Samsung SSD 860 EVO 250GB - major sudden corruption of random files - advice needed

[Shadowized]

After lots of prime95 testing, it seems that with 2 DIMMs it does not throw large FFTs errors. With 3 and 4 DIMMs, it throws.
I think I'm at 1.05V for PCH and VCCSA, RAM is at 1333MHz. What should I do? 1.1V is 10%, is it safe? Will I fry some chips?
GSkill didn't reply with voltage/config suggestions yet.

Sounds like increasing the VCCSA/VCCIO might help, what CPU/chipset are you on?

 

[eternalsadness]

3770k chipset Z77.

It happens that prime95 application crashes itself after a while. Yesterday it crashed during the day 1-2 times, over night it crashed again. What should I conclude?
But no memory errors...

 

[Shadowized]

Yesterday it crashed during the day 1-2 times, over night it crashed again.

Did those crashes occur with just 2 sticks installed, or was it with 4 still? Try looking for a BIOS update if one is available, then set the VCCSA at 1.1v (I wouldn't bother going higher) and try testing again.

 

[eternalsadness]

it has the latest BIOS. Those crashes occur with 2 sticks. With 4 sticks it gives errors in the workers, or black screen followed by reset.

 

[Shadowized]

Sounds like your CPU is degraded then, could try increasing that VCCSA and some of the other voltages (VTT, VCCIO) and if it still happens just buy a used one on ebay or something should be super cheap.

 

[eternalsadness]

I put VCCSA at 1.1V, I inserted 3 DIMMs and left it over night. One worker failed after 6 hours with the error that it could not allocate memory and that one possible cause for that is my swap memory being too small. And 2 hours later, another worker died with incorrect calculation.

 

[Shadowized]

I guess a memory leak somewhere like a browser or a task running in the background pushes prime95 over the limits it starts off with, I wouldn't pay too much mind to that specific error. The other worker crashing out though seems to me to be a CPU problem specially since you've gotten errors with just two DIMMs installed. How are the temps looking when running prime, any throttling?

 

[eternalsadness]

I guess a memory leak somewhere like a browser or a task running in the background pushes prime95 over the limits it starts off with, I wouldn't pay too much mind to that specific error. The other worker crashing out though seems to me to be a CPU problem specially since you've gotten errors with just two DIMMs installed. How are the temps looking when running prime, any throttling?

3 DIMMs. With 2 DIMMs I don't get errors. CPU is at 60C, I have a Scythe Mugen.

 

[Shadowized]

So with 3-4 DIMMs you get worker errors + crashes, and with 2 DIMMs you just get crashes? Are any memory dumps being written on the system, error codes in the event viewer with information about those crashes? Could be useful to debugging the issue.

 

[eternalsadness]

I will check for longs. Those crashes could be because I saw that swap was actually completely deactivated.

 

[Shadowized]

Yeah that's a possibility, the pagefile should should always be enabled even if it's set to a small size because some applications require it. I would stay with 2 DIMMs until you can figure out the cause of the worker errors though.

 

[eternalsadness]

Asus just replied to me that my RAM sticks are not on their QVL list and that's it.
They don't want to share any suggestions on how to improve the stability and get rid of the RAM errors.

But they work fine with either 2 DIMMs, but not with 4 DIMMs, so it's clearly a motherboard issue.

They gave me this link: https://dlcdnets.asus.com/pub/ASUS/mb/LGA1155/P8Z77-V_LX/P8Z77-V_LX_DRAM_QVL.pdf?model=P8Z77-V LX

 

[Shadowized]

That's not surprising, When your chipset was launched 8GB sticks were rare, I don't even think they existed outside of on servers with ECC. You can see this reflected in the QVL that the largest size sticks they tested were 4GB. Trying to drive 32GB in a 4x8config on that board is just asking for trouble because the IMC is so weak.