NUC8i5 Fan Speed after BIOS Update 0071

[Trial-Member]

Hi all.

Today I updated my BIOS to version 0071.

The initial version was 0066, which was withdrawn by Intel a few weeks ago (without comment or information to the customers).

Unfortunately Intel is cowardly and tries to sweep bugs under the carpet. So of course the release notes were removed instead of leaving them as they are and with a note that and why Intel decided to withdraw BIOS version 0066 (and at least one more), Chapeau Intel!

Seems that I will need to beginn archiving BIOS versions and release notes, the Intel homepage is apparently not a safe source for information. Even here in the forum Intel employees had tried to deny the existence of version 0066 until users proved the opposite.

Why I write is the following:

I did the update with the Power Button method, it ran without error message. However, I now notice that the fan turns up fully from time to time. There is no reason recognizable for me.

I use my system without high load (one to two browser windows, a text pad). In the browser, normal pages (news pages) are open, no video streams, no music streams. The CPU load varies between 2% and 5%, the CPU temperature is about 51 °C to 53 °C (read with Core Temp).

As you can see, nothing really exciting for an i5.

The fan then turns abruptly to full speed, remains constant for some seconds and then slowly decreases again. After a few minutes of rest the whole thing starts all over again ... abruptly full rpm, slow decrease.

This turning up of the fan is not accompanied by a significant increase of the temperature. Actually then also the fan should rise slowly, so it was at least so far, this abruptly rise does not fit to my settings.

These are my current settings:

I already reset my BIOS settings to the default values and reset them, unfortunately without success. Does anyone else have any ideas?

Trial-Member

[LeonWaksman]

Hi @Trial-Member​ 

1. Enter Bios settings Advanced > Performance > Processor and uncheck Intel Turbo Boost Technology.
2. Press F10 followed by "Y" to save settings and exit Bios.
3. Check how this settings affects fluctuation of blower RPM.

Leon

[Trial-Member]

Hi Leon,

I will try this. However, I somehow hope that this is not the solution. Why should a customer buy a NUC with an i5 processor and then cut its performance to keep the fan quiet? Then I could have bought a (cheaper) i3 right away.

Just as a reminder:

The problem didn't exist with version 0066 of the BIOS, it has only occurred since 0071.

Although in the past there were always rises in speed, but not these rises to full speed within a second. In the past, the speed slowly increased with rising temperature and then (as now) decreased even more slowly when the temperature dropped.

Directly after the start of Windows (a few seconds later) the first peak in the rpm is reached. At that time no applications are started (apart from the normal background processes, but this was absolutely identical before the BIOS update, there were no changes on my part).

I've changed the CPU settings in the past when watching video streams for a long time.

I noticed that the SkyGO app under Windows 10 is apparently problematic.

It takes the temperature to absurd heights (sometimes above 95 °C), although no special performance is actually required. Other streams (Netflix, Youtube) are far less hungry. Thats why I had experimented a bit and found that SkyGO runs great (and relatively quiet) when I deactivate Intel Turbo Boost Technology and limit the CPU cores to 3 (with 2 cores the image starts to jitter).

I admit that the change in the BIOS annoys me, but if I not only briefly look at a report, but would like to have a film or a sports broadcast while working (on another Computer) the effort is worth.

As I said, I like to try it out once I'm back home. But this can't be a final solution.

[LeonWaksman]

Hi @Trial-Member​ ,

1. You are right, disabling the Turbo Boost Technology is not a good solution. I agree with Scott that including a filter in software for temperature spikes is not difficult. However I'm not sure that this is the only fix that need to be implemented.
2. In my NUC7i7DNHE, I've suffered long time (more that half year), from noisy and fluctuating blower. The NUC become quiet and the blower RPM didn't fluctuate only after disabling the Turbo Boost.
3. On May 3 2019, Intel has release again Bios version for NUC7i7DNHE (knowing the blower problem long time ago). Now the situation is better, however the NUC generate blower noise from time to time. With Turbo Boost disabled, NUC is completely quiet. See the attached image showing graphs, taken while NUC was working in idle and during some Internet browsing - the Turbo Boost is enabled.

