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]

ra5040 I use Coollaboratory liquid metal TIM for all my delidded CPUs. I've got both the Ultra and Pro variants, honestly don't recall which one I used for this particular chip, I've delidded a fair number of them since I started doing this for my personal builds and clients back when Ivy Bridge came out (5 years now). I use that between the die and the underside of the IHS and then a top notch paste type TIM between the IHS and water block cold plate. In this instance that was Phobya Nanogrease Extreme.

[CCapt]

@ ra5040,

By "sustained" I mean specifically as in days or weeks; not as in minutes or hours.

Concerning spiking, since you're selectively quoting from my post on page 40, here's what I wrote:

• Concerning the spiking problem, I noticed one item in particular that's been overlooked in this thread, which is "Speed Shift".

Spiking has always been present, however, as a few of you have already pointed out, it hasn't been so obvious on prior Generations due to several variables. Intel's specification for Digital Thermal Sensor (DTS) response time is 256 milliseconds, or about 1/4th of a second, so Core temperatures respond instantly to changes in load. Intel introduced Speed Shift technology with Skylake, then kicked it up a notch with Kaby Lake. Speed Shift was intended as an improvement over SpeedStep, so Speed Shift responds faster to changes in workload demands while being more energy efficient, but the downside is spiking. Also, excessive spiking is sometimes caused by unnecessary tray startups, processes and services, so it help to keep your software cleaned up.

Here's two links to our sister website, AnandTech, which describes Speed Shift:

Examining Intel's New Speed Shift Tech on Skylake: More Responsive Processors - http://www.anandtech.com/show/9751/examining-intel-skylake-speed-shift-more-responsive-processors Examining Intel's New Speed Shift Tech on Skylake: More Responsive Processors

Speed Shift v2: Speed Harder - http://www.anandtech.com/show/10959/intel-launches-7th-generation-kaby-lake-i7-7700k-i5-7600k-i3-7350k/3 Speed Shift v2: Speed Harder - Intel Launches 7th Generation Kaby Lake: 15W/28W with Iris, 35-91W Desktop and Mobile Xeo…

Speed Shift is also described in the Datasheets for 6th and 7th Generation processors in Section 4 - Power Management:

http://www.intel.com/content/dam/www/public/us/en/documents/datasheets/desktop-6th-gen-core-family-datasheet-vol-1.pdf http://www.intel.com/content/dam/www/public/us/en/documents/datasheets/desktop-6th-gen-core-family-datasheet-vol-1.pdf

http://www.intel.com/content/www/us/en/processors/core/7th-gen-core-family-desktop-s-processor-lines-datasheet-vol-1.html Datasheet, Vol. 1: 7th Gen Intel® Processor Family for S Platforms

It's interesting to note that as there's no distinction offered in BIOS between Speed Shift and SpeedStep, it can't be disabled or adjusted separately from SpeedStep. It may, however, be possible for Intel and it's motherboard partners to deploy microcode and BIOS updates to provide for different Speed Shift response levels and / or disable."

Concerning which liquid metal TIM used on my 7700K, it was Liquid Ultra.

CompuTronix

[TGrab1]

Hi CompuTronix,

I did read over your post a few pages back and my lack of response was more that I could not add anything further. I agreed from what I remember with the whole post and didn't feel I could add anything of meaning to a very informative post packed with useful information for people in this thread.

As to the question on liquid Ultra or pro. I would go with Ultra as it is much easier to remove and seems to be easier to work with in general.

I think that part of the problem is we are constantly shrinking die size and also increasing the ghz they are pushing. This creates a much more condense heat source that also has less surface area to dissipate that heat. These chips are being manufactured to withstand temps up to 100C.. so while coming from other chips that those huge temp spikes appear to be pretty alarming.. I just don't see them being dangerous to the lifespan of the chip. That is unless you are pushing 90C+ 24/7.

The problem is that this will reduce the amount of thermal head room for people looking to overclock. So in that case delidding is always and option or two play the silicon lottery until you get a good sample with a nice TIM application from factory.

At this point though overclocking is just a badge of honor. There are gains from the boost to 5.1ghz, but they aren't something you would ever notice in daily operation.. the only time you would notice would be in a CPU bound game which would provide a slight FPS boost or in a benchmarks of any sort. I guess my point is that currently there is nothing out there that you can run at 5ghz that wouldn't be equally as good at 4.5ghz.

