I’ve spent the last week struggling with the components for my PC. I’ve taken the minimum components that I need for an operating PC so that I could transfer the data from my old laptop to it. Then, after holidays, I would get additional RAM, graphics card, and what not. I was looking forward to installing Linux on the new box as it seemed to have advanced much more than Windows in the past several years.
However, I started to have troubles from the very beginning. The video signal would disappear at random intervals and the machine would freeze. I could still operate the DVD drive but that was it. The power was still on but the computer would not react to anything but power off or reset switch. After reset it would continue normally for an arbitrary period of time. The strangest thing was that, while freezing mid any OS installation, I managed to install Linux when I’d select Fail Safe kernel settings. At the same time, simply running memtest – either from the Linux installation DVD or a flash USB card – would never complete.
My first suspicion was the integrated GPU. So, the first thing I did was to get a separate video card. Tried running the memtest after that and at least the video would not go off this time. The freezes continued, though. The normal PC operation would go anywhere from 2 minutes to half an hour and that was it. Then it would freeze.
When I was talking to the techs in the shop where I got the video card, they expressed concerns that it would be faulty RAM that is causing the issue. I still thought it was the motherboard and the chipset on it. So I got a new motherboard. Tried running the system with motherboard replaced but the behavior was exactly the same! RAM situation was weird because I could not test either of the RAM sticks completely so I thought it would have been a strange coincidence that both sticks were bad. Anyway, I went and bought a new RAM stick. I could not finish the memtest with that one, either.
I was beginning to lose hope. The only things I haven’t replaced were processor and the Power Supply Unit (PSU). The processor would make the whole system simply be dead, rather than freezing. Also, neither processor, nor the integrated chipset cooler, nor the PSU were ever hot. So, a temperature was OK. I also checked the voltage reported by BIOS and everything seemed OK.
The next thing I decided to change was the PSU. The processor was the most expensive item and I left that for the very end. To test the PSU, I tried disconnecting the DVD and the hard drive. Did not help. Removed also the external video card to reduce the power consumption by the devices. No help, either. So I bought a new PSU. When I compared the new PSU to the old one I was amazed! It was much heavier and had heaps more connectors. And it was completely silent, which is a useful addition. Now the desktop PC was as quiet as the laptop. The first thing I noticed was that the system was much more responsive even on the boot screen. It would enter the BIOS instantly, which was never the case with the old one. Then I ran the memtest and appeared to be running much faster than ever before. The faith was starting to come back. Anticipation while looking at the progress bar and wondering whether it will suddenly stop and prolong my misery was awesome…
And, yes – at 41% the machine froze again. :( The only thing remaining is the processor. Tomorrow is the last day of the 7-day warranty and I think I’ll have to bring the processor back although I have not confirmed that it is the exact problem as I don’t have another one to try out.
Edit: Yes, as the Murphy’s Law states, the component that is faulty is to be the LAST one remaining and also the most expensive one. Yes, the processor was the problem the whole time.