Computer randomly restarting when idle.

[iameatingjam]

It doesn't happen very often which makes it very hard to diagnose. A couple times it happened 3 times in one day, but sometimes it doesn't happen for two weeks at a time.

At first I thought it was the cpu as it came at around the same time the max vcore seemed to increase, and my previous configs that allowed me to keep it under 1.4 while maintaining max boost in games no longer worked. The thing is, I don't recall it being right after a ucode/bios update, which could have explained it, but it was definitely near to the time of one, as there's been a few lately. Maybe it needed a couple restarts for the effects to complete, or something, idk. But regardless, those two things together definitely made me suspect the cpu, especially since most of its life was spent on the supposed buggy microcodes.

But anyway.... So I got my replacement cpu and neither problem is solved... sigh..... But one at a time.

So, with a fresh cpu, I started looking at ram. I reduced clock speed. Tried increasing cpu side voltages, then dram voltage itself. It passes memtest86 and anta777 extreme. I don't think its the memory.

Another thing I'm thinking is.... maybe some of these c-states intel says to keep on are causing problems? I don't know much about the different c-states and how they work. Any suggestions about ones I could disable that might help in this situation would be helpful. I definitely do not want to turn them all off permanently, if it can be helped. If voltage is going too low I could perhaps try performance mode. I know thats another thing intel says not to do, but it should at least help the cpu from winding down too much if that is indeed the problem (as it has not happened under load even once). For now I've disabled all c-states I guess except c1e which intel now enforces. We'll see if that makes a difference and if it does maybe I can narrow it down from there.

I really want to avoid spending money if I don't have to, and if I do have to, I really don't want to be spending money on anything lga1700. If its a motherboard problem.... I don't know what I'll do. But... it could be the psu. But wouldn't a psu usually fail.... during periods of high load? Anyway I could probably stomach a psu upgrade, since at least its something that can be transferred to a new build.

Before each crash I get an event viewer message something like this, the numbers change, but its always one of these, event id 125.

Oh yeah I've also tried resetting cmos and running sfc /scannow and dism /online /cleanup-image /restorehealth a few times as well, with no luck. I just reinstalled windows last month. Really not in the mood to again.

 

[chrcoluk]

You can try disabling C6 state, if that stabilises the problem I would then be suspecting the PSU.

 

[Caring1]

This site basically reiterates what you already know:

The Kernel-Power Event Provider

It's a thermal event of some kind that causes a shutdown.
Check the NVME drives temps if possible

 

[eidairaman1]

Just save up is the suggestion at this rate since you dont want to pour anything into the setup you have

 

[iameatingjam]

You can try disabling C6 state, if that stabilises the problem I would then be suspecting the PSU.

Thanks... could you explain your thought process behind that? I don't doubt you, just want to understand. Is this something that could be tested with a psu tester or something...? I guess probably not eh?

Anyway I'll do that, as there were quite a few shutdowns yesterday, which is unusual in itself, but none since I turned all c-states off (except c1e). So I'll try just c6.

I do have a backup psu but its only 600w so I'll have to be real careful with the 4090 in there.

This site basically reiterates what you already know:

The Kernel-Power Event Provider

It's a thermal event of some kind that causes a shutdown.
Check the NVME drives temps if possible

Its got to be something else. Nothing is even close to being too hot when under load, and especially not when idle. I do keep an eye on it all. I have 6 ssds. They are between 25 - 35c. One is 52c but I think the sensor in that one is busted or something as it always says the same thing, be it summer or winter, with heatsink or without. But its the one running 3x1 right now through an adapter so I don't think it even has bandwidth enough to get too hot even if it wanted to.

Thats another thing I've been wondering about actually. My mobo has got a lot of shit connecting to it. The 6 ssds like I mentioned, the dvd drive, the LED light strip, the aio, 4 more fans and all the usb devices. The pcie to m.2 adapter. 3/4 sata connectors used. Every usb 2 header used, plus the usb 3. Maybe its too much for... something.

 

[ShrimpBrime]

I'll be honest I stared and thought for a long ass time before making this reply.

14700K. 230w or you've lost performance under-volting to
make up for the cooling issue, likely the
ALFII 240 cooler.
HUGE video graphics processing power at load, by it's self,
over half the peak rating of the
""Gigabyte 850w""

So here's you're specs. If you had read a forum user had asked this same very question, what are the 2 things you'd suspect first? I'm willing to bet you're under volting this cpu isn't doing favors anymore.
OK - SO

My thought is heat makes leakage. Leakage is compensated by more power, back and forth. You're under-volting, but it's just not quite stable. v-core VID under-volt is likely the problem. It was created over time perhaps. Can't call it degradation, you're not even running at spec yet.

