Broadwell/Skylake/Zen voltages and overclocking.

[eton975]

I understand that as the transistors in our processors get smaller and smaller, less and less voltage is needed to cause permanent physical degradation. I've heard horror stories about people who've fried Haswell 22nm FINFET chips by pushing too far above 1.3V, yet I hear lots of stories of people running 32nm SB at 1.5V perfectly fine.

Will the safe voltage limit for these chips be much lower than Haswell, and will it impact how we OC?

 

[eton975]

Didn't old chips like Pentiums and stuff run at super-high voltages too? (3.3/5V) Could you get insane overclocks bc of that?

 

[LTC8K6]

Have to go way back to get to those really high voltages.

https://en.wikipedia.org/wiki/P5_(microarchitecture)#Models_and_variants

P3 Katmai of 1999 vintage was 1.93V minimum and 2.07V maximum recommended voltage, for example.

By the time we get to the Tualatin P3 of 2001, we are down to 1.5V.

 

[sm625]

Been running a G3258 at 1.4V for the past year.

 

[DrMrLordX]

I understand that as the transistors in our processors get smaller and smaller, less and less voltage is needed to cause permanent physical degradation. I've heard horror stories about people who've fried Haswell 22nm FINFET chips by pushing too far above 1.3V, yet I hear lots of stories of people running 32nm SB at 1.5V perfectly fine.

Will the safe voltage limit for these chips be much lower than Haswell, and will it impact how we OC?

I don't think it's a given that smaller transistors necessitates lower voltage. Leakage characteristics play a role as well. I think it's more true to say that smaller transistors require less power to operate overall; additionally, it would take a smaller increase in power draw from overvolting/overclocking to cause degradation than an older process featuring larger transistors.