[SOLVED] Intel i7 9700k - Can't pass any stress test due to high temps with Dark rock pro 4

[xKuLee]

"𝐈 𝐡𝐚𝐯𝐞 𝐡𝐚𝐝 𝐭𝐡𝐢𝐬 𝐜𝐩𝐮 𝐟𝐨𝐫 𝐪𝐮𝐢𝐭𝐞 𝐚 𝐰𝐡𝐢𝐥𝐞 𝐰𝐢𝐭𝐡 𝐝𝐚𝐫𝐤 𝐫𝐨𝐜𝐤 𝐩𝐫𝐨 𝟒 𝐚𝐧𝐝 𝐚𝐫𝐜𝐭𝐢𝐜 𝐦𝐱𝟒
𝐈𝐧 𝐭𝐡𝐞 𝐩𝐚𝐬𝐭 𝐢 𝐡𝐚𝐝 𝟒𝟕𝟗𝟎𝐤 𝐚𝐧𝐝 𝐢 𝐧𝐞𝐯𝐞𝐫 𝐡𝐚𝐝 𝐬𝐨 𝐦𝐚𝐧𝐲 𝐢𝐬𝐬𝐮𝐞𝐬 𝟗𝟕𝟎𝟎𝐤 𝐚𝐥𝐰𝐚𝐲𝐬 𝐠𝐢𝐯𝐞𝐬 𝐦𝐞 𝐰𝐞𝐢𝐫𝐝 𝐩𝐞𝐫𝐟𝐨𝐫𝐦𝐚𝐧𝐜𝐞 𝐡𝐢𝐭𝐭𝐢𝐧𝐠 𝟏𝟎𝟎% 𝐜𝐩𝐮 𝐮𝐬𝐚𝐠𝐞 𝐢𝐧 𝐬𝐨𝐦𝐞 𝐠𝐚𝐦𝐞𝐬 𝐚𝐧𝐝 𝐬𝐮𝐜𝐡

𝐑𝐞𝐜𝐞𝐧𝐭𝐥𝐲 𝐢 𝐡𝐚𝐯𝐞 𝐟𝐨𝐮𝐧𝐝 𝐬𝐨𝐦𝐞𝐭𝐡𝐢𝐧𝐠 𝐰𝐞𝐢𝐫𝐝 𝐢𝐧 𝐠𝐭𝐚 𝐯 𝐢 𝐰𝐨𝐮𝐥𝐝 𝐠𝐞𝐭 𝟗𝟎-𝟏𝟎𝟎% 𝐜𝐩𝐮 𝐮𝐬𝐚𝐠𝐞 𝐰𝐢𝐭𝐡 𝐭𝐮𝐫𝐛𝐨 𝐛𝐨𝐨𝐬𝐭 𝟒.𝟔𝐠𝐡𝐳+ 𝐛𝐮𝐭 𝐰𝐡𝐞𝐧 𝐢 𝐝𝐢𝐬𝐚𝐛𝐥𝐞 𝐭𝐮𝐫𝐛𝐨 𝐛𝐨𝐨𝐬𝐭 𝐢 𝐰𝐨𝐮𝐥𝐝 𝐠𝐞𝐭 𝟓𝟎-𝟔𝟎% 𝐜𝐩𝐮 𝐮𝐬𝐚𝐠𝐞 𝐭𝐡𝐚𝐭 𝐝𝐫𝐨𝐯𝐞 𝐦𝐞 𝐜𝐫𝐚𝐳𝐲 𝐢 𝐜𝐨𝐮𝐥𝐝𝐧'𝐭 𝐮𝐧𝐝𝐞𝐫𝐬𝐭𝐚𝐧𝐝 𝐰𝐡𝐲"

𝐖𝐡𝐲 𝐰𝐡𝐞𝐧 𝐦𝐲 𝐜𝐩𝐮 𝐫𝐮𝐧𝐬 𝐟𝐚𝐬𝐭 𝐠𝐢𝐯𝐞𝐬 𝐦𝐞 𝟏𝟎𝟎% 𝐜𝐩𝐮 𝐮𝐬𝐚𝐠𝐞 𝐚𝐧𝐝 𝐰𝐡𝐞𝐧 𝐢 𝐬𝐥𝐨𝐰 𝐢𝐭 𝐝𝐨𝐰𝐧 𝐭𝐨 𝐬𝐭𝐨𝐜𝐤 𝟑.𝟔𝐆𝐡𝐳 𝐧𝐨 𝐛𝐨𝐨𝐬𝐭 𝐢𝐭 𝐰𝐨𝐫𝐤𝐬 𝐰𝐢𝐭𝐡 𝟓𝟎-𝟔𝟎% 𝐜𝐩𝐮 𝐮𝐬𝐚𝐠𝐞"

