Gigabyte Z68AP-D3 Vcore stability problem

[emrz]

Hi everyone

First of all, you should know that i do know very little about overclocking

Well, i tried to oc my system according to the this thread "Beginners: How to set your 25/6/700K to 4.5Ghz" and i did whatever he said .And my problem is that I set vcore to 1.35v but it is not fixed and seen as 1.368v in bios and cpu-z, 1.38 in coretemp app.
I couldn't figure out why it changes this much. So if you can tell me what i did wrong , i would be very happy ps: I did not run the prime95 yet.
Thanks for your helps in advance

My sytem specs

i5 2500k
Gigabyte Z68AP-D3
Hyper 212 evo
Kingston 8 gb 1600 Mhz DDR3
Gtx 1050ti
Sufficient power supply (forgot the name )

 

[trents]

Set your LLC to a lower level. LLC (Load Line Calibration) supplements the vcore to try and keep it from drooping but if LLC is set too high it will actually raise the vcore from what you set it to manually.

But you need to first stress test the system to see if the vcore drops down under load before you change anything. Use Prime95 and run for 20 minutes. This will also give you a rough idea if the sytem is stable or not. Watch temps when you stress test. Don't let core temps exceed 90c. Use HWMonitor for monitoring temps and voltages during stress testing.

I would also set your RAM frequency and timings for optimum performance by enabling XMP.

 

[emrz]

I set my ram timings and enabled xmp. Then , i lower the llc to level 3 . Afterthat, i run the prim95 blend test , after 1 minute i got bluescreen i monitored the vcore it was between 1.38-1.40 even though i set it to 1.35v. i dont know where the problem is

 

[trents]

This is the problem with taking someone else's values off the internet and plugging them into your bios when overclocking.

The bluescreen could have been caused by several factors:

1. CPU core ratio being too high for the amount of vcore.
2. CPU core ratio being too high for the capability of that particular CPU. Not every 2500k is the same. There are variations in the quality of the silicon and the build process from batch to batch at the factory.
3. Motherboard VRM (voltage regulation module) being too weak and overheating
4. Temp threshold of the CPU cores is being exceeded.

You are doing this all wrong. Don't start at 45x. You should reset your core ratio to stock values then stress test and monitor temps . If it passes 20 minutes of Prime95 blend the bump the core ratio up by 1x and retest. Keep increasing the core ratio by 1x until you fail the test. Once that failure happens, increase your vcore by .01 increments and retest. If I recall correctly, you are safe in running a maximum 24/7 vcore of 1.45 on that generation Intel.

You will eventually hit a wall where either you reach the limit of the CPU or the motherboard or temps begin to exceed 90c. Make sure you have HWMonitor open while running the stress test.

 

[EarthDog]

I set my ram timings and enabled xmp. Then , i lower the llc to level 3 . Afterthat, i run the prim95 blend test , after 1 minute i got bluescreen i monitored the vcore it was between 1.38-1.40 even though i set it to 1.35v. i dont know where the problem is

1. Don't touch your RAM timings and then set XMP. Just set XMP... it does that for you!!!!!!


The reason why your voltage is fluctauting is likely the LLC as trents said.... mess with that.

 

[emrz]

okey trents, i'll go step by step and try to find the limits of my system, thanks for that. But , i still dont understand the reason of vcore unstability beacuse , even i lowered the llc , voltage spike didn't go away. Then i changed LLC to 5/10, raise the speed to 45 (vcore is set 1.35v) and i ran the prime95 blend test for 5 hours, did not get a errors or warnings. But , my core 1 hit 82 C (others in the inverval of 73-82). and voltages changed between 1.416-1.241. Idle voltage is 1.381v right now(regardles of the LLC level, it was alwasy 1.381v at idle, only if LLC level was below the level 3 , i got bluescreen immediately) Do you think these values are acceptable and can there be something else cause vcore unstability instead of LLC ?

 

[EarthDog]

Do Coretemp and CPUz and Hwmonitor agree on the core voltage? CPUz seems to match what you have. There will always be SOME variability, but not 1.24 to 1.35 to 1.41...