Question Will 106-115c fry my cpu?

[Grouch19]

So I am not sure if I have thermal shutdown disabled in bios or what but hwinfo showed my max temp 106c on cores and 115c on package while running p95. And it did not shut down. It was probably like that 4-5 minutes before I noticed and stopped the test. And tbh this has happened twice now. The last time was with another utility. Is there anything to worry about going forward assuming I make sure nothing like this eve happens again?

 

[COLGeek]

That is way too hot. What are the specs for this system (make/model of all components)?

Is this a AMD Ryzen based system?

 

[Grouch19]

9700k 5ghz @ 1.275v Gigabyte z370 Gaming 7. Evga 1080ti sc2 hybrid. Corsair h115i gt cooler.
r15 it hits 84-86c max. LLC is set to extreme. On turbo there is vdroop so it's set to extreme. CPU-Z reads 1.272-1.296. Very seldomly it'll hit 1.308v but that's the max readout. It's fine until about 16 minutes in then it just goes crazy with the temps. I ran Aida 64 extreme cpu only for 12 hours and it maxxed out at 81c but p95 even the 26.6 version just gets so hot it's crazy. I just did a remount about an hour ago with liquid metal on the ihs. I had gc extreme tim before and my latest application was too thick. I couldn't run r15 without hitting 100c so I cleaned it off and used LM and now r15 doesnt hit 90c ever but ya 15-20 min into p95 it's hot af

Anyway thanks for the quick reply maybe I'll lap the ihs and cooler tomorrow and reapply. I work graveyard and i'm leaving right now to hit the gym before work so I won't be online again until morning.

 

[COLGeek]

The mid-80s are okay. 100+ is not.

Seems your OCs are fairly aggressive and are contributing to the heat. How hot is your GPU getting?

 

[Grouch19]

Gpu with a +75 core and 496 memory it never goes below 2000mhz. It usually sustains between 2012-2025 and stays well below 50c. This cooler sucks, it's very old, I'm about to build a loop just been working too much to find the time. I have just about everything I need. Sure 5ghz is fairly aggressive but I'm at a reasonably low voltage so I shouldn't be getting those kinds of temps on a blend test staying under 1.3v

 

[COLGeek]

Gpu with a +75 core and 496 memory it never goes below 2000mhz. It usually sustains between 2012-2025 and stays well below 50c. This cooler sucks, it's very old, I'm about to build a loop just been working too much to find the time. I have just about everything I need. Sure 5ghz is fairly aggressive but I'm at a reasonably low voltage so I shouldn't be getting those kinds of temps on a blend test staying under 1.3v

I recommend dropping the OC back a bit to not hit those temps again until you can improve your cooling situation. Those temps can lead to permanent damage. What is your airflow situation (number and orientation of case fans)?

 

[CompuTronix]

You're exceeding Throttle temperature above 100°C. The only way that's possible is if CPU Thermal Protection is disabled in BIOS, which should never be done. Check BIOS to make sure this extremely critical feature is not disabled.

Since you mentioned P95 v26.6 which is a valid 100% workload and doesn't have AVX, then AVX can't be causing the problem. Later P95 versions with AVX/2 can push 130% workload. CineBench R15 peeks at about 90% workload. Also, your Vcore is well within the maximum recommended value for 14 nanometer processors, which is 1.4 volts, so that's not the problem either.

Keep in mind that P95 v26.6 Blend changes FFT size at 16 minutes from memory bound to CPU bound, so it's normal for Core temperatures to increase at that point. AIDA64 has 4 CPU related stress test selections which have 15 possible combinations that yield 15 different Core temperatures. That's a lot of variables. The individual CPU test is only about 70% workload, whereas the CPU/FPU test combination is about 100% workload.

Most utilities have partial, fluctuating workloads with fluctuating Core temperatures. “Stress” tests vary widely and can be characterized into two categories; stability tests which are fluctuating workloads, and thermal tests which are steady workloads. Prime95 Small FFT's (without AVX) is ideally suited for testing thermal performance, because it conforms to Intel's Datasheets as a steady 100% workload with steady Core temperatures. No other utility can so closely replicate Intel's thermal test workload.

Utilities that don't overload or underload your processor will give you a valid thermal baseline. Here’s a comparison of utilities grouped as thermal and stability tests according to % of TDP, averaged across six processor Generations at stock settings rounded to the nearest 5%:

​

Although these tests range from 70% to 130% TDP workload, Windows Task Manager interprets every test as 100% CPU Utilization, which is processor resource activity, not actual workload. Core temperatures respond directly to power consumption (Watts), which is driven by workload. Prime95 Small FFT’s (without AVX) provides a steady 100% TDP workload.

Just for comparison, you might want to run CPU-Z > Bench > Stress CPU, which is a steady 80% workload. However, I suspect that your H115i is failing, which is all too common. Fortunately, Corsair has a 5 year warranty, so the option of an RMA replacement is available.

In the interim, if you have a decent air cooler, it can help you troubleshoot and rule out the variables.

