Be Excellent To Each Other

Right this one has got me stumped. (And yes please do feel free to snicker at my PC problems......)

Anyway, my PC has been running in an unchanged configuration for ages, last change I made was adding a GTX670 a few months ago, but the CPU/mobo/RAM have been unchanged for about two and a half years.

It's an Intel Core i7-920, oc/ed to 3.4GHz on an Arctic Cooler Pro Rev 2 - literally no change for some two and a half years.

It has been rock-solid stable the entire time I've had it, never missed a beat. The CPU did get pretty toasty during a GTAIV marathon one summer's day (nosing up to 90C due to the high ambient temperatures) but still short of the Tjunction of 100C.

Most recently I had a three hour plus Battlefield 3 session on Saturday night, where it was chucking the game around fully ULTRAed out at 2560x1440 the entire time. No problems encountered.

I've been using the PC lightly since then with no issues. (Was watching a DVD last night for example, and usual web-based stuff.)

Today when I got home from work I turned the PC on as usual, then went upstairs to get changed and put my feet up on the bed for about 30 minutes 'cause I'm an old fucker and that's what I do these days.

When I came downstairs my PC was turned off. First thought was to ask AE Jnr and she told me she hadn't touched it, and TBH I believed her as she knows not to mess with my computer. Turned the PC back on and got the 'Your PC was not shut down properly' message, which I thought was odd, but let it boot up again and it seemed fine.

So I went down to the kitchen, got my tea ready, put the bin out, that sort of thing, came back upstairs and my PC was turned off again. Turned it back on, got the same 'Your PC was not shut down properly' message so definitely knew something was up at this point, booted back into Windows and fired up Core Temp, wondering if it could possibly be overheating.

Well, fuck me, even idling in Windows all four cores were dancing between 90C and 95C (!!!!), so it was pretty obvious it'd hit the Tjunction and just powered itself down the previous two times, and would soon do so again.

Pre-empting this I turned it off myself, yanked it out, got the side off and checked the HSF, seemed to be attached fine, no play or anything like that, fan was working fine. Into the BIOS which said the fan was spinning at 2000RPM which is correct, but even in the BIOS the CPU Temp was showing as around 75C.

Deciding there wasn't much else to do but re-seat the HSF I pulled it all off, unscrewed the mount etc, cleaned off all the TIM (I've got proper cleaner for it), got both the CPU spreader and HSF spreader shiny clean, did a nice careful re-apply of fresh TIM, and put it all back together making sure it was absolutely perfect and all solidly in place. (TBH I just put it back exactly how it already had been, as I'd made sure I did a very good job when I originally did the build.)

Booted back into Windows and the temps are better, but still awful.

It's idling with all four cores at 63-68C, which is fucking stupid, even just lightly loading it with a couple of online slots causes the temps to start peaking up towards 90C and gradually get higher, I daren't chuck something like Orthos at it as it'll just fucking fry.

Any ideas? Could it be the actual temperature probe on the CPU is misreporting, but if that's the case, why did yanking off the HSF and replacing the TIM make any difference at all?

I'll power it down and remove the overclock, but even that wouldn't explain why it's fucked today and it was fine yesterday.

I'm stumped!

Screenshot below is when I just chucked a couple of online slots at it.....

lol pcs, get a 360, have you tried water, lol

Now that is out of the way, I had a similar issue a while ago, (although my PC is no where as good as yours) but it came down to there being dust in the heat sinks, when I cleaned them out the PC was idling about about 10C lower than before. you have probably thought of this though.
Also, wrt the temp being misreported, aren't there about 20 probes in your PC, it seems unlikely all of them are wrong at the same time.

Yeah I did all that, got all the dust out of the HSF although TBH it wasn't that bad anyway as all the intake fans on my PC have dust filters on them anyway.

Thing is Bobby, we're not talking about an incremental change here, there's no way on earth the PC would have survived Saturday night's BF3 session in the state it is now.

I've just gone back into the BIOS and removed the overclock so it's now running at stock speed, it is a little bit cooler than it was before as you'd expect, but even just running a couple of online slots all four cores are pushing up past 80C. It'd last about 30 seconds running BF3.

Something major has changed between last night and today, and I just can't for the life of me think what it might be.

A short somewhere?

Malc

So I took the PC apart again, dismantled the HSF assembly again, totally scratching my head as to what the fuck was the matter, when I saw a tiny little piece of plastic stuck under one of the motherboard's VRMs, like the size of a child's little finger nail.

Wondering what the bloody hell it was I picked it out, there should be no loose pieces of plastic inside a PC.

I then checked all the mounting posts for the HSF bracket, and one of them was ever so slightly loose, it was still in place but not totally 'firm', I gave it a little 'twist and tug' and it completely came out (this shouldn't happen). I then pulled the other three out (this took a lot more force as should be the case), then pulled the HSF mounting bracket off the motherboard, and fuck me if there wasn't a tiny little bit missing off one of the corners.

