Potential motherboard problems?

[Durixx]

Hello once again,

this forum has been very helpful recently in overcoming my PC building shortcomings, I'm turning to your advice once again.

I've tried to give my CPU a mild overclock of 4,5GHz just to squeeze in that extra power and to see if my cooler is capable of handling it. I've raised the vCore to 1.2V and tested it with Prime95. After a few minutes, the CPU underclocked itself to 3.5GHz, the temperatures at that time were steadily reaching 71*C.

After searching the internet for a while and trying various things (Turning off XMP, raising the voltage, lowering the clocks, turning off SpeedStep...) I've come across a similiar post which hinted at raising the Long and Short duration power limit in the bios to max. I've done just that and ran Prime95 for an hour or so.

This seemes to resolve the problem. The clocks stay at 4500MHz and the highest reached temperature is 79*C which - while not optimal - is more than expected considering my not so powerful cooler. What's more, after reaching that thermal apex, the temperatures seemed to lower down to 69*-72*C and stabilized around that range.

Although the problems seems resolved, questions remain:

Can I hurt my motherboard or CPU by raising the power limits? Is maximum of 79*C in Prime95 comfortable? I'm a complete beginner to overclocking so excuse the basal questions.

P.S.: As of now, I have the Intel SpeedStep turned off, should I turn it back on?

Specs:

MB: MSI Z370-A PRO
CPU: i5 8600k @ 4.5GHz
Cooler: SilentiumPC Fera 3
GPU: Gigabyte 1070 Ti
RAM: Kingston HyperX Predator DDR4 16GB (2x8GB) 3200MHz CL16
HDD: Seagate Barracuda 1TB
PSU: Corsair RM750x

tldr: Had to set my Long/Short duration power limit in my BIOS to max in order to not let my CPU underclock itself while stresstesting. Voltage is relatively low (1.2V) and OC is very moderate (4.5GHz). Will it cause me any problems in the future?

 

[androbourne]

79c is perfectly fine for that CPU. I have a 8700k OC'ed to 4.5ghz and get similar temps as well.

You really don't need to worry bout the CPU temps until you get around 85c or higher.

Raising the voltage generally causes more heat in the system which reduces lifespan. But you are barely adding voltage and keeping it at decent temps. You shouldn't have any problems.

 

[Darkbreeze]

79 is within tolerance, but in my opinion is too high.

I would RE-test using Prime95 version 26.6, choosing the Small FFT option. Newer versions with AVX instructions run unrealistic workloads and are not suitable for thermal testing. Normal AVX usage applications would never normally use the kind of workloads seen in Prime versions that use AVX instruction sets.

For stability, 8 hours of Realbench. Choose stress test, choose the amount of RAM you have installed, choose 8 hours, test. Anything less than a pass means there is an issue AND you want to do EVERYTHING you need to do with your CPU, BEFORE you ever change the memory settings from the default SPD configuration of 2133mhz and whatever else it defaults to. After the CPU is fully validated at it's configuration, THEN, and only then, should you enable the XMP settings or configure the memory manually.

Also, I've not had to make any of those adjustments you described on a basic overclock like yours. That should be reserved for very high end overclocks or possibly later fine tuning.

I'd reset to default BIOS settings and reconfigure, starting at whatever the max all core turbo boost is for your CPU and working your way up from there.

If your system is throttling, like you say it is, at only a 4.5GHz OC then there are likely other configuration problems or a problem with the VRMs. Maybe one of the VRM heatsinks is loose. Maybe the cooler is not pasted correctly. Maybe the case airflow is poor. That is actually where I'd look first and I'd download HWinfo, install it, run Prime and look at the VRM and other motherboard thermal sensors to see if something is jumping out. You can post screenshots here. Usually takes about three screenshots to catch all the sensor values and you want to run "sensors only" when you open HWinfo.

 

[SgtScream]

As long as you don't come close to 100c and your not exceeding max voltage for the CPU, you're absolutely fine.

 

[Durixx]

I've downloaded the 26.6 version, I'll post the HWInfo screens later today.

Just a side note that may be of interest - the post that has hinted on raising the power limits was about the very same CPU - Motherboard setup. Maybe a wider manufacturing defect?

 

[Durixx]

Here's the specs after running Small FFT on 26.6 version for 1 hour.
(Not really sure if my motherboard has VRM sensors)

https://imgur.com/a/TO5oAku

 

[Darkbreeze]

As long as you don't come close to 100c and your not exceeding max voltage for the CPU, you're absolutely fine.

