It was bound to happen eventually. I mean, it wouldn't be my life if it weren't Hell, would it? Since getting most of my woes with Windows XP fixed, I've enjoyed a mostly functional system. Despite its flaws, Windows XP is the best operating system Microsoft has made to date. Nevertheless, there were some things that never quite worked for me. For example, I could never get one of my all-time-favorite games, Dungeon Keeper II (DK2), to run. Just over a month ago another all-time-favorite game, Clive Barker's Undying, stopped working for no apparent reason. Still, it was serviceable. I could live with the limitations, and I certainly appreciated the additional stability.
At least, I appreciated it until recently. Wednesday, October 2, 2002, was much like any other day of late. I worked at home all day. After my wife went to bed in the evening, I played a video game or two and joined her. When I got up the next morning, however, I discovered that the laws of nature and all reason had been suspended at some point during the evening while I slept, which was really quite disconcerting. You see, the very same computer that had worked just fine the previous night, the very same computer that had powered down with nary a whiff of anything unusual, the very same computer that had been mostly functional for roughly nine months simply stopped working. And as is so often the case, all Hell broke loose at once.
I hit the power switch early Thursday morning and walked away to get the coffee going. When I returned, I was quite surprised to find Windows XP running at 640 x 480 x 8 bpp resolution. My desktop the previous night had been set to 1600 x 1200 x 32 bpp, you see, and nobody had touched the computer during the night. Needless to say, I found that odd. After poking around for a moment, it became obvious why Windows XP was running at such an awful resolution, namely, my video card, a GeForce3 Ti200, was no longer in the system. Or at least, that's what Windows XP thought. It told me that a stock, VGA-compatible PCI card was in place. That made no sense to me for two reasons: (1) nothing had changed since the previous evening, and (2) my video card was plugged into the AGP slot.
Nevertheless, I wasn't worried (yet) about how things had gone wrong. I just wanted my computer to work—a common refrain around here—so I concentrated on getting the proper drivers reinstalled. That was when I discovered that my network wasn't working either. Yes, that's right: the very same network that had worked just fine the previous night, the very same network that no less than three other machines were currently using without a hitch, was no longer working either. In this case, the device manager reported that the network card was in my system, but it wasn't functioning properly. I thought that was pretty odd too at the time.
Well, one thing at a time, I figured. I dug out the driver CD that came with my network card, and I started to update the drivers. While in the process, I started noticing an unpleasant odor, and it was fortunately one that I recognized. You see, a few months ago I got up in the morning and found my system completely dead. When I opened it up, I discovered that the CPU cooling fan clip had come loose during the night as I slept. Naturally, my Athlon chip committed suicide shortly thereafter, presumably in protest of the awful working conditions. The odor that was presently assailing my nostrils was all too familiar, namely, the smell of thermal paste and silicon overheating.
I practically ripped the side panel off my case and peered inside with a flashlight. Sure enough, the CPU fan wasn't turning at all. I reached around the case as quickly as I could and hard-powered the system down immediately. Gosh, the day sure was shaping up "nicely", wasn't it? "How many more things can possibly go wrong?", I wondered stupidly and prophetically. After a few minutes of digging around, I discovered why the fan wasn't working.
One of the fan blades had somehow become broken, and it had been bent underneath the others, thus gumming up the whole fan. The motor couldn't put out the kind of power needed to turn that fan with that blade where it was, so I grabbed a pair of pliers and snapped off the offending blade. I figured I could replace the CPU fan later. What I needed at the moment was to get the system powered back up, so I could make sure the CPU was still intact, get it cooled down, and get my network up and running in order to fix the video drivers.
When I turned the system on this time, I watched the startup process carefully. The CPU fan was again turning, my motherboard passed its power-on self test (POST), the usual litany of initialization messages flashed by, and I was soon back at the Windows XP desktop (albeit at 640 x 480 x 8 bpp). This was a good sign, I figured, as I could now get my network drivers installed.
I placed my driver CD in the CD-ROM drive, and I started to update the drivers when it hit me: the system hadn't made a single sound during boot up. Normally, Windows XP makes a very pleasing noise as my logon is processed, but not this time. At this point, I opened my journal on my laptop computer and started taking notes. Clearly, my day was far from over, and I figured I'd better start logging the problems, lest I simply forget them in light of how fast and thick they were coming.
