Thermal sensor issue i7-7700k?

[CK7]

I have a brand new build; EVERYTHING NEW. i7-7700k is running at stock speeds. I have the RAM set to XMP for DDR4-2666. Motherboard is Asus Maximus IX Hero Z270.

I have found that the i7-7700k reports a momentary (a second or less) temperature spike +25 > 35 degrees Celsius anytime a program is opened, a webpage is opened, a background app runs etc. The temperature blip cascades through the cores in random order; not the same every time. This causes my heatsink fan to constantly cycle up and down. Temperatures otherwise report as steady, normal increases. Peak temperature under Prime95 blend test is 71 degrees Celsius.

Attempted solutions:

I have re-installed my heatsink and thermal paste with no change.

I have tried to manually set my fan speed in the bios. The only setting that avoids this issue is setting the temperature / fan at a constant (and loud) 80-100%. I've tried PWM and DC mode.

I have found a few user reports elsewhere on the web, all reasoning that it's just the way it is. I don't accept that. Opening a folder or browser should not spike temps +30 degrees. Not only is the fan cycling annoying, it puts undue stress on my fan; possibly shortening its lifespan.

What's the answer, if any? RMA?

[JP_2]

Ok so like everyone else I !had! this problem of temperature spiking up to even 80 °C while doing barely anything and that was causing fans to ramp up up and down unnecessarily. I didn't mind the temp jump as I could believe this is possible but was most annoying was the fan jumps, 0% -> 30% -> 50% -> 0% in the span of 3 sec and that went on repeat while doing anything like light browsing.

What helped at least in my case: DELID. I was scared while doing it, but it worked, even though I saw 1 spot where I probably accidently brushed PCB green color but hey, it still works. (phew)

Temps before delid in Realbench: 92 °C @ 4.8 @ 1.28 (auto)

After delid: 62 °C. 30 °C drop in heavy benchmark.

Now what's more important here, 80 °C 1 sec spikes are gone! After booting up I opened RealTemp, came here, registered, opened a bunch of various tabs and now it shows max temp jump of 57 °C.

80 °C -> 57 °C is good for my sanity, fans will finally shut up for good in idle. God I love my 0 dB setup, lol.

Anyway, delid is nerve racking but in the end it's very worth for this generation of CPU's. Intel might do something different for new stuff as Ryzen is totally soldered and we finally have competition but there's only 1 solution now and sadly Intel will not be able to fix this.

So if you want to remove the instant temp spikes and lower temperatures overall -> DELID!

[idata]

Unless you're using a water cooler, liquid nitrogen etc. (and not overclocking) the only solution appears to be a delid, otherwise you can expect high temps. I still get the spikes and the computer fans spin up. Only a delid would solve that, but it's too risky! I personally don't agree with overclocking, my data is too important to me and I believe overclocking always causes something to go wrong at some point. Even if the computer has been stress-tested for literally days non-stop and passes all tests, something will go wrong in future - data corruption is inevitable; overclocking isn't worth it for just a 30 percent performance increase.

The point I'm making is a delid seems to be the only permanent fix to these problems. If anyone reading is willing to take the risk of delidding, I hope you can post your temps before and after the delid - using Prime95 Small FFT's for at least 30 minutes on each test.

[ASušt]

I can't call delidding a fix, because it doesn't solve the problem entirely.

It makes spikes smaller, but still a spike from 30 to 60 will cause fans to ramp up their rpms.

Plus of course the risk you take and the warranty you loose.

[idata]

Out of interest ... do you have any hard evidence to back up what you say regarding overclocking and inevitable data corruption? And wouldn't you think that delidding would also increase the risk of CPU failure and so potentially increase the risk of data corruption?

Like you my data is critical ... also, I'm not at all interested in gaming, and so stability is more important to me than performance. But from my tests I can run my CPU at 4.9GHz, Cache at 4.5GHz and memory at 3.2GHz at 1.275 volts, and the max temperature with Prime95 (over several hours) is 78C. If I drop the CPU speed to 4.8GHz the max temperature drops to 74C (in both cases the liquid temperature is stable at 32C). And this is quite extreme stress testing. Under normal operation with programs like Photoshop or Camtasia the temperatures are lower and the peaks are for short periods only ... so I don't see any real risk in running at either 4.8 or 4.9GHz. If, as I hope, Intel offer a fix to our problem, then the temperatures with Prime95 could drop down 10-20C, so overheating would not be a concern at all.

Using the Intel Extreme Tuning Utility it's quite easy to limit the CPU current and power, as well as the voltage ... and of course the CPU itself has an over-temperature throttle. Which would make me think that a relatively mild overclocking of 9% (4.5 to 4.9) should be quite safe, especially as when we overclock we can tune the settings so that the CPU runs cooler than at stock with default settings.

[idata]

ra5040 wrote:

Out of interest ... do you have any hard evidence to back up what you say regarding overclocking and inevitable data corruption? And wouldn't you think that delidding would also increase the risk of CPU failure and so potentially increase the risk of data corruption?