F7 Load optimized bios defaults will fail primetest at 100c temp throttling

Lowering LLC too much gives cpu errors on primetest

Lowering clocks to 4.3Ghz and Lowest LLC/ADC/DC can run at arround 1.20v idle and 1.14 under prime load this test passed over 1h at 82c

Running no intel boost at 3.6GHZ runs at 1v and pass all tests at 50-60c

I see people using this cpu at 5GHZ 4.9 4.6 and i can barely run it at 4.3ghz anything above 1.14 under load will give me 100c
I am using dark rock pro 4 a super beefy and known cooler and arctic mx 4 what is going on with my temps guys?
Also have you ever heard higher clocks giving you 100% cpu usage than lower clocks?
z390 Aorus Elite - intel i7 9700k - GTX 1660 ti - 16GB DDR4 3200mhz - 750w plat+ corsair psu

 

[Koekieezz]

"𝐈 𝐡𝐚𝐯𝐞 𝐡𝐚𝐝 𝐭𝐡𝐢𝐬 𝐜𝐩𝐮 𝐟𝐨𝐫 𝐪𝐮𝐢𝐭𝐞 𝐚 𝐰𝐡𝐢𝐥𝐞 𝐰𝐢𝐭𝐡 𝐝𝐚𝐫𝐤 𝐫𝐨𝐜𝐤 𝐩𝐫𝐨 𝟒 𝐚𝐧𝐝 𝐚𝐫𝐜𝐭𝐢𝐜 𝐦𝐱𝟒
𝐈𝐧 𝐭𝐡𝐞 𝐩𝐚𝐬𝐭 𝐢 𝐡𝐚𝐝 𝟒𝟕𝟗𝟎𝐤 𝐚𝐧𝐝 𝐢 𝐧𝐞𝐯𝐞𝐫 𝐡𝐚𝐝 𝐬𝐨 𝐦𝐚𝐧𝐲 𝐢𝐬𝐬𝐮𝐞𝐬 𝟗𝟕𝟎𝟎𝐤 𝐚𝐥𝐰𝐚𝐲𝐬 𝐠𝐢𝐯𝐞𝐬 𝐦𝐞 𝐰𝐞𝐢𝐫𝐝 𝐩𝐞𝐫𝐟𝐨𝐫𝐦𝐚𝐧𝐜𝐞 𝐡𝐢𝐭𝐭𝐢𝐧𝐠 𝟏𝟎𝟎% 𝐜𝐩𝐮 𝐮𝐬𝐚𝐠𝐞 𝐢𝐧 𝐬𝐨𝐦𝐞 𝐠𝐚𝐦𝐞𝐬 𝐚𝐧𝐝 𝐬𝐮𝐜𝐡

𝐑𝐞𝐜𝐞𝐧𝐭𝐥𝐲 𝐢 𝐡𝐚𝐯𝐞 𝐟𝐨𝐮𝐧𝐝 𝐬𝐨𝐦𝐞𝐭𝐡𝐢𝐧𝐠 𝐰𝐞𝐢𝐫𝐝 𝐢𝐧 𝐠𝐭𝐚 𝐯 𝐢 𝐰𝐨𝐮𝐥𝐝 𝐠𝐞𝐭 𝟗𝟎-𝟏𝟎𝟎% 𝐜𝐩𝐮 𝐮𝐬𝐚𝐠𝐞 𝐰𝐢𝐭𝐡 𝐭𝐮𝐫𝐛𝐨 𝐛𝐨𝐨𝐬𝐭 𝟒.𝟔𝐠𝐡𝐳+ 𝐛𝐮𝐭 𝐰𝐡𝐞𝐧 𝐢 𝐝𝐢𝐬𝐚𝐛𝐥𝐞 𝐭𝐮𝐫𝐛𝐨 𝐛𝐨𝐨𝐬𝐭 𝐢 𝐰𝐨𝐮𝐥𝐝 𝐠𝐞𝐭 𝟓𝟎-𝟔𝟎% 𝐜𝐩𝐮 𝐮𝐬𝐚𝐠𝐞 𝐭𝐡𝐚𝐭 𝐝𝐫𝐨𝐯𝐞 𝐦𝐞 𝐜𝐫𝐚𝐳𝐲 𝐢 𝐜𝐨𝐮𝐥𝐝𝐧'𝐭 𝐮𝐧𝐝𝐞𝐫𝐬𝐭𝐚𝐧𝐝 𝐰𝐡𝐲"

