Memory generating bit-errors (not always) thus BSOD or app crash

[fslateef]

I will try this also, but previously IIRC when I increased CPU speed in NORMAL voltage mode its going up-to 1.49 and was increasing CPU temperature drastically.

Also I want to increase CPU-NB frequency also from 2G to 2.6G (or probably 2.4G at-least) as I noticed performance increase when changing CPU-NB like at-default I was getting 5GBps RAM speed in memtest86+ and with 2600 I am getting 5.9GBps even benchmark ratings and windows rating increased and noticed some frame increases in Crysis also.

I will try to set things at NORMAL and increase CPU to 3.8 and CPU-NB to 2.4 or 2.6 which your recommended voltages and will get back to you.

Thanks

 

[jaquith]

Then set the vCore manually; there's no need to over-volt/heat. The increase in CPU-NB was probably causing the errors -- the increase in CPU NB VID Control was to compensate.

Bottom-Line - what ever 'Works Best' --> use!

 

[fslateef]

You are 100% right Thanks for clearing my every confusion. And also after going through that 3-steps overclocking guide my knowledge increased a lot.

 

[jaquith]

Here's what I learned - there NO 'Exact' settings per MOBO/CPU/RAM/etc. When I give you or anyone a range e.g '±0.1~±0.2v' it's because if you bought 5 matching MOBO/CPU/RAM/etc -- none of them will be the same. I know the ranges, but haven't a clue of the outcome until it's tested.

I have a 'make it simple - stupid' until there's a need to 'fix it.'

OC'ing is a tedious task of 'Trial by Error'; there's Thermal variations, Voltage Variations, etc all of which is maddening. Oddly, sometimes a higher OC works better than a lower one?!

Phases, a MOBO with higher phases e.g. UD5 or UD7 typically can run with lower voltages e.g. vCore the UD3 is okay but how you OC a UD3 vs UD7 IS different -- vDroop.

 

[fslateef]

A quick update:

Yesterday I played with some more configurations and then I increased CPU freq to 3.8GHz and Voltage to 1.4250v (+0.050v) and its stable although during Prime95 temperatures was touching 60C and mostly on 58C. This means previously I was hitting maximum voltage settings for CPU frequency and increasing voltage by 0.050v I broken that barrier and can go beyond 3.7GHz (I will try 4GHz also today; just to see if that voltage can hadle 4GHz and temperature can reamin at-most at maximum 60C).

Just for information: for 2.6GHz of CPU-NB I can't reduce CPU-NB-VID voltage any more (even if I increase CPU vcore) thus for this 2.6GHz 1.175v is good. Less than that generates memory-errors which I was seeing before.

I think now I understand things really well. Thanks.

 

[jaquith]

You're getting a 'feel' for your MOBO and most importantly seem to be making good progress dialing-in your OC with improved Temps & most importantly overall stability.

Please keep the updates coming and good luck!

 

[fslateef]

A update and question:

My system is still not behaving well. Its crashing random applications, host-processes or games; due to again memory error 🙁 This seems to happen after playing Crysis for some hours but with Prime95 its fine even with custom Prime95 test using around 6.5GB RAM in it and remaining was already use by system and web-browser.

Now I reduced CPU-NB frequency to 2.4GHz (from 2.6GHz) and CPU frequency to 3.6GHz (from 3.7GHz); left the CPU-NB voltage as it is and increase CPU vcore to 1.425v. Things again seems stable and within some hours of Crysis nothing happened.

I will update status soon.

Now the question: Can this happening due to CPU issues/malfunction (means my CPU is broken like its IMC) ?

 

[fslateef]

I was browsing some forums and in one CPU IMC overclocking guide its written that Prim95 blend test is enough for testing CPU IMC; is this correct ? If yes then seems like I have some locations in a RAM module is bad. I will try to run memtest86+ again on single module at a time if my system/application crashes with the configuration mentioned above.

Thanks.

 

[fslateef]

I found the issue which was with one of my RAM module (have some bad cell locations I think thats why there was random BIT-ERRORS which triggers crash/BSOD if somehow application/system write and read something from those bad locations).

Thanks for help.

 

[Guest]

This topic has been closed by Grumpy9117