Congratulations in the exploration but can you try this? It's from a recent post I did.
--------In the WEAH errors say Kernel-Power----------
If your Event Viewer shows WHEA errors for CPU (repairable hardware errors and parity errors), your OC is not stable, even you run Prime for a month. In Sandy CPUs, that check was very weak and you needed hours of stressing with different applications to prove stability. With IVY CPUs things are better and errors reported in a better and more sensitive way. Download IntelBurnTest, start it, set 8 threads, choose Maximum stress test and ALL your available memory and press run. Run a couple of cycles and if you dont get a BSOD or a crash of the application, stop it, go to EVENT Viewer and check for WHEA errors. Increase your Vcore (manual or with offset), in any case of failure, with steps of 0.005 and repeat the stress test until you WHEA errors dissapear. Only then you can run your prime and be in a more safe zone of a good stability status. If you get an error message when you try to start the test about memory, change from drop down menu to High and again to Maximum a couple of times until it starts.
In parallel, also run Real temp to watch your temps and CPUz to check the variation of your Vcore. If it happens to have in your desktop the widget for memory usage, from windows, you will see that as the stress test go on, the memory usage increase until ALL is in use. The same time you will see that your Vcore decrease, and reach the lowest value for the settings you have used (depending your LLC value, Vcore and others).
This happens because at that time your CPU cores are under 100% load, but also the memory controller (inside the CPU, your Northbridge) draws the maximum power and current. So that's the point you reach the lowest Vcore and if this, is not enough you get WHEA errors. If it goes further down, you get an application crash. If the Vcore is way to low you get a nice BSOD.
If your Event Viewer shows WHEA errors for CPU (repairable hardware errors and parity errors), your OC is not stable, even you run Prime for a month. In Sandy CPUs, that check was very weak and you needed hours of stressing with different applications to prove stability. With IVY CPUs things are better and errors reported in a better and more sensitive way. Download IntelBurnTest, start it, set 8 threads, choose Maximum stress test and ALL your available memory and press run. Run a couple of cycles and if you dont get a BSOD or a crash of the application, stop it, go to EVENT Viewer and check for WHEA errors. Increase your Vcore (manual or with offset), in any case of failure, with steps of 0.005 and repeat the stress test until you WHEA errors dissapear. Only then you can run your prime and be in a more safe zone of a good stability status. If you get an error message when you try to start the test about memory, change from drop down menu to High and again to Maximum a couple of times until it starts.
In parallel, also run Real temp to watch your temps and CPUz to check the variation of your Vcore. If it happens to have in your desktop the widget for memory usage, from windows, you will see that as the stress test go on, the memory usage increase until ALL is in use. The same time you will see that your Vcore decrease, and reach the lowest value for the settings you have used (depending your LLC value, Vcore and others).
This happens because at that time your CPU cores are under 100% load, but also the memory controller (inside the CPU, your Northbridge) draws the maximum power and current. So that's the point you reach the lowest Vcore and if this, is not enough you get WHEA errors. If it goes further down, you get an application crash. If the Vcore is way to low you get a nice BSOD.
Hey fbm211,
You don't have to run stress-tests if your rig is stable for what you need... - anyway, passing a stress-test does not mean anything more than your system can pass it...
You don't have to run stress-tests if your rig is stable for what you need... - anyway, passing a stress-test does not mean anything more than your system can pass it...
Don't get me wrong guys. I am always happy with everyone's achievement. I just posted that, because in the beginning, I had some good OCs that later proved very optimistic. After I started using the above method, never had an issue. Again Congratulations 😄
nikosa43 wrote:
Don't get me wrong guys. I am always happy with everyone's achievement. I just posted that, because in the beginning, I had some good OCs that later proved very optimistic. After I started using the above method, never had an issue. Again Congratulations 😄
No problem, mate, I do agree with you! 🙂 And I was using in excess stress-testing to get my system stable... - until I arrived on this Forum and get involved in similar discussions and then tried what I learned here on my system...
For me, the main conclusion is: the best test method is everyday use. No matter how many stress-tests/benchmarks and how many hours/days you're running, that will prove only that your system can pass those tests... my point is that is no reason to expose your CPU to extreme heat for so long, when you can not be sure that that your settings are stable enough even in case of passing...
