Denna wrote:

When running a stress test, how can you tell which function needs to be adjusted when the system locks up, blue screens or reboots ?

Costas wrote:
Many times its not a clear cut answer - So you need to adjust via trial/testing etc...

There are specific tests which predominately test the various subsystems so these can help to some degree in tracking down what is causing the lockup eg mem Vs cpu etc.

Denna wrote:
Chino,

Let's say the Core, Uncore, System Agent or something else voltage related needs an adjustment.

When running a stress test, how can you tell which function needs to be adjusted when the system locks up, blue screens or reboots ?

Denna wrote:
Chino,

Let's say you have a stable overclock for daily use.

Then you run a stress test or benchmark that causes a crash.

Where do you start ?

It seems like overclocking is less scientific and more like random button mashing in a game.

I used to watch voltage changes of various components in AIDA64's stress test thinking that an insufficient voltage might be indicated during a stop of the stress test. But modifying voltages rarely produced an increase in stability. What that method did produce was a lot of unnecessary voltage increases - which were later reduced.

Raja@ASUS wrote:
Use a workflow like this to cut out some of the guesswork. This isn't down to the level of a complete amateur, but anyone with a bit of nous should understand:

http://edgeup.asus.com/2016/06/17/broadwell-e-overclocking-guide/6/

The rest of being good and efficient at overclocking is down to experience and understanding of subsystems. The type of stuff that needs a solid foundation in electronics that cannot be taught in the casual format of an overclocking forum. However, most people don't have that level of understanding but get by on the little correct hearsay there is posted by users on forums and manage to overclock just fine.