Asus Cerberus 1070ti OC crashing

[RonnieLima]

Good Afternoon LTT,

 

I have been using an Asus Cerberus 1070ti for a little over 3 months now, been very happy thus far.  I have OC the card using MSI Afterburner with +135 Core Clock and +335 on Memory Clock, it was working perfectly up until about 1 week ago.  I have the Voltage and Power Limit maxed out, and has never caused an issue in the past.  The temp on the card rarely reaches 70C under load with the custom fan curve I have on, when the card does reach 70C or 71C the game crashes.  I have opened event viewer in Windows and it shows the NVIDIA driver has failed. I have not experienced any problems up until now, has anyone experienced this?  Do you think the GPU is not getting enough power from the PSU?

[TH3R34P3R]

i mean, it could be that it isnt getting enough power, but what happens when you remove the overclock and fan curve? does it still crash at 70c?

[RonnieLima]

The card typically dosnt reach those temperatures at stock, I have not experienced any crashes at stock at all.  

[-rascal-]

Your overclock *might* be not 110% stable.

Revert back to stock, and start overclocking one thing (GPU Core of Memory) at a time -- NOT both at once.

I am thinking the GPU Core overclock is unstable, rather than the Memory, but who knows.

 

I am not too familiar with the cooling design of the Cerbrus cards, but the VRMs might be running hotter than the GPU Core itself.

Core could be 70*C, but the VRMs (or the VRAM chips maybe) could be 90*C+ 

[Velcade]

How did you test your overclock?  Are you sure it's stable?  Run something like 3DMark Firestrike and see if it finishes

[RonnieLima]

51 minutes ago, -rascal- said:

Your overclock *might* be not 110% stable.

Revert back to stock, and start overclocking one thing (GPU Core of Memory) at a time -- NOT both at once.

I am thinking the GPU Core overclock is unstable, rather than the Memory, but who knows.

 

I am not too familiar with the cooling design of the Cerbrus cards, but the VRMs might be running hotter than the GPU Core itself.

Core could be 70*C, but the VRMs (or the VRAM chips maybe) could be 90*C+ 

This is what I thought, it was just weird that it only started happening.  Should I max out the Core Voltage and Power Limit?

[RonnieLima]

49 minutes ago, Velcade said:

How did you test your overclock?  Are you sure it's stable?  Run something like 3DMark Firestrike and see if it finishes

I run a full test with Unigine Heaven every time I change the clock, how do you suggest I do it?  Max out the Core Voltage and Power Limit and then do Core Clock and Memory Clock +5 at a time? Should I do +5 Core, run test, then +5 Memory?  Or find the max core first?  It was weird because this only started happening