bravegag Posted November 1, 2022 Share Posted November 1, 2022 Hi everyone, I have the most powerful AIO Corsair sells, the mammoth h170i mounted to the front side so the coolest possible configuration. Yet, my i9-12900KS spikes at 100C with CineBenchR23.2 .. how come? I'm not using the stock h170i thermal compound but the Thermal Grizzly Kryonaut Extreme instead with full spread application procedure so can't get any better than that. Link to comment Share on other sites More sharing options...
c-attack Posted November 1, 2022 Share Posted November 1, 2022 (edited) This is going to be either physical contact or voltage. With cpu temp spikes it’s not the radiator, fans, or pump speed which would cause the cpu temp to slowly become too high over several minutes of time. Start with the easy one — voltage assessment. Check the max sustained Vcore when R23 is running, in CUE and a secondary program. CUE has had some inaccuracies for voltage readings in the past. You might also note the peak when doing normal stuff on the desktop. The R23 load should be lower with Vdroop under max load (LLC setting dependent). Another thing to do is run another program to test. Bad contact is bad contact and anything you do always ends with terrible temps. Try running the “Bench Stress Test” in CPU-Z. It’s relatively mild and linear in load making it a good assessor for contact issues. If it’s massively lower, then you are looking for an AVX or bios level setup instruction to keep things under control for R23 or similar synthetics. If it’s 100C instantly as well, then that’s likely a physical contact issue (bracket, standoffs, mounting pressure, etc). Edited November 1, 2022 by c-attack 1 Link to comment Share on other sites More sharing options...
bravegag Posted November 1, 2022 Author Share Posted November 1, 2022 Hi c-attack! Thanks a lot once more for your timely support! Indeed I ran the CPUZ Ver 2.03.0.x64 and all cores were pretty cool during the Bench CPU test around 30-40C, then on submitting the results it spiked to 80C funny that on submit it spikes and not during the benchmark but anyways all temps were in reasonable ranges so I'm happy: https://valid.x86.fr/gxm1sm The ran the Stress CPU and temps maxed at 88C which I think is still OK-ish not getting close to 100C. Am I out of the woods not having a contact issue and then tick down the Volt usage in BIOS? I am surprised because I left everything in default settings only the XMPII OC configuration. TIA! Link to comment Share on other sites More sharing options...
Solution c-attack Posted November 1, 2022 Solution Share Posted November 1, 2022 (edited) A lot of the MBs will disable normal Intel power limits and bump clocks up to the single core max for all cores. The voltage curve is much more intelligent than it used to be, but it still will always err on the heavy side to make sure you don't crash out of the box. I can't see your load Vcore in the data, but your multi-core score in CPU-Z suggests the board is reaching above and beyond a vanilla 12900K. See if you can get a Vcore reading from HWMonitor or HWiNFO. Now the other thing is this may not matter if you don't run these types of loads. If your normal gaming/video/productivity temps max out at comfortable levels, this may be a byproduct of the instructions on the synthetic test. I do think this is voltage levels and not an actual physical contact problem. The weird spike on upload is from the instruction type used to validate your gear. I see this all the time on 3DMark where the SysInfo part of the program registers the highest temps I ever record, since it uses some form of AVX2 instruction. It's a bit ironic this occurs in the analyzing phase and not the actual max load, but this is common for most of us not using a negative AVX offset to prevent it. Those that use programs with AVX or other complex instructions typically need to set their multiplier down a bit from where the sport gaming overclockers can hang out. Edited November 15, 2022 by c-attack Link to comment Share on other sites More sharing options...
bravegag Posted November 2, 2022 Author Share Posted November 2, 2022 Thanks a lot c-attack! indeed dialing down on the AVXs is a good plan. btw how about this one? Thermal Grizzly Contact frame for Intel 12th Gen CPUs (LGA 1700). https://www.amazon.de/-/en/gp/product/B09YYQBXQY/ref=ox_sc_act_title_1?smid=A2AEWZ68B7P4J7&psc=1 It is supposed to significantly reduce temps 10C under heavy loads. I wonder how well it will work with the h170i ... Link to comment Share on other sites More sharing options...
LeDoyen Posted November 2, 2022 Share Posted November 2, 2022 it will work the same regardless of the cooler since it's tackling contact issue. You can see more details there 1 Link to comment Share on other sites More sharing options...
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now