[SOLVED] 9900K Stock Config Hitting temps of 97C

[smithcs2345]

Hi guys,

I recently built a new PC for music production and 3D VFX (taking a course in Houdini), and have been dealing with extremely high temps up to 97C when rendering. This is on the stock configuration with no overclock. Is something wrong, or did I simply lose the silicone lottery? Was thinking about overclocking but not at these temps. Also seeing CPU Vcore as high as 1.368. Completely stock Bios besides enabling XMP. Any thoughts? I have already tried re-seating with Kryonaut.

Gigabyte z390 Designare Mobo
Intel 9900K CPU
GSkillz Tridentz 3000 32gb
Corsair h115iPro AIO
Corsair Rm750x PSU
Sapphire rx580 GPU
Samsung 970 Evo 500gb SSD
Fractal Meshify C

 

[Rodrigodrt]

I would have no clue why this chip is set @ 1.368v at stock settings, so i'd lower that to 1.20-1.25v right off the bat if you are trully all default settings, in fact, i'd keep lowering and find the stock sweet spot, no wonder things are so hot.

now rendering is probably what comes the closest to stress testing so cpu usage and temps will skyrocket, specially with that voltage settings. lower and check.

 

[The Original Ralph]

i render quite a bit on my 9900K, with cpu showing 97-100% load and don't see those temps - actually with my air cooler i'm seeing 69C to 75C across all the cores with occasional spikes to 77 or even 78, and temp when i stress test with Prime 95 are much higher (by about 10C) than when rendering

for whatever reason, most of the Motherboard mfgrs seem to be overvolting their boards BIOS
Voltage - i'd start around 1.235V, and run Prime95 V26.6 (that specific version) for about 15-20 mins, if you don't get a BSOD, you can move it down to 1.230V and test again with P95, the idea being to find what your lowest voltage is that the computer is still stable with. Once you get a BSOD, at whatever voltage, bump the voltage back up 1 or 2 of the increments you were adjusting it in. And run P95 v26.6 select "small FFTs otherwise it'll run AVX instructions (more about AVX below)

The idea is, the lower the voltage, the lower the temps

there's a real decent OC by the numbers guide, over on the MSI board - even though you aren't interested in OCing, if you go thru the guide it will help you understand what's going on. Plus, if you take that cpu up a bit in clock frequency while maintaining decent temps, you'll be surprised at how much time you'll save or how much faster files will render

here's a link if interested https://www.msi.com/blog/intel-9th-cpu-overclocking-5ghz-with-z390-motherboards

One aspect you might want to consider, setting your AVX offset at -2 or -3, which means whenever a cpu core runs into a workload involving AVX instructions (which it will when rendering) it will drop the clock ratio (or multiplier) by 2-300 MHz. AVX workloads are more stressful loads on the cpu. Adjusting your AVX offset is covered in that MSI guide

 

[smithcs2345]

Thanks guys! If I can get the temps under control I would definitely like to overclock and get some extra performance and lower render times. Quick questions though, when you run prime95 do you mean the Blend test, the small FFTS, or Large FFTS?

Thanks!

 

[smithcs2345]

Update- Got BSOD running stock clock @ 1.235V after about 3 minutes using Prime95 V26.6 on the small FFTS torture test.

 

[poopflinger]

Update- Got BSOD running stock clock @ 1.235V after about 3 minutes using Prime95 V26.6 on the small FFTS torture test.

First off, check your LLC (Load Line Calibration) in the bios. It should not be set to the minimum. Try a higher setting and run the test again. If that doesn't do the trick then up the voltage to 1.240 and try again. If it fails again, take it 1.245. Keep doing this until it's stable. If you find that you can't keep thermals under control using stock clocks, you may need to contact Intel and get a replacement CPU sent out to you under their warranty.

Another issue with these soldered CPUs is the surface of the heat-spreaders is sometimes not flat and causes very poor contact with the cooler, even with a good thermal paste application. You can check this by removing your CPU from the mobo, placing it on a truly flat surface with the heat-spreader down, and spinning it like top. If it spins freely (like a top) you have a problem with the heat spreader and may need to sand it down to a flatter shape. There are many videos on youtube that demonstrate this.

 

[The Original Ralph]

to the OP, take a look at that MSI OCing guide - it will make a lot plain as day. If you raise your LLC you'll be raising voltage considerably and raising temps. For the time being leave LLC on auto

raise your voltage in increments of .005V Ie to 1.240V and rerun P95 as you described earlier - see if you still get the BSOD - if not, raise the voltage another .005 just for a little "warm and fuzzy" factor

The lower your voltage the cooler your CPU temps will be

I'm at 4.9, all cores, with a Vcore of 1.245v. My temps rendering video files, with CPU showing 97-99.5% load run in the range of 68-76 across all the cores, with occasional spikes to 78. When i run P95 V26.6 small FFTs, i see temps of 84-90C, with occasional spikes to 92C. My idle temps are 30-34C I have no presets enabled, (ieno XMP, TUrbo Boost or Enhanced boost), and LLC is still on auto

check my sig for my rig