Leon

[n_scott_pearson]

First of all, Intel did explain (after a few days) that the drivers were retracted because they caused an issue where passwords placed on drives would interlock. The problem is that Intel does not have a standard place on the forums site where announcements about issues with the products can be placed and thus folks can easily miss them when they are made in responses to other folks questions. This is a procedural problem that can be easily fixed - if Intel has the gumption to do so. Personally, I think that the defect database should be visible to customers so that they know all of the issues that are outstanding and how long they have been outstanding. We'll never see this, of course...

As for the fan speed control, there is definitely a problem. When I hear the blower(s) accelerating on seemingly idle systems, I quickly look at temperatures and they are all low. If there was a spike of activity that caused the temperature to spike, this didn't last long enough to warrant the response that is occurring. Intel needs to add a dampening capability to the solution. A simple smoothing algorithm will suffice. This is nothing to implement (I know).

...S

[Trial-Member]

Hi Scott,

I think we both have a different understanding of open communication with the customer.

What Intel has done, you're right, is to write a forum post in which they pointed out in as few words as possible a problem that can occur if you activate the password protection function. But they did it only reluctantly, not proactively.

I am of the opinion that a forum like this is not a really good place to publish bug reports (startet by Intel), at least if you don't have an own section for it, which has to be structured according to model. After all, Intel (and any other company) should enable the customer to get this information quickly.

I rather think that Intel should publish this information in two places to make sure that the information is to find as easy as possible. The first place would be the overview page of the IDSA (which shows the status of the drivers etc.). But Intel has tried to hide that BIOS versions have been withdrawn. What was shown to the customer in the overview was the information that the system was up to date. Only those who clicked further through the entries could be surprised that they had a higher BIOS version than now offered. And even then there was no information about the dangers of the errors in the withdrawn versions. The second place would be the release notes. Here you can simply leave the previous information and add a note to the affected versions that and why the BIOS versions were withdrawn. The same could be done for drivers, if Intel has to withdraw a certain driver.

Now I can't compare for example which FW versions Intel ME etc. had with BIOS 0066 to find out where something was changed. I think it's wrong that Intel now simply hides the existence of 0066 (and at least one more). The version exists and is installed on an unknown number of devices. So everyone who still has 0066 is still affected by the bug, but can't easily figure it out. Not all customers participate in this forum.

Would it be enough to publish a Known Error Database? I'm not sure that would depend on how technical the descriptions are. For me, good communication with the customer is based on three pillars:

1. Do good and talk about it

2. Acknowledge your mistakes and

3. Speak the language (*) of your customers - and these are not always just the cracks. Intel should remember that the NUCs are also sold as completely finished systems, which are interesting for all users who decide against a tower PC, for example, because they have no room for a tin can.

(*) Of course, this does not refer to a geographical language, but to the way in which information is formulated.

[n_scott_pearson]

I disagree with almost nothing that you've said, but, after having worked at Intel for 21 years, I do know what isn't going to change (unfortunately) and I avoid (or have given up) suggesting them -- and, in some cases, I avoid even talking about them (too depressing?). In fact, I shouldn't even have asked for database exposure because I know that this will never happen -- and you're right; most folks wouldn't understand its contents at all (but I would!).

The real issue with the BIOS retraction is that it was a cluster all around. Retracting a BIOS that has a significant issue can be good, but when there are that many BIOS releases that need to be retracted, the percentage of users that will have at least one of these releases already installed is going to be significantly higher and thus this, at least as a single action, is not the right answer. Putting up an announcement indicating that this issue exists (and that folks should not use the password feature until it is fixed) is also good, but Intel has no (existing) way to ensure that all users will see the notification. A thought: perhaps, instead of retracting them, they lock them so that they cannot be downloaded (but still seen, including release notes) and information as to why they cannot be downloaded placed on its download page.