Like you my data is critical ... also, I'm not at all interested in gaming, and so stability is more important to me than performance. But from my tests I can run my CPU at 4.9GHz, Cache at 4.5GHz and memory at 3.2GHz at 1.275 volts, and the max temperature with Prime95 (over several hours) is 78C. If I drop the CPU speed to 4.8GHz the max temperature drops to 74C (in both cases the liquid temperature is stable at 32C). And this is quite extreme stress testing. Under normal operation with programs like Photoshop or Camtasia the temperatures are lower and the peaks are for short periods only ... so I don't see any real risk in running at either 4.8 or 4.9GHz. If, as I hope, Intel offer a fix to our problem, then the temperatures with Prime95 could drop down 10-20C, so overheating would not be a concern at all.

Using the Intel Extreme Tuning Utility it's quite easy to limit the CPU current and power, as well as the voltage ... and of course the CPU itself has an over-temperature throttle. Which would make me think that a relatively mild overclocking of 9% (4.5 to 4.9) should be quite safe, especially as when we overclock we can tune the settings so that the CPU runs cooler than at stock with default settings.

It's well known that overclocking is not without risks. Overclocking means running the CPU in a way it wasn't designed. The warranty will be voided. Strange problems will likely crop up in the future etc etc.

As for delidding, yes this will void the warranty and its risky, but with reports of ~20C drop in temps (look on YouTube for delidding videos), delidding and using a better thermal compound does reduce the temps. It's very unlikely that Intel will replace people's problematic processors and acknowledge that the rubbish thermal material is causing the high temps.

[AArth1]

Dave494 wrote:

It's well known that overclocking is not without risks. Overclocking means running the CPU in a way it wasn't designed. The warranty will be voided. Strange problems will likely crop up in the future etc etc.

I have to disagree with the way that you're stating this. You're stating that overclocking WILL cause problems and, in a prior post of yours, data corruption. The risks of this happening are highly dependent on a large number of variables. If overclocking were so risky then why on earth do so many people do it - me included. To date I have never had an issue related to overclocking my CPU. My last setup was an Intel i-875k (2.93GHz) CPU which was overclocked to 3.85GHz for 5 years using a standard AIO cooler. I NEVER had any issues or data corruption. Yes - overclocking carries risks, but please don't make statements that may mislead people into believing that this will inevitably lead to issues and data corruption.

[idata]

I agree with ikrananka ... of course overclocking involves risks as it's possible to fry the CPU if you don't know what you are doing. However even Intel do not discourage it as they offer an 'overclocking warranty' at the very reasonable price of $30. They surely would not do that if they believed that there is a serious, widespread risk of people destroying their CPUs by overclocking. Furthermore, they explicitly sell unlocked CPUs (and all of AMDs are unlocked) when they could easily lock them all: and so even though they do not encourage overclocking, they do not discourage it either.

I can see no reason why a conservatively overclocked CPU is more likely to result in data loss than one running at stock (with no optimization). On the contrary, overclocking allows one to tune the settings so that the CPU operates at it's optimum performance (taking AVX into account) without overheating. As we know, each CPU is different, so running at default settings could well be over-stressing some CPUs, while other CPUs could be under-performing (and in fact are, in my experience).

On the other hand, delidding is very risky and does void the warranty. It sounds OK, but we have no idea of the sort of mechanical stresses that the CPU is subjected to during the delidding, and whether or not these will cause a malfunction (or death).

A CPU malfunction is very unlikely to lead to data loss, or if it does it would only be to cached data ... and the best way around that is to make sure your data is well backed up (and the OS too, so that if there is a crash you can easily get back to where you were). That way any data loss should be minimal, irrespective of the cause.

[ASušt]

Intel said that they can easily replace the CPU, because of 3 years warranty.

They may not star a total recall company but all of us can demand an individual replacement.

Maybe let's start mass-return ourselves?

Guys, yes, OC is a risk. But why the heck do we need all these OC stuff then? Those 200$ coolers, 500$ MB's and RAM?

Let's sit stock with 50% less expensive hardware. I'm not a fan of liqud nitrogen OC, but for all the money I have spent, I guess, I have deseved a 200-300Mhz present.

I think, delidding if more dangerous than OCing by using some fool-proof utility from motherboard manufacturer.

There IS a little suspicion, that with previous version of the BIOS there were no heat spikes.

But I haven't remarked them on purpous, so there is no confidence.

[idata]

The conversation here has become a little embarrassing. Talking about CPU's causing data corruption? Jesus. People that believe something like that should not be building pc's or even be commenting on them.

Onto de-lidding.. obviously it has risks, but if done could be very beneficial, I've not needed to do mine as I said previously, the Z270 Pro Gaming Carbon by MSI has pretty much none of these issues, and I will continue to state that a lot of the issues are the motherboards, everyone I've seen with the ASUS Strix is having issues (if not de-lidded) - Yet i've fixed everyones issues with the pro gaming carbon. YEs there is still minor spikes, but as my previous post stated, these aren't a 'problem' just not ideal.