I'm sorry, but this is completely inaccurate. You apparently need to read the Intel temperature guide and reconfigure your thought process. TjMax is 100 degrees. That means you've ALREADY gotten to the temperature that is BAD. Throttling is supposed to begin at that point, but if you reach TjMax it's more than just likely that you've already exceeded recommended VRM temperatures and VRM throttling has begun or has been happening.

Intel’s specification for DTS response time is 256 milliseconds, or about 1/4th of a second. Since Windows has dozens of Processes and Services running in the background, it’s normal to see rapid and random Core temperature “spikes” or fluctuations, especially during the first few minutes after startup, which should eventually settle. Any software activity will show some percentage of CPU Utilization in Task Manager, where unnecessary Tray items, Startups, Processes and Services that contribute to excessive or continued spiking can be disabled. For details see Section 13, Note 1.

Here's the nominal operating range for Core temperature:

Core temperatures above 85°C are not recommended.

Core temperatures below 80°C are ideal.

Section 6 - Throttle Temperature

Also called "Tj Max" (Temperature Junction Maximum), this is the Thermal Specification that defines the Core temperature limit at which the processor will Throttle (reduce Core speed and voltage) to protect against thermal damage. Processors that reach Throttle temperature can cause momentary hesitations in applications and frame stuttering in games.

Throttle temperatures are shown below for several popular processors, including "TDP" (Thermal Design Power) and idle Power, which are expressed in Watts (W).

Core

8th Generation 14 nanometer i7 8700K / i5 8600K (TDP 95W / Idle 2W),
7th Generation 14 nanometer i7 7700K / i5 7600K (TDP 91W / Idle 2W),
6th Generation 14 nanometer i7 6700K / i5 6600K (TDP 91W / Idle 2W):
Tj Max (Throttle temperature) = 100°C

5th Generation 14 nanometer i7 5775C / i5 5675C (TDP 65W / Idle 2W):
Tj Max (Throttle temperature) = 96°C

4th Generation 22 nanometer i7 4790K / i5 4690K (TDP 88W / Idle 2W),
4th Generation 22 nanometer i7 4770K / i5 4670K (TDP 84W / Idle 2W):
Tj Max (Throttle temperature) = 100°C

Legacy Core

3rd Generation 22 nanometer i7 3770K / i5 3570K (TDP 77W / Idle 4W):
Tj Max (Throttle temperature) = 105°C

2nd Generation 32 nanometer i7 2600K / i5 2500K (TDP 95W / Idle 4W):
Tj Max (Throttle temperature) = 98°C

1st Generation 45 nanometer i7 860 / i5 750 (TDP 95W / Idle 12W),
1st Generation 45 nanometer i7 920 D0 (TDP 130W / Idle 12W):
Tj Max (Throttle temperature) = 100°C

Core 2 45 nanometer Q9550 E0 (TDP 95W / Idle 16W),
Core 2 65 nanometer Q6600 G0 (TDP 95W / Idle 24W):
Tj Max (Throttle temperature) = 100°C

 

[Darkbreeze]

When you started HWinfo did you get a warning about a specific sensor potentially causing problems, and then choose to disable that sensor?

 

[Durixx]

No, nothing like that. I'd let you know otherwise, of course. Moreover, I've done some more digging, especially in regards to potential VRM problems and as it stands now, apparently, it's all about AVX Workload. This is the second thread regarding the MSI Z370 that recommends raising the power limits. What do you make of it?

Edit: Not really sure if I can post links to other sites. I can provide pictures of the thread otherwise.

Edit.2: The other thread is in my native language, but essentialy saying the same thing. The person, however, is trying to push 5.0GHz and increases voltage significantly. Although the increase of power limits helped, many people believed that if he wants to push it that hard, he should get himself a better board.

https://forum-en.msi.com/index.php?topic=297763.0

 

[SgtScream]

As long as you don't come close to 100c and your not exceeding max voltage for the CPU, you're absolutely fine.

I'm sorry, but this is completely inaccurate. You apparently need to read the Intel temperature guide and reconfigure your thought process. TjMax is 100 degrees. That means you've ALREADY gotten to the temperature that is BAD. Throttling is supposed to begin at that point, but if you reach TjMax it's more than just likely that you've already exceeded recommended VRM temperatures and VRM throttling has begun or has been happening.

Intel’s specification for DTS response time is 256 milliseconds, or about 1/4th of a second. Since Windows has dozens of Processes and Services running in the background, it’s normal to see rapid and random Core temperature “spikes” or fluctuations, especially during the first few minutes after startup, which should eventually settle. Any software activity will show some percentage of CPU Utilization in Task Manager, where unnecessary Tray items, Startups, Processes and Services that contribute to excessive or continued spiking can be disabled. For details see Section 13, Note 1.

Here's the nominal operating range for Core temperature:

Core temperatures above 85°C are not recommended.

