ASUS ROG STRIX Z370-F + i7-8700K (New to Intel, need help setting up and OC)

[pnd]

Something is not right here guys. I only set Vcore voltage to 1.275 and started at 4,4 GHz and it jumps right into 100 degrees and starts thermal throttling right after starting Prime95.


Could I have installed something incorrectly?
I used the CPU installation tool exactly like this video:

The lever that you push down and lock felt a little bit harder to push down than I expected, but didn't feel like I was breaking anything.
Then I installed my Kraken x62 on it, without reapplying thermal paste as there were plenty on there from my last installation. The shop that sold me the motherboard and CPU said I should just do it that way.

Temperatures seem fine in idle. Mid 30's to 40. I was just playing PUBG where I had a major FPS boost from my Ryzen 5 - was running at around 70 degrees. (Without OC, just the stock settings with XMP for RAM)

 

[wingman99]

From your screenshot the Vcore is at 1.312v that is to high for 4.3GHz and you should reapply the thermal paste with only a pea sized drop in the middle.

 

[pnd]

From your screenshot the Vcore is at 1.312v that is to high for 4.3GHz and you should reapply the thermal paste with only a pea sized drop in the middle.

I figured that might be the case, so just tried with 4.5, 4.7 and 4.8. Same thing happens.

Could it be with this motherboard something is enabled I need to disable or something? I only set XMP on, cpu core voltage to 1,275, sync all cores and then the mentioned clock speeds. When I search for OC videos of this board, they are changing all kinds of things, but it might be to get to next level?

I will buy new thermal paste tomorrow and reapply. But even without thermal paste, this seems wrong doesn’t it?

I am a bit unsure about how hard to tighten the cooler head. I cross tightened it as best I could, and kind of tightened it kind of lightly with a screwdriver. Hard to explain, but I hope you get what I mean.

 

[wingman99]

4.3 to 4.8 will create more heat because the clock speed is higher. You need to reduce the core voltage for lower clock speeds to reduce the heat. Your prime95 FMA stock default temperatures with XMP is about right, it could just be the thermal paste application that is not efficient.

How to Apply Thermal Interface Material (TIM) LINK: https://www.intel.com/content/www/us/en/support/articles/000005576/processors.html tighten cooler head screws till stop.

 

[EarthDog]

1.3v+ is around where those clc coolers tend to crap out. 4.8 shouldnt need more than that. Likely 1.275v or less.

That said, you did put thermal paste on, right? Your words concern me a bit...

Perhaps addjust LLC from auto to its lowest level and see if that sto0s that voltage from raising on load.



Ill have this board up in friday just in case things dont get straight by then.

 

[pnd]

Hello guys,

Thank you for your help so far

Today I've bought Noctua NT-H1 thermal paste, and reseated my NZXT Kraken x62 with it, after thorough cleaning with isopropyl alcohol. I ran completely stock in the beginning to check everything was ok. However when I booted, the temperatures were NUTS. 100 degrees even, and the pump were making a weird sound. So I decided to pull the USB cable out of the pump, which immediately spun the pump to 100% along with 100% on the cooler fans. Temperatures quickly fell to about 30 degrees. Replugged the USB cable in the pump and I hear everything spin down, no weird sounds, and about 35 degrees. So something is a bit weird here? Think the AIO is dying or?

-------------------------------------------------------------

I might also have figured out why I could not overclock before. It seems if I have ASUS Multicore enhancements enabled, which is default, nothing works. So the settings I am testing out atm are:

XMP: Enabled
ASUS Multicore Enhancement: Disabled
CPU Core Ratio: 48
CPU SVID support: Disabled
Load Line Calibration: Level 6
Intel Speedstep: Disabled
Long Duration PKG Power Limit: 4095
Short Duration PKG Power Limit: 4095
CPU Core/cache voltage: 1.25v

So everything, expect Prime95 seems to run fine now. Cinebench does it's job, gaming works great. But I get fatal error in Prime95. Can you guys check this picture and point me in the right direction from here?