I'll never know if Windows XP was going to update its network drivers or not. I say that because as I was going through the various steps involved in selecting a driver to use for the update, a horrible grinding or buzzing noise started emanating from the innards of my computer. Fearing that something else had gone wrong with the fan, I didn't even take the time to peek inside. I simply reacted. I hard-powered down the system before I could finish the curse on my lips—now that's fast! After a quick check of the CPU fan, and some other poking around, I discovered yet another problem: the fan on my motherboard's VIA chipset was loose. I couldn't tell why it was coming loose, mostly because I couldn't tell how it was fastened, but I could tell that its being loose was allowing it to vibrate in such a manner as to produce the awful grinding/buzzing noise I had heard.
At that point, I took a quick inventory. The video wasn't working right. The sound wasn't working at all. The network wasn't working at all. A blade on my CPU fan had broken. And finally, the fan on my motherboard chipset was about to fall off. All of this had occurred within the span of a few minutes. The next few hours were simply unbelievable, as just about anything that could go wrong did. At one point, I was on the telephone with technical support, trying to figure out why my video card drivers wouldn't install, when my wireless phone, which had incidentally been sitting on the charger all night, inexplicably went dead. That was about par for the course, don't you think?
Perhaps the most frustrating event of all involved Windows XP activation. At one point, the technician suggested that I should get rid of the Advanced Configuration and Power Interface (ACPI) from my Windows XP configuration. He suspected that all the IRQ sharing that goes on when using ACPI was likely the source of my problem. This made no sense to me insofar as the system had worked fine for months, but I was, as you might imagine, quite desperate at that point.
Microsoft had told me previously that it was impossible to ditch ACPI without reinstalling Windows XP completely, and I wasn't yet prepared to do that, but the technician taught me a sneaky trick. If you go into device manager and expand the "Computer" tree, you'll probably see something that says "ACPI Uniprocessor PC". If you then select the item and examine its properties, you can update the driver and ditch ACPI completely.
Unfortunately, that didn't do anything to solve my many problems. Worse, and unmentioned by the technician, it was a one-way trip. That is, once you've used that trick to switch to a standard (i.e., non-ACPI) PC, the option to update the driver no longer appears. Oops. The net result of that shift not only failed to fix my existing problems, it also guaranteed that my computer could do nothing with its own power management. It couldn't even power itself down, which was less than optimal to say the least. Adding insult to injury, the first time I powered up the system as a standard PC, Windows XP started griping that my system needed to be activated immediately. When I told it I didn't want to reactivate, it happily refused to let me log on.
I simply couldn't get past that stupid activation wizard, so I figured my only hope was to go ahead and reactivate Windows XP. Naturally, the attempt over the Internet failed—my network still wasn't working, remember?—so my only hope was to activate by telephone. But of course, my wireless phone had gone dead. Not to worry, I figured. I could always use my mobile phone. Unfortunately, in the middle of receiving the fifth or sixth group of digits in order to activate Windows XP, I inexplicably lost my PCS connection. Ever since Sprint upgraded their service in this area, I've always gotten the very strongest possible signal at home, but it simply had to fail right in the middle of activating, didn't it? (sigh)
Fortunately, my wireless phone had recovered enough of a charge for me to get the job done before it died again. With Windows XP activated, I could finally log on. Windows XP spent roughly forever re-detecting all my hardware and configuring itself. Three reboots later, I was just about finished when the Windows XP activation wizard popped up again. It told me that my system had changed significantly and that I now had three days to reactivate. Office XP gave me the very same crap the first time I ran any of its applications for testing purposes. Were Bill Gates present in my home that day, I can promise you he would surely have met his demise.
While I'm on the subject of activation, here is my open statement to Microsoft: If I ever find a program that will let me crack Windows XP activation and/or Office XP activation, I will use it. I paid for every copy of your software that I use, and I will not subject myself to your ridiculously annoying whims if I can avoid it by any means, legal or otherwise. I recognize that it is in your interests to stop the pirates from illegitimately using your software, but punishing your honest customers is not the way to do it, thank you very much! I said as much to the poor gal on the telephone activation line, and she said she would pass my comments along up the food chain at Microsoft. I doubt seriously whether anything will change, but who knows? Maybe if enough people express their displeasure, Microsoft will get a clue and stop treating its customers like criminals.
