The GTX 1070 may not retain data from previous systems on a new setup?
The GTX 1070 may not retain data from previous systems on a new setup?
Hey there, I only share updates when things go wrong. Usually I manage everything myself, have built quite a few systems, and I think my motherboard might be behaving oddly with this card. Hope someone can assist me.
I’m using a Gigabyte xtreme gaming 1070, paired with an i9 10900k, Z490 MSI Tomahawk, 970 Pro NVMe, and the Noctua NH-DH-15 cooling solution. The only other component removed from my previous setup was the Cooler Master V650 PSU.
I remember when I first tried overclocking this GPU on an older system (i7 6700K Z270 PMCATE), changing voltage or power limits didn’t affect my clocks at all. That was surprising because this card is designed for overclocking and is one of the few 1070 models needing 8+6-pin connectors. Eventually, I realized I’d forgotten to replace the memory—it could handle more than 700 MHz, but only around 50 MHz on the core, regardless of voltage.
Now the card won’t run smoothly without stock clocks; it keeps crashing, as if it suddenly demands a different voltage. I haven’t added it yet because this seems unusual—sometimes it won’t even hold 300 MHz in memory or 30 MHz in the core. Could this be my motherboard trying to behave differently, or is the old Z270 PMCATe involved? I’m really confused.
Gpu is an add-on card that operates independently. The motherboard doesn’t influence the range of memory clicks or internal gpu voltages. Only specific OC software updates for Afterburner or Gpu Boost affect these settings. I owned an Asus Strix 970 DC2 that worked smoothly at 14% power but became unstable above 10% or 15%. That’s unusual, but it seemed to prefer that setting. You adjusted the card, swapped PSUs, changed PCIe versions, modified voltage LLC settings from the previous system. It could be a power stability issue, or simply the gpu not adapting well to the new configuration.
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.
Thank you for your detailed reply. However, as I mentioned earlier, it was maintaining these clocks well, then after two hours the performance dropped significantly (new system). I don’t believe any long-term damage occurred during the two-hour build period. Temperatures have stayed within normal ranges, reaching a peak of 71 at the end of the Heaven benchmark, while in-game they remain much lower.
Update: I maximized the power, voltage, and temperature settings (all set to default on the old system). It passed Heaven with 600 on memory, almost passed with 700, but still didn’t exceed 25 cores.
Gpu is an add-on card that operates independently. The motherboard doesn’t influence the range of memory clicks or internal gpu voltages. Only specific OC software updates for Afterburner or Gpu Boost affect these settings. I owned an Asus Strix 970 DC2 that worked smoothly at 14% power but became unstable above 10% or 15%. That’s unusual, but it seemed to prefer that setting. You adjusted the card, swapped PSUs, changed PCIe versions, modified voltage LLC settings from the previous system. It could be a power stability issue, or simply the GPU reacting poorly to the new configuration.