And just so we know, what is your ambient temperature?

CT ​

 

[Grouch19]

I can’t tell you ambient room temp. I will look further into the bios tomorrow I’m not home anymore I’m at the gym lol. I specifically have my thermal throttling temps adjusted to 96c in bios so I don’t imagine I’d have the shutdown disabled but I may have. Aida just the cpu box checked. With fpu it got too hot. But as stated previously I remounted tonight and my r15 temps have decreased about 15c so I will try Aida with fpu/cpu tomorrow. Fans, the aio is up front of my define s2 in push/pull. Two ml 120 pro up top and two on the gpu rad in push/pull meaning my knly
Exhaust fan is on the rad. But all the pcie slots are vented so there should be plenty ways for the air to get out. I gotta go will explore further tomorrow and test more. Thank you

 

[Phaaze88]

LLC is set to extreme.

That's contributing to your temp problem. Turn that down to medium.
The high and extreme LLC settings are for LN2 users, or those who are liquid cooling their motherboard's VRMs, period.
If the cpu core is that hot, then the mobo's VRMs would be just as hot.

On turbo there is vdroop so it's set to extreme. CPU-Z reads 1.272-1.296. Very seldomly it'll hit 1.308v but that's the max readout.

Doubt it.
Cpu-Z can only refresh so often... it won't catch all the voltage spikes. The same goes for the lowest point of the vdroop.

 

[Grouch19]

So obviously I'm not going to do any more stress testing until I have better cooling. hwinfo64 read my vrm max at 88c. This cooler sucks, I walked away from p95 for a few minutes because everything looked good then after 16 min it got really hot and when I walked in with it at 19 minutes I stopped the test. I am leaving the load line where it's at. I have never used a setting where there is vdroop under load that's backward thinking to me. I am of the mindset that idling at a higher voltage than it needs under load is backwards. I have always had my cpu's idle at a lower vcore and have it increase under load as needed. But thanks.

No monitoring software is going to be perfect, what happened happened and since it happened I was pretty much just looking for someone to hold my hand and tell me everything is going to be ok. It won't happen again. At medium LLC there is about .04-.05v decrease under load. Turbo about .02-.025. Extreme is an increase of .02-.03v. Gigabyte often works that way. My vcore in bios is 1.275v. LLC alone with that low of vcore isn't going to fry anything. The day that happens I will buy a new cpu and motherboard and light it on fire in my living room.

 

[CompuTronix]

Grouch19,

You might want to reconsider your position concerning LLC. I am well experienced with the Gigabyte Aorus Z370 Gaming 7 motherboard. Although I agree that Gigabyte's implementation of LLC in this instance leaves a bit to be desired, the Vdroop value in Turbo of about 20 to 25 millivolts is not a bad thing, which can be compensated for by using an offset.

Not to insult your intelligence, but respectfully, and for the benefit of other Members and our visiting readers, the purpose of LLC is to maintain Vcore when at 100% workload, so it's consistent with the Vcore setting in BIOS. When LLC is set properly, Vcore in Windows should never "overshoot" BIOS, but should instead minimize Vdroop to within 24mV (millivolts) of BIOS or closer, as excessive Vdroop causes BSOD's when overclocking.

You might want to watch this YouTube: Explaining Load-Line Calibration (LLC) & Not Killing Your CPU

Also, I edited my previous post to include some additional information for you, so please re-read it.

CT

 

[Grouch19]

CT,

Thank you. I watch every GN video they upload along with a few more of my favorite tubers. I only recently went from 1.32v with a turbo llc to 1.275v with an extreme setting. I've only had this chip installed about 6 weeks and have been working too much to really get it dialed in. I will continue to find the settings, and play with an offset. My mindset was even if I had an increase of 100mV (which is pretty unrealistic) I would still be well within safe voltage specs. I switched my settings about a week ago.

I appreciate the video, I was gone for a couple years and didn't catch that. The last thing I would ever want to do is mislead and or misinform someone who is new to this stuff. What I do to my own stuff is my responsibility. That being said, I appreciate everyone's replies. Maybe this topic is better off on another thread. The Corsair Link ran in the background even after closing it. I opened the graph and it showed a max cpu temp of 98c . HWiNFO64 read the same in the third column which heads CPU (#0): Intel Core i7-9700k : DTS. The very next column labeled (enhanced) read a max of 100c for cpu package. 106c for cpu ia cores and gt cores. VRM MOS read a max 88c. I'm sure the cpu overheating may have contributed to everything else getting a little hotter than it normally would have.

Even right this second after only having my pc on for hours doing absolutely nothing with it but watching you tube videos, my max package temp is 56c. But I scroll down to about the fifth section in hw64 under the motherboard section out of nowhere it says cpu current 30c/min 25c/ max 112c. So that is an obvious misread right there. Again no software is perfect. I was pretty much looking to see if anyone else had once or twice let their cpu get this hot for a matter of a couple minutes and had any noticeable degradation. That video was helpful I will take heed.