Well, to make what is already a long story shorter than it otherwise would be, I spent the next few hours working on my system without any success. Eventually, I had to drop it and get cleaned up and head to campus to teach. By the time I had left the house, I had already made my decision: Friday was going to be the day I rebuilt my system. I'm in the middle of writing my dissertation right now, and I don't need these kinds of hassles. What I need is a working system.
As with most mornings, Friday, October 4, 2002, came too early. I arose, showered, had breakfast, and headed straight to my office. There I made a list of the components I was going to buy. After my work yesterday, I was suspicious of my power supply, my motherboard, my video card, my network card, and my sound card. I didn't know what was malfunctioning, but I knew all of those separate systems were all affected. Thus, I was determined to purchase enough components from local stores that I could narrow down the problem and return back what I didn't need. It's at times like those that I thank God for Fry's Electronics.
I actually made two trips that morning. Since I had time for roughly zero research, I walked into my local Fry's and simply asked one of their tech guys what I should buy. I walked out a few minutes later with an Antec Plus 1080AMG case, a Gigabyte GA-7VAXP motherboard, 512 MB of DDR 333 MHz. Mushkin RAM (1 strip), and a laser pointer. Ok, the laser pointer wasn't really necessary (it's for playing with my dogs), but I'd been meaning to buy one for a while, and Fry's just happened to have them on the cheap.
On the way home, I swung by Best Buy and used a coupon I had received to obtain a 20% discount on a brand new ATI Radeon 9700 Pro video card. Sure it was expensive, but I figured as long as I was being forced to do something with my system, I might as well build something that's going to last for a while. The 20% discount sure didn't hurt, and I had been wondering about that card since I first heard of it trumping anything NVIDIA had to offer (grin).
Upon arriving home, I carefully examined all of my purchases, read the various getting-started manuals, and so forth. I wanted to make very sure that I understood my new hardware. While doing so, I had a pretty good idea: I remembered that I had an extra hard drive in my office server, and it occurred to me that this could be my big chance. Ever since I installed Windows XP the last time so many months ago, I had been unsatisfied with the results. To be more specific, there were five reasons for my general unhappiness with the system.
Given that I was already going to have to spend only-God-knows how many hours fixing my hardware problems, what's another only-God-knows how many hours reinstalling software? I figured I might as well get it right from the outset. Thus, I decided to make my spare hard drive a sort of "sandbox" with which I could assemble a new system. With that goal in mind, I used my office server to wipe that spare hard drive clean and break it into three partitions: (1) an 8 GB boot partition for Windows 98, (2) an 18 MB partition for applications, and (3) a 14 MB partition for general working files. After formatting those partitions, and rendering the first partition both bootable and active, I copied the Windows 98 setup files to it, powered the server down, and removed the hard drive.
I then turned my attention to my new case. I had settled on the Antec case because it had the largest power supply and best cooling features of any case at Fry's. In retrospect, I think I got very lucky that day (more on that later). After fussing about with the Antec case for a few minutes, it became blatantly obvious to me that my previous case purchases were very poor decisions by comparison. Several of my systems use Fong-Kai FK-320 cases, which were recommended highly by Anandtech. They seemed like nice cases compared to what I'd worked with in the past, though I did have lots of trouble with the case side panels not wanting to go on after coming off.
Nevertheless, the Antec case puts them all to shame. Seriously, I can't convey how impressed I am with that case. I've been building computers for years, and that thing is simply wonderful. My drives are all easily accessible, sliding out from the front thanks to easily-attached plastic rails with metal release clips. The internal drive bays are all easily removed simply by pushing some levers. The case comes standard with no less than three cooling fans and room for more. The internal layout is simply a joy, and the system even has front-accessible USB and Firewire ports if needed. The power supply is a whopping 430W. In a word, wow. I won't be buying cases from anyone but Antec for the foreseeable future.
