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?

[idata]

Disabling CPU SVID in the BIOS will indeed lower the temps (Asus BIOS), but then Turbo Boost no longer works, unless you also choose the option to synchronise all cores, which - when combined with disabling CPU SVID support - causes blue screen crashes or Prime95 to fail!

No hope with my i7 6700k. Using default BIOS settings, gets up to 81C in Prime95 with Corsair H75 water cooling. It also has temperature spikes and the CPU temp is almost 70C when gaming.

[TGrab1]

SVID is basically the sensor in your CPU that tells the motherboard how much power it needs to operate. This has never been exact in the way of VID, but it is how intel has designed these chips to make sure they have the power they need at all stages.

When you leave everything and set SVID to disabled your motherboard's auto feature doesn't actually work as it uses the SVID to generate the voltage needed at different frequency levels. So if you DO disable Svid, you need to setup your voltage manually as both offset and adaptive both off your VID.

That is probably why you are getting a blue screen, when the SVID is disabled and everything else left on auto your ASUS motherboard reverts back to it's lowest voltage which is somewhere between .50 and .79 if I remember correctly. So those numbers are nowhere close to the amount of Voltage your CPU needs to boot into windows.

Hope that helps.

[idata]

TGrable wrote:

SVID is basically the sensor in your CPU that tells the motherboard how much power it needs to operate. This has never been exact in the way of VID, but it is how intel has designed these chips to make sure they have the power they need at all stages.

When you leave everything and set SVID to disabled your motherboard's auto feature doesn't actually work as it uses the SVID to generate the voltage needed at different frequency levels. So if you DO disable Svid, you need to setup your voltage manually as both offset and adaptive both off your VID.

That is probably why you are getting a blue screen, when the SVID is disabled and everything else left on auto your ASUS motherboard reverts back to it's lowest voltage which is somewhere between .50 and .79 if I remember correctly. So those numbers are nowhere close to the amount of Voltage your CPU needs to boot into windows.

Hope that helps.

Thanks for the explanation. The CPU does run much cooler at lower voltages. I would have no idea how to manually set the voltages.

I see many others are having similar problems with the i7 temps. I'm assuming the i7 is getting too much voltage somehow on the default BIOS settings and clock speeds, hence the high temps being reported?

[AS19]

Let's call it as it is guys, there is no way in the world Intel did not know about this before releasing the chip. Unless they have never tested a single CPU before releasing them all. There is no "testing" being done. There is no "fix" coming. All they do now, is assessing how is that going to affect their business (read: profit). Is the storm big enough to react to it yet? Or just delete this thread (and others like it) and claim ignorance. Whatever the outcome, I think we are stuck with the issue... unless you are prepared to return the chip + motherboard, and go for something else.

Fixed voltage did not fix the spiking for me. It has reduced the voltage and the temps a little, but the spiking is still there as it was before.

[TGrab1]

Ok so I decided I wanted to see if I noticed the spikes on a different intel chip. The answer is yes/no. I do see spikes, but the temp difference is much less noticeable and might not even be true "spikes".

This was done on a 2600k at 4.8ghz using a old Hyper 212 lol.

Here are the screenshots of the results. I ran idle - Stress Test - Idle

I honestly feel these spikes happen on most cpu's, just the 7700k runs hotter than most and it just makes it more noticeable. Remember this is a smaller die that has less die space to dissipate heat. Also has less surface area in contact with the IHS and a crappy Thermal Interface bridging the two. I honestly don't find the spikes alarming. I find the overall temps and quality of the TIM alarming, but the spikes I see as normal just with the other issues making it more noticeable.