𝐖𝐡𝐲 𝐰𝐡𝐞𝐧 𝐦𝐲 𝐜𝐩𝐮 𝐫𝐮𝐧𝐬 𝐟𝐚𝐬𝐭 𝐠𝐢𝐯𝐞𝐬 𝐦𝐞 𝟏𝟎𝟎% 𝐜𝐩𝐮 𝐮𝐬𝐚𝐠𝐞 𝐚𝐧𝐝 𝐰𝐡𝐞𝐧 𝐢 𝐬𝐥𝐨𝐰 𝐢𝐭 𝐝𝐨𝐰𝐧 𝐭𝐨 𝐬𝐭𝐨𝐜𝐤 𝟑.𝟔𝐆𝐡𝐳 𝐧𝐨 𝐛𝐨𝐨𝐬𝐭 𝐢𝐭 𝐰𝐨𝐫𝐤𝐬 𝐰𝐢𝐭𝐡 𝟓𝟎-𝟔𝟎% 𝐜𝐩𝐮 𝐮𝐬𝐚𝐠𝐞"

F7 Load optimized bios defaults will fail primetest at 100c temp throttling

Lowering LLC too much gives cpu errors on primetest

Lowering clocks to 4.3Ghz and Lowest LLC/ADC/DC can run at arround 1.20v idle and 1.14 under prime load this test passed over 1h at 82c

Running no intel boost at 3.6GHZ runs at 1v and pass all tests at 50-60c

I see people using this cpu at 5GHZ 4.9 4.6 and i can barely run it at 4.3ghz anything above 1.14 under load will give me 100c
I am using dark rock pro 4 a super beefy and known cooler and arctic mx 4 what is going on with my temps guys?
Also have you ever heard higher clocks giving you 100% cpu usage than lower clocks?
z390 Aorus Elite - intel i7 9700k - GTX 1660 ti - 16GB DDR4 3200mhz - 750w plat+ corsair psu

simply u need a better cooler for a 8 CORES cpu, because your 4790K is 4 CORES only.

 

[xKuLee]

simply u need a better cooler for a 8 CORES cpu, because your 4790K is 4 CORES only.

I just mentioned 4790k cuz its the last cpu i used i don't use the same cooler as back then, i am using Dark Rock pro 4 you think that is not enough? thanks for your reply

 

[Phaaze88]

Whoa... probably best you just start all over.

1)Restore bios defaults. After, go back in and set Gigabyte's Enhanced Multi Core Performance to disabled.
Found in Advanced Mode, under the Tweaker tab.

2)Was Windows reinstalled after the major upgrade? If it wasn't reinstalled, you need to do that.
After that, install/update the motherboard chipset, lan, audio drivers from the board maker's site.

3)Leave Prime alone for the time being.

4)We need to understand the cooling situation a little better.
What is the make and model of the chassis?
What is the fan arrangement? For reference:

Spoiler

Might be clearer to take some pictures and post them here. [Use imgur.com to upload images, and remove the side glass panel before taking pics; reflections are unwanted.]


5)𝟏𝟎𝟎% 𝐜𝐩𝐮 𝐮𝐬𝐚𝐠𝐞
𝟓𝟎-𝟔𝟎% 𝐜𝐩𝐮 𝐮𝐬𝐚𝐠𝐞 (turbo boost disabled)
Don't look at cpu usage like that - the first one is an exception, though. There are multiple cores/threads, and these must be examined separately. Games are bound by a single core, which handles basically everything off the game engine - even commands to other cores!
With turbo boost disabled, you slowed down the Primary Thread. It is likely bumping into 100%.