Yet despite how cool the case is, it doesn't begin to compare to the motherboard. In the recent past, I've used the ASUS P3V4X, the ABit KT7A-RAID, and the ASUS A7V133, the first two of which were recommended highly by Anandtech. In case you're wondering why I keep mentioning that, be patient with me; there is a good reason. I've had nothing but problems with two of those motherboards. The P3V4X was problematic at best. I never got the performance I should have seen with it, and it had lots of compatibility problems. Worse, the ABit KT7A-RAID was an utter disaster. That board gave me nothing but pain and suffering. I ditched it as fast as possible in favor of an ASUS A7V133, which I bought largely because the guy at my local Fry's said it was a good board.
In comparison to the others, the Gigabyte GA-7VAXP motherboard is a geek's wet dream. True, it requires the overclocker to fuss around with jumpers, and it uses the Award BIOS instead of the oft-preferred AMI BIOS, but it gets just about everything else right. For starters, it supports USB 2.0 and Firewire right off the bat. It has a Promise RAID controller. It has an on-board 10/100 Ethernet connection from RealTek. It has the most impressive on-board six-channel AC97 sound I've ever heard. And, of course, it has all the other features that differentiate the Gigabyte line of boards (e.g., dual BIOS chips for a wonderful safety margin). Again, wow. Putting the system together was a piece of cake, thanks to the great case, and getting it up and running was similarly simple thanks to the great motherboard. I just didn't have the kind of ridiculous compatibility problems I've had in the past.
Once I got the new motherboard in place, I installed only the new video card and migrated the newly-prepped hard drive along with my CD-RW, DVD, and floppy drives. The system started up without a hitch the very first time and booted into DOS. Less than half an hour later, I had an entirely functional installation of Windows 98 up and running. Whereas earlier that morning I had a mostly useless lump of metal for a computer, my system had been reborn thanks to some new parts. It wasn't quite like watching the Phoenix rise from the ashes, but it was pretty satisfying nevertheless.
With Windows 98 working nicely on the first partition of the first hard drive, I powered down the system, added my old, much-larger drive as the slave on the primary IDE channel, and powered the system up again. After a quick install of the latest version of Partition Magic—believe me, if you ever have to fuss with your partition table, you need this utility—I was able to twaddle the partitions on my big drive non-destructively until I had a 15 GB partition for a fresh installation of Windows XP and a 65 GB partition for other files and games.
I gritted my teeth, pulled out one of my Windows XP installation CDs, and got down to business. The setup process went pretty well. Windows XP failed to detect and/or install drivers for several of the features of my new motherboard, but I was able to take care of those lapses relatively quickly with the relevant driver CD. Within an hour or two, I had the proper video drivers in place, the on-board networking up and running, and the on-board audio was making relatively pleasant sounds. After far too many trips to the Windows Update site, my copy of Windows XP was as updated as it gets and largely functional. The only good thing I can say about the activation wizard is that I again had a full thirty days after the install. Believe me, I wasn't about to activate again until I was positive the system was good to go.
I'm both pleased and surprised to say that I had a few nice discoveries in store from me. For starters, Clive Barker's Undying ran the first time I tried it. That was a nice, though not unexpected, change. What absolutely stunned me, however, was that DK2 also ran just fine! I tried everything to make that game work with my old system, but I had exactly zero success. With my new system, it ran the first time I tried it, and it has been running just fine ever since. That was a serious thorn in my side, and the rebuilding process solved it, which counts for a lot in my book. It also makes me wonder: what was so horribly wrong with my original installation of Windows XP?
Moving on, because I had a stock dual-boot setup (i.e., the Microsoft-supported dual-boot configuration), I was able to reinstall GoBack, which is another must-have utility in my book. With GoBack in place, I sure feel a lot safer. Installing software, updating drivers, etc. no longer worry me because I know that I can always revert my drives if needed. That kind of peace of mind is worth far more than the purchase price of GoBack. Since that time, I've reinstalled my applications, development tools, and a number of games. I'm happy to report that, as of today, everything works just as it should.
There were a few problems along the way that needed solving, however, and they merit a little discussion. First, several games that worked just fine with my old GeForce3 Ti200 card didn't work well at all with the new ATI Radeon 9700 Pro. The most serious malcontents were Unreal Tournament (UT), Unreal Tournament 2003 (UT2k3), and Tribes 2 (T2). In each case, the fixes took quite a while to find, so I'll describe both the problems and the solutions in the hope that others can benefit from my efforts.