[cmack2]

hi team,

after considerable testing i have returned my Z270 TUF Mark 1 to ASUS defaults via BIOS 0801 (4.52GHz)...

always temp spikes using HWMonitor (shows consistently the highest temps) & HWInfo...

may i suggest that anyone who is concerned with temperature performance to run 4 very short tests in AIDA64:-

Julia

Mandel

RAY32 & RAY64

these tests will quickly get to the heart of your 7700K processor/temperature/cooling performance...

these 4 tests will also give you an accurate (using HWMonitor & HWInfo) measur of CPU Package POWER - that no-one seems to be mentioning - surely very high CPU power will over the long term have an effect on CPU degradation...

i initially made the mistake of using the ASUS Fast Tuning via AI Suite III TPU & also EZ Tuning via BIOS @ both EveryDay settings & Gaming settings - these 3 ASUS tuning methods seemed BRUTE FORCE & not particularly nuanced as far as sensible overclock settings go - very dissapointing & led to high temperatures (& HIGH POWER) on stress tests...

@ BIOS standard settings i can now game, video edit, stress test - plus intensive multitasking all below 80 degrees C (ambient 25 degrees C) & absolutely rock solid stable with my AIO H115i on minimal 'Quiet' setting...

the originally reported temp spikes seem to be a feature of the 7700k...

cheers, craig

[idata]

'the originally reported temp spikes seem to be a feature of the 7700k...'

I don't really think that spikes to 80 C when opening notepad can be considered as a feature. I can't really do much about voltage in my system because i have motherboard witch B150 chip. Can't change the VCORE. I can't afford new motherboard (Z270) for now just for the sake of testing Intel 'features'.

I just want Intel to state whether this is working as designed or there is something that we should do about it.

[idata]

Regarding the power consumption ... I've checked using both HWMonitor and AIDA64 and the reported package power while under AIDA64 stress testing is about 12.2W (running at 4.9GHz). Clearly this can't be the correct figure as I would expect something around 100W. The temperatures are averaging at around 65 with maximum peaks of 79. This is using TGrable's settings.

What figures were you getting and am I interpreting the values incorrectly?

Thanks

[TGrab1]

Sorry Ra5040 been pretty busy past few days.

Do you have SVID enabled or disabled?

Try getting HWINFO and looking on it for the following Setting

As you can see mine is running at only 1.136 atm which is practically idle, but will shoot up to 1.3 ish when I actually start pushing it.

As for the settings you are using did you just throw in 1.285 and leave that or have you been taking off .005 and testing to see when it becomes unstable then just adding back .001 until you get it stable again?

I say this because if you can get your voltage down further from that baseline your temps will get better. Another Factor I haven't even mention is Vdrop, because it can really make things more confusing when playing with LLC. That could also be something we can do to get your voltage down a little more while making sure your max voltage is still where it needs to be.

Btw, the main reason I asked about SVID is because when you start using offset and adaptive move that is what your motherboard uses. So lets say your computer is on auto. Your VID is asking for 1.285 v, your settings of auto means motherboard will try to supply it with 1.285 volts (but can be lower or higher as the motherboard tries to compensate for Vdrop)

Now lets say you turn OFF SVID and leave it on auto. Now your CPU is requesting 1.285 v but your cpu is not listening to it since SVID allows your motherboard to read the VID sensor on the CPU.

So that same thing is factored when using offset. the offset applies to the SVID reading. So if you have a -.020 offset and the VID was reporting 1.285 your motherboard will only supply 1.265 (once again this might be lower because of Vdrop)

That is how it works in a nutshell. So if you have SVID disabled try renabling it again and playing with offset.

Also I would leave your other settings on auto for the time being. The only things I would be changing atm would be Multiplier for your turbo speed. The voltage for your CPU and the mode for modifying it auto/offset/manual/adaptive/ etc. Also if you aren't already set your RAM to use it's XMP Profile.

[idata]

Hi Tgrable,

Please don't apologize … you have already been hugely helpful and I really shouldn't be bothering you for more help!