My thought at this time would be to do all testing at system defaults, then evaluate the user settings before implementing them again. Maybe start the under-volt over again. I would just leave v-core and VID on auto and set LLC low and be done with it. You'll have to sacrifice performance with that cooler. Probably even if you just turned off e-cores and leaft P-cores on auto, that cooler might not struggle as bad, but that CPU LOVES everybit of 1.450v statically.

Oh, and I have a 14100F I can send you to test with. Ask in PM bout it. You won't have a thermal constraint and I've posted the CPU only twice. Essentially, it's brand new tested....

System Name	BarnacleMan
Processor	14700KF
Motherboard	Gigabyte B760 Aorus Elite Ax DDR5
Cooling	ARCTIC Liquid Freezer II 240 + P12 Max Fans
Memory	32GB Kingston Fury Beast
Video Card(s)	Asus Tuf 4090 24GB
Storage	4TB sn850x, 2TB sn850x, 2TB Netac Nv7000 + 2TB p5 plus, 4TB MX500 * 2 = 18TB. Plus dvd burner.
Display(s)	Dell 23.5" 1440P IPS panel
Case	Lian Li LANCOOL II MESH Performance Mid-Tower
Audio Device(s)	Logitech Z623
Power Supply	Gigabyte 850w

 

[iameatingjam]

I'll be honest I stared and thought for a long ass time before making this reply.

14700K. 230w or you've lost performance under-volting to
make up for the cooling issue, likely the
ALFII 240 cooler.
HUGE video graphics processing power at load, by it's self,
over half the peak rating of the
""Gigabyte 850w""

So here's you're specs. If you had read a forum user had asked this same very question, what are the 2 things you'd suspect first? I'm willing to bet you're under volting this cpu isn't doing favors anymore.
OK - SO

My thought is heat makes leakage. Leakage is compensated by more power, back and forth. You're under-volting, but it's just not quite stable. v-core VID under-volt is likely the problem. It was created over time perhaps. Can't call it degradation, you're not even running at spec yet.

My thought at this time would be to do all testing at system defaults, then evaluate the user settings before implementing them again. Maybe start the under-volt over again. I would just leave v-core and VID on auto and set LLC low and be done with it. You'll have to sacrifice performance with that cooler. Probably even if you just turned off e-cores and leaft P-cores on auto, that cooler might not struggle as bad, but that CPU LOVES everybit of 1.450v statically.

Oh, and I have a 14100F I can send you to test with. Ask in PM bout it. You won't have a thermal constraint and I've posted the CPU only twice. Essentially, it's brand new tested....

System Name	BarnacleMan
Processor	14700KF
Motherboard	Gigabyte B760 Aorus Elite Ax DDR5
Cooling	ARCTIC Liquid Freezer II 240 + P12 Max Fans
Memory	32GB Kingston Fury Beast
Video Card(s)	Asus Tuf 4090 24GB
Storage	4TB sn850x, 2TB sn850x, 2TB Netac Nv7000 + 2TB p5 plus, 4TB MX500 * 2 = 18TB. Plus dvd burner.
Display(s)	Dell 23.5" 1440P IPS panel
Case	Lian Li LANCOOL II MESH Performance Mid-Tower
Audio Device(s)	Logitech Z623
Power Supply	Gigabyte 850w

Yeah that was the first thing I did, almost to the letter. Everything is system defaults.... well except a few things like ram. But AC/DC is the board default of 90. No offsets. I was thinking of trying to go up to 110 if the c-states thing don't bear any fruit. Yeah the 4090 is undervolted. But I don't think thats relevant here. That wouldn't result in a reset. I'd get a black screen or artifacts or something.

Anyway its a new CPU and completely reset cmos so I don't think my configs on the old cpu still hold any bearing....- would they?

And oh thank you very much for the offer. but I actually have a 12100 non f as a backup worst comes to worst. Anyway heat really isn't really a problem for me. This cooler is very good, the fans spin 3300rpm if they need to and the radiator is nice and thick. Not to mention the cpu rarely has to work that hard as I cap all my games at 60 fps. I game at 4k but its usually either a light game, or I use dlss. The restrictions are moreso as to not pass the 850w power limit, and more recently, to preserve the life of the cpu.