Regarding UT, the problem was that the game played in a rather jerky fashion, at any resolution. Even when playing the single-player aspect of the game, the view would jerk noticeably every second or so. It wasn't enough to render the game unplayable, but it was definitely enough to throw off my aim significantly. For those who know what it "feels" like, it was very similar to playing on a server with intermittent packet loss.
Bizarrely enough, the fix was simply to force the vertical-sync setting off in the Direct3D options panel of the ATI drivers. I say that's bizarre because that's how UT was already configured, which was apparently being ignored for some reason. All I know is that the problem went away completely once I forced vertical sync off. With the fix, I can now play UT at 1280 x 1024 x 32 bpp resolution with all the graphical goodies and still average in excess of 100 frames per second (FPS) on big maps. The ATI Radeon 9700 Pro is a beast of a card.
Regarding UT2k3, the problem was that the system wasn't performing as well as the previous incarnation of my system, which had only a GeForce3 Ti200 card. The solution was to update the VIA chipset drivers used by Windows XP with the latest 4-in-1 pack supplied on my motherboard's driver CD. As soon as I updated the chipset drivers, the difference was astonishing. My average frame rate roughly doubled in UT2k3, though I don't have any hard benchmarking data to support that claim. I still use the Quake III Arena demo for benchmarking purposes, and my frame rate with that also jumped significantly (from 96.9 to 146.4), and that's at 1280 x 1024 x 32 bpp with all the graphical goodies enabled. Not bad, eh? That surely underscores the importance of having the latest chipset drivers.
Finally, regarding T2, the problem was much more difficult to diagnose. The problem was twofold: (1) I was getting utterly crappy frame rates with the game, and (2) on some maps (e.g., Recalescence) I was seeing utterly bizarre visual artifacts as well. On Katabatic map, for example, I was getting a whopping 15 - 17 FPS compared to the 50 - 60 FPS I saw previously with my old video card. The numbers got even worse on other maps with the visual artifacts, sometimes dropping as low as 1 - 2 FPS. After hours of fussing about with the driver and game configuration, I eventually discovered that disabling the "Interior Textured Fog" in-game setting did the trick, fixing both problems. With that setting enabled, my new system plays much worse than my old system. With that setting disabled, my new system is now capable of averaging better than 60 FPS at 1280 x 1024 x 32 bpp resolution on even the biggest maps. Again, that's not bad at all. It took forever to find, but T2 now runs like a dream on the Radeon card.
The sound problems have been the more substantive, nagging, and as-of-yet unsolved issue. When I first assembled the new incarnation of my system, I left the on-board AC97 audio enabled, largely because I was interested in evaluating it. While it was pretty impressive, it just wasn't up to par with what I used to get from my Creative Labs SB Live! card. To be fair, however, I should note that I have pretty good ears. I've been playing various musical instruments since I was about six years old, I've had formal training in both sound reinforcement and sound recording techniques, and despite my complete lack of any decent stereo, I'm definitely an audiophile. In short, I'm kind of fussy about audio.
For most people, the on-board audio supplied by the Gigabyte board will be more than sufficient. For those with ears to hear it, though, the dynamic range is significantly compressed, there is more distortion present, and it sounds to me as if relatively cheap digital-to-analog converters are in use in light of the funky harmonics I was hearing in the upper frequencies. Still, it's very good for an on-board solution. If I didn't have an SB Live! card sitting around, I would probably have left it alone. As it was, I decided to reinstall my SB Live! card and disable the on-board audio.
The problem was that my audio "stuttered". Though it was most noticeable when playing back video files with Quick Time or Windows Media Player, it could also be heard while playing back MP3 files. To describe the problem accurately and precisely is rather difficult, but I'll give it my best. Imagine that you are listening to a piece of music. Then imagine that a snippet of that music roughly 0.1 seconds in duration repeats over and over, back to back, for roughly 0.25 - 0.75 seconds before normal playback continues. That's what my system was doing, and it was very disconcerting.
