[SOLVED] GTX 1070 won't hold anything near same clocks on new system ?

[bishopi5]

Hey guys, i only post when i have odd issues, i can usually figure all this out on my own, built many many systems, i suspect my motherboard is just playing differently with my card? Im hoping someone can help

Its a gigabyte xtreme gaming 1070, my new system is i9 10900k, z490 MSI tomahawk, 970 pro nvme, noctuah ndh-15 the only other part that was taken from old system other then this GPU is the coolermaster v650 PSU

I noticed a long time ago when i first started overclocking this GPU on my old system (i7 6700k z270 pcmate) that changing voltage / power limit / anything would not affect my clocks, not at all. I found this strange as this card is specifically built for overclocking and is one of a few 1070's that require 8+6pin. I soon got over it as i realised i had samsumg memory and it would hold 700+ on memory, and measly 50 on the core (voltage or no voltage)


Now the card will not pass heaven without stock clocks, it keeps crashing, its like all of sudden it wants voltage? i have not started adding it yet as i find this weird, it literally will not even hold 300 on memory or 30 on core (seperately), could this be just my motherboard wanting to act differently? or the old z270 pcmate screwing around? im very confused

 

[Karadjgne]

OC isn't a guarantee. The only guarantee is the actual stock clocks and the silicon used is good enough to hold onto those clocks for the expected physical life of the gpu.

But you OC'd the card and possibly pushed the limits of the silicon used, maybe even lost integrity through electromigration, like the cathode in a florescent light bulb that turns purple as it degrades. Most ppl confuse Voltage with Power, it's not. Voltage is the multiplier. Amperage is the brute force. When you raise clocks and draw more power through the card, it's not the voltage that gets raised from stock, it's the amperage. Which equals a lot more heat.

With vram and VRM heat comes thermal pad degradation too, they doo get less resilient and less efficient with higher heat loads, and it isn't hard at all to imagine either or both hitting well above 90°C, even if the gpu (processor itself) is only registering in the 70's, simply from the added amperage the power used demanded.

You might want to look into replacing the pads and repasting and that might help, but failing that, it's a very real possibility that the silicon chips have suffered enough damage over the life of the card that you'll only now get whatever you can get.

There's a very big difference between 'specifically Built for OC' and 'specifically Marketed for OC', if you think about an Evga FTW, it's not specifically built for OC, it's specifically built for its factory OC, so includes the higher power plan, better cooling, custom pcb etc. But whether you get just a little more or a lot more out of the card isn't dependent upon its build design, or it's bigger power connectors, but on the chips themselves.

Specifically built for OC would be an Evga KO that has a 6+8pin and giant cooler vs a Evga SC that has a 6pin and tiny cooler, yet have identical clocks and performance at stock Evga settings.

 

[bishopi5]

Thankyou for the detailed resonse, But as i said, it was holding these clocks fine, then 2 hours later wasn't (New system) so i dont think long term damage has occured during the 2 hours build time. Temps have never been high, gets to 71 at the end of heaven benchmark, in game its much lower

Update : I maxed the power / voltage / temp sliders (all of which were on default on old system)

It passed Heaven with 600 on them memory, nearly passed with 700, but still won't go above 25 core

 

[Karadjgne]

Gpu is an add on card, it's self governing. The motherboard has nothing to do with how high or low memory clicks can be set at or internal gpu voltages.

Only different OC software versions of Afterburner or Gpu Boost etc have any input to those.

I had an Asus Strix 970 DC2 that purred like a kitten at 14% power, but was unstable at anything past 10% or 15% otherwise. Really wierd, but that's what it liked.

You moved the card, changed psus, changed pcie versions, changed voltage LLC etc from the old pc. It maybe just a power stability thing, no telling. It might be nothing more than the gpu simply doesn't like the new setup power delivery.