Cpu relationship between volts, heat and stability

revolultrablue

Distinguished
Dec 18, 2010
253
0
18,790
I know volts are huge when it comes to stability, that's obvious. Less obvious is the relationship of heat to volts in it's relationship to stability.

Will less heat and more cooling allow for less volts along with more stability. Or is heat just potentially damaging and has no relationship to the amounts of volts needed to run a stable overclock?
 
Solution
This is a better question than you might think.

Talking about Intel current gen processors:

When overclocking, one wants to raise the multiplier. on all cores
Raising the multiplier requires more vcore to maintain stability.
This can usually be done via motherboard auto voltage settings.
Monitor this with cpu-z while stress testing.
The chips can tolerate perhaps 1.4v
Specs say 1.5v max, I think, and the long term damage of 1.4v is uncertain.
Speculation is that at 1.4v, the chip will be obsolete before it fails.
Higher vcore increases the heat generated in the cpu chip.
A hot cpu chip needs to be cooled sufficiently to keep from throttling.
.
The cpu will slow down or throttle if it detects a dangerous temperature.
That point is...
This is a better question than you might think.

Talking about Intel current gen processors:

When overclocking, one wants to raise the multiplier. on all cores
Raising the multiplier requires more vcore to maintain stability.
This can usually be done via motherboard auto voltage settings.
Monitor this with cpu-z while stress testing.
The chips can tolerate perhaps 1.4v
Specs say 1.5v max, I think, and the long term damage of 1.4v is uncertain.
Speculation is that at 1.4v, the chip will be obsolete before it fails.
Higher vcore increases the heat generated in the cpu chip.
A hot cpu chip needs to be cooled sufficiently to keep from throttling.
.
The cpu will slow down or throttle if it detects a dangerous temperature.
That point is about 100c
If you keep stress test temperatures at 85c. you will be ok.

Operating normally will be much less stressful than a stress test.
One can implement speedstep and adaptive voltage.
That will reduce the vcore and multiplier when the cpu has little to do.

Os overclocking worth it?
Absolutely, otherwise you are leaving some 25% performance on the table.

 
Solution

CompuTronix

Intel Master
Moderator
revolultrablue,

Your thoughts are correct. There are definite relationships between power, temperature and stability. Core voltage proportionately affects Power (Watts), which is expressed mathematically in "Watt's Law" as W = E x I; where "W" is Watts, "E" is volts and "I" is current (amps).

No two processors are identical; each processor is unique in voltage tolerance, thermal behavior and overclocking potential. As Core speed (MHz) is increased, Core voltage (Vcore) must also be increased to maintain stability. This increases Power dissipation (Watts) which increases Core temperatures.

Additionally, as Core temperatures increase, transistor "leakage current" also increases, which in turn increases power dissipation (Watts), resulting in a further increase of Core temperatures. In order to minimize this thermal "spiral", increased cooling capacity is required.

We know that excessive heat over time damages electronics, so excessive Vcore and Core temperature can result in accelerated electromigration, which is actually caused by current flow (amps).

The consensus among well informed and highly experienced system builders, overclockers and reviewers, is that cooler is better for ultimate stability, performance and longevity.

Check out our Guide: Intel Temperature Guide - http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html

Sections 7 & 8 have the info you want.

CT :sol: