There are a lot of reasons why a computer will die on you, and I think I may have experienced most of them. The truly frustrating thing about it—for me, anyway—is that so many of those reasons look exactly the same at the beginning. A hard drive dying can screw up applications the same way that bad ram can screw up applications the same way that a cracked motherboard can screw up applications the same way that a bad video card can screw up applications. When you take the time to troubleshoot the problem, you can with time and effort figure out which problem it is—either by swapping out components until it starts working again, thus winnowing out the bad component (“aha! Everything started working when I removed Windows!”) or by running a very fancy program that tests everything for you, and three days later tells you the problem is expensive.
Yes, those are your only two choices. All the other possibilities are variations of those two choices.
The problem is that everything takes time. Troubleshooting the problem takes time. Once you figure out what needs to be fixed, actually fixing the problem takes time.
When my work computer died on me—the computer that I use to write fiction, to miss deadlines, and to do contracting work that pays for my ability to write fiction and miss deadlines—a day before Christmas Vacation started, I did not have the luxury of time on my side. I had roughly a week to get my computer back in order, and then I needed to get back to work.
I wasn't confident of my ability to successfully troubleshoot the problem, because the symptoms were strange. On the Windows side, it was acting like the ram was bad, or the hard drive was bad, but on the Linux side it was just being really... really... slooooooow. In my experience Linux was usually the OS that reacted immediately to defective components, while Windows would usually shrug it's shoulders, watch its house burn to the ground around it and wonder when someone was going to get around to fixing that. So it was an intriguing problem, but the problem was so severe on one side that it wouldn't run—and unfortunately, I still need Windows around to get paid.
Getting paid is important, so I had to figure out what I was going to.
The work computer was an ASUS G74SX—a beast of a laptop three years ago, and still pretty damned impressive even today. I loved that thing to bits (literally—I've had to replace multiple keys on the keyboard) but it was long past its warranty, so any corrective action was either going to come from me, or from a third party. Sending it off to a third party for repairs over Christmas wasn't going to get it back to me in time for work, and trying to fix it myself... well, I was pretty sure the problem was catastrophic failure of some part or other, but I didn't know which part, and I didn't want to just start ordering laptop parts because those suckers are expensive, and I was pretty sure doing that over the Christmas holidays wasn't going to get it back to me in time for work.
The option that gave me the best opportunity to have a working computer in time for work was to buy a new one, but man, I hated that. The Christmas holidays is not a good time for me to have an extra, unexpected expense, and if I was going to replace that laptop it needed to be something at least equivalent in power because while I'm not actually rendering gorgeous 3d images or anything fancy like that, I am starting to do things that require a lot of RAM and storage.
Laptops like that are expensive, even on the PC side of things. And I didn't want to spend that kind of money. That's when a little thought went skipping merrily through my head: you don't have to buy a laptop. You could always buy, you know, an actual PC. Better yet, you could BUILD ONE. It'll be cheaper...
I haven't actually built a computer since 2004 or 2005. And that was just a partial upgrade. It was also when I swore I'd never do it again, because I'd been recently diagnosed with Essential Tremor (hand tremors) and my fine manipulation skills were really not great (it's very difficult to put those little fucking screws in a PC when your hands keep shaking). So I'd been out of the build-your-own-PC game for about ten years. This is an important thing to keep in mind.
So I went to NewEgg and started looking at prices, and yes indeed, you could buy a very powerful PC for significantly less than a competitively beefy laptop. This didn't surprise me, but it convinced me. If I needed to replace my computer fast, and I wanted the power to be equivalent, but I also needed to not break the bank, then I needed to buy an actual PC.
I started with a bare-bones system instead of buying all components individually (so I wouldn't have to attach the motherboard to the chassis) then bought extra ram, an extra hard drive, and a standalone graphics card. When all the components were tallied, it cost less than half what I paid for my monster laptop in 2010 or 2011. NewEgg shipped everything to me on Christmas Eve—I was astounded! So I had that weekend to put everything together.
It had worked! I was saved! I was going to get my computer up and running, and get back to work, and other than the panic and the scars, and the extra money I had to spend that I didn't want to I'd be no worse for wear!
Well, no. Not quite.
Worse for wear
Reacting quickly under pressure is one of those things you just have to do from time to time, but in my experience it doesn't usually give you the best solution. It gives you a solution, and if the person is quick and clever enough it can give you a workable solution, but usually given time constraints it's the solution you live with, rather than the solution you prefer. That's what I thought I was doing: I was pursuing an inefficient but workable solution in order to meet a very compressed deadline.
