PDA

View Full Version : Realbench overheating cpu



Cannon.19
12-04-2020, 10:31 PM
I'm still testing my new setup to see how low I can undervolt while stable in prime95, realbench, etc...
My setup:
CPU: i9-10900K
Cooler: bequiet Dark Rock Pro 4 (with both fans installed)
Thermal paste: Thermal Grizzly Kryonaut
GPU: Asus TUF RTX 3080 OC
RAM: 2x16GB 3600MHz HyperX Fury
Motherboard: Asus TUF Z490-PLUS
Case: Asus TUF GT501 (3 intake fans at the front panel, 1 exhaust fan on the back)
BIOS settings:
XMP: XMP I (DDR4-3603 17-21-21-39-1.35V)
ASUS MultiCore Enhancement: Enabled - Remove all limits
SVID Behavior: Auto
AVX Instruction Core Ratio Negative Offset: 0
CPU Core Ratio: Auto
Power-saving & Performance Mode: Auto
CPU Load-line Calibration: Level 4 - Recommended for OC
BCLK Aware Adaptive Voltage: Enabled
CPU Core/Cache Voltage: Manual Mode
CPU Core Voltage Override: 1.370 V
DRAM Voltage: 1.35000 V
CPU Core/Cache Current Limit Max: 255.75 A
Long Duration Package Power Limit: 4095 Watt
Package Power Time Window: 56 Sec
Short Duration Package Power Limit: 4095 Watt
Intel SpeedStep: Enabled
Intel Speed Shift Technology: Enabled
Intel Turbo Boost Max Technology 3.0: Enabled
CPU C-states: Enabled

So the clock frequency right now is 4.9GHz on all cores.
Vcore voltage under load (during realbench stress test) is 1.288v. I'm still not 100% sure if it's stable, but at 1.279v I had WHEA errors in hwinfo. I'll have to test 1.288v some more before I'm ceratin it's stable.
However, I couldn't believe how high the realbench pushes the temperature (it goes to 99C in 5 minutes), especially considering I used to run realbench for 6 hours on my gaming laptop (i7-4720hq|gtx 980m) and temps were never even close to this. I already checked the cooler mounting pressure and thermal paste spread, it was perfectly fine (then I reapplied the thermal paste of course).
I also tested with case side panel off and temperature was only 1C lower than when I tested with side panel on (in blender), so airflow is not an issue either.

geneo
12-04-2020, 11:52 PM
Well you have 2.5x the number of cores running compared to your old system, and it is air cooled so this is not a surprise. Even taking into account your Haswell was 22nm vs 14nm on your comet lake.

I have a Noctua NH-D15 on my delidded 10700k, which is only 8-core, and at 1.3 vcore 5.1 GHz it hits mid-80s on Realbench. So 1.28v all cores on a 10900k air cooled should be quite hot. Reason I got a 10700k instead of a 10900k is I didn't want to water cool.

Cannon.19
12-05-2020, 11:01 AM
Well you have 2.5x the number of cores running compared to your old system, and it is air cooled so this is not a surprise. Even taking into account your Haswell was 22nm vs 14nm on your comet lake.

I have a Noctua NH-D15 on my delidded 10700k, which is only 8-core, and at 1.3 vcore 5.1 GHz it hits mid-80s on Realbench. So 1.28v all cores on a 10900k air cooled should be quite hot. Reason I got a 10700k instead of a 10900k is I didn't want to water cool.

I just realized I forgot to write what my temperatures are in my original post :D . It goes to 98-99C within 5 minutes of realbench.
I'm debating with myself right now whether I should switch to water cooling and get something like kraken z73 or arctic lf2. Or maybe even the intel's sub-zero cooler (https://landing.coolermaster.com/pages/ml360-sub-zero-cryo-clocking/)?
What's also weird to me is that I saw a guy on youtube running 5.2GHz on his 10900k with only 1.33v load vcore in cinebench, and it remains stable. I don't think I'd be able to run 5.2GHz at anything below 1.4v. Did I mess up something with my settings in bios or what, I have no idea? Also, that same guy peaked at about 70C during cinebench (he was using kraken z73 I think), while I get really close to 90C at 4.9GHz during cinebench.

Silent Scone@ROG
12-11-2020, 11:31 AM
I just realized I forgot to write what my temperatures are in my original post :D . It goes to 98-99C within 5 minutes of realbench.
I'm debating with myself right now whether I should switch to water cooling and get something like kraken z73 or arctic lf2. Or maybe even the intel's sub-zero cooler (https://landing.coolermaster.com/pages/ml360-sub-zero-cryo-clocking/)?
What's also weird to me is that I saw a guy on youtube running 5.2GHz on his 10900k with only 1.33v load vcore in cinebench, and it remains stable. I don't think I'd be able to run 5.2GHz at anything below 1.4v. Did I mess up something with my settings in bios or what, I have no idea? Also, that same guy peaked at about 70C during cinebench (he was using kraken z73 I think), while I get really close to 90C at 4.9GHz during cinebench.


Hello,

Reachbench makes use of X256 encoding which uses AVX2, these extensions are known for pulling more current than conventional stress test routines. In regards to clocks and temperature, you may find your ceiling for AVX overclock will improve with more adequate cooling, as these CPUs respond well to temperature.

Cannon.19
12-12-2020, 07:04 PM
Hello,

Reachbench makes use of X256 encoding which uses AVX2, these extensions are known for pulling more current than conventional stress test routines. In regards to clocks and temperature, you may find your ceiling for AVX overclock will improve with more adequate cooling, as these CPUs respond well to temperature.
Hey,
Yeah, the kraken z73 finally arrived yesterday. I was blown away by how much cooler the cpu runs now. I was able to run realbench for 1 hour now, peak temperature was 77C while average was 73-74. Considering I was throttling at 100C with drp4 within 5 minutes, this could be well over 23C difference. I was also able to lower the load voltage during realbench stress test from 1.288v to 1.243 and remain stable (I guess the cooler the cpu is the less voltage it needs). At 1.234v however, I started getting whea errors in hwinfo, so 1.243v should be the limit for 4.9GHz here. Is that a normal result? I mean compared to other 10900k chips out there? I feel like it should be able to go lower at 4.9GHz, somewhere around 1.2v, or is that possible only for the lucky golden sample chips?