Both fortunately and unfortunately there is a fix available for the problem. Fortunately, if I let Windows Update install the Microsoft "Creative Labs Media Driver Version 5.1.2602.0" update, then my stuttering problem is fixed. Unfortunately, the presence of this media driver renders all of the Creative Labs utilities useless. That's a rather steep cost in my book, as I used to use those utilities to tweak volume settings, equalizer settings, and even EAX settings for games that don't have built-in support. I used those utilities a lot, but they won't run with the Microsoft driver, and the latest drivers from Creative Labs (dated 07/17/02 as of this writing) are simply unusable in light of the stuttering.
I've taken the problem up with Creative Labs support, but to date they're blaming my BIOS. This makes no sense, of course, so I'm hoping they'll eventually come around. As it stands, I can load the "Fail-Safe Defaults" for my system BIOS, set all of the timings to their slowest, and I'll still get the stuttering problem with the Creative Labs drivers. In contrast, I can load the "Optimized Defaults" for my system BIOS, tweak all of the settings to their fastest, and I don't get any stuttering with the Microsoft drivers—even when I'm pushing the system ridiculously hard (i.e., playing back a video file while simultaneously defragging that hard drive, burning a CD in the background, and copying files). My "solution", for the moment, is to settle for the paucity of configuration features the Microsoft drivers provide in exchange for the reliable playback. Hopefully, Creative Labs will eventually fix this issue. They've been pretty good about such things in the past, so I have yet to abandon hope.
It's been a long, ugly road. Frankly, it's always a long, ugly road to rebuild a system and reinstall any flavor of Windows. Microsoft just makes it much worse with their stupid and useless activation garbage. People like me who use their system a lot, installing and uninstalling a lot of different hardware and software, have to walk that road far too often. Worse, I haven't reached any conclusions as to why my old system just suddenly crapped out. My suspicion is that something went wrong with the power supply or the motherboard, but I can't prove that, and I have neither the time nor the energy right now to investigate further.
I have learned a few things throughout the process, however, that are worth noting. First, I am never taking another recommendation from Anandtech seriously. Sure, I'll still read their articles, and I'll still read their reviews. But without exception, every single piece of hardware I've bought based on their recommendation has been a royal pain in the ass from day one. The Fong Kai case seemed nice at first, but only because I was accustomed to working with something far more baroque. The Antec case easily puts the Fong Kai models to shame. The ASUS P3V4X and ABit KT7A-RAID were huge time-wasters for me. In contrast, the motherboards (i.e., the ASUS A7V133 and Gigabyte GA-7VAXP) and the case recommended by some regular guy at Fry's have all been great. In the future, I'll be buying Antec and Gigabyte until Joe Techguy at my local Fry's gives me different advice.
Second, I'm going to be buying the biggest power supplies I can afford going forward. The reason I suspect the 300W power supply in my old case is because I tried the Radeon card in my old system and discovered two things: (1) it's flaky as hell in a 4x AGP motherboard, and (2) the system would just hang hard in any game that pushed a lot of polygons. In short, I think the 300W power supply in my old case was just barely adequate for my Athlon/GeForce combination. It would certainly explain a lot of otherwise inexplicable lockups. As it stands, the reincarnation of my desktop system hasn't had even one bizarre lockup since building it, and that is completely unheard of around here. I've never bought such a substantial power supply in the past, so it gives me reason to wonder whether it might have something to do with that level of stability.
Third, ATI has finally arrived. I'm not going to take the time to write about it here (this article is already too long anyway), but ATI has finally put forth a video card that not only competes with but actually beats NVIDIA. True, the Radeon is not without its flaws, but it's the video card to buy these days if you want the best. As for me, I'm definitely enjoying the improved performance, and I'm enjoying the much richer color balance still more.
In closing, I have to say that I'm happy with my system. In fact, I'm about as happy as I've been with a computer in almost a decade. I seem to have far more than my "fair" share of problems with technology, so much so, in fact, that it gives me reason to believe in luck—or bad luck anyway. As of today, however, I have a system that not only functions, that not only runs every program and game I've thrown at it, but runs them all well. That's all I really want from a computer; i.e., that it works, that it runs the stuff I want to run. Hopefully, this system will stay that way for a long time.
10/16/2002