Neither my cpu or gpu never really pass 60c unless I'm doing a benchmark. Usually my 4090 doesn't go over 200w. As the undervolt was at 2400mhz. But since ff16 is so demanding I've lifted the restrictions a |
bit and upped it to 2800mhz. Power still doesn't pass 230w or so at most, generally more in 1xx range. Its gigantic heatsink keeps it pretty cool.

--------

No resets YET today with state c6 and c7 disabled ( they were bundled together). But like I said sometimes I do something and think its fixed and then it shows up again 2 weeks later so it could take a long time to truly figure it out. The CPUs idle voltage and watts remain low and so does its temperature. I'm almost wondering... did anything change?

I could definitely notice the difference when I disabled all c-states. But I I'll try this first. Don't want to unnecessarily stress the cpu.

----------

Been going through the system logs and have noticed something that I thought was worth mentioning. This event, 125, does seem to come up occasionally and not be followed by a crash.

 

[Caring1]

Try the CPU voltage at default, or a slight increase as degradation might have affected it and it will require more to run stable.

 

[ShrimpBrime]

Yeah that was the first thing I did, almost to the letter. Everything is system defaults.... well except a few things like ram. But AC/DC is the board default of 90. No offsets. I was thinking of trying to go up to 110 if the c-states thing don't bear any fruit. Yeah the 4090 is undervolted. But I don't think thats relevant here. That wouldn't result in a reset. I'd get a black screen or artifacts or something.

Anyway its a new CPU and completely reset cmos so I don't think my configs on the old cpu still hold any bearing....- would they?

And oh thank you very much for the offer. but I actually have a 12100 non f as a backup worst comes to worst. Anyway heat really isn't really a problem for me. This cooler is very good, the fans spin 3300rpm if they need to and the radiator is nice and thick. Not to mention the cpu rarely has to work that hard as I cap all my games at 60 fps. I game at 4k but its usually either a light game, or I use dlss. The restrictions are moreso as to not pass the 850w power limit, and more recently, to preserve the life of the cpu.

Neither my cpu or gpu never really pass 60c unless I'm doing a benchmark. Usually my 4090 doesn't go over 200w. As the undervolt was at 2400mhz. But since ff16 is so demanding I've lifted the restrictions a |
bit and upped it to 2800mhz. Power still doesn't pass 230w or so at most, generally more in 1xx range. Its gigantic heatsink keeps it pretty cool.

--------

No resets YET today with state c6 and c7 disabled ( they were bundled together). But like I said sometimes I do something and think its fixed and then it shows up again 2 weeks later so it could take a long time to truly figure it out. The CPUs idle voltage and watts remain low and so does its temperature. I'm almost wondering... did anything change?

I could definitely notice the difference when I disabled all c-states. But I I'll try this first. Don't want to unnecessarily stress the cpu.

----------

Been going through the system logs and have noticed something that I thought was worth mentioning. This event, 125, does seem to come up occasionally and not be followed by a crash.

Well in the past, this was an issue with AMD 3000 series and ddr4 Vengeance memory kits running hynix memory. I do not think this is a very common issue with ddr5 and certain processors, but ya never know right? Just knowing this possibility, Power Down mode was a remedy of sorts. Not a cure, but the Kernel event ID 41 is a sudden loss of power, system was unexpectedly shut down.

4090 at 200w. My 4070 S pulls more than that stock lol. Are the tasks not demanding enough? Or are you really that low on the clocks and power(underclocked?)? TimeSpy Extreme, that card should be hitting 400-450w no problem

Cpu memory controller is running off the v-core I believe? Have you underclocked the cache frequency for the under-volt?

I am not running a radiator. I can game Cyber Punk 20 minutes and then I'm heat soaked. The modded waterblock weighs in at 360 grams all copper. It's vastly better than the HeatKiller IV (non pro, just a plate with plastic top) which the CPU hits 100c in maybe 10 to 20 seconds because the copper thermal storage is used up that quickly. Of course during a benchmark. Full load. That's the only way to test a coolers peak capability anyways.

You do have an interesting issue though. Frustrating I'm certain.

 

[iameatingjam]

Well in the past, this was an issue with AMD 3000 series and ddr4 Vengeance memory kits running hynix memory. I do not think this is a very common issue with ddr5 and certain processors, but ya never know right? Just knowing this possibility, Power Down mode was a remedy of sorts. Not a cure, but the Kernel event ID 41 is a sudden loss of power, system was unexpectedly shut down.