I placed the HSF bracket back onto the motherboard, put all the mounting posts back in, and yes, where there was a tiny little chunk missing off the corner of the bracket there was a tiny little bit of movement in the mount itself, less than maybe a millimetre of movement, but that's enough, and that explained the massive bump in temperatures, as with the weight of the HSF hanging on that bracket, that sub-millimetre of movement would be enough to create a gap between the HSF and the CPU heat spreader.

So that's the old HSF fucked, but as luck would have it I still had a stock Intel LGA1366 HSF hanging around, it's absolutely fucking awful of course, and uses that unspeakably dreadful pushpin design - but I got it fitted, put the PC back together, wired it all up and it's now a lot happier temperature wise. Just had a quick run round a map in Counter Strike and all was fine.

It runs hot as the Intel HSF is just awful, but it's well within safe temperatures and it won't overheat. Overclocking is out of the question until I get a new aftermarket HSF of course, but at least the PC is up and running again.

As for how it happened, my guess would be that AE Jnr has given the PC a hell of a whack when she's been playing in here (she's bashed into it more than once), or Mrs AE has hit the PC with the hoover or something. That plastic bracket will get very hot and very cold, very hot and very cold, and so on and so on for two and a half years, so it'll be very brittle, I reckon one good 'CRACK' into the PC with something could be enough to convey enough energy to knock a little bit of the bracket off like that.

Pain in the arse but at least the riddle is solved, I hate it when computers confuse me.

Glad you got it sorted dude. I came in to see if I could be any use as you were so helpful to me recently but thankfully you've already solved it as I doubt I'd have been much use.

I had that twice. Bracket just broke. I thought it was the kids. The 2nd time it broke the kids had been away a week.

That is why I got a new pc

'Someone' knocked into it? Yeah right... You should be less vigorous when rubbing yourself up against the PC, AE.

The first time (actually, so far the only time) I built a PC I was quite apprehensive about clipping this big, heavy feeling heatsink and fan onto the seemingly flimsy motherboard but it all held together and is still going strong some 6 years or more later.

Good gawd this Intel HSF is fucking rubbish, even at stock speed one of the cores was dancing around 90C during a BF3 session.

The C0 stepping of the i7-920 is a notoriously hot-running chip and the stock Intel HSF was well known to be just about up to the job of taming it, but even so, pretty pants.

Ebuyer have confirmed the despatch of my new Arctic Cooler so hopefully Mr Parcelforce will bring it tomorrow. I also treated myself to a new case fan and some high quality TIM.

£23 order plus a £13 delivery charge.....

New Arctic Cooler arrived from Ebuyer today, along with some MX-4 compound.

Got the INCREDIBLY SHITTY stock Intel HSF off that I'd been making do with (whoever came up with that pushpin design needs to be shot to death), cleaned the pre-applied compound off the new Arctic Cooler and polished it down, cleaned off the CPU heat spreader and then used some of the preparation fluid I have on both CPU and contact plate on the HSF (not sure it makes any difference but it makes me feel better ), got a nice application of the MX-4 on and got it all mounted OK.

Overclocked back up to 3.4GHz, and even running the torture mode of the IntelBurnTest the hottest core only just tipped up to 75C (on the stock Intel HSF it was hitting the mid 90s at stock speed!). Could probably get it up to 3.5GHz, perhaps on the way to 3.6GHz, but it's always been 100% stable at 3.4GHz so I'll just leave it there.

I think this new cooler must be a bit better than the old one (even though it's exactly the same model), or maybe this superior TIM is making the difference and/or I've just fitted it a bit better, but it's about 5C cooler on this one than it was on the old one, and the TIM hasn't even cured yet.

Should you be running it before the TIM has cured? I thought that was a no no?

Was it ill?

I had CoreTemp running at all times so could have stopped the burn test at any point, if it had got into toasty territory (90C+ on any core) I'd have stopped it.

After a nice cold relax in the lounge overnight, it should be cured and hopefully running a bit cooler tomorrow, but TBH if it can hold the hottest core at 75C in the burn test, it's all good already.

Overclocked to 3.4GHz it's running cooler in a high-stress torture test than it hit on the Intel HSF at stock speeds just playing BF3 - which really goes to show how shit the Intel HSF is.

See computer discussion in a computer thread is great.

look, kovacs, I know ae can take things too far at times, but that is pretty damn patronising. furthermore, I don't believe there are any rules stating you must only talk about PCs in pc threads.

@ae, I thought running it before it cured meant you got thermal expansion during the curing process, leading to it being less efficient when it set. this could be complete bullshit though.

Well it's hitting exactly the same temperature in the torture test today as it did yesterday, so it's possible I may have messed up the curing process but I'm no worse off than I was last night.

It's 5C cooler than it was on the old Arctic Cooler so either way it's a reasonable result, and the new case fan is quieter than the old one. Everything's coming up AE.

Sorry I had had a lot to drink....