cpu getting hotter as it gets older...

[eccerr0r]

Curious if anyone has seen this behavior or not, or could it be simply explained by a very obvious reason...

... is it common for a cpu at the same clock rate and voltage to get hotter (perhaps, draw more current) as it gets older?

The obvious reason... fan clogging up with dust or heatsink compound drying up, of course. Or perhaps it's possible for the heatsink to not be as effective as before, should heatsinks' fins be cleaned with more than just compressed air/vacuum? Perhaps dunked into a ultrasonic bath? Then perhaps the heatsink should be replaced (with like, since it had been working fine for many years prior)?

Having a weird situation, heatsink had been working just fine for many years but now, even after cleaning with an air compressor, is no longer providing enough cooling. Grr...

[stj]

Re: cpu getting hotter as it gets older...

cpu's do not get hotter with age,
thermal paste, dirt etc wont help, but also the caps on the vrm going out of spec will cause increased heat.

one other thing i just remembered - heat pipes can lose fluid from corroded pinholes.
i'v seen a laptop heatsink fail like that - easy test is dunk one end in a cup of boiling water and see how fast you get burned at the other end!

[petehall347]

Re: cpu getting hotter as it gets older...

might be working harder than when then system was new .

[eccerr0r]

Re: cpu getting hotter as it gets older...

Yeah I didn't think they should but I only know theory (and theory vs reality have always differed) and have a very small sample of real cpus, and figure you guys have seen a lot more.

AFAIK I have solid polymer caps on my MB so it shouldn't be that, tried out a few PSUs, and my HSF is a copper core aluminum block so have to rule out heat pipe.

Also I don't think it's working any harder than before, I've maxxed out the CPU in running stuff several times in the past (Gentoo user!). Just that recently it's been bumping the heat limits much more frequently than before and can't figure out why.

Another thing I thought of - what about the IHS... though I don't think it's loose, I wonder if the thermal bond of that can be disrupted?

[Topcat]

Re: cpu getting hotter as it gets older...

Check thermal paste. Check that the fan & heatsink aren't clogged with dust. Remove the sink and wash it with simple green & hot water, dust coatings even if thin can inhibit cooling. Check caps in the PSU and motherboard of course. Check the OS and make sure there isn't something weird going on in the background.

[stj]

Re: cpu getting hotter as it gets older...

is the cpu an exposed die, or a big metal cap type?

[eccerr0r]

Re: cpu getting hotter as it gets older...

Yep, discounting dust and heatsink compound failure as I've air compressor blown out and replaced heatsink compound several times to no avail. Whether there's an invisible sheen is also a question I can't answer hence the earlier idea to stick it in an ultrasonic cleaner. There's no way to efficiently clean between the fins, they're too close together. And no I don't smoke... and the computer is fairly far away from the kitchen so I have a hard time believing this machine and not the others that I even run 24/7. This machine is only on usually 12 hours a day, instead.

Yes, as said, it has an IHS and it does not feel loose. Question remains is whether they break off frequently or not? That could explain things...

This particular machine is an i7 and stays 13°C above ambient when (mostly) idling. I don't recall what this number was when it was new, but I've hammered all cores before and sometimes but infrequently hit 90°C. However now it easily hits that and higher, unless I downclock. Downclocking is really killing performance which is why I'm kind of concerned...

[redwire]

Re: cpu getting hotter as it gets older...

Some Intel i7 has poor heat transfer, inside they used thermal paste to the integrated heat spreader. I imagine it ages and maybe time to delid...

What heatsink do you have? Maybe a heat-pipe style leaked. I think use some water inside. Or it's no longer mounting flush and straight to the CPU lid.

[eccerr0r]

Re: cpu getting hotter as it gets older...

sandybridge...apparently they have soldered IHS so I think I have to rule this out?
I have the stock heatsink that came with the cpu. Again it was working fine for many years when it was new, but now it just doesn't seem to work right at the same operating conditions...