4090 at 200w. My 4070 S pulls more than that stock lol. Are the tasks not demanding enough? Or are you really that low on the clocks and power(underclocked?)? TimeSpy Extreme, that card should be hitting 400-450w no problem

Cpu memory controller is running off the v-core I believe? Have you underclocked the cache frequency for the under-volt?

I am not running a radiator. I can game Cyber Punk 20 minutes and then I'm heat soaked. The modded waterblock weighs in at 360 grams all copper. It's vastly better than the HeatKiller IV (non pro, just a plate with plastic top) which the CPU hits 100c in maybe 10 to 20 seconds because the copper thermal storage is used up that quickly. Of course during a benchmark. Full load. That's the only way to test a coolers peak capability anyways.

You do have an interesting issue though. Frustrating I'm certain.

Yeah, it is. I'm sure hoping all thats needed is some bios flip switch and not new components. Cause I'm not in a good position for that right now. I bet turning speed shift off would fix it (AKA go to performance mode). But I don't want to do that if I don't have to.

As for why my 4090 consumes so little power. Its just because I don't ask that much of it + undervolt and usually an underclock too. I cap my games at 60. I use dlss. I usually use graphics presets one down from the highest. I don't like to push any harder than I need to for a good experience. It certainly can use 450w if I remove the undervolt and remove the frame cap. but I don't know why I would want to do that, when it serves my needs very well as it is. I think the most watts I ever needed for a good experience at 4k60 was ~300w on alan wake II but I could be misremembering.

Anyway, what exactly was the problem with amd 3000 and ddr4? System restarts at idle? What do you mean by power down mode?

------

@ShrimpBrime

Well I looked it up and I see it is a power saving feature for ddr4, which I don't have in my bios. HOWEVER. It reminded me that I noticed in a recent bios revision, a ram setting 'memory timing mode' was changed from static to dynamic. I figured it was because of the instability and was there to give the imc a break, so figured it was a good thing since it didn't seem to affect memory speed while under load. But... sounds kinda similar. So I just turned it to fixed. Maybe that will help.

Try the CPU voltage at default, or a slight increase as degradation might have affected it and it will require more to run stable.

Even for a brand new cpu?

 

[ShrimpBrime]

Yeah, it is. I'm sure hoping all thats needed is some bios flip switch and not new components. Cause I'm not in a good position for that right now. I bet turning speed shift off would fix it (AKA go to performance mode). But I don't want to do that if I don't have to.

As for why my 4090 consumes so little power. Its just because I don't ask that much of it + undervolt and usually an underclock too. I cap my games at 60. I use dlss. I usually use graphics presets one down from the highest. I don't like to push any harder than I need to for a good experience. It certainly can use 450w if I remove the undervolt and remove the frame cap. but I don't know why I would want to do that, when it serves my needs very well as it is. I think the most watts I ever needed for a good experience at 4k60 was ~300w on alan wake II but I could be misremembering.

Anyway, what exactly was the problem with amd 3000 and ddr4? System restarts at idle? What do you mean by power down mode?

------

@ShrimpBrime

Well I looked it up and I see it is a power saving feature for ddr4, which I don't have in my bios. HOWEVER. It reminded me that I noticed in a recent bios revision, a ram setting 'memory timing mode' was changed from static to dynamic. I figured it was because of the instability and was there to give the imc a break, so figured it was a good thing since it didn't seem to affect memory speed while under load. But... sounds kinda similar. So I just turned it to fixed. Maybe that will help.


Even for a brand new cpu?

The power down mode deal was just an indicator the system didn't like the memory installed in older Ryzen systems. If you disabled power down mode, it prevents the memory from going into power saving mode. In some cases a little extra v-dimm or IMC voltage increase would help.

Anyhow, I wish to be more of a help in diagnostics, but without saying to fresh install windows to removing all drives except C drive, trying different memory settings or even a different set of memory, to even a different key board and mouse. Because literally anything plugged into the PC could cause the Event ID 41 to happen. Start with stuff you have to work with I suppose. Use the old spare keyboard for a week, could be as simple as that. Have seen stranger things......

Just got done playing War thunder myself. Havent played Alan Wake in a hot minute though.

 

[iameatingjam]

The power down mode deal was just an indicator the system didn't like the memory installed in older Ryzen systems. If you disabled power down mode, it prevents the memory from going into power saving mode. In some cases a little extra v-dimm or IMC voltage increase would help.

