Spinor Info

If killing a motherboard was a criminal act, I guess it’d have a fancy Latin name, too, like matrotabulicide or something.

Fortunately, it is not criminal to murder a computer motherboard just yet, at least not if it is one that you own. Today, sadly, I managed to accomplish just this, prematurely ending the life of a computer I only recently put together for my wife. No, it was not first-degree murder, more like man-, or rather, motherboardslaughter: I did not set out to do damage, on the contrary, I was hoping to get rid of an annoying little problem (no keyboard after the system came back from hibernation) by upgrading its BIOS.

The consequences of the attempt are, sadly, known to many: after the BIOS upgrade program happily reported success and attempted a reboot (I did everything by the book, and there were no signs of any trouble), the system became completely non-responsive, suggesting a failure of the BIOS boot sequence.

Now usually, there are ways to recover from a failed BIOS upgrade. Many BIOSes recognize a corrupt image and drop into an emergency boot flash mode that allows recovery from, for instance, a floppy disk. I’ve done this before, more than once. However, in this case the process did not work: even when I manually “forced” a boot into the boot flash mode by shorting some of the BIOS pins, it just stubbornly refused to attempt to read anything from a floppy, the hard drive, or a CD-ROM. I even tried booting using a POST diagnostic card, but I became none the wiser.

So that’s it. It is time to move on. It wasn’t a new motherboard, but it still had plenty of useful life left in it, so I am sad to see it go. But investing further effort into it is just pointless. I could order an identical motherboard off eBay for about $80 total, including shipping from China, but why would I want to spend that kind of money on 7-year old technology, when I can buy a decent current motherboard, dual-core CPU, and 2GB memory for less than three times that amount?

The said thing is that it means not only junking this motherboard, but also the accompanying Socket 754 single-core CPU and 1 GB DDR memory, as I have no other Socket 754 or DDR motherboards. I hate contributing to the world’s landfills. I wonder just how much otherwise still flawless electronics ends up in those landfills because of BIOS failures?

It was not a virus after all. I have no idea what made my old workstation unstable, but after a monster 30-hour scan of its hard drives, all I found was 4 potential threats, none of which was an actual infection. One was a malicious Java applet that never had a chance to do harm because it relied on an old version of the JVM, and I tend to keep my software up-to-date. Another was a 20-year old joke program, pretending to hijack an unsuspecting victim’s computer, which for some reason Microsoft Security Essentials marked as a threat. The third was the user registration program in an old Iomega installation kit, whereas the fourth was the remote control software VNC, which can be harmful if someone puts it on your computer without your knowledge, but that wasn’t the case here.

So then, what was it? My guess is ailing hardware, perhaps the video card. But… it no longer really matters anymore. I’ve done the deed, swapped computers, and successfully set up my new quad-core system to work with essentially all of my peripheral hardware (no mean feat, considering that peripheral hardware, in this case, includes instruments like spectrum analyzers connected via GPIB.)

Now, I can actually get back to work…

One lousy plastic bit. That’s all that was needed to take out my main workstation while I was in Hungary.

Fortunately, it did not interfere with my ability to access my data, as I was set up with a backup workstation that I could access using Remote Desktop while traveling. But still, it was most annoying when my wife called and asked if it was me that was messing with that computer, as it was at first making some strange noises, and then went dark.

The plastic bit in question was one of the bits to which the CPU heat sink is latched. It broke. The heat sink didn’t quite fall off, but as it was no longer pushed securely against the CPU, the CPU overheated and shut down. (A good thing, too. Some older generation CPUs failed to shut down under such circumstances and instead, burned a nice black hole in the middle of the motherboard.)

Grumble, I am old enough to remember when TVs had tubes and the repairman was a regular visitor. Back then, solid state electronics held a promise: it was supposed to last forever. But that’s because nobody at the time foresaw all the innovative ways modern computers have come up with to die at the most inopportune moments. What worries me is that the same thing, a computer disabled because of one lousy plastic bit, can happen anytime anywhere… even if said computer, say, is the one running the fly-by-wire system of a modern jetliner or controls the X-ray dose you receive in a medical device.

Of course those systems are redundant. But so was mine… I did have that backup workstation, after all. Nonetheless… I wish these things were a tad more reliable.

Today is the second Tuesday of the month, and it’s Windows Update day. Not my favorite day of the month.

So here’s a curious question: why is it that I was able to upgrade Microsoft Virtual PC from version 2004 to version 2007 and it did NOT require a reboot… but when I installed a security update to version 2007 (making sure that Virtual PC as well as all its drivers are actually shut down) a reboot was necessary?