I have been pushing for a place where announcements can be put on the forums site, but I realize that this really isn't the best answer. There are simply too many users who have never visited (or rarely visit) the forums site and thus they wouldn't be reached. Well, there is one other possibility that comes (ok, just came) to mind: the Intel Driver and Support Assistant (IDSA) application. In the same way that it can notify folks of driver and BIOS updates that are available for their particular NUC model, it could notify folks of announcements related to their particular NUC model. I have to be honest: I do not use IDSA. I have samples of most NUCs and Compute Sticks and I keep them running all the time so I can watch for issues related to longer-term (powered on) operation and issues with updates from both Intel and (especially) Microsoft. It makes more sense for me to be downloading driver releases in bulk since I have so many units to update. As the same time, I forced myself early on to use IDSA to help with its issues, but reached the point of being so jaded with it that I stopped. If IDSA has this announcement system -- and the NUC/ICS team was actually using it -- I would then need to use it all the time (well, on at least one unit)...

Some comments on your list:

1. If only they had time for that (yes, *very* sarcastic remark).
2. But that's just giving additional fodder to the folks suing Intel (yes, also sarcastic - and a major societal problem).
3. If only they knew how to do so. Sometimes, even I forget and bombard folks with TLAs (that's Three Letter Acronyms) - which is an Intel thing (but I do this on purpose sometimes <g>). It's true though, the shipping of L10 systems (that's TLA for level 10 manufacturing - see https://www.amax.com/blog/?p=668 (sorry, best summary I could find) for more details) is a relatively new thing for the NUC organization and they, um, don't handle it well.

...S

[Trial-Member · Accepted solution]

@LWAKSMAN​ , @n.scott.pearson​ 

Disabling Intel Turbo Boost did not fix the problem.

Now I loaded the default values again, switched off the NUC and disconnected it from the power for a minute.

After that I restarted the NUC.

The high speeds didn't happen (but the default values for the fan are already bad enough).

I have now made my settings again. So far the NUC is now as quiet as before the BIOS update.

I'm curious if this will work.

Will post an update soon.

Trial-Member

[Trial-Member]

@LWAKSMAN​, @n.scott.pearson​ 

So, after a few hours of operation I can now say that the fans reacts as expected and doesn't turn up to full power any more without any reason.

Its not noiseless (but NUCs aren't that anyway), but just like before the update from BIOS 0066 to BIOS 0071.

Seems that a simple loading of the default values in the BIOS was not enough and it needed to be disconnected from the power supply.

Thank you very much for your help!

Trial-Member

[LeonWaksman]

1. So what are the setting you decide to go with? Default?
2. The Fan off capability have you leaved ON?
3. Can you attach an image? May be helpful for other users.

Leon

[Trial-Member]

I am currently using these settings:

(The screenshot is a few days old, but the settings are exactly these)

Minimum Duty Cycle 20 is ok and makes the NUC a little bit more silent during normal operation.

Fan off capability has no definable effect for me (the fan should switch off below 48 °C, the normal temperature during light use is 52 °C).

As I said, the NUC isn't completely silent, but I've realized in the meantime that I can't expect this from a NUC

[LeonWaksman]

Hi @Trial-Member​ 

1. I'm glad that your NUC runs more quietly now. It is also good that the problem was not caused by the Turbo Boost Technology function.
2. I'm attaching an image of cooling settings from my NUC. With those settings the NUC blower runs noiseless. Pay attention that I've disabled the Fan off capability and therefor blower doesn't stop at all. So, there are now temperature changes. Try this.

Leon

 Edit:

Forgotten to attach the image:

[Trial-Member]

Not far away from my settings (especially the Minimum Duty Cycle).

Due to your Fan Off Temperature the Fan should never stop even if you activate the Fan off capability unless you place your NUC inside of a Refrigerator 😉.

Edit: Typo

[AUser9]

I want an always-on silent box. This is what I have done with satisfaction in the BIOS settings:

- Turbo boost disabled (but I suspect this is not significant)

- Low power mode

- fan off at 45°C (but this is ignored, in fact the fan never stops)

CPU temperature: min 40°C, speed min 1%, increment 2%/°C

Memory  temper.: min 35°C, speed min 1%, increment 3%/°C

On an idle box at ambient temperature 21°C I read a core temperature around 42°C