magazine with a PC/Internet flavor. PC Magazine seems to be getting smaller. I haven't actually counted pages, but there seems to be less content with each passing issue. More ominously, the magazine itself seems to be getting noticeably thinner, presumably because they're having trouble selling ads. That's what killed BYTE, and it may well kill PC Magazine in the near future. If so, I for one won't mourn it. * * * * *
Thanks. I visited the site but couldn't see anything. Presumably that was because I have ActiveX disabled. At any rate, IE threw up a warning to that effect. Which brings me to another rant. I keep ActiveX, Java, JavaScript, and persistent cookies disabled on my browser for the Internet zone. That's certainly not an unusual configuration for a knowledgeable web user. In fact, most sensible people do the same if they are even vaguely aware of the security holes that all of these elements present. And yet, there are many sites whose webmasters are so stupid that they have created sites that require one or more of those elements to run. That's senseless. There's no need to use any of those things to create a good web site, and the downsides to using them are quite clear. Not only are all of them horrible security holes, but many users, like me, will be unable to access their sites. And the ActiveX warning is another thing that seriously pisses me off. It's entirely gratuitous in the first place. That would be bad enough, but what's worse is that Microsoft provides no way that I can find to turn off the warning. They're essentially saying that if I choose not to use ActiveX, that's my prerogative, but I'll sure regret that decision because I'll have to manually click OK on all of their damned warning dialogs. It's so bad that you can't even back out of a site without clicking OK for the damned warning at each page on your way back out. It's often easier for me to return to my links page and go back in forward instead of trying to back out.
|
|
Search [tips] |
Wednesday, 9 February 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Not time for much here this morning. I spent the time that I'd ordinarily spend writing an update here on doing a quick tech review pass for Tom Syroid on one of the chapters for Outlook in a Nutshell. Lots of mail, and I don't even have time to read all of it right now, let alone respond to it, let alone post it. My apologies. I'll get to it when I can. Things are backing up around here, and it hasn't been helped by the widespread data storms on the Internet recently. I'm still getting messages that were sent Monday, which adds to the confusion if one doesn't realize that they aren't current mail. More later, as I have time.
|
|
Search [tips] |
Thursday, 10 February 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Got another chapter sent off to my editor yesterday, this one on Cases & Power Supplies. As usual, when I sat down to start writing it, I thought I'd be hard-pressed to turn out a dozen pages. It ended up three times that, and is pretty tight even at that size. Chris Ward-Johnson asked me for a pithy quote about Windows 2000, to be run as part of an article on Windows 2000 he's writing which will appear in The Times next week. I, to use Chris's term, "slagged" Windows 2000, so I suppose I can expect a Microsoft hit team to show up at my door sometime soon. I need to get to work on another chapter, so it's short shrift time on mail. * * * * *
Thanks. FrontPage screwed me once again. It has this nasty habit of changing cases on file names and then deleting "unused" files. Both the main Pegasus image file and the thumbnail image file for it were present on my local copy of the web, so I couldn't see anything wrong until I went and looked at the actual copy on my web server. I hate Microsoft. * * * * *
Ah, I see. Thanks. But I don't do plug-ins, either. If my browser by itself with very secure settings won't display a page, I just don't look at that page. Period. * * * * *
No attitude? Come, now. Netscape is at least as obnoxious as Microsoft in that regard. Ask anyone who's struggled to get rid of the evil little AOL icon that they didn't ask for and didn't want. And at least IE is a modern web browser. Navigator is junk. * * * * *
That's a method I'd not have thought of. I'd probably have just checked page counts on a few random issues. Of course, they may also have switched to more condensed type, thinner paper and so on, so short of a word count there's no way to be completely sure. But this does confirm what I suspected. Thanks. * * * * *
I'm surprised that the DMA entry had any effect on SCSI. DMA is simply an alternative ATA mode which is faster than PIO mode. As to arbitrarily changing all the *DMA=0 registry entries to *DMA=1, you're a braver man than I. I won't run the ATI AIW 128 in my main system simply because it is sacrosanct. I have to have one stable system that I can depend on. Also, it wouldn't do any good, because the AIW 128 (or the Matrox G400/TV for that matter) requires Windows 98 to support the TV functions, without which it's just another fast video card.