[CCapt]

It appears that MSI has done a better job than many competing motherboard manufacturers concerning fan control.

The MSI Z270 Gaming M7 has 6 fully controllable fan headers which allow PWM or DC Modes, as well as 4 individual temperature setpoints from 0°C to 100°C with RPM from 0% to 100%. Fan speeds can also be adjusted by voltage, and have step up delay times at 0.1, 0.3, 0.5 and 0.7 seconds, and step down delay times at 0.1, 0.3, 0.5, 0.7 and 1.0 seconds.

I've tested the following settings with no fluctuation in fan RPM:

Fans headers 1 thru 6 step up & down at 0.1 seconds:

Setpoint 1 - 40% RPM @ 70°C

Setpoint 2 - 60% RPM @ 73°C

Setpoint 3 - 80% RPM @ 76°C

Setpoint 4 - 100% RPM @ 79°C

Although the test 7700K is delidded under high-end air, the worst spikes seldom reach 65°C. So for those who are experiencing spiking problems causing fan cycling issues on motherboards which are still eligible for RMA exchange, the MSI boards may be an option.

In the future, hopefully other motherboard manufacturers will deploy BIOS updates which provide more flexible fan controls than those currently available.

CompuTronix

[idata]

Thanks. I've read that Liquid Ultra hardens and degrades over a period of a year or so. Have you observed this?

Could you tell me why you didn't also use Liquid Pro or Ultra for your heatsink? This is what these are intended for, after all.

Cheers

Robert

[idata]

I'm surprised that you would think that it is OK to run the CPUs at near TjMax for hours when you also say that sustained temperatures over 80C are not recommended for reasons of stability and reliability and longevity.

At any rate I take it that your answer to my question is that repeated spikes near TjMax are nothing to worry about.

[AElib]

I've read that Liquid Ultra hardens and degrades over a period of a year or so.

Bs.

Me and some friends of mine had 3770K Delidded and OC'ed at 4.8 my was 5.0 and we never noticed degradation, even some of them still have their 3770K.

Note: i am talking about 3770K delidded 4/5 years ago.

[idata]

I hope you're right. Have you tried delidding any of these CPUs? Here is a user who wouldn't agree with you:

http://www.tomshardware.com/reviews/thermal-paste-performance-benchmark,3616-2.html Coollaboratory Liquid Ultra: Application - Thermal Paste Comparison, Part Two: 39 Products Get Tested (look for the comment from DANWAT1234, who says:

"Coollaboratory Liquid Ultra isn't all that good after a year of hard use. In fact, it completely hardens / dries. On my X9100 after 9 months of nearly 24/7 100% load, I started seeing high temps and after 1 year auto shut downs while crunching. Turns out it was shutting off because it hit the 105 C thermal protection.

Opened it up; thermal compound was as hard as a rock. has to pocket knife blade and sand it down.

So for longevity it sucks. That is something to consider, not just initial performance, but performance months and years down the road. Especially for laptops that aren't designed to be opened up frequently for repasting.

After trying Liquid Ultra many times and having it fail on me, I've put on Arctic MX-2 that has a supposed 8 year durability rating. Initial performance is great, we'll see how it lasts (been 3 weeks so far)."

[idata]

"At this point though overclocking is just a badge of honor. There are gains from the boost to 5.1ghz, but they aren't something you would ever notice in daily operation.. the only time you would notice would be in a CPU bound game which would provide a slight FPS boost or in a benchmarks of any sort. I guess my point is that currently there is nothing out there that you can run at 5ghz that wouldn't be equally as good at 4.5ghz."

This may well be true of many apps, but it's not true of image processing. For example, I've tested generating 1:1 previews of 47 images in Lightroom (from 42MPx raw images). At 4.0GHz the rendering took 240 seconds. At 4.8 GHz it took 210 seconds, so 14% for 8GHz. Assuming that the processing performance is linear (which it may not be, of course), an increase from 4.5GHz to 5.0GHz should give just under 9% decrease in rendering time. I would typically generate 1:1 previews for 100s of images at a time, so the time saving from a higher clock speed is very significant.