Core temperatures below 80°C are ideal.

Section 6 - Throttle Temperature

Also called "Tj Max" (Temperature Junction Maximum), this is the Thermal Specification that defines the Core temperature limit at which the processor will Throttle (reduce Core speed and voltage) to protect against thermal damage. Processors that reach Throttle temperature can cause momentary hesitations in applications and frame stuttering in games.

Throttle temperatures are shown below for several popular processors, including "TDP" (Thermal Design Power) and idle Power, which are expressed in Watts (W).

Core

8th Generation 14 nanometer i7 8700K / i5 8600K (TDP 95W / Idle 2W),
7th Generation 14 nanometer i7 7700K / i5 7600K (TDP 91W / Idle 2W),
6th Generation 14 nanometer i7 6700K / i5 6600K (TDP 91W / Idle 2W):
Tj Max (Throttle temperature) = 100°C

5th Generation 14 nanometer i7 5775C / i5 5675C (TDP 65W / Idle 2W):
Tj Max (Throttle temperature) = 96°C

4th Generation 22 nanometer i7 4790K / i5 4690K (TDP 88W / Idle 2W),
4th Generation 22 nanometer i7 4770K / i5 4670K (TDP 84W / Idle 2W):
Tj Max (Throttle temperature) = 100°C

Legacy Core

3rd Generation 22 nanometer i7 3770K / i5 3570K (TDP 77W / Idle 4W):
Tj Max (Throttle temperature) = 105°C

2nd Generation 32 nanometer i7 2600K / i5 2500K (TDP 95W / Idle 4W):
Tj Max (Throttle temperature) = 98°C

1st Generation 45 nanometer i7 860 / i5 750 (TDP 95W / Idle 12W),
1st Generation 45 nanometer i7 920 D0 (TDP 130W / Idle 12W):
Tj Max (Throttle temperature) = 100°C

Core 2 45 nanometer Q9550 E0 (TDP 95W / Idle 16W),
Core 2 65 nanometer Q6600 G0 (TDP 95W / Idle 24W):
Tj Max (Throttle temperature) = 100°C

I agree with the data you posted because that data states exactly what I stated. "As long as you don't come close to 100c". 85c Isn't recommended because it's starting to come close to 100c, which is correct. Intel chips run hot due to their crappy TIM. Hopefully the 9th gen chips will be soldered like intel claims and we won't have to deal with forum post after forum post of people thinking their processors are running hot when they're not. His temperatures are fine for that cooler, running that particular stress test, with that overclock. The sole purpose of his build isn't to be running stress tests 24/7. It's probably just gaming and as you know gaming doesn't place anywhere near the same amount of stress on the CPU/VRMs that a prime 95 stress test does. I seriously doubt that quality MSI Z370-A motherboard's vrms are overheating, as It's a good quality motherboard. I think you're making a big deal over his temps, as most other people do when this topic of discussion comes up.

 

[Darkbreeze]

I'm not making a "big deal" out of it, I'm simply saying that it's higher than you would OPTIMALLY like to see. I realize he has an additional two cores over most historical i7's, even though his is an i5, but even my 6700k with it's four hyperthreads that his does not have (Which of course increases the workload on the cores that ARE there) and most i7's that I've worked with or on, were able to stay well below 75°C at 4.5Ghz.

So, regardless of the additional cores, it's still right at the point where you don't want to see it going any higher. The fact that his vcore is so low almost assures me that the OC is not stable. Especially if he hasn't even assigned a line load calibration yet to stabilize it when the voltage drops under heavy loads.

Again, I'd run Realbench for 8 hours. Choose the "Up to 16GB" option. If it passes, great, If not, you're going to likely need to increase the vcore and THAT is going to push thermals beyond where you could say "Yeah, it's still ok".

What I'd LIKE to know, is what case he's using, how many case fans are in use, what size they are, where they are populated and in what configuration they are in. A lack of case airflow on an otherwise within spec system, on a good board or not, especially when using a tower cooler (Rather than a top down model) is a prime candidate for VRM throttling regardless of whether you have in spec core temps or not. If there is not much airflow being exchanged through the case, we already know VRMs will throttle whether they are good enough for the OC or not because the CPU cooler, which in this case isn't a great example anyhow, isn't going to provide enough airflow over the heatsinks by itself.

Plus, without really decent flow through the case, the CPU cooler tends to just be pushing hot air around inside the case.

Or it might be totally fine and I don't know what the hell I'm talking about. I sort of doubt it though.