 

[xKuLee]

h1MmhEZ.png (900×405) (imgur.com)

I got a picture for you all fans shoot air towards left the case is Corsair Carbide Clear 400C

20160202133919_15c77e66.jpeg (1099×1163) (scdn.gr)

thanks for your reply

 

[xKuLee]

Also i set the fans full speed 100% (since i found out the overheating issue) but no difference

Also i haven't done an upgrade now i build the computer new ssd new installs etc and used it for 1-2 years i just now realised the temps issue i never stressed it i just had some hikups in games that annoyed me but never realised it could be hardware issue since some games are badly optimized i just realised i cant pass stress tests

 

[xKuLee]

5)𝟏𝟎𝟎% 𝐜𝐩𝐮 𝐮𝐬𝐚𝐠𝐞
𝟓𝟎-𝟔𝟎% 𝐜𝐩𝐮 𝐮𝐬𝐚𝐠𝐞 (turbo boost disabled)
Don't look at cpu usage like that - the first one is an exception, though. There are multiple cores/threads, and these must be examined separately. Games are bound by a single core, which handles basically everything off the game engine - even commands to other cores!
With turbo boost disabled, you slowed down the Primary Thread. It is likely bumping into 100%.

Also i didn't mean that i get 100% cpu usage when i disable turbo boost its vice versa my cpu usage drops when i disable turbo boost so gta doesn't stutter anymore but when i turbo boost or overclock to 4.6+ i get the 100% cpu usage but that could be completely random and irrelevant

As about temps i haven't noticed any weird temps while gaming just the stress tests XTU and prime and OCCT

 

[xKuLee]

no9Qdll.png (1920×1080) (imgur.com)

look at this i just run the game im not moving im not doing anything tasking and the cpu is already almost maxing

 

[uWebb429]

@xKuLee
The first two tabs of the Task Manager report CPU Utilization. When Intel CPUs have Turbo Boost enabled, this is not the same as CPU usage. The Task Manager graphs are Utilization not usage.

Only the Details tab shows CPU usage data.

MX-4 can pump out at high temperatures. Try Noctua NT-H2 or a similar paste.

 

[xKuLee]

@xKuLee
The first two tabs of the Task Manager report CPU Utilization. When Intel CPUs have Turbo Boost enabled, this is not the same as CPU usage. The Task Manager graphs are Utilization not usage.

Only the Details tab shows CPU usage data.

MX-4 can pump out at high temperatures. Try Noctua NT-H2 or a similar paste.

Thanks for your reply i noticed that mx4 dries out a lot i remember using thermal paste once a year or more nowdays mx4 dries out easy thanks for the recommendations

 

[Karadjgne]

I've never known MX-4 to dry out. AS5 yes, MX-4 no. That paste is good enough that it should last 6+ years before it starts loosing thermal conductivity due to chemical breakdown.

Dried out paste is just as good as when it was wet, as long as it's applied wet. The 'paste' part doesn't actually have any thermal conductivity properties, if anything it's closer to an insulator. What it really is is a medium to contain the silicate, silver, carbons, diamonds or whatever other substrates are used that actually do transfer heat. The wet or dry of the paste doesn't change the ability of the substrates to pass heat.

The only issue with dried out pastes (like AS5) is when the cooler suffers a hit and the dried paste looses whatever surface cohesivness it had. Dry won't reattach, wet will.

A 9700k at stock can't overpower a DRP4. It takes stupidity on the part of the motherboard vendor by auto setting MCE default enabled to come close. The 9700k is @ 125w with turbo enabled under all core full load. The DRP4 is a 250w cooler. It takes OC to 5.0GHz+ all core to hit the 200w mark

Stop messing around with LLC. Leave it on Auto if not overclocking. LLC is an added voltage. In the instant after the cpu grabs a ton of power, there's a power drop (vdroop), LLC adds a small-large amount of voltage to cover that gap, so when the cpu grabs the next amount of power, it's covered and you don't crash due to lack of voltage. The side affect is that LLC is an Added voltage, it gets applied to the peak draw as well as the low, so if LLC is 0.3v and your vcore is set for 1.3v, instead of the cpu hitting 1.0v, LLC covers the gap and cpu remains at 1.3v. But the peak, when vcore Is 1.3v, the cpu gets 1.6v instead. Which drives temps seriously high.

So leave the cpu to decide for itself how much LLC is necessary for stability.

My best guess as to temps cause? Motherboard is set to PL2 on a semi-permanent basis (performance mode) combined with cooler mount issues and high cpu voltages due to LLC or other voltage tampering.

 

[xKuLee]

I've never known MX-4 to dry out. AS5 yes, MX-4 no. That paste is good enough that it should last 6+ years before it starts loosing thermal conductivity due to chemical breakdown.