I don’t think I’ll ever figure out this Microsoft reboot misery.

This morning, I decided to clean my main workstation, as it’s been noisier than usual these last couple of days. Sure enough, it had plenty of dust and cat hair in it, but I also noted that a little motherboard fan was not working. I had spares, so I decided to replace it. Removing it is not easy, as it is in an awkward location, and I also had to resolder its connector onto the new fan.

That having been done, I rebooted my machine. Or, I tried to anyhow, but it failed to boot. It failed to do anything after the BIOS initialization. After a while, I realized that for some weird reason, it was trying to boot from an attached USB device. I disconnected the device, but still nothing… worse than nothing, actually, as it told me that my system disk is unbootable.

Actually, I should say disks, as I am using a dual drive mirror configuration in this computer. After poking around in it some more, I realized that I have inadvertently knocked out a cable from one of the drives. Unfortunately, reconnecting this cable didn’t help… the system was still unbootable.

And I’ve seen this before! A while back, a similar thing happened… one drive was disconnected, and when I reconnected it, the stupid RAID controller somehow managed to wipe the boot sector from both. What on Earth it did that for, I have no idea. But this is why I have a spare computer with identical drives and RAID controller… once I understood what was happening, the solution was to simply remove both drives, put them into USB cases, connect them to the other computer, and then copy over the boot sector using a raw disk editor. Easier said than done, of course, especially as I couldn’t find a raw disk editor on that spare machine (I swear I used to have one installed!) but eventually downloaded HxD, a very fine free disk editor that I have used before.

Two hours after I began, I had a working computer again. Still, it’s not a fun way to start a morning when you need both a soldering iron and a raw disk editor to get your computer working again…

I built myself this dual-core computer several years ago, and it has been my main workstation ever since. By and large, it’s a good and reliable machine (not counting a broken fan and some fried capacitors on its motherboard… never mind, I now have spares of the same motherboard just to be safe) but there have always been a few minor glitches.

One of those glitches concerned MIDI files… if I played back a MIDI file on this computer, its tempo was all off. That is, unless I set the processor affinity of the MIDI playback program to play back only on a specific CPU core. Go figure. If that’s the worst problem I have on a machine that otherwise runs for months without a reboot despite being used for everything from software development to video editing, I guess I can call myself lucky.

But now, years later, I ran into another curious problem. There’s this computer game from the 90s, Duke Nukem, that I, for some reason, still find quite enjoyable. I usually run it on an old Windows 98 box. Recently I found out that there is an open source effort to develop and maintain a multiplatform Duke Nukem executable. I downloaded it and tested it on a test machine… it worked fine. So the other day, I put it on my main machine. It didn’t work fine… if I hit a key on the keyboard, it registered as a large number of very rapid keystrokes to the program.

I’ve been searching for a solution and came across a comment about CPU drivers and processor affinities. Whoops! Restricting EDUKE32 to a single CPU did the trick as a workaround. And then, I remembered that two years ago, I chose not to download a processor driver update from Microsoft Update, following the good old principle of “don’t fix it if it ain’t broken”. So now I took a deep breath and downloaded this update (okay, I wasn’t too worried, since this update already ran fine on a test computer with nearly identical hardware.)

Bingo! EDUKE32 runs like a charm and guess what… so do MIDI files? A years old mystery solved. Now I can happily shoot some heavily pixelated aliens in post-apocalyptic L.A…

I have to thank a fellow blogger (ouch, does that make me a blogger, too? I still can’t stand this word, but I suppose it’s now inevitably part of the English vocabulary) for an excellent post that helped me out: during the install of VISTA SP2 on my laptop, the machine failed with the error code 0xC0190001 associated with explorer.exe. The first thing I did was to try Google, and the first Google hit I found was the above-mentioned blog entry, advising me to reboot into Safe Mode, allow Windows to do its thing and reboot again, and presto: VISTA is back, with SP2 properly (I hope) installed, and I saved myself a significant amount of unpleasantness associated with a system reinstall.

Or maybe I’ll have to reinstall something in the end… because although SP2 came up just fine, for some reason I lost the Aero desktop altogether, and I am back to a standard Windows 2000 style theme, the Aero theme nowhere to be found. Curious. Good thing that laptop is not mission critical, except when I am traveling, which I am not planning to do anytime soon.

Ah. Stupid service pack upgrade disabled the Themes service. It also monkeyed with one of the VMWare services, but now that I started everything that needed starting, things seem to be working fine.