I have the 7920X and my cores are exactly the same. I can see differentials of up to 20°C between different cores even with the the frequencies and voltages all set the same. On idle i can see that 2 specific cores run hotter than the others and then on full load those cores are not the hottest anymore. I suspected that there was an issue with the application of the liquid metal on the die so i delidded the CPU again and redid it. I did notice some small lumps of shat looked like hardened flakes on the IHS and the original application of the product by the company who delidded it was a bit excessive but other than that i found no issues and had exactly the same varieance in temperatures after testing again. I did however use liquid metal on the interface between die and IHS and between IHS and waterblock when i reassembled and saw a massive drop in temperatures across all cores. In the beginning i could not run at full load without thermal throttling and now i can run faster at full load with the fans only at around 60%. I am not sure if it is a typical feature of the skylake X CPU's that there is such a variance in core temps and have asked around myself what others are seeing but am getting no other feedback either.
The only place i ever saw any temperatures was i this thread on another forum https://forums.overclockers.co.uk/threads/8pack-10-18-core-i9-including-7980xe-testing-bundles-and-b... There is no graph for temps but from the basic data i could see from the benchmarking of the guy who wrote the post and does all of the delidding and testing of pre binned CPU's for that company there seemed to be quite a variance between core temperatures with the indicator being the variance in max temperature recorded. I know your CPU is not on there but it may be an indicator of general performance with Skylake X CPUS?
The only place i ever saw any temperatures was i this thread on another forum https://forums.overclockers.co.uk/threads/8pack-10-18-core-i9-including-7980xe-testing-bundles-and-b... There is no graph for temps but from the basic data i could see from the benchmarking of the guy who wrote the post and does all of the delidding and testing of pre binned CPU's for that company there seemed to be quite a variance between core temperatures with the indicator being the variance in max temperature recorded. I know your CPU is not on there but it may be an indicator of general performance with Skylake X CPUS?
I have the same problem with my 7900x (delided by caseking)
wln123 wrote:
I've overclocked with offset mode for each core up to 4.5GHz
In my case in BIOS core #1 and #7 are marked with *
My other OC settings:
CPU input voltage - 1.850
CPU Load-line Calibration - Level 7
CPU Current Capability - 200%
AVX/AVX-512 - 3
energy savings - off
memory working XMP 3200MHz for F4-3200C16Q-32GVK
I think that's all changes that I've made.
My knowledge about OC is very basic so maybe my settings are just simply bad.
On stock settings, core #5 also have bigger temperatures around 12 degrees but with OC'd that difference is HUGE almost 20 degrees. I dont think it's normal.
Can't tell what is the source of this problem.
Don't have any problems with stability on my system, not a single BSOD, rebooting around every 2 weeks or forced by windows updates.
I really would like to push my CPU little higher but I'm afraid because of this stupid #5 core.
Temps on screenshot after around 15 minutes of using no_avx prime95 and Kraken X62 (tried preapplied NZXT compound and right now using Kryonaut)
Like you i am no expert overclocker.
CPU cores all set to 4600MHz
Currently i have a manual override core voltage set to 1.22V (i did try running with an adaptive negative offset to achieve the same result and noted individual core voltages were really all over the place and i get lower temperatures this way even though the voltage on average is the same?????)
I have the CPU input voltage set to 1.85V
Load Line calibration set to 3
CPU Current Capacity 200% although 140% may have been enough
CPU power phase control set to extreme
Spread spectrum Disabled
I have My AVX offsets at 2 and 3
I originally had the stock EK TIM that came with the monoblock but i have replaced this with conductonaut and it has made such a huge difference and the below is what i see now with Prime95 in non AVX mode.
No matter what i do, like you i still get a huge difference across the cores.
.jpg "Marko911"){kind=avatar}
