You are talking about the R7 2700X. OP has a Non X like mine and our cpus normally need higher voltage at a given oc and may not oc as high as the X version. I use 1.35v to be stable at 4GHz. Like 2600ryzen said, PBO does not really pertain to the Non X Gen 1+.
The known 1.32v ideal limit for CPU Vcore is commonly used for GEN 2. I've had my 2700 Non X at 1.39v for 4.1GHz for most of its life ( 2yrs). Only recently that i reset to 4GHz cos my PSU is 10 yrs old
Trying to lessen the strain.
The 2700X is what we call a "leaky" CPU
It's binned for higher voltage, same as the 2600X, 1800X, 1700X, 1600X & 3800X
These CPUs need more voltage and a stronger loadline ~ which results in less voltage
1700,2600,3600, 1920X aren't leaky CPUs
These barely need any loadline, are binned out of factory for low voltage & efficient operation
X units on Zen one had a differentiation
2nd gen doesn't follow these rules
PBO limits work on all of them, up to SMU and bios/CBS the possibilities varry
What we call "PBO" , is more than one feature combined
The voltage I mentioned is accurate, I didn't make a mistake and focused on nm node
It just appears to be very similar to 7nm, while 7nm only allows similar behavior by it's different boost shuffle design
12nm shouldn't exceed 1.325v as constant voltage
As mentioned, it has a FIT module too and can prevent the CPU from dying and so faking voltages and MHz
I'm sadly not able to confirm if clock stretching existed also on 2nd gen
Tho the limits are accurate
Definition of safe voltage and stability changes between people, but these are values tested by The Stilt according to how AMD designed the FIT module on these units
The only thing that will vary between leaky and non leaky CPUs, is the running loadline and how or "how much" of PBO you can utilize
The overdrive remains to be working, but we can't expect 4.35Ghz upper limits on non X units
EDIT:
I have to correct myself
7nm safe allcore voltage is variable up to silicon
1.325v is
wrong for 7nm, only 1.48v upper limit remains coincidentally identical
Actual safe voltage for 7nm is
around 1.29625v
Sometimes it goes down to 1.2875v if it has thermal constrains and hits above 80c where SMU will start to hard throttle
Real safe voltage for 7nm is around that
The spread 1.325v [7nm] is plain wrong and depends on silicon characteristics
While 2nd gen coincidentally has the same allcore limits while being more of a hard limit, as silicon sampling is not done fast enough and voltage is not shifting/updating fast enough
1.325v [12nm] remains accurate