 

[SgtScream]

What makes you think a mild overclock of 200mhz would be unstable and have the necessity to alter load line calibration? He's barely even pushing the CPU and felt the need to start this post. The hardware he's using was designed handle more than a mild overclock and the CPU cooler he's using is designed to withstand up to 180w. From what I've read an overclocked i5 8600k to 5ghz barely hits 100w, so he must have inadequate case air flow, however at this point 200mhz really doesn't matter. It's Friday night. He probably got board, forgot about this thread and is too busy playing his games that don't push his hardware enough to be a threat thermally. Enjoy the rest of your weekend!

 

[Darkbreeze]

You're not actually that dense are you? I know you realize that the base clock for that CPU is 3.6Ghz. That means it's a 900mhz OC, not a 200mhz OC.

Nearly a full 1Ghz. A full time overclock, ANY full time overclock, absolutely (Insert expletive) requires stability testing. Saying anything to the contrary shows a fundamental lack of understanding for the actual process involved and the potential ramifications from an unstable system beyond simply having blue screens or other common and obvious indicators. Systemic corruption from micro-errors and potential loss of data are always factors. It doesn't matter if it's a 100mhz OC, it still requires verification of stability.

"But I just game". I hear this all the time. It doesn't matter. Micro errors can corrupt not just your OS, but all of your personal files if they are moved due to editing, or defragmentation or backup purposes. There are a cart full of reasons why it's incredibly rash to suggest that any OC is less, anything, than any other overclock. Even XMP memory configurations need to be tested. Just because they are reported to have been "mostly" stable configurations from the manufacturer doesn't mean that they are stable on YOUR specific chipset, motherboard or system. Anything higher than the JEDEC SPD configuration is technically an overclock of some form, and lately some of those can go as high as 2400mhz, so basically if your memory is 2666mhz or higher and you've enabled the XMP profile, it needs to be vigorously tested with Memtest86 and Prime Blend or Custom modes to make sure it's stable.

Manufacturers are WELL known for grossly overexaggerating the TDP capabilities of their coolers. That's a small single finstack, 120mm cooler. If that cooler can handle a 180w TDP cpu, well, I seriously doubt it can so I won't even bother with the rest. Just, too much nonsense here to even try and correct it all.

 

[Ilijas Ramic]

Prime is just for stability issues you will never push your cpu to 100% mostly of the time even while gaming soo its not a problem 79 is still acceptable but thats only prime temp. Try overclocking ram

 

[Durixx]

Look, I have turned off the XMP profile and I'll run RealBench for 12 hours on monday - to see if anything's wrong. So far I've found five threads solving the same issue on the very same motherboard, although that little note might have gone ignored. Such a common constant amidst so many variables leaves me buffled even more as to what the problem might be. As I said, the temperatures are fine. 72*C after hour of Prime95 is more than acceptable for me, especially considering the cooler and paste.

Now, just for the disclousure:

Case: Zalman R1
Airflow: 2 intake's in the front (stock Zalman, 120mm), 1 exhaust on the back (Cooler Master, 120mm). Have one spare but decided not to use it in order to keep positive air pressure.

Again, the temperatures are fine.

CPU @ 4,5GHz: 72*C (74*C) max after 1h in Prime95 (26.6)
GPU OC @ 150MHz cc, 350MHz mc: 69*C after 2h in Heaven Benchmark (1440P, everything on max)


To be honest, after reading all that, I'll probably drop the OC alltogether, not really worth the struggle nor my time. Just sorry I've wasted the additional expenses on 8600K when I could have grabbed the 8400.

 

[Darkbreeze]

Prime is just for stability issues you will never push your cpu to 100% mostly of the time even while gaming soo its not a problem 79 is still acceptable but thats only prime temp. Try overclocking ram

LOL. Ok. If you say so. I'm done arguing here. It's clear nobody is listening anyhow.

 

[Durixx]

If anyone's still interested:

https://imgur.com/a/F77ssnx

The temp screen was taken about 7,5 hours in.

 

[Ilijas Ramic]

Its ok idk why are you worried about it. You are fine. What cooler are you using ?

 

[Darkbreeze]

Yes, those temps are fine and that overclock can be considered stable.

Now you can move on to the memory since it's currently running at 2400mhz instead of 3200mhz which is it's advertised speed. Download Memtest86 (From Passmark) and create bootable USB media. Restart, go into the bios and enable the XMP profile or manually configure the memory timings, speed and voltage to those that match the advertised values or your own choices, then set your USB drive as the one time or primary boot device and run four passes of Memtest86.

Any errors, at all, means the memory configuration is not stable. No errors and it is fine.

 

[Darkbreeze]

Its ok idk why are you worried about it. You are fine. What cooler are you using ?

I'll tell you why he's worried about it, because he didn't take your trollish advice earlier to heart, which is good. You should shove off and go troll someplace else.