Dried out paste is just as good as when it was wet, as long as it's applied wet. The 'paste' part doesn't actually have any thermal conductivity properties, if anything it's closer to an insulator. What it really is is a medium to contain the silicate, silver, carbons, diamonds or whatever other substrates are used that actually do transfer heat. The wet or dry of the paste doesn't change the ability of the substrates to pass heat.

The only issue with dried out pastes (like AS5) is when the cooler suffers a hit and the dried paste looses whatever surface cohesivness it had. Dry won't reattach, wet will.

A 9700k at stock can't overpower a DRP4. It takes stupidity on the part of the motherboard vendor by auto setting MCE default enabled to come close. The 9700k is @ 125w with turbo enabled under all core full load. The DRP4 is a 250w cooler. It takes OC to 5.0GHz+ all core to hit the 200w mark

Stop messing around with LLC. Leave it on Auto if not overclocking. LLC is an added voltage. In the instant after the cpu grabs a ton of power, there's a power drop (vdroop), LLC adds a small-large amount of voltage to cover that gap, so when the cpu grabs the next amount of power, it's covered and you don't crash due to lack of voltage. The side affect is that LLC is an Added voltage, it gets applied to the peak draw as well as the low, so if LLC is 0.3v and your vcore is set for 1.3v, instead of the cpu hitting 1.0v, LLC covers the gap and cpu remains at 1.3v. But the peak, when vcore Is 1.3v, the cpu gets 1.6v instead. Which drives temps seriously high.

So leave the cpu to decide for itself how much LLC is necessary for stability.

My best guess as to temps cause? Motherboard is set to PL2 on a semi-permanent basis (performance mode) combined with cooler mount issues and high cpu voltages due to LLC or other voltage tampering.

The only reason i play with LLC is to decrease voltage and not to increase as on auto my cpu asks ridiculous voltages 1.38-1.45 at stock

Thanks for that long reply its helpful

 

[Karadjgne]

You won't decrease voltage with LLC. To decrease voltage use VID/offset. And 1.38-1.45 isn't rediculous, single cores will use higher voltages than multi cores when no static OC is present because the boost is higher.

Silicon varies, each cpu has unique properties when it comes to voltages actually used, what's demanded, what's needed, so both Intel and Amd err on the side of caution by setting defaults higher than absolutely necessary, this means it doesn't matter which cpu, they are all covered and guaranteed to remain stable. So you'll see 1.45v demand even if the cpu actually only uses 1.28v.

Vcore is use, vid is supply. By lowering vid/negative offset, it lowers supply and forces the cpu to demand less. So you'd see 1.35v with a -0.1v offset but cpu still only use 1.28v.

Using LLC, if you did a -0.2v offset, that'd put supply at 1.25v, so LLC would be applied (auto) to bring voltage up higher than 1.28v. If you set a low LLC, it might not cover the gap, and pc crashes.

It's trixy, and time consuming, especially when testing gets involved, but you'd need to know exactly what the cpu actually uses, which changes per load. But now you are basically into OC territory.

 

[xKuLee]

You won't decrease voltage with LLC. To decrease voltage use VID/offset. And 1.38-1.45 isn't rediculous, single cores will use higher voltages than multi cores when no static OC is present because the boost is higher.

Silicon varies, each cpu has unique properties when it comes to voltages actually used, what's demanded, what's needed, so both Intel and Amd err on the side of caution by setting defaults higher than absolutely necessary, this means it doesn't matter which cpu, they are all covered and guaranteed to remain stable. So you'll see 1.45v demand even if the cpu actually only uses 1.28v.

Vcore is use, vid is supply. By lowering vid/negative offset, it lowers supply and forces the cpu to demand less. So you'd see 1.35v with a -0.1v offset but cpu still only use 1.28v.

Using LLC, if you did a -0.2v offset, that'd put supply at 1.25v, so LLC would be applied (auto) to bring voltage up higher than 1.28v. If you set a low LLC, it might not cover the gap, and pc crashes.

It's trixy, and time consuming, especially when testing gets involved, but you'd need to know exactly what the cpu actually uses, which changes per load. But now you are basically into OC territory.

I see, thank you

 

[xKuLee]

You won't decrease voltage with LLC. To decrease voltage use VID/offset. And 1.38-1.45 isn't rediculous, single cores will use higher voltages than multi cores when no static OC is present because the boost is higher.

Silicon varies, each cpu has unique properties when it comes to voltages actually used, what's demanded, what's needed, so both Intel and Amd err on the side of caution by setting defaults higher than absolutely necessary, this means it doesn't matter which cpu, they are all covered and guaranteed to remain stable. So you'll see 1.45v demand even if the cpu actually only uses 1.28v.