[PGarc12]

Can you share how did you fixed it? I have a z270 pro carbon and I have the issues of the temps spikes... and Idle is around 30-40C degrees

[idata]

CPU Clock: 50 (5GHZ)

EIST: Disabled

Extreme Memory Profile: Enabled

CPU Core Voltage: 1.325 (1.328V displaying in BIOS)

CPU SA Voltage: 1.200 (1.208V displaying in BIOS as shown in the image after Core Voltage) - This value was set to AUTO previously, and over 1.3V

CPU IO Voltage: 1.150 (1.160V displaying in BIOS as shown in the image after Core Voltage) - This value was set to AUTO previously, and over 1.3V

CPU PLL OC Voltage: 1.150

That's for a 5GHz setup, but even for 4.5 GHz as standard, you just need to lower the CPU Core Voltage down to like 1.2, and see if you can keep dropping until it crashes etc.

[idata]

What max temperatures are you getting when you run Prime95 (26.5 no AVX) and current (with AVX) for 20 minutes? Are you getting any cache errors showing on HWInfo64?

Thanks

[TGrab1]

RitchieDrama wrote:

CPU Clock: 50 (5GHZ)

EIST: Disabled

Extreme Memory Profile: Enabled

CPU Core Voltage: 1.325 (1.328V displaying in BIOS)

CPU SA Voltage: 1.200 (1.208V displaying in BIOS as shown in the image after Core Voltage) - This value was set to AUTO previously, and over 1.3V

CPU IO Voltage: 1.150 (1.160V displaying in BIOS as shown in the image after Core Voltage) - This value was set to AUTO previously, and over 1.3V

CPU PLL OC Voltage: 1.150

That's for a 5GHz setup, but even for 4.5 GHz as standard, you just need to lower the CPU Core Voltage down to like 1.2, and see if you can keep dropping until it crashes etc.

My guess would be the change to EIST being disabled.

This prevents the processor from entering different states.. so keeps the voltage/clock consistent. You might see some slight changes do to Vdroop, but overall you would not see spikes because the processor is only changing the amount of load now and not the states. So you might be idle, but you aren't in an actual idle state.

Temps would still rise with higher loads, but it wouldn't be a drastic idle 28c to 60c spike when it shifts from idle to full throttle.

[idata]

Even if EIST is on, makes no difference. It's just my temperatures are under control due to the MSI board.

[idata]

When you say that your temperatures are under control, it would be nice to see what these are running Prime95 Small FFT (with and without AFX) for 20 minutes or so (at 5GHz).

I would also like to see a graph from HWInfo64 of the package temperature when you repeatedly open and close Chrome, say.

If the problem is indeed the ASUS motherboard, as you claim, then we need to pin this down ... but it would surprise me if it was, as surely Intel would have come out and said so by now.

Cheers

Robert

[DBux]

Please... Mine is Gigabyte Gaming 7 (as I said few times in this threat) and have the same spiking temps issue.

[idata]

There you go!

This is an interesting article re enabling/disabling EIST etc.: http://www.dungeoner.com/en/overclocking-and-intels-power-management-settings-eist-c-states-turbo-boost/ Overclocking and Intel's Power Management Settings (EIST, C-States, Turbo Boost) | Dungeoner

One option is to turn set the Power settings in Windows to Performance mode it you need the max performance, then change it back to Balanced for normal every-day use. As far as I can see Performance mode effectively disables EIST, Turbo Boost and C1 up.

[ACozo]

I lowered the core voltage to 1.2v whereas in auto it was set at 1.32v approximately. It ran stable for a few days but then it started to hang. So I changed it back to auto.

When at 1.2V the temperatures where between 75 to 85°C under full load. Now they are between 85 to 95°C. Everything works stable now.

What do you guys think, am I ok like this?

[idata]

No ... you shouldn't be running at 85-90 at stock frequencies ... in fact you shouldn't be running at these temperatures anyway. When you say 'full load', what do you mean? Stress testing with Prime95, or some games, or what?

[JLamb8]

Not who you replied to, but I am currently seeing 90+ in Prime95 at 1.3v. I thought I was crazy until I found this thread. I get 80*C at 1.2v right now. I'm going to try another cooler I have tonight but I am seriously considering returning this chip for a new one now.

[ACozo]

No ... you shouldn't be running at 85-90 at stock frequencies ... in fact you shouldn't be running at these temperatures anyway. When you say 'full load', what do you mean? Stress testing with Prime95, or some games, or what?

With full load I mean encoding a 4K clip in premiere pro. All 8 threads are at 100%. Max temp reached is 96°C.

Intel offered me a replacement chip but I will wait until Intel fixes the problem. I don't want to have the same problem again.