 

[EarthDog]

Cinebench runs for like 30 seconds... lol.. that isn't a measure of stability (but good it can pass it).

P95 is only using 6 of 12 threads... be sure you input 12 threads there where asked so it tests all of the CPU. As far as passing it, you likely need more voltage to do so. But we should really get the temps in order first...
EDIT: I see, workers dropped. Yeah, that isn't stable.

I would set it to 1.275V and see if that helps...

What RPM does it say your pump is running at when it is plugged in? You need to screw on that pump/cpu block pretty hard. It should be VERY hand tight. It should be a struggle to get off the screws by hand.

 

[pnd]

Cinebench runs for like 30 seconds... lol.. that isn't a measure of stability (but good it can pass it).

P95 is only using 6 of 12 threads... be sure you input 12 threads there where asked so it tests all of the CPU. As far as passing it, you likely need more voltage to do so. But we should really get the temps in order first...
EDIT: I see, workers dropped. Yeah, that isn't stable.

I would set it to 1.275V and see if that helps...

What RPM does it say your pump is running at when it is plugged in? You need to screw on that pump/cpu block pretty hard. It should be VERY hand tight. It should be a struggle to get off the screws by hand.

Na I know Cinebench isn't THE test, but I figured it was a good indicator whether I was at least getting somewhere.

I stepped up voltage to 1.275 and afterwards 1.3, same thing happens in P95. But everything else I tested runs with all those voltages 1.25-1.3.

Seems to me like my temperatures are fine now. I just ran ASUS RealBench for 15 minutes, and it was topping around 85 degrees. Or do you think I am still having thermal issues? I set the pump to run at 100%, I don't know where I can see RPM though. Fans are running around 1000 RPM at idle.

 

[EarthDog]

The pump shouldn't need to run at 100%.. somewhere less should be fine.

 

[wingman99]

@EarthDog the pump should auto regulate at default setings. Is there something wrong with it?

 

[EarthDog]

Not sure... I was going off of this.......

I ran completely stock in the beginning to check everything was ok. However when I booted, the temperatures were NUTS. 100 degrees even, and the pump were making a weird sound. So I decided to pull the USB cable out of the pump, which immediately spun the pump to 100% along with 100% on the cooler fans.

I think his pump is running 100%, but without more information from the OP, I have no idea if it is ramping up there or constant at 100%.

Yes, the issue seems to be resolved, but... it shouldn't run at 100% and we are not sure if it ramped down.

 

[Johan45]

Early on in the thread he was having issues with that CAM software, ould be related

 

[pnd]

Not sure... I was going off of this.......

I think his pump is running 100%, but without more information from the OP, I have no idea if it is ramping up there or constant at 100%.

Yes, the issue seems to be resolved, but... it shouldn't run at 100% and we are not sure if it ramped down.

Sorry I did not make that very clear. I set the CAM software to performance on the pump and fans. This means that the pump runs 100% from 60 degrees, ramps down under that. Since I only play games like 2-3 hours a day about 3 days a week, and never use the PC for anything else, I thought that would be good measure.

My thermals seem solid now though topping out around 85 degrees in P95.
However it took me a Core voltage of 1.35 to reach 4.8 GHz stable for 10 minutes in P95.

For some reason, core clock jumps between 4700 and 4800 in CPU-Z. Under full load it just stays at 4700. (Set to 4800 in BIOS)

Where should I go from here?

- - - Updated - - -

Okay damn, just realized that it is only when running blend that I get good result. When running the small packages it thermal throttles instantly at 100 deg.

 

[wingman99]

Do you mean Prime95 small FFT? If that is what your saying 100c temp would be because of the prime95 utilization speed I was describing early on in the thread, blend gives the processor time to cool off it runs larg FFT swaps with memory small FFT every 3 minutes. Then small FFT runs fast all the time. Things you can do is reduce the clock speed or use AVX offset so when any AVX program uses AVX it will reduce the clock speed or run prime95 without AVX or run blend and watch your temperature 10 minutes.

