more on the PC

So yes, I was getting loads of error messages which suggested RAM, but running Memtest 86 showed up no problems.

There's another testing tool called Prime 95. This can be run in what's called a torture test and it's a way of testing a mix of things. You can do three test, which test the CPU and RAM to a greater or lesser extent.

Running this the machine completely flaked out on all three tests. This actually suggested there might be a power problem, as in the system wasn't getting enough juice. A quick trip to the BIOS revealed that something called Intel Turbo Boost was turned on by default.

This is meant to increase the clock speed of your CPU with no effort, but often overclocking requires more power, so I thought maybe that was the problem and I turned it off. Doing so meant that Prime95 now passed two of the tests, but the third still failed. This third test was one that mainly tested RAM.

But given Memtest 86 had passed I was now thinking that the mobo/PSU combo couldn't supply enough juice to run all four sticks of RAM. So last week I took the brave decision of buying a new mobo (it's actually about the most expensive you can buy) and a new, high quality 1,000W PSU.

This last weekend I rebuilt the system with this new mobo and PSU and the initial results were good. I spent most all of Saturday building the system, installing Windows 7 and all the updates (this was the bit on the previous system that was a crash-fest of epic proportions) and then playing Anno 1404 and CoD: Black Ops with not a single crash.

But then on Sunday, it wasn't happy. However, I checked the RAM modules and one of them was definitely slightly loose (I'd touched them the day before after I'd shut down to see how hot they'd got). So, I pushed it back in and the machine ran fine all day.

But then I switched it on this last Monday evening and last night and it initially seemed to be having the same problems.

Actually, I should note I'm skipping a point here. See, the weird thing about the BSODs is that they really only occurred from what's known as cold boot - literally booting up after the machine has been switched off for a good while and has gone cold.

And that was one of the problems I've been having working out what's wrong. See, I'd boot it up, it'd crash and then I'd start investigating. However, as I'd do more testing (swapping the RAM modules, for example) the system would no longer be booting from cold, so the BSODs would clear up.

But also there's another aspect there I mentioned - swapping out the RAM modules. See, the weirdest thing is that the problems on Monday and last night were simply solved by me pushing on the RAM modules. Now they weren't loose that I could tell, but I applied some pressure to them and the machine was totally fine after that.

It's almost like as one or more of the RAM modules cool down they become slightly loose in the slots and so as they warm up from use they expand a little and then when I reseat them ever so slightly by applying pressure they then run fine.

To me, this seems like a bonkers explanation, but it's the best I've got. It also suggest I've spent a lot of additional cash on a mobo and PSU when perhaps the RAM really is the culprit. Problem is, I'm still not sure. See, a complicating factor is that the memory controller is actually located on the CPU.

So it could be the processor and my bizarre conclusion above is simple coincidence.