That's not what I was doing, though. What I was doing was making a decision based on information that was ten years out of date.
The bare-bones system came with a 500w power supply. I looked at that power supply and I was, quite frankly, aghast. The last time I owned a desktop computer it had a 250w power supply, and I remember thinking “that's kind of excessive” since it had previously had a 150w power supply. “Surely 500 watts will be sufficient to run everything,” I said to myself, and didn't think much more about it.
In truth, 500 watts would probably have been sufficient to run everything with room to spare, if it hadn't been for the graphics card.
I wound up buying a high-end graphics card. I didn't know it was a high-end card at the time—I thought it was one of those cards AMD put out that was based on the high-end cards, but was actually a mid-range card because it didn't have the amped-up components. It must have been on sale, and I obviously got all the names flipped around in my head, and what arrived on Christmas eve was a monstrosity that looked like two video cards glued to seventy pounds of heat sink that required a dedicated miniature nuclear reactor to power it.
I didn't know it had ridiculously high power requirements, of course. No! I looked at it, wondered if I could get it into my computer without wrenching my back, said something stupid like “gee, these mid-range cards sure look impressive,” then put it in my mid tower case.
So on Christmas Day I spent some of my time putting the computer together and installing software. Lots and lots of software—all the software I'd need for work. Then the computer started downloading updates, and that's when it started rebooting for no reason that I could tell. Weird, random reboots that made no sense. So started googling and the common answer for “weird, random computer reboots” was “your graphics card needs a larger power supply.”
I scoffed at the wisdom of the crowds. After all, my power supply was a MIGHTY 500 WATTS. It was more power than was proper to put in a computer! But I researched further and found that apparently the minimum recommended power supply for my card was 600w. Minimum! And most of the people who used the card felt that was insufficient.
Shortly after discovering this my power supply exploded.
I don't remember exactly what the computer was doing when it happened. I think it was doing something that actually made the video card work—maybe running a graphics demo. It was the first time the video card actually had to say “Scotty, I need more power!” and the power supply said “I KINNAE GIVE Y A ANNYMORE POWER CAPTIN, I'M GIVVIN 'ER ALL SHE'S GOT” and then the video card said “I SAID MORE POWER SCOTTY” and then the power supply said “WE'VE LOST CONTAINMENT” and the next thing I knew there was a very, very, very, very, very loud BANG and everything smelled like ozone and I was frantically unplugging every electronic device I had from the wall hoping that my computer wasn't on fire and then when I opened the case I saw... well, I saw a dead power supply. The chassis was intact—the explosion hadn't ripped open the case—but it rattled.
That was day one.
The day after Christmas Day I found myself on the hunt for a new power supply. I found one at Best Buy—750w. It was, I thought, far more than I needed, but I wasn't in a position to argue. I bought it, installed it, and the computer booted up apparently none the worse for wear.
Yay! Problem solved! Well, no. It turns out that 750w was not enough power for this abomination, because it kept rebooting. IT KEPT REBOOTING. I did more research. I stumbled across a web calculator where you entered every single component in your computer and it gave you an estimated power consumption, which promptly told me that my computer required at least an 1100w power supply in order to function, so I ignored that calculator and never went back—but then I came across another that said “well, you might want to consider 850w” and I remembered a few forum posts of people who had this card who also recommended something in the 800-850 range...
...but I couldn't find an 800-850w power supply anywhere. I wound up buying a 900w power supply from Amazon. But I wouldn't get it until the Monday after the New Year.
So my first week back at work was spent like this: type-save-type-save-type-save-reboot-type-save-type-save-type-save-reboot-type-reboot-dammit-recovery-save-type-save-reboot and then finally, finally, the 900w power supply arrived. I was, to be honest, absolutely horrified that I was actually putting this thing in my computer. It was a power supply that would run THREE OF THE COMPUTERS I HAD IN 2006 WITH ROOM TO SPARE FOR THE COMPUTER I HAD IN 2003 and I was putting it into my computer to run a single fucking graphics card. Well, two graphics card fused together in an unholy union of technology. I felt unclean.
I installed the power supply.
I turned on the computer.
My house did not burn down.
The computer rebooted almost immediately.
After I stopped swearing I noticed it was Thursday. I took off the chassis and tried to see if I could see what was wrong. That's when I noticed the far end of the PCI card was actually tilted about fifteen or twenty degrees down, because the entire card was so massive that it really shouldn't have been put in a mid-tower case where it had to rest on its side. It as actually twisting out of the PCI slot.
