Can't get any stable overclock on Strix GTX 970

[Puntero]

Hello,

I would appreciate some expert tips regarding my attempts to overclock my Strix GTX 970 card. It seems like I can't keep my GPU stable under any significant overclock at all.
Last settings I applied were as follows:
Power: 120%
Core Clock offset: +125 (experiencing artifacts on +175, +150 is just not stable, but no artifacts).
Memory Clock offset: 0 (I've read that on cards like mine there is no decent cooling on VRAM, any comments on that, please?)
Core Voltage offset: 0 (don't want to increase it yet, although I'm considering it in general. Just not sure what values are safe and what are not).

With those values I'm still experiencing "hard" freezes after like half an hour playing some GPU-consuming games. Is it worth trying to get a stable overclock with <100 MHz Core Clock offset? I've configured my GPU fans so they keep my temp under 60-62 degrees under full load. My GPU's ASIC quality is 69.9%, if it matters somehow.

Thanks in advance!

UPD: My system specs:
CPU: i5-3570K @ 4.1 GHz (with auto voltage)
CPU Cooler: DeepCool IceEdge 300
MB: ASUS P8Z77-M
GPU: ASUS Strix GTX 970
RAM: 16 GB DDR3
PSU: AeroCool VP-650 (650 W)
Case: AeroCool Predator X1

I'm using MSI Afterburner, although I tried EVGA Precision X and ASUS GPU Tweak with same outcome.

 

[Johan45]

How's the airflow in your case? Seems like it runs OK for a while. Could be heat build up in the case since those cards vent inside not outside the box.

 

[Puntero]

I have fans on both front and rear panels of my case, it's not supercool in there but no overheating I guess.

 

[Johan45]

What SW are you using? And that thing about the memory is bunk. You should be able to get that up to about 8000 effective speed. Have you added any voltage?
What is your case.
Actually a good list of system parts would help alot

 

[Puntero]

Of course, I'm sorry.
Here you go:

CPU: i5-3570K @ 4.1 GHz (with auto voltage)
MB: ASUS P8Z77-M
GPU: ASUS Strix GTX 970
RAM: 16 GB DDR3
PSU: AeroCool VP-650 (650 W)
Case: AeroCool Predator X1

I'm using MSI Afterburner, although I tried EVGA Precision X and ASUS GPU Tweak with same outcome.
I added no voltage on GPU Core.

 

[EarthDog]

Aerocool psu? What model exactly? I don't recall those be I doubt it's your issue, but Imy curious anyway.


Add voltage bud. Just make sure you aren't hitting the power limit.

 

[Puntero]

www.aerocool.com.tw/de/psu/value-series/vp-650w
I can't access their website atm (522), but I guess that should be a correct link.
I'm not sure I understand what you mean by "what model exactly", I thought VP-650 was the model.

 

[EarthDog]

My fault. Missed the 'vp' part. I think those are Andyson units but cheapened.

 

[Puntero]

I added 20 mV and increased clocks +150 core / +400 memory, right now I'm "stress testing" it in Witcher 3 with all settings maxed out, temp is ~60 degrees. It hasn't crashed yet, I'll report back when it does

 

[Puntero]

It took surprisingly long for my PC to freeze but it did happen in the end. I'm afraid I can damage my computer doing so many hard resets during the last days.
The highest temperature I witnessed during my gameplay was 63 degrees on GPU and around 67 degrees on CPU. What do you guys suggest to do next?

 

[EarthDog]

Increase the voltage more... if it lasted a lot longer, clearly voltage is doing something.

What are you ACTUAL boost clocks? You can see this in the graph in msi ab...

Did you hit the power limit? I assume it's raised to the highest?

 

[Puntero]

The max power limit was 115% with +37 mV voltage according to the graph.

My clocks are 1466 MHz Core / 3903 MHz Memory (also from the graph).

Yes, I set Power Limit to 120% (max available value).

 

[Puntero]

Aaaand it hanged with +37 mV as well. It always happens some time after GPU temp reaches 63 degrees, so IMO VRAM accumulating heat is quite likely, but the same thing happens when I don't overclock memory at all.

 

[Johan45]

Two more things, well one's a repeat. What do you have cooling the CPU since from what I understand witcher3 hit the CPU pretty hard and I still think you could use more ventilation.
Try running with the side panel off your case and see if it runs longer .
One other question, have you ever managed to get it to run with a 100 MHz core increase or did you just go for broke?
I find with newer cards putting the voltage to max isn't always the best bet.

 

[EarthDog]

CPU hit 67C Johan...intel, so he's good there. You talking airflow around vrm?

 

[Johan45]

Just in general air flow with two fans??. The only temps I have seen mentioned are for GPU. Depending on CPU cooling all the hot air could be system lock up not card. Hard to say but running with the side off may show something and it's easy.

 

[EarthDog]

Post 10... 67C on cpu


Not disagreeing with what you are saying to do, note.

 

[Johan45]

I see it now thanks. Could be Auto voltage isn't giving the CPU enough under load too.

 

[Puntero]

My CPU cooler is DeepCool IceEdge 300, I usually don't experience CPU overheating (I think) since I'm not overclocking it too much and my fan is configured to spin fast and loud, I don't mind that. I also monitor the temps ingame.

I can take off the side panel but the problem is my PC is located in a way that there isn't much space by that side.

And no, I haven't tried to play with +100 MHz core clock since I believe cards like mine should do better, so I thought I was doing something wrong.

Also I hadn't experienced any stability issues playing same games before I tried to OC my GPU, so I don't think CPU auto voltage is to be blamed.

 

[EarthDog]

You still shouldn't use auto voltage.