Right now I have the system running using your basic settings, adjusted to my system, in other words:

- Clock: 4.9 GHz

- SVID: Disabled

- CPU Core Cache Voltage: Manual

- CPU Core Voltage Override: 1.26

- CPU VCCIO: 1.10

- CPU System I/O Voltage: 1.10

Under heavy load the temperature is holding at 70 and below which I think is OK, although of course it would be good to get it lower. At any rate it's a hell of a lot better than what I was getting with the ASUS overclocking at 4.8GHZ!

What I was trying to do was to re-introduce a variable voltage, in offset mode ... but I haven't been successful in doing this. I don't know what I'm doing and what I really need to do is to read up on the whole topic so that I can at least understand what you guys are saying.

I thought these article were very useful:

http://www.tomshardware.com/reviews/cpu-overclocking-guide,4593.html CPU Overclocking Guide: How (and Why) to Tweak Your Processor

http://www.anandtech.com/show/5763/undervolting-and-overclocking-on-ivy-bridge Undervolting and Overclocking on Ivy Bridge

http://www.anandtech.com/show/2404/5 Intel Processor Power Delivery Guidelines - Overclocking Intel's New 45nm QX9650: The Rules Have Changed

http://www.anandtech.com/show/2727/evga-x58-classified-first-look/6 More BIOS Stuff - EVGA X58 Classified - First Look

but I need to read them again carefully. Also some of the acronyms don't appear to match up to the ASUS ones ... for example I don't see any Vdroop, but I guess that's the same as LLC.

This article also looks promising, but I haven't read it yet:

https://uk.hardware.info/reviews/4855/5/workshop-how-to-overclock-haswell-processors-basic-settings https://uk.hardware.info/reviews/4855/5/workshop-how-to-overclock-haswell-processors-basic-settings

If you can suggest an article that clearly explains all of the variables that can be tuned it would be great ... especially if it refers to the ASUS BIOS and Kaby Lake.

At any rate, it's now clear to me that I do need to get the VID down as low as possible, so I'll go back to 4.8GHz (which I think is quite high enough) and tune down the voltage. Alternatively I could set the voltage to 1.22, say, and adjust the multiplier up from 42 until I lose stability (which seems to be the way that Anandtech recommend in their article).

Do you think it's important to move away from manual VID setting to offset mode in order to reduce the power consumption at low loads?

Thanks!

Robert

[TGrab1]

Just a few things.

VID- Is the voltage the CPU is asking for.

Vcore- Is How much Voltage you are actually giving. This is what changes when you set manual/offset/adaptive.

SVID - Is the sensor that tells the motherboard what Voltage the CPU is Requesting via VID

If you want to mess with offset voltage you will need to ENABLE SVID. As it makes its adjustments to the VID your Cpu is requesting.

So what i'm trying to say is getting your VID down, isn't the aim. It is your actual VCORE that matters. As VID is just the basic voltage the CPU is Requesting at any given time and not what it actually needs.

Here is a really long guide talking about overclocking the 7700k specifically for the ASUS boards. It goes into a little detail on all the things it looks like from my quick glance at it. Give it a read and see if it helps.

http://edgeup.asus.com/2017/01/31/kaby-lake-overclocking-guide/ The Kaby Lake overclocking guide - Edge Up

[idata]

Yes, as you can see I haven't even got the right terms plugged into the right boxes yet! I'll have a read of the guide you suggested now ...

It all takes a bit of getting used to as I have never tried to overclock before, but I'm slowly getting a sense of how it all hangs together. Now I need to fill in the blanks and get a more in-depth understanding.

I take it that you have not messed with your offset voltage and left it at a fixed VCore? I would be happy to leave it at that myself ... but is it worth doing going for offset/adaptive?, or is it better just to get a fixed VCore down as low as possible? Perhaps this would give a more stable system, with no (or less) problems with voltage overshoots and undershoots?

Cheers

Robert

[TGrab1]

I use offset mode actually, but I can't just give you settings for it as they would probably be vastly different lol.