[RJARRRPCGP]

Re: cpu getting hotter as it gets older...

sandybridge...apparently they have soldered IHS

And, Intel decided to cheap out with Ivy Bridge, the gen right after Sandy Bridge. Reports of the IHS TIM on 3rd-gens, being poop! While I got words that AMD, with FX, was using solder IHS TIM, IIRC. (or at least something better)

[diif]

Re: cpu getting hotter as it gets older...

You've been posting in the weather thread how it's been 40c outside, now you're telling us the cpu fans are spinning up more than usual....that's not a coincidence.

[stj]

Re: cpu getting hotter as it gets older...

he's american, he probably has 10KW of aircon running!

[eccerr0r]

Re: cpu getting hotter as it gets older...

Actually despite ambient temperatures are high, it's still happening when the temperatures are around 30°C. In fact it doesn't seem to get better until ambient gets around 20°C or so -- but the machine was working just fine years ago at 30°C.

No I don't run AC on my computers...

[stj]

Re: cpu getting hotter as it gets older...

is the psu in good condition?
ripple makes heat.

[RJARRRPCGP]

Re: cpu getting hotter as it gets older...

No I don't run AC on my computers...

I usually do on my most powerful ones, or they'll get flaky.
My PCs are much like those vaccines!

But, I only run a tiny one now. I left my most favorite AC behind in 2016.

[eccerr0r]

Re: cpu getting hotter as it gets older...

is the psu in good condition?
ripple makes heat.

ripple makes heat... where?

I'd imagine the onboard PSU SMPS MOSFETs possibly but not the CPU itself?
If the onboard regulator passes the ripple onto the CPU, then it's not doing its job... also if it indeed is ripple there I should see the machine crashing a lot which is not the case, the only problem I'm seeing is high coretemp temperatures and PECI alerting that CPU temperatures are high, but otherwise machine is stable.

---

more stuff:
The machine was in a small case that had a questionable 120mm "quiet" case fan. I moved everything into a mini tower with functional, high speed, noisy 80mm fans. I can definitely feel the exhaust is warmer than ambient, but that's no different than before - the 120mm was pumping heat out. However, inside case temperatures is noticeably cooler when closed up, the old case is significantly warmer (when I open the case and put my hand to feel the inside air, old case is warmer for a while before the heat escapes due to the open case.) The 120mm was clearly not pumping as much heat out.

... but no change in behavior, it's still overheating when I run all cores full bore at full speed, have to decrease clock speed when running all cores to make sure it doesn't hit 90°C coretemp (80°C PECI).

TBH the "new" application I'm running is a java app. Traditionally my stress test was running g++ compiling programs, but I'm pretty sure it will be affected as well.

And another new variable that I did add to the machine that I forgot to mention: the machine now has 3x as much RAM though it was running dual channel before as well. I wonder if the memory controller are contributing to more heat due to having more memory hooked up to them? Doesn't seem it should contribute that much.

---

Another weird observation... I tried undervolting my CPU to 1.2V... and it fails to boot. So I cleared CMOS. However something changed: when I was running i7z to detect CPU speeds, prior to CMOS reset, it had turboboost enabled for all cores even if all cores were being used(!) That would use more power... After CMOS cleared, now it went back to downclocking a bit if all cores were running. Alas, CPU still heats up way too much for the cooling.

Thing that really bugs me about these "turboboost" CPUs that actually changes voltages as it changes clock speeds "automatically" ... can never tell what it's setting it to.

[clearchris]

Re: cpu getting hotter as it gets older...

Is the RPM of the fan monitored? Fans often fail in mysterious ways and if you have the original cpu heat sink and fan, the fan may be nearing the end of it's useful life. Sometimes you can peel the sticker and add a drop of oil, that may give you a little more life out of if, but not much.

[eccerr0r]

Re: cpu getting hotter as it gets older...

yeah it's working, seems still pretty strong and quiet, one of those stock fans (i've never had a s775 or newer stock fan fail, just those stock s370s are bad...) Still changing RPMs as needed too - it's quiet when cool and can hear it spin up, but currently with the panaflo 80mm case fan, the cpu fan is completely drowned out in noise...