CPU Temperature rises to 90C at idle from boot (1st,3rd...)

Hi,

i did not want to put this problem with my other topic as I believe it to be a separate issue.

Intel 4670K haswell CPU
Gigabyte Z87X-OC Motherboard.

After every cold boot, e.g. first boot of the day, i watch the 4 core temperatures of my CPU rise under idle from 32C or under to past 60 or 70C in 15 minutes. I've witnessed it get to 96C on all cores (100C TJ Max).

Here is the strange thing. After an immediate reboot which lasts about 15 to 30 seconds, I open CPu Temp, or RealTemp or HWMonitor and they are back down to under 30C and do not rise out of the ordinary.

Then a further reboot will cause the temp rise issue again (1st, 3rd, 5th....).

I do not feel any hint of warm air from exhaust when they are at 90c plus.

How confident can I be that this is not a genuine heat issue, thermal compound issue with the heatsink or fan issue, given that temperatures return to sub 30C from 96C in under 30 seconds?

It has to be a bios, or temp sensor issue right?

I'm quite concerned about this. There is something unsettling about seeing cpu temp readings hit 96C :)

At a minimum, I'd see if the issue persists if you use several different temperature monitoring tools.

I use HWInfo64; there are others.

To confirm: this happens ONLY on a cold boot from the off position and occurs EVERY cold boot? Never happens on ANY ordinary warm reboot (simple restart when already running)?

Do your fans ramp up in speed as they might if the temperature rise was actually occurring?

Hi,

This has persisted for several months now. I have used all the know monitor tools both as 64 bit and 32 bit versions. All indicate the same readings as each other.

This always occurs on first boot and not the 2nd reboot. but it also occurs on the third reboot which is strange.

It always follows this behavior.

The fan settings in my bios are a bit primitive but i'm sure on the setting where it ramps them up depending on temp, it never actually did. I've been looking to get a manual fan controller when this issue has been sorted out.

My motherboard is know for its crappy fan controls in bios. It has Full speed (runs at max), normal, silent and manual. Its set to normal which would ramp the temps but it never does.

Is it not almost impossible for core temps to drop from 90C plus to around 30c so quickly? Dust doesnt seem too bad as I did this about a month ago. went through with an air canister while it was off and open.

If you boot into the BIOS, what do the temps show there?

I just read your other memory related question and disagree that they are not related.

IME which goes back to the late 1990's, memory/driver/ IRQ erros are indicating you have motherboard problems, possibly imminent failure. Do you have intermiitent sound weirdness occasionally? I don't recall but seem to think that you did not indicate as to whether or not this box is overclocked. I would recommend a system restart and reset the BIOS back to stock. Giga-byte has decent documentation on this specific action. Even if you haven't OC'd, reset the BIOS back to stock. I would presume too that the PSU is new or as new as the rest?

BTW if possible check your RAM in a different Known Good box as the results you get from this machine could be unreliable.

Hi,

Thanks for the replies.

Just to clarify some things.

I ran MemTest86 and it did not detect any errors under 4 passes of all test procedures.

The CPU temp in the MemTest software indicated 99C/100C all through the test. No heat from exhaust fan and my pc does not seem to be fried in any way.

Still have to the odd pattern of behaviour with reboots causing the temp rise issue at idle. I watched this just get to a reading of just under 80C on all four cores, restarted pc and into bios (30 seconds or so from restart) and the CPU temp reading is 37C (it nearly always is this in the BIOS). There is no breakdown of the individual core temps in bios.

Again, these can not be genuine temperatures can they as my system would have been melted and ruined if the cpu had run at 100C for 20 hours during mentest?

Not had any sound related issues

This is an overclock capable motherboard with an overclock profile configured by a reputable retailer, with warranty on the oc, however I have always been using the stock profile as I have not started gaming on it yet and do not need to use the OC right now.

PSU should be fine, not new but it's a corsair modular PSU that is still under warranty and should be more than suitable for the spec I'm running.

I think I may have to get in touch with the retailer as it is covered by support warranty. I wanted to see if I could first rule things out, or try and solve it without hassle or lots of time spent testing.

I had wondered whether this is a bios bug that could be remedied with an update (there is one but i'll void my warranty)

This is a pain...

Sorry for the late reply, (a thing called 'Real Life' just happened to me).

I am wondering if there might be an issue with 'dirty' power, as in a circuit that has many devices on it that would cause a power drain. This could be part of the cause of the BSODs. This could be fixed up with an UPS with a power regulator built in, (many have this built in nowadays).

I remember from back in WinXP days that the file 'lsass.exe' would use 100% of the CPU from time to time and that this was a sign of profile corruption. I haven't seen this behaviour since switching to W7, but you might consider checking with Task Manager and see which process is using so much CPU.

As to the BIOS update, this could be one of the things included in the mysterious 'microcode update'. Crappy way to write a log.

A further thought on the PSU would be to suggest a PSU tester be applied to it. I have one that I run regularly on my own and any other boxes that come to me. It has proven to be worthy of its cost.

Just had to register just to inform that I have the EXACT same issue, though I'm using a gigabyte x99-ud3 motherboard with the latest i7 CPU.

I solved the issue by disabling the internal motherboard LED (what they call the Audio LED). Not sure if you have that on your board as well?