Krunkkracker

The safe voltage level comes from Stilt's research on overclock.net and 1.325 is the voltage usually used under load at stock settings: https://www.overclock.net/forum/10-amd-cpus/1728758-strictly-technical-matisse-not-really.html To solve your first issue, you could have just adjusted your fan profile in your BIOS to only ramp up after 50c or so. Idle temperatures mean very little. 1.3v core voltage is very reasonably safe, especially for only 4.2ghz where the current draw will be minor. Degradation happens all the time but 1.3v/4.2ghz shouldn't be noticeable increase over stock, if at all.

It's also hosted on their website. Cinebench r20 is a great for a quick, very short term stability/temperature check. In preferences, you can set a time duration if you want to loop the test automatically.

CPU-Z is rather lethargic as far as stress testing goes. Use Realbench, Prime95 29.8 non-avx, OCCT, or Intel Burn Test on 8 threads if you really want to test stability. Max temperatures at 78c in your last test seem pretty decent, gaming won't get nearly that hot. My 9900k does 5ghz at around 1.34v VR VOUT load voltage with Prime95 AVX but that's extreme and I can get away with 1.24~ load voltage if not using Prime95 AVX. What's your VR VOUT in HWINFO under load? Also what LLC level are you using? Also, going from a 280mm to a 360mm is mostly a side-grade since the radiator area is very similar, with a slight lead for the 360mm. Don't worry about temperatures under stress tests unless they hit around 90-95c since that's throttle range. They won't get nearly that high unless you run blender or other high intensity programs, in which case a lower clockspeed may be required. Edit: If you don't have your radiator set as an intake, that can improve coolant temperatures which in turn improves cpu temperatures. Gpu and other parts may warm up but the impact shouldn't be as large since the 9900k's temperature is usually the bottleneck.

Double check that you're not power limit throttling like this: Also always use HWInfo for monitoring. The VROUT reading for voltage. Also it's true that some motherboards don't always measure vrm or backside vrm temperatures. Only other thing I can think of is if the higher performance plan somehow got set to something below 100% cpu speed in the advanced settings.

I'd say it entirely depends on if you need the extra performance. Are you having fps drops in cpu bound games or need more performance in workstation oriented tasks? For gaming, buying amd will allow you to spend more money on the gpu. For other things, it depends on the applications itself.

I have the aorus wifi pro with an 9900k at 5ghz and 1.32 volts and the VRMs get up to maybe 80c under sustained load from prime95. Plenty of headroom.

For XMP on a 9600k, you may need to change the VCCIO and VCCSA voltage. For my 9900k, I had to increase them both to 1.15v for my 3733mhz kit to work.

I would only use an avx offset if it is unstable with an avx stress test. Games these days use it pretty commonly so you want that performance. Avx and non avx would be perfect. With your temperatures, you can even use prime95 with avx which many can't run because it thermal throttles but it will definitely show if there are any instabilities.

I say leave it running for 8 hours. Not sure what cpu you have but the point of stress testing is to STRESS. Running a single stress test for an hour won't always find instability and instability can cause a whole host of problems if it manages to corrupt an update.

By enabling MCE, you have unlocked the TDP limits of the processor which was probably causing it to throttle. You can also find wattage/amp/etc limitations in your bios and set those to max settings to do the same thing. To accurately see the voltage value, use the VROUT sensor reading on HWinfo. Drop the voltage if you can but definitely check that your system is stable with realbench/latest prime95/aida for several hours a piece. Also only drop load line calibration is already stable since there's no reason to give it more voltage than it needs. Just be aware that silicon lottery needs 1.3v at avx offset at -2 just to be stable for their 2nd best 9900k's. The last thing you want is one of the many games that use AVX to crash your system or a windows update corrupting your system. Your temps look great and that should give you more headroom than others have.

Hey there! I had the same processor running at stock with an 280mm all in one and it still hit 80c+ especially when I don't run my window AC. Before you drop 70 dollars or more on a cooler, try undervolting and looking up some guides. My 9900k was hitting 1.35volts or higher with stock motherboard settings and even the VCCIO and VCCSA was at 1.3volts for some reason. I settled on 1.22volts for the cpu at 4.7ghz and dropped the other two voltage settings to 1.10volts. My 9900k loads around 70-75 celcius when my apartment is at its hottest. Just food for thought before you spend all that money.