Vcore is use, vid is supply. By lowering vid/negative offset, it lowers supply and forces the cpu to demand less. So you'd see 1.35v with a -0.1v offset but cpu still only use 1.28v.

Using LLC, if you did a -0.2v offset, that'd put supply at 1.25v, so LLC would be applied (auto) to bring voltage up higher than 1.28v. If you set a low LLC, it might not cover the gap, and pc crashes.

It's trixy, and time consuming, especially when testing gets involved, but you'd need to know exactly what the cpu actually uses, which changes per load. But now you are basically into OC territory.

Im trying OCCT stress test the Power supply one (it stresses everything gpu cpu etc) so i run this with avx2
all bios stock and it reaches 90c do you think thats still high for a DRP4? stock 9700k

 

[Karadjgne]

Turn off AVX totally. AVX, AVX2 and AVX-512 are advanced vector analysis tools. There's maybe a small amount of AVX used in a few games, it's faster than standard instructions for calculating particles after explosions etc, but is brutal on a cpu. AVX2/AVX-512 is used by ppl like LucasArts in their corporate machines during game creation, not in game itself. There's no games I've ever heard of that use AVX2.

So all you'd be doing is running a simulated 135ish% load on a 100% actual load.

It's like if you lift your arm, that's a 100% load. Now make your bicep pop up, you still only need that 100% to lift your arm, but added an extra % for no reason just to make the bicep swell up. Pointless exercise.

You are pushing 135ish% heat out of a cpu that'll only run to 100%. Occt also uses a linpack varient, and if you look at the usage graph, it's not consistent outputs as stuff gets rotated. So temps will rise and fall as Occt bounces from @ 80% cpu to 135% cpu.

Occt and Aida64 are useful stability tools, but useless for thermal testing as such. For thermals, use Prime95 Small FFT (not smallest, AVX tech disabled) as that's a 100% consistent load, basically a worst case scenario for a cpu. Gaming temps cpu will be somewhere underneath that. For gpu, use Firestrike or MSI Kombuster donut.

 

[xKuLee]

Turn off AVX totally. AVX, AVX2 and AVX-512 are advanced vector analysis tools. There's maybe a small amount of AVX used in a few games, it's faster than standard instructions for calculating particles after explosions etc, but is brutal on a cpu. AVX2/AVX-512 is used by ppl like LucasArts in their corporate machines during game creation, not in game itself. There's no games I've ever heard of that use AVX2.

So all you'd be doing is running a simulated 135ish% load on a 100% actual load.

It's like if you lift your arm, that's a 100% load. Now make your bicep pop up, you still only need that 100% to lift your arm, but added an extra % for no reason just to make the bicep swell up. Pointless exercise.

You are pushing 135ish% heat out of a cpu that'll only run to 100%. Occt also uses a linpack varient, and if you look at the usage graph, it's not consistent outputs as stuff gets rotated. So temps will rise and fall as Occt bounces from @ 80% cpu to 135% cpu.

Occt and Aida64 are useful stability tools, but useless for thermal testing as such. For thermals, use Prime95 Small FFT (not smallest, AVX tech disabled) as that's a 100% consistent load, basically a worst case scenario for a cpu. Gaming temps cpu will be somewhere underneath that. For gpu, use Firestrike or MSI Kombuster donut.

Alright i see thank you,

The only reason i was worried is that in the past 4790k and DRP3 i could pass prime with avx with normal temps with 9700k DRP4 i cannot

Also what worries me a bit is that in some games i get super unstable performance but it could be the games RUST specifically drops every second but other games do not

But also i been struggling for months with an issue where if my pc is OFF for many hours lets say overnight 15+ hours first boot of the morning would not pass it would spin all fans like pc is working but red cpu light would be on and no signal in monitor then restarting pc would work and every other boot would work

That was consistant every morning now its not even that much consistant past week every first boot worked so i thought i fixed it by replacing cmos battery but it very soon came back

So im thinking either the cmos is draning fast for some reason i don't know or either the cpu is acting up

My motherboard is already sent to techs and tested i already replaced PSU with a new platinum+ 750w corsair

So i guess cpu and ram would be next to blame or maybe SSD i have a 5 years old crucial mx500 that shows 65% health maybe its acting up sometimes but i dont care replacing it cpu scares me