Anyhow, I wish to be more of a help in diagnostics, but without saying to fresh install windows to removing all drives except C drive, trying different memory settings or even a different set of memory, to even a different key board and mouse. Because literally anything plugged into the PC could cause the Event ID 41 to happen. Start with stuff you have to work with I suppose. Use the old spare keyboard for a week, could be as simple as that. Have seen stranger things......

Just got done playing War thunder myself. Havent played Alan Wake in a hot minute though.

So have I. Turned out it was my printer that was causing my pc to basically have a seizures every time it woke up. I thought it was some dldsr driver bug. Nope... it was the printer.

Well, its only been a day but there has been no reset since disabling c-state 6 and changing that memory setting, when there were several yesterday. However, that could just be chance since its so random. But anyway I'm not sure if its even working?

This is what the settings look like for me:

And then here's inside package c state:

This is what hwinfo shows me

Looks like C6 and C7 are still working... BUT even if I change package c-state to 10, and turn windows to power saving mode, it still doesn't show anything past 7. So I wonder if C6 and C7 both mean something deeper in this case. Is there better software for monitoring c-states?

Or I wonder if C8 and deeper simply doesn't apply to desktop computers and is more a laptop thing.... I really don't know. But as long as this setting doesn't result in a crash, I'll keep it how it is. As soon as there's' a crash, I'll try something else. Man this may take months to work out.

I know if I go to package and select C6, then only cstates below 6 will show in hwinfo. So if this current method fails, thats something else I could try I suppose. Though I would lose all benefit from ucode 0x12B.

Though on your point - yeah, you're right, my computer has so much shit attached to it, including all 6 ssds and the dvd drive, and so many usb devices. I could take it back to basics, but I paid good money for those ssds. I guess if it comes to that yeah, I'll try swapping more inconsequential stuff first, like the keyboard, I think I have a generic logitech hanging around somewhere, the paint is coming off this rgb POS anyway. I could probably go without the led light strip and dvd drive too ( even though I had to mod my case to make it fit -_-)

Anyway.... thanks for your input. I'll keep messing around until something works.

 

[BoggledBeagle]

It doesn't happen very often which makes it very hard to diagnose.

I'll keep messing around until something works.

Building a new computer is usually the easiest solution. With many parts reused and other sold it is not expensive at all.

Also - do not add all the old parts in the same time, add them slowly, one by one - for example the SSDs.

 

[chrcoluk]

Thanks... could you explain your thought process behind that? I don't doubt you, just want to understand. Is this something that could be tested with a psu tester or something...? I guess probably not eh?

Anyway I'll do that, as there were quite a few shutdowns yesterday, which is unusual in itself, but none since I turned all c-states off (except c1e). So I'll try just c6.

I do have a backup psu but its only 600w so I'll have to be real careful with the 4090 in there.




Its got to be something else. Nothing is even close to being too hot when under load, and especially not when idle. I do keep an eye on it all. I have 6 ssds. They are between 25 - 35c. One is 52c but I think the sensor in that one is busted or something as it always says the same thing, be it summer or winter, with heatsink or without. But its the one running 3x1 right now through an adapter so I don't think it even has bandwidth enough to get too hot even if it wanted to.

Thats another thing I've been wonder about actually. My mobo has got a lot of shit connecting to it. The 6 ssds like I mentioned, the dvd drive, the LED light strip, the aio, 4 more fans and all the usb devices. The pcie to m.2 adapter. 3/4 sata connectors used. Every usb 2 header used, plus the usb 3. Maybe its too much for... something.

When CPUs first came out with C6 support, it was known some PSUs couldnt handle it, its also the prime reason AMD added a C6 toggle in their AGESA options. C6 reduces the voltage to a very low level. Disable those C8/C10 options as well that are visible on your bios. Sadly these days you cant turn off whilst keeping C3 on which means its not a good day to day config, but useful to diagnose if the low voltages are triggering the problem.

Package c-states I would keep off anyway, the main thing it does (if C7 is enabled on CPU) is flush your l3 cache which hurts performance with very little power/heat saving. So package off, core C6 on for daily use, but turn core C6 off for the testing to see if the low idle voltages are the problem. If C6 off stabilises it, you then can either run daily with it off or put in a replacement PSU, at least the PSU could be reused for a future platform update and isnt LGA1700 specific.