|
|
Search [tips] |
Friday, 11 February 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] FrontPage 2000 behaved very oddly this morning. As usual, I fired it up and then clicked on File - Recent Files and choose this page, 20000207.html. Ordinarily, the page loads instantly, but this morning I got an hourglass. After at least 30 seconds of staring at the hourglass, I went out to NT Explorer to make sure the server wasn't down or something. It was up and responding normally. After a total of at least a minute, I finally used Task Manager--End Task to kill FP2K. I brought it up again, and this time, it took forever to load the current web, which is normally almost instant. Fearing that perhaps something was corrupted with the web, I ran the Recalculate Hyperlinks tool. Ordinarily, despite the warning that running it may take a while, running Recalculate Hyperlinks takes 30 seconds or less. This time I watched the hourglass for literally five minutes. At that point, I minimized FP2K and went out to check some other stuff. When I maximized FP2K again, the hourglass was gone. There was no indication of either success or failure, so I closed FP2K, fired it up again, and tried to load this page. It came up normally that time, although it took perhaps five seconds to appear, which is very slow. At any rate, everything seems to be normal now. I just wish I knew what had caused that problem. Very strange. FedEx showed up yesterday with a box that contained an Intel VC820 "Vancouver" motherboard and a CC820 "Cape Cod" motherboard. These motherboards (or "desktop boards" as Intel now calls them) are near twins, except that the VC820 uses RDRAM RIMMs and the CC820 uses standard SDRAM. When Intel offered me the VC820 board for review, I told them that I didn't have any RDRAM memory and didn't want to buy any. It costs about five times as much as SDRAM, literally. They kindly included a 64 MB RDRAM RIMM with the VC820. These motherboards are in very short supply, and Intel told me they'd need them back shortly, but they didn't say exactly when. I'm also waiting for an Engineering Sample (ES) Pentium III/733. ES processors are very nice for testing because they are not multiplier-locked. That means I can run the ES Pentium III/733 at 733, 667, 600, or 533 MHz, covering the entire range of 0.18 micron 133 MHz FSB Intel processors except the 800 MHz model, which is very difficult to obtain. The Intel 820 chipset has gotten a lot of bad press, initially over the three versus two memory slot mess, and later because various hobbyist sites reported that its performance was not up to the 440BX. In particular, some sites reported that when using SDRAM via the Memory Translator Hub (MTH), memory performance was degraded by as much as 40%. I think those numbers are suspect--kind of like PC Magazine's report that W2K was "up to 24% faster" than NT4--but we'll see. My guess is that the VC820 with RDRAM will perform comparably to the 440BX with SDRAM, and the CC820 with SDRAM may be a few percent slower. Despite what the enthusiast sites would have you believe, minor performance differences are probably the least important aspect of a chipset. I don't have a 440BX board that will run at 133 MHz FSB, but fortunately the VC820 and CC820 also support 100 MHz FSB Pentium II/III CPUs, so I'll be able to do a direct comparison with a Pentium III/450 at 100 MHz FSB. Not that I'm particularly interested in the exact results. I just want to verify how badly, if at all, the CC820's use of the MTH affects SDRAM performance. If it's anything close to the 40% figure quoted elsewhere, I'll be very surprised. Despite my concerns about the objectivity of PC Magazine, I will use the WinBench benchmarks, which I believe are reliable. Here I again depart from the enthusiast sites, most of whom give greater weight to WinStone. WinBench is a so-called synthetic benchmark, which many people regard as a Bad Thing. Synthetic benchmarks have code written specifically to test various aspects of performance, but do not mimic applications. Barring manufacturers "spoofing" the benchmark (which has been known to happen), a synthetic benchmark provides reasonably reproducible results which can be compared across products. WinStone is an application-based benchmark. App-based benchmarks, particularly composite ones, are essentially useless unless you use exactly the applications that form the benchmark, and in the same mix. For example, an app-based benchmark might use WordMunger 2000, PhotoChop 5.5, ExEll 2000, and HyperCAD 17.3. If those aren't the applications you use, the benchmark is useless to you. Even if you do use those applications, the weighting given to each may not correspond with your usage patterns. If, for example, the benchmark assigns a 25% weight to each, but you spend 90% of your time in WordMunger 2000 and only 10% total in PhotoChop 5.5, ExEll 2000, and HyperCAD 17.3, the results of the benchmark are still meaningless to you because very good results from PhotoChop, ExEll, and HyperCAD, which are unjustifiably highly-weighted relative to your usage pattern, may overwhelm very poor results from WordMunger, which is what you really care about. There's mail, but I'm out of time.
|
|
Search [tips] |
Saturday, 12 February 2000 [Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] ARRRRRGGGGGHHHH.
|
|
Search [tips] |
[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] Interesting. I can make outrageous observations on politics and technology and get only a few messages in response. When I posted my FreeCell loss yesterday, the messages flooded in. About fifty so far, running the gamut from accusing me of cheating (I don't) to suggesting I cheat by hacking the Registry (I won't). I've started over, and am now up to 7-0. Interestingly, one reader sent me a link to a site that demonstrates that not every FreeCell game is winnable, contrary to what the help file for the program suggests. Actually, I seem to remember that Windows 95 in a Nutshell includes an actual FreeCell game that is not winnable. Not that that makes any difference, because that's not the one I lost. I replayed that game immediately and won it. Many messages also suggested that I have too much time on my hands, but that's not the case. My 1731 wins came over a period of a year or so--about five games per day. And, yes, I do hack the registry to transfer my current score from one machine to the next when I upgrade my system, but that's not cheating. I don't know about this left brain/right brain stuff, but I've found empirically that, when I'm struggling to phrase a sentence properly, playing a game of FreeCell sometimes lets me take the Logical Processing Unit out of gear and the proper phrasing arrives unbidden. Chris Ward-Johnson, over at Chateau Keyboard, tells me that he does the same thing with SimCity. Of course, it takes him five hours to play a game, by which time he's forgotten what it was he was writing about. M. Praeger has been sending me numerous emails detailing his experiments with DMA, including many not related to IDE hard disks. I've told him that, unfortunately, I don't even have time to fully understand what he's talking about, let alone respond meaningfully to it. I also suggested that he format that material and post it on his new web site. I don't know if he's done that yet, but the material looked interesting. Speaking of interesting stuff, J.H. Ricketson sent me email a few days ago saying that he wanted to add an IDE hard disk to an all-SCSI system. I responded with a couple of pointers, and then received this.
Congratulations. Looks like you dodged all the bullets.
[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week] |