I went to the Home Depot, bought a right-angle bracket, and screwed it to the hard drive chassis so that the heat sink/fan of the Graphics Card On Steroids would rest nicely on it.
I turned on the computer.
The computer ran for an hour then rebooted.
After I stopped swearing I noticed it was Saturday. I didn't know what to do. I'd updated all the drivers. I'd replaced the power supply twice. I'd installed a right-angle bracket. That was the extent of my knowhow, and the computer I'd bought and put together to replace the computer that no longer worked still wasn't working AND I DIDN'T KNOW WHY AND IT WASN'T EVEN MOCKING ME IT WAS JUST FUCKING REBOOTING.
I was losing hope, but I decided to Google one last time... and I found the answer sitting in the Tom's Hardware forums (thank you Tom's Hardware).
It seems that in order to cut down on the heat that the Bestial Graphics Card of Infinite Remorse puts out, the manufacturers underclocked the 2d processors. And it seems that, as a result, the 2d processors will, when they're not doing much, be so inactive that the computer can't find them. And when this happens, apparently the computer panics (OH MY GOD WHERE ARE THE 2D PROCESSORS HAS ANYONE SEEN THEM I SWEAR I WAS AWAY FOR JUST A SECOND) and then it decides the 2d processors actually not working and it reboots in an effort to save the 2d processor's life (LIVE, DAMN YOU! LIIIIIIIIIIVE!) which is why it would reboot whenever. The solution was to get an application like ASUS GPU Tweak and ever-so-slightly overclock the 2d processors and damned if that didn't work.
(It didn't fix the problem in Linux, because there are no overclocking tools in Linux that I can find that will overclock only the 2d processors. However, I did find that the open source ATI drivers don't have this problem at all. So the solution on Linux was “go GNU,” which, to be honest, is very often the solution to a problem on Linux.)
After I realized the fix was to overclock the card, I immediately wondered if I needed the 900w power supply after all. So I switched back to the 750w power supply, turned on the computer, and... it started rebooting, even with the overclocking. Well, part of me really wanted those 150 watts back, but another part of me was relieved that I hadn't bought it for nothing. Pride makes you relieved about stupid things.
Climbing out of the wreckage
I now had a working work computer. This was a good thing. It was also about two weeks into January, which was a bad thing. I now had to spend time getting all my data back on the computer and setting it up for the OTHER things I do. The stories. The comics. The podcasts. The (few and far between) videos. That took a weekend, but I was finally on top of things. I was, finally recovering from my temporary detour into silicon-induced despair.
And that, of course, was when the knife twisted.
The Icing on the Cake
I needed some files on the laptop—the dead laptop that caused this entire mess to begin with. I figured there was a good chance it would still boot on the Linux side, so one Friday night I figured, “eh, I'll spend a few hours trying to recover those files.”
So I collected all the pieces, plugged it in (in a room not my office so as not to disturb the 900w TerrorBeast prowling its too-thin mid-tower cage) and fat-fingered grub, so I wound up booting directly into Windows instead of Linux like I'd planned.
It booted perfectly. I logged in without incident. Everything worked.
That's not actually unheard of when components go bad. Sometimes you'll get days when the component works, and there aren't any problems. I figured this was one of those days, so I recovered the files I needed, sent them off to network storage, then just messed around with the laptop a bit. The next day, out of curiosity, I booted into Windows again, and this time it was all screwed up like it had been before, throwing out obscure error after obscure error and generally acting like it was getting ready to die.
But I noticed something I hadn't before. In the middle of the slew of errors that would show up during login, it would open and then close a program that I didn't recognize. Suddenly I wondered if maybe the problem wasn't hardware-related, but was a virus. So I kept trying to log in until I finally managed to do it, then looked at all the apps installed on my computer.
Well, it wasn't a virus. It was a very badly written piece of software, which is almost like a virus but it's unintentional. It was a program that was supposed to allow Windows 8 to read ext4 partitions as native file systems. And each time Windows 8 tried to load it, it apparently died an ugly death and took as much of the OS with it as it could manage. When I uninstalled the program, my Windows partition immediately returned to normal—no disk errors, no random crashes, no strange Pool Error BSOD messages, no issues at all whatsoever. Also, my Linux partition returned to its normal speed. I don't know exactly how this Windows application was affecting Linux, but my uneducated guess is that whatever it was doing to try to read ext4 made Linux spend a little more effort to clean up after the mess it was making. I have no proof. All I know is that simply by removing a piece of software I forgot I'd installed on the system I completely fixed the computer I thought was utterly and completely dead.
I still haven't stopped swearing.