 

[iameatingjam]

I was feeling cautiously optimistic this morning, which would have made 72 hours without a restart. But as soon as I woke my computer up, restart again. This is with all c-states on. I didn't want to lose the benefit of 0x12B and was hoping something else, like changing the ram timing mode, taking out the buggy usb 2 connectors, or the second cmos reset would magically be enough, but I guess not.

However, I have seen some people complaining about the same thing on the gigabyte forums

Here's an example:

https://shorturl.at/ae7e1

but I also saw it come up in a different thread. So it seems to be affecting others as well. That doesn't necessarily preclude a psu problem for all those affected, but still, now I'm wondering about it just being a bios bug.

Anyway, now I've disabled just C8 and C10. We'll see what happens now. I mean it still sucks. Now voltage goes down to .7 instead of .19.... but still... .7 can't be that harmful.... can it?

 

[eidairaman1]

Sounds like that mobo is a pita

 

[iameatingjam]

Sounds like that mobo is a pita

No doubt about that. Next time I'm definitely getting a better mobo but there's no way I'm pouring more money into lga1700 so for now I'm just gonna do what I have to, to make it work and ride it out as long as I can :/

Currently survived one day with c10 and c8 off. I think once I get to 3 weeks, I'll be confident its the solution. Will report back.

 

[iameatingjam]

One week no resets with c8 and c10 off. I think its probably the c8 one that really matters.

 

[eidairaman1]

One week no resets with c8 and c10 off. I think its probably the c8 one that really matters.

You might consider contacting gigabyte to fix it

 

[alcoholbob]

I had this happen to a DDR4 12700K system that I had for over a year at one point. I replaced transplanted the mobo + CPU to another working system including swapping the RAM, and it kept happening. I swapped the mobo, and it kept happening. I ended up changing the CPU to a 13900K and getting a contact frame. I always suspected it was excessive CPU bending from the ILM that was causing my issues.

 

[iameatingjam]

You might consider contacting gigabyte to fix it

I have been... don't even get me started on that ugh, they are impossible to talk to. Every case only gets one answer, and if you ask a new question/ or rather I was just trying to make a bug report, and I included the previous case numbers for context, they don't read them. Because after I already explained it was brand new chip in my first query, on my third, they suggested contacting intel for an rma. In my last one I just begged them to shut up and please just give the information to the team that works the bioses.

They also say things like " you shouldn't have to make any changes to get it working at default settings"

Like no shit thats why I said I was making a BUG REPORT.

-----

I have been scouring forums for people with this problem and asking them to report their problems directly to gigabyte. Seems people really like to complain on reddit but not actually make it to formal report. But I've at least convinced two people to do it. So thats something I guess.

 

[Waldorf]

not sure why you use stnby/sleep/hibernation at all.
less than 30 min idle just leave it running, more than that shutdown, since your using a desktop, not laptop.

fast boot setting off as well.

 

[iameatingjam]

not sure why you use stnby/sleep/hibernation at all.
less than 30 min idle just leave it running, more than that shutdown, since your using a desktop, not laptop.

fast boot setting off as well.

yeah fast boot is off. And its not just about idle, its also about light loads. Like say, posting on these forums. Thats actually when a lot of the restarts happened.

 

[Waldorf]

unless already done, remove any under/over on anything, add some little bit of Vs where it might be needed (like soc on amd), and see what happens.
the more stuff isnt "stock", the more possible points we have to verify

 

[ShrimpBrime]

unless already done, remove any under/over on anything, add some little bit of Vs where it might be needed (like soc on amd), and see what happens.
the more stuff isnt "stock", the more possible points we have to verify

14700KF is such a fast power hungry CPU, that I only encounter issues with trying to run UV, or decreased LLC at Lvl1. It hates Lvl 1 LLC. Random restarts and hangs seems to be mostly the issue, It'll pass a benchmark then hang right after or right at the end, say CBR24. Anything less than "auto" V-core, seems to be finicky with any LLC lower than LLC Lvl 3.

The idea is to work Loads at 85c or less, and with this particular CPU, it seems impossible with Liquid Freezer II 240 and just leaving it stock. Even on my custom loop, power limits raised or removed 100c is easy to accomplish on short order. But then gaming, doesn't ever hit that temp.

If this was mine, just avoid heavy loads and enjoy for gaming. Increase LLC to Lvl4 OC with XMP enabled and should be able to coax 4.6Ghz Cache from that, will bring some extra performance.

TLDR, more power not less.