Skylake, kabylake and coffee lake maximum vcore voltages per temperatures 115C to 0C for longevity.

[UnidentifiedOW]

1.13V <115C (RIP CPU)
1.22V <110C (really not recommended)
1.315V <100C (not recommended)
1.36V <90C (not recommended)
1.39V <80C
1.42V <70C
1.445V <60C
1.465V <50C
1.475V <40C
1.485V <30C
1.49V <20C*
1.495V <10C*
1.50V <0C*
^ Package temps in cb20.
Take off 20mv if your motherboard is bad.
* 20C and lower is very conservatively estimated and not compiled of actual real data. Use at your own risk.

 

Random other things that are related.

 

Highest I recommend for benchmarking is 1.5V. Do not blame me if your CPU dies though. My friend ran a 8350k at 1.52V for 2hours at 5.53ghz and it didn't degrade. My other friend ran a 7600k at 1.535V 5.36ghz for 1 hour until it crashed due to degradation. (Not software readings, actual hardware ones) (I'm saying 1.5 and not 1.52 because every CPU is different and I like to be safe.) Edit > Due to confusion on this point I must add that 1.5V isn't safe for much time at all. Only run this if you really want that amazing cinebench score or something. No longer then 6 hours. 

 

To all you people running 1.45V, this isn't Sandy bridge, you can't run 1.45V at 80C for 9+ years on skylake. Please acknowledge this.  

 

If anyone has had a skylake, kabylake or coffee lake CPU die at any listed settings, please inform me so I can update this list (does not include people running trash motherboards).

 

If you want to tell me that these numbers are too conservative, I really couldn't care less because being safe is more important then being clocked 100mhz higher.  
 

[7Hertz]

3 minutes ago, UnidentifiedOW said:

1.13V <115C (RIP CPU)
1.22V <110C (really not recommended)
1.315V <100C (not recommended)
1.36V <90C (not recommended)
1.39V <80C
1.42V <70C
1.445V <60C
1.465V <50C
1.475V <40C
1.485V <30C
1.49V <20C*
1.495V <10C*
1.50V <0C*
^ Package temps in cb20.
Take off 20mv if your motherboard is bad.
* 20C and lower is very conservatively estimated and not compiled of actual real data. Use at your own risk.

 

Random other things that are related.

 

Highest I recommend for benchmarking is 1.5V. Do not blame me if your CPU dies though. My friend ran a 8350k at 1.52V for 2hours at 5.53ghz and it didn't degrade. My other friend ran a 7600k at 1.535V 5.36ghz for 1 hour until it crashed due to degradation. (Not software readings, actual hardware ones) (I'm saying 1.5 and not 1.52 because every CPU is different and I like to be safe.)

 

To all you people running 1.45V, this isn't Sandy bridge, you can't run 1.45V at 80C for 9+ years on skylake. Please acknowledge this.  

 

If anyone has had a skylake, kabylake or coffee lake CPU die at any listed settings, please inform me so I can update this list (does not include people running trash motherboards).

 

If you want to tell me that these numbers are too conservative, I really couldn't care less because being safe is more important then being clocked 100mhz higher.  
 

Your friends PC crashed after an hour meaning it was not stable =/= degradation.

You reaching 5.53ghz and it not crashing =/= it being fully stable.

 

This is not Sandy Bridge that is correct.

 

A lot of logically baseless statements.

 

[UnidentifiedOW]

1 minute ago, 7Hertz said:

Your friends PC crashed after an hour meaning it was not stable =/= degradation.

You reaching 5.53ghz and it not crashing =/= it being fully stable.

 

This is not Sandy Bridge that is correct.

 

A lot of logically baseless statements.

 

After being run at 1.535V for the hour the CPU was unable to run for 1 minute at an overclock that could run it for 6hours+ that is degradation.

[7Hertz]

Just now, UnidentifiedOW said:

After being run at 1.535V for the hour the CPU was unable to run for 1 minute at an overclock that could run it for 6hours+ that is degradation.

Without proof it is still baseless my friend.

[GoldenLag]

3 minutes ago, UnidentifiedOW said:

After being run at 1.535V for the hour the CPU was unable to run for 1 minute at an overclock that could run it for 6hours+ that is degradation.

1,535 volts is lethal..........

 

Like it will kill your CPU rapidly. 

[chckovsky]

11 minutes ago, UnidentifiedOW said:

these numbers are too conservative

I thought concensus was 1.4-1.425 was max safe voltages for Intel 14nm. 1.5 could never be considered "safe" from what I've gathered.

BTW: interestingsemi-related video:

 

[UnidentifiedOW]

1 minute ago, GoldenLag said:

1,535 volts is lethal..........

 

Like it will kill your CPU rapidly. 

Yes and It did, that CPU could run 5.0 at 1.33V and now it needs 1.37V for the same clock.

[7Hertz]

Just now, GoldenLag said:

1,535 volts is lethal..........

 

Like it will kill your CPU rapidly. 

It is heat that normally kills a CPU faster not voltage, and i have seen this with various GPU's i have owned, i have never killed a CPU. Ran 32NM Xeon's at 1.5v pushing 4.6ghz without degradation, max temps were around 75-80c and this is with recommendations of not going above 1.35v.

1.45V was safe recommended for Phenom 2's and ran one for 7 years at 1.55v at 4.2ghz just fine.

 

 

[UnidentifiedOW]

Just now, chckovsky said:

I thought concensus was 1.4-1.425 was max safe voltages for Intel 14nm. 1.5 could never be considered "safe" from what I've gathered.

BTW: interestingsemi-related video:

 

1.5V IS NOT SAFE. It can be run for a very short amount of time.

[UnidentifiedOW]

2 minutes ago, 7Hertz said:

It is heat that normally kills a CPU faster not voltage, and i have seen this with various GPU's i have owned, i have never killed a CPU. Ran 32NM Xeon's at 1.5v pushing 4.6ghz without degradation, max temps were around 75-80c and this is with recommendations of not going above 1.35v.

1.45V was safe recommended for Phenom 2's and ran one for 7 years at 1.55v at 4.2ghz just fine.

 

 

We are talking about skylake here.

[7Hertz]

Just now, UnidentifiedOW said:

We are talking about skylake here.

Ok.

Skylake is magically different i get it.

[GoldenLag]

2 minutes ago, UnidentifiedOW said:

1.5V IS NOT SAFE. It can be run for a very short amount of time.

Then what id the point of this list?

[GoldenLag]

Just now, 7Hertz said:

Ok.

Skylake is magically different i get it.

Voltage and degredation has to do with the node. The larger the node the higher voltage you can theoretically use. 

[7Hertz]

Just now, GoldenLag said:

Voltage and degredation has to do with the node. The larger the node the higher voltage you can theoretically use. 

Yes, theory is not reality though.

 

Theory has to be tested to become reality, i trust reality and what buildzoid posts over someone's mix up of heat vs voltage though.

[UnidentifiedOW]

3 minutes ago, GoldenLag said:

Then what id the point of this list?

Max safe voltage for long term usage based on temperature of the CPU under cb20 load.

[GoldenLag]

57 minutes ago, UnidentifiedOW said:

Max safe voltage for long term usage based on temperature of the CPU under cb20 load.

But it doesnt paint a picture over safe voltages at all. Its honestly just a but missleading.