I haven't done tests myself, but if you look at tests by Puget Systems, this one, for example: https://www.pugetsystems.com/labs/articles/Adobe-Photoshop-CC-2017-AMD-Ryzen-7-1700X-1800X-Performance-907/ https://www.pugetsystems.com/labs/articles/Adobe-Photoshop-CC-2017-AMD-Ryzen-7-1700X-1800X-Performance-907/ you will see that overall, the 7700K beats the Ryzen 1800X on most tests, except for rendering and exporting, where the 1800X has an advantage because of its extra cores. However the advantage, even in Premiere Pro, is not that great:

A performance increase of 9% would equalize the 1800X and 7700K on both render and export, leaving the 7700K clearly ahead.

The extra speed would also make a difference to common tasks like filters as these are very compute-intensive and can be quite slow. It isn't just a question of having to wait a few extra minutes: the responsiveness (and therefore usability) of the application becomes much better.

Cheers,

Robert

[LRibe2]

Hi.

And that's not the only case here.

People keep saying that Ryzen is awesome and has a lot of cores.. and has great multi-core benchmarks and bla bla bla... But in real life situations it fails. Even besides gaming.

I'm a musician and I use Pro tools and Cubase. This following article crearly shows that besides the benchmarks, where everything looks sweet to ryzen 7, the real-time performance is a mess.

http://www.scanproaudio.info/2017/03/02/amd-ryzen-first-look-for-audio/ AMD Ryzen first look for audio production | Scan Pro Audio

Ryzen has architecture glitchs that bottlenecks performance using small buffers. This means that even with those multi core benchmarks on fire, when recording and listenig for example, a full band, ryzen will mess things up adding latency.. and you won´t get the job done as you would expect.

And I have this theory that the spikes everyone is talking about is linked with the ultra quick response this 7700k has wich is a good thing for real-time work lines.

Just wanted to leave this example so that people have in mind that more cores and benchmark results dont mean every thing about a CPU capabilities.

The real question to me is, do I prefer to save a few seconds while decompressing files or rendering video or do I prefer to have better performance in real-time, like gaming or produce video and audio with real-time export? I think a few seconds more wont bug me that much while rendering oe work large files.

Regards.

[AlHill]

Yes, the Ryzen has issues as well, as I posted here:

Doc

[TGrab1]

I have not ran into an issue of it drying out, but I only used liquid ultra between DIE and IHS. If you use it between IHS and block you run into several other issues. One being it can pump out, the second being higher chance of it making contact with something on your motherboard and causing a short. The last would be you have to make sure you go with a brass block, copper will work too, but aluminum will be ruined shortly with the application of liquid metal.. copper also tends to make the liquid dry out for whatever reason.

That is why most people just use the liquid metal for the die and something like thermal grizzly for the ihs>block.

[AElib]

Sorry for the late replay.

I hope you're right. Have you tried delidding any of these CPUs? Here is a user who wouldn't agree with you:

Yes i am right, as i already said the CPUs were Delidded and after 5 years the temps are the same, so about that user that wouldnt agree with me, maybe he had bad luck.

TGrable confirmed that he never ran into an issue of it drying out, i can can say the same thing, he also said he used liquid ultra between DIE and IHS, If you use it between IHS and block you run into several other issues.

he is right, liquid ultra should be used just between DIE and IHS those that use it between IHS and block are Noobs, between IHS and block i use Arctic MX-4 some others prefer Grizzly.

Edit: i was thinking that maybe and i say maybe it could dry depending if you use Watercooling or Aircooling, but i cant confirm that.

[idata]

Yes, it could be that this particular user had a copper block and this made the Liquid Ultra dry out (he also had high temperatures).

BTW ... ZeoxZariX ... the fact is that both Liquid Pro and Liquid Ultra are marketed specifically for IHS to block, not die to IHS, so it's not really fair to people who use it between the block and IHS to call them Noobs ... although the marketing literature really SHOULD say that the pastes should only be applied by expert users. Liquid Pro does appear to give the best result of all pastes if properly applied, with Liquid Ultra very close behind.

But having said that, the really excellent Tom's Hardware articles I posted on the previous page points out very well the dangers of using either paste for the cooler for a number of reasons ... and it also demonstrates that even very well applied, they do not give much of a performance increase over a paste like GC-Extreme from Gelid Solutions. Which is partly why I questioned the use Liquid Ultra/Pro between the die and IHS. It would seem to me that if the pastes are problematic between the block and IHS then they may also be problematic between the die and IHS (although not for reasons of aluminium/copper or shorting).

[TGrab1]

Mostly right, but there are some key differences.