 

[pnd]

For the life of me, I can't get anything else stable than 4.8 GHz @ 1.3v.
I set AVX offset to 300 MHz too.
Am I doing something wrong, or did I loose the silicon lottery?

 

[wingman99]

For the life of me, I can't get anything else stable than 4.8 GHz @ 1.3v.
I set AVX offset to 300 MHz too.
Am I doing something wrong, or did I loose the silicon lottery?

What do you mean by anything else stable than 4.8GHz? Show a screenshot of what is going on. Also give details what your running and settings during stability failure.

 

[pnd]

Alright I'll try to provide as much as I can here:

To be honest, I have been messing so much around with settings. Looking at what you guys were saying, Googling everywhere etc. And I have been trial and erroring like that crazy too. This seems to be the closest to stable I can get. Been running fine for gaming.
On a side note, we are having extremely hot temperatures in Denmark at the moment. I probably got like 30 degrees in my computer room.

Can you guys have a look at my BIOS and results and guide me in the right direction maybe?

BIOS Settings:







Cinebench results:


RealBench Stress testing:


RealBench Crash during Benchmark:



Prime95 Blend:


Prime95 Small FFTs:

 

[wingman99]

Thanks for all the screenshots and information, your doing just fine with all your settings. looks like your processor needs increased Vcore over 1.3v+ for passing the stress tests at 4.8GHz if temperature will allow. Otherwise you could just test with RealBench, you have the old version without AVX, download the new version 2.56v. Prime95 is a lot harder to pass than RealBench with AVX. Your stock voltage for 4.3GHz was ~1.216v for prime95 so you would need increase AVX offset past 300Mhz and the voltage needed without AVX will still be high so it does not help much.

If your only going to game and possibly run more demanding programs like RealBench then you can make that choice how stable it should be for applications.

 

[pnd]

Thanks for all the screenshots and information, your doing just fine with all your settings. looks like your processor needs increased Vcore over 1.3v+ for passing the stress tests at 4.8GHz if temperature will allow. Otherwise you could just test with RealBench, you have the old version without AVX, download the new version 2.56v. Prime95 is a lot harder to pass than RealBench with AVX. Your stock voltage for 4.3GHz was ~1.216v for prime95 so you would need increase AVX offset past 300Mhz and the voltage needed without AVX will still be high so it does not help much.

If your only going to game and possibly run more demanding programs like RealBench then you can make that choice how stable it should be for applications.

Alright. I tried with 2.56. The benchmarking part of that version gives me Files modified or corrupted, no matter how I try to unzip the files. Seems like I am not the only one though. But stress testing works.

I reverted to everything completely stock in BIOS and ran everything. In P95 Small FTTs, I get weird warnings that my temperatures are thousands of degrees. This ONLY happens in P95 Small FTTs. I feel that test might be a little unnecessary, as it pushes everything so much further than all my games and benchmark softwares and everything?

Can you please check that everything looks ok for stock? Just to try and figure out why I can't really push this system.

Idle after boot:


RealBench 2.56 Stress Testing:


Prime95 Blend Testing:


Prime95 Small FTTs:

 

[Johan45]

First, looking through your BIOS setting, you have way too many things set manually, 95% of the time Intel will run fine with only the XMP(say yes to ASUS MCE), multiplier, and core voltage set accordingly. If it were me I would restore BIOS with F5 then set ONLY what I just suggested. Next, I still think you have some type of software problem which is causing those erroneous readings that Windows is reporting. I would remove any type of monitoring SW you have installed including CAM. You don't need it for the pump/fans to work, at least you shouldn't and only use HWMonitor to check temps/voltages. Personally I would wipe the SSD and start over but that's me.
Next you need to see if that 125° reading at TMPIN4 actually is real, if you have some compressed air in can for cleaning the PC that works or a fan will just not as quickly. Start with the heatsinks around the CPU and spray air(point a fan) on them then watch the temp if it goes down you have a VRM overheat issue and need a fan or more air through the PC. IF it doesn't change then it's likely not a real reading.

EDIT: Just had a thought, is that the same SSD that you were using on AMD and then just plugged it into your Intel Platform?