Next it is 100% worst finding your lowest stable vcore. The lower you can get it the less heat you are generating and wasting on a chip that doesn't need it.

So like I said take the manual voltage you are at now and reduce it by .005. Do a stress test. If it passes do another .005. Do this until the stress test fails. Then add .005 back to get it stable again. You could also dial this down even more exact by ADDing .001 and testing again. If it is stable you are at the correct voltage. If it fails add another .001. rinse repeat until stable.

Dialing it down as much as .001 isn't actually needed but does help for those trying to maximize a overclock when pushing the heat envelope.

[idata]

Thanks

I've just read the tutorial you suggested and it is very useful. I'll now do some tuning and see where I get ... and after that perhaps have a go at offset mode (btw ... the tutorial recommends adaptive mode ... what are the pros and cons? or are they just different ways of achieving the same result?).

[TGrab1]

The difference between the two is this.

Offset sets the curve throughout the frequency ranges.

Adaptive only changes the turbo mode voltage. So your cpu would basically be in auto when not being turbo'd but when it starts to turbo to 4.8 or 4.9 it would use the voltage specified (use the one you found was stable in manual for this)

Hope that helps.

[idata]

Yes, it does help (as usual!).

Of course I'm not sure what 'Auto' is in this case (I assume the CPU requests the voltage that it thinks it needs based on a table, presumably, and this voltage is likely to be higher than what the CPU actually needs).

When you say that offset sets the curve throughout the frequency range it sounds as though the offset is not applied linearly as VID request goes lower. Is that right?

But at any rate it would seem that offset would be a better option (but probably harder to set ... as you would need to make sure the voltage at low loads didn't go too low, presumably?). If you have the time it would very useful to see an example of an offset setting based (on made up) full load manual values (and no load values?).

Cheers,

Robert

[cmack2]

yes we all have a 7700K thermal/spike issue...

this is quickly becoming an overclocking forum...

HWMonitor records higher temps & power compared to HWInfo (all on the same parallel processes - i wonder why?)...

this is my readout running AIDA64 Julia, Mandel, RAYTrace32 & RAYTrace 64 benchmarks (these benchmarks give very quick temperature results along with total wattage)...

all tests ran with minimal performance AIO cooler settings...

cheers, craig

[TGrab1]

The intel rep here will be fast to tell you those temps are in the safe range, but I personally find them too high.

The bigger problem is you are running that cpu at stock, so unless you are in a really hot ambient during these tests I would say that chip is having major issues with TIM contact somewhere.

Actually on second look your motherboard is running your Vcore way higher than it needs to be. At stock settings you could easily run a Vcore in the 1.15 Range

[PYoon1]

Hi,

I'm having the same random thermal spike issue on my 7700k. It seems near the end of the thread people are trying to overclock their computer to fix the issue or make it better. I honestly have zero experience in overclocking, so I'm afraid to do anything because I have no idea what people are talking about when they're saying they overclocked their PC and made it better.

Is there another solution to this or do we just have to wait to hear from Intel?

If someone could help assist with perhaps overclocking my PC, I could give it a try.

I have the following -

Processor Model: i7 7700k 4.20GHZ

Motherboard Brand: Asus RoG

Motherboard Model: Maximus IX Hero Z270

BIOS Version: V0906

RAM Part Number: 2x8Gb Corsair Dominator Platinum DDR4 DRAM 3000Mhz C15 Memory Kit

OS installed: Win10 Home

Video Card: Geforce GTX1080 FTW Hybrid

[RonaldM_Intel]

Hello Everyone,

We appreciate the data provided and all your feedback.

We are continuing to investigate this report.

We will provide updates as we progress with our review.

Regards,

[AArth1]

Ronald,

While the update is appreciated, I believe that what most people here would appreciate more is a commitment to provide a regular update (e.g. weekly) with at least some details of the progress made. Ad-hoc updates just leave us all in a void wondering when, and most of the time IF, a further update is ever coming.

Thanks