12900K safe voltages?

[techanalyst]

What are the safe voltage for the 12900k? Bios I set max 1.375 ok or is that going to kill the CPU in 3 months

[Hopper64]

Jays 2 cents has a video about OC on this platform which was enlightening. I’m using AI OC setting in the bios with a 100mV negative offset to decrease the peak voltage. Just benched mine with a maximus extreme, 12900k and highest temp was 80C in HWInfo64. Pcores were 5.5 to 5.3 and max Vcore was 1.4V. Using Adaptive Voltage with the negative offset. Negative offset worked quite well. It was stable throughout. I’m still testing mine currently.

*https://m.youtube.com/watch?v=y2HvCcb-eTc

[Chainbold]

Hopper64 wrote:
Jays 2 cents has a video about OC on this platform which was enlightening. I’m using AI OC setting in the bios with a 100mV negative offset to decrease the peak voltage. Just benched mine with a maximus extreme, 12900k and highest temp was 80C in HWInfo64. Pcores were 5.5 to 5.3 and max Vcore was 1.4V. Using Adaptive Voltage with the negative offset. Negative offset worked quite well. It was stable throughout. I’m still testing mine currently.

*https://m.youtube.com/watch?v=y2HvCcb-eTc

5.5. ~ 5.3 with all cores running and HT enabled at 1.4 volt? You are lucky, you got a super silicon. 🙂 At 1.4 volt, I can just make it to 53-52 with E cores and HT disabled. Running the blend test in P95, I need to go down to 52-51.

[techanalyst]

Chainbold wrote:
5.5. ~ 5.3 with all cores running and HT enabled at 1.4 volt? You are lucky, you got a super silicon. 🙂 At 1.4 volt, I can just make it to 53-52 with E cores and HT disabled. Running the blend test in P95, I need to go down to 52-51.

I stopped trusting those I believe in broadwell days, new machines out of box would BSOD on prime95 (HP, Dell, Alienware laptops etc), so the high core count processors, custom or oem responded well to prime anymore especially

BTW 0811 bios that was put up, everything is reporting correctly

Also noticed a more consistent scoring on CPU benches, just ran a test run limiting vcore to 1.375 then 1.4 just using the AI overclocking and I'm scoring/getting more consistent core clocks vs using XTU

Considering the system will drop all to 52 or 53 (depending on what multi Im testing) the "highest" doesnt really matter much 😛 getting 2/3 cores at 55 is fine but that getting all to 53 is a bit harder

[Hopper64]

Chainbold wrote:
5.5. ~ 5.3 with all cores running and HT enabled at 1.4 volt? You are lucky, you got a super silicon. 🙂 At 1.4 volt, I can just make it to 53-52 with E cores and HT disabled. Running the blend test in P95, I need to go down to 52-51.

Here's the data I collected running Cinebench on this new machine. Just benching a little on my test bench for stability.










I'm thinking a peak voltage around 1.4 V is likely not any issue for longevity. Especially since it's in adaptive mode. It was quite stable in cinebench. I was using bios 0002 at the time.

[Chainbold]

Hopper64 wrote:
Here's the data I collected running Cinebench on this new machine. Just benching a little on my test bench for stability.

I'm thinking a peak voltage around 1.4 V is likely not any issue for longevity. Especially since it's in adaptive mode. It was quite stable in cinebench. I was using bios 0002 at the time.

It seems - according to the reading you posted - your peak voltage is almost 1.46 though.

I cannot get my system stable at anything above 53-52 - using adaptive.

But even when running at x53-52 (with an average of x52) I get in Cinebench the same score like you: 28373



It is not stable though when running P95. I need to revert to x51 to run the blend text for at least 10 minutes without errors

[Silent_Scone]

Chainbold wrote:
It seems - according to the reading you posted - your peak voltage is almost 1.46 though.

Hello, see my posts

[Chainbold]

Silent Scone@ROG wrote:
Hello, see my posts

well, I was referring to the reading from the screenshot that he had posted (not your screenshot).

I understand though that with previous BIOS vcore readings were somehow not correct.

I wonder by the way why Shamino has posted updated BIOS for all Z690 boards - with the exception of the Z690-F. So, I am still running 0803 - which apparently is giving misleading voltage readings.

[Silent_Scone]

Chainbold wrote:
well, I was referring to the reading from the screenshot that he had posted (not your screenshot).

I understand though that with previous BIOS vcore readings were somehow not correct.

Hello, Hopper clearly shows the vcore reading. As I showed in my USB Osc screenshot, the vcore reading from the onboard sensor array is what you should be paying attention to, not the core VID. The vcore readings have always been accurate.

[Chainbold]

Silent Scone@ROG wrote:
Hello, Hopper clearly shows the vcore reading. As I showed in my USB Osc screenshot, the vcore reading from the onboard sensor array is what you should be paying attention to, not the core VID. The vcore readings have always been accurate.

Ok , finally I get it. 😮 I can see the difference in my HWInfo readout. But, what exactly is the difference between "core VID" and "vcore"?

Meanwhile I got BIOS 0811. Works well.

[Hopper64]

Chainbold wrote:
It seems - according to the reading you posted - your peak voltage is almost 1.46 though.

I cannot get my system stable at anything above 53-52 - using adaptive.

But even when running at x53-52 (with an average of x52) I get in Cinebench the same score like you: 28373



It is not stable though when running P95. I need to revert to x51 to run the blend text for at least 10 minutes without errors

Yes. Correct, guess I’m still used to Rocket Lake where 1.5 was my peak under adaptive mode. This operates at a lower voltage so I guess I was referring to “about 1.4Vâ€�. It really works well. Some have recommended avoiding P95 altogether here. Â*