There is a big difference between the characteristics needed for a paste on top of the IHS as apposed to below it.

The DIE as (most have been complaining about on the 7700k) heats up and cools down very quickly from one extreme to the other. These extreme temps shifts is really hard on normal TIM and causes it to break down and dry out rather quickly.

Then you have the fact that regular TIM is still prone to air pockets and doesn't do a great job filling all the micro crevices.

The next factor is thickness of the TIM which creates more thermal resistance.

The liquid metal does not have these weaknesses. It is able to withstand the temp shifts with little to no consequence. It is a liquid so air pockets and bubbles are not a big concern as they work their way out in the event you have one. The last is that it does a good job filling all the micro crevices. That is why when you look at the bottom of an IHS it will be stained from the liquid metal.. it is because the metal works into all those micro pores and that is why it looks stained.

Also remember liquid metal has a much higher thermal transfer. So on a surface like a die with very little contact area you need as much transfer as you can get.

[idata]

Thanks ... that makes sense. However I expect that the difficulty of application is still going to be a problem, as will air pockets. As it isn't really possible (I don't think?) to apply much pressure and as the IHS is not flat (according to the Tom's Hardware article), applying the right amount of liquid to properly fill the space between the die and IHS must be quite difficult. That's probably the main reason why I'm not too keen to try it myself.

[TGrab1]

To be perfect honest liquid ultra is pretty darn easy to apply. It isn't as easy as just putting a pea sized drop and putting on the block, but it isn't far off.

The thing with applying it to the die is just to use a little presetup. Put some clear finger nail polish or liquid electrical tape on all the short points around the die. Then what I do is get some clear tape and tape off around the die.. so that if I do brush some over the side it just goes on the tape

From there it is just down to applying a tiny little drop and covering the die by brushing it around with the brush.

One you are done peal off the tape and put on the IHS. I use a silicon gasket maker to bond the IHS and Chip back together... but others have used superglue (which i don't recommend) or they just let the socket clamp hold it on.

[idata]

Thanks for the info ... I might give it a go when I'm feeling brave .

This method: https://www.youtube.com/watch?v=y6KFzoFwYrE Kaby Lake 7700k delid - Delidding and applying liquid metal instead Intel TIM - YouTube seems very good to me. I've wondered about applying the liquid metal to both die and IHS and this is the first example I've seen doing it. He also leaves a small space for air expansion in the glue, which makes sense. What he doesn't do is to apply insulation to the short points like you do ... that would be a good idea.

He also doesn't use the cotton tip to spread the liquid metal, but the plastic shaft instead ... it seems to work well.

Cheers

Robert

[ssote]

TGrable wrote:

Hi CompuTronix,

I did read over your post a few pages back and my lack of response was more that I could not add anything further. I agreed from what I remember with the whole post and didn't feel I could add anything of meaning to a very informative post packed with useful information for people in this thread.

As to the question on liquid Ultra or pro. I would go with Ultra as it is much easier to remove and seems to be easier to work with in general.

I think that part of the problem is we are constantly shrinking die size and also increasing the ghz they are pushing. This creates a much more condense heat source that also has less surface area to dissipate that heat. These chips are being manufactured to withstand temps up to 100C.. so while coming from other chips that those huge temp spikes appear to be pretty alarming.. I just don't see them being dangerous to the lifespan of the chip. That is unless you are pushing 90C+ 24/7.

The problem is that this will reduce the amount of thermal head room for people looking to overclock. So in that case delidding is always and option or two play the silicon lottery until you get a good sample with a nice TIM application from factory.

At this point though overclocking is just a badge of honor. There are gains from the boost to 5.1ghz, but they aren't something you would ever notice in daily operation.. the only time you would notice would be in a CPU bound game which would provide a slight FPS boost or in a benchmarks of any sort. I guess my point is that currently there is nothing out there that you can run at 5ghz that wouldn't be equally as good at 4.5ghz.

Folks can run the processor at 99c 24/7 for up to 3 years, the processor is built for it and Intel stands behind it with there warranty. The i7 7700k processor is designed to throttle to 4.2GHz at 100c, that is why Intel advertise's 4.2GHz up to Max Turbo Frequency 4.5 GHz.

[DS3]

I actually opened this thread for my problem, the answer I've got there is that its fine.

[AElib]

and i bet it was intel_corp the jerk who answered your question.