Voltage confusion (965 BE)

[TheDustyHeatsink]

Hiii, I'm at a loss to find information on what is considered to be "safe" voltages? I'm certainly not a novice when it comes to computers, however as increasing the voltage ramps the temps up like crazy, is there considered to be any dangers of going pretty high other than temperatures? Say I use a ridiculous cooler, not sub-zero temps, but would there actually be any danger with putting the voltages to what are considered to be "pretty high", as long as the temperatures are okay? If not, how do I know where to draw the line? It seems like a bit of a grey area to me. I don't practically have this as a problem for overclocking my own set up, but its just been bugging me!

 

Dusty.

[-rascal-]

Hiii, I'm at a loss to find information on what is considered to be "safe" voltages? I'm certainly not a novice when it comes to computers, however as increasing the voltage ramps the temps up like crazy, is there considered to be any dangers of going pretty high other than temperatures? Say I use a ridiculous cooler, not sub-zero temps, but would there actually be any danger with putting the voltages to what are considered to be "pretty high", as long as the temperatures are okay? If not, how do I know where to draw the line? It seems like a bit of a grey area to me. I don't practically have this as a problem for overclocking my own set up, but its just been bugging me!

Dusty.

For air cooling, the suggested maximum is below 1.5V (I.e. 1.45V / 1.47V).

Depending on what CPU cooler you are using, you may hit the temperature limit before the "recommended" voltage limit.

[TheDustyHeatsink]

For air cooling, the suggested maximum is below 1.5V (I.e. 1.45V / 1.47V).

Depending on what CPU cooler you are using, you may hit the temperature limit before the "recommended" voltage limit.

Yeah but talking theoretically here, what if I exceed the "suggested" voltage way before I'm even close to the temperature limit? For example, pushing 1.6/1.7V on my Phenom 965. What would be the next dangerous factor? (if any)... such a grey area.

[Speedbird]

Yeah but talking theoretically here, what if I exceed the "suggested" voltage way before I'm even close to the temperature limit? For example, pushing 1.6/1.7V on my Phenom 965. What would be the next dangerous factor? (if any)... such a grey area.

It could damage the CPU... A voltage too high can kill it. 1.6V is the ABSOLUTE MAXIMUM on any CPU and NEVER go over that.

[TheDustyHeatsink]

It could damage the CPU... A voltage too high can kill it. 1.6V is the ABSOLUTE MAXIMUM on any CPU and NEVER go over that.

That's all well and good but its still "it could"...what makes 1.6V the maximum? CPU's cooled on various liquid's easily pass the 1.6 mark and they don't instantly self destruct... Intel themselves actually have recommended in the past a max of 1.9V on ln2 for ivy bridge, so 1.6 is definitely not the "maximum" if you remove cooling as a factor..

[Speedbird]

That's all well and good but its still "it could"...what makes 1.6V the maximum? CPU's cooled on various liquid's easily pass the 1.6 mark and they don't instantly self destruct... Intel themselves actually have recommended in the past a max of 1.9V on ln2 for ivy bridge, so 1.6 is definitely not the "maximum" if you remove cooling as a factor..

Well, most of us don't have LN2 cooling, so I put 1.6 here.

[Linux User]

I remember on Linus's Intel stream, the Intel guy said even if you have infinite cooling, voltage will kill the chip

[-rascal-]

Yeah but talking theoretically here, what if I exceed the "suggested" voltage way before I'm even close to the temperature limit? For example, pushing 1.6/1.7V on my Phenom 965. What would be the next dangerous factor? (if any)... such a grey area.

 

The temperature scaling with increased voltage is NOT linear. It's exponential. There will be a point where a small amount in voltage adjustment will yield a HIGH temperature increase.

 

I remember on Linus's Intel stream, the Intel guy said even if you have infinite cooling, voltage will kill the chip

 

Correct.

 

 

 

Putting excessive voltage through a electronic device will cause Electronmigration.

This, over time, will cause permanent irreversible damage to the device / part.

 

This is why you see people with CPUs with overclocks that will all of a sudden become unstable.

 

Example:

Someone has their Phenom II X4 965 running a 4.2GHz for 3 years. At the time, it passed all vigorous stability tests. No problems for 3 straight years. Then out of the blue, the CPU decides to BSOD with the overclock settings -- or now requires more voltage to keep 4.2GHz stable.

[Tronnic]

The temperature scaling with increased voltage is NOT linear. It's exponential. There will be a point where a small amount in voltage adjustment will yield a HIGH temperature increase.

 

 

Correct.

 

 

 

Putting excessive voltage through a electronic device will cause Electronmigration.

This, over time, will cause permanent irreversible damage to the device / part.

 

This is why you see people with CPUs with overclocks that will all of a sudden become unstable.

 

Example:

Someone has their Phenom II X4 965 running a 4.2GHz for 3 years. At the time, it passed all vigorous stability tests. No problems for 3 straight years. Then out of the blue, the CPU decides to BSOD with the overclock settings -- or now requires more voltage to keep 4.2GHz stable.

 

Thanks! This helped me understanding it.

 

I also do have an question and I hope it's okay to ask here.

 

 

I have an i7 4790k. It just won't go higher then 4.4GHz (which is stock) without increasing the voltage. My default (auto) voltage at 4.4GHz is 1.252V. Is it save to go up to 1.4? Because when I ramp up the multiplier while leaving my voltage at auto it will ramp up to 1.375-1.400V at 4.8GHz. I once accidentally set the multiplier to 5GHz which was also stable but after I saw that it was running at 5GHz with voltage on auto I immediately shut down the system. Didn't even look at the voltage, lol.

 

Will "Electromigration" happen when it's running at 1.4V? What can I consider a "not damaging (or over time damaging)" voltage? My chip should last a few years

 

Edit: I guess that also happend to my old i7 2600k. It was running 3 years on 4.5GHz and all of a sudden 4.4GHz was all it could get.

[TheDustyHeatsink]

The temperature scaling with increased voltage is NOT linear. It's exponential. There will be a point where a small amount in voltage adjustment will yield a HIGH temperature increase.

 

 

Correct.

 

 

 

Putting excessive voltage through a electronic device will cause Electronmigration.

This, over time, will cause permanent irreversible damage to the device / part.

 

This is why you see people with CPUs with overclocks that will all of a sudden become unstable.

 

Example:

Someone has their Phenom II X4 965 running a 4.2GHz for 3 years. At the time, it passed all vigorous stability tests. No problems for 3 straight years. Then out of the blue, the CPU decides to BSOD with the overclock settings -- or now requires more voltage to keep 4.2GHz stable.

That's great, thanks for the help! 

[DildorTheDecent]

Hiii, I'm at a loss to find information on what is considered to be "safe" voltages? I'm certainly not a novice when it comes to computers, however as increasing the voltage ramps the temps up like crazy, is there considered to be any dangers of going pretty high other than temperatures? Say I use a ridiculous cooler, not sub-zero temps, but would there actually be any danger with putting the voltages to what are considered to be "pretty high", as long as the temperatures are okay? If not, how do I know where to draw the line? It seems like a bit of a grey area to me. I don't practically have this as a problem for overclocking my own set up, but its just been bugging me!

 

Dusty.

Up to 1.55V (absolute max)  for the AMD Phenom II x4 965 BE. Done a bit of research and some guys will do 4.5GHz at 1.55V. Yours may or may not do that (luck of the draw). 

 

The 140W model has a usual voltage range of: 0.850-1.425V

The 120W is lower at: 0.825-1.40V

 

There you go.