Adjusting performance settings on the 'Gigabyte GeForce GTX 1070 G1 Gaming' (GV-N1070G1-GAMING)
Adjusting performance settings on the 'Gigabyte GeForce GTX 1070 G1 Gaming' (GV-N1070G1-GAMING)
I recently upgraded to this system and have been focusing on maximizing its capabilities. I ensured all previous drivers were removed and replaced with the latest versions during installation. I also enabled OC mode via Gigabyte XTREME GAMING software. After conducting several benchmarks, the results appeared modest compared to other systems online.
Begin at a low frequency, staying under 100MHz initially though that's a significant step. It's wiser to progress gradually than to jump immediately. This approach helps reduce the risk of hardware issues and OS corruption from repeated crashes. If the +100MHz phase remains stable, then advance in 25MHz steps. Make sure to verify performance with the games that were problematic beforehand. I'm aware that BF (3 and above) titles often encounter instability on both GPU and CPU.
I was thinking about focusing on the core overclock and keeping the memory unchanged. You should probably see almost no improvement in performance from tweaking the VRAM. This could help steady things if the issue started there. Usually, memory problems are the main culprit. Have you increased the power limit on your card? That would allow Turbo Boost to work more effectively. Increasing the GPU core voltage is generally bad for Turbo Boost because it can push the card closer to its power ceiling, leading to lower boost speeds. This applies to both Maxwell and Pascal cards. Temperature management is key here. Based on what you're observing, the card seems mostly stable at first, but after some time it starts to show artifacts and crashes. Leaving your memory clock untouched might help a bit, though it probably won't fully fix the crashes. Running a more aggressive fan profile could also help keep things cooler.
Techgeek would suggest focusing on the main overclock rather than altering the memory settings. Expect minimal performance gains from VRAM overclocking, which could help if instability comes from that area. Most issues seem to stem from memory problems. Check if you've increased the power limit for your card; this allows Turbo Boost more flexibility. Raising the GPU core voltage is generally harmful since Turbo Boost relies on it. It often leads to reaching power limits faster, causing lower boost speeds. This applies to both Maxwell and Pascal cards. Temperature management is crucial—your card stays under 70°C even at full load, which is impressive. Consider testing without memory overclocking to see if crashes decrease. For core clock settings, aim for a higher value if possible.
Begin at a low frequency, staying under 100MHz initially though that's a significant step. It's wiser to progress gradually than to jump immediately. This approach helps reduce the risk of hardware malfunctions and OS corruption from repeated crashes. If adding another 100MHz brings stability, proceed in 25MHz steps. Make sure to verify performance with the games that were problematic beforehand. I'm aware that BF (3 and above) titles often encounter instability issues on both GPU and CPU.
Start with a lower frequency, I wouldn't exceed 100MHz at first, though that's a significant jump overall. It's wiser to progress gradually instead of jumping immediately. This approach helps reduce the risk of hardware failure and OS corruption from repeated crashes. If +100MHz proves stable, then increase by 25MHz increments. Make sure to test with the games that were causing problems before. I notice that BF (3 and up) titles often struggle with instability on both GPU and CPU.
Remember that Heaven / Valley / 3DMark are completed in just a few minutes. This prevents the card from heating up. Although a card might stay stable during a quick warm-up up to 70°C, it could lose stability if left there for longer periods. Once this duration is reached, almost all parts beneath the heatsink become warm, including the PCB itself. This heat spreads to other components that generate little warmth of their own.
I’d guess that if you let Heaven / Valley run continuously for hours (without benchmarking), it would begin to act unpredictably over time.
Techgeek:
Keep in mind that Heaven / Valley / 3DMark finish quickly, which prevents the card from heating up. Even if it stays stable during a short warm-up to 70°C, prolonged exposure will heat up the entire setup, including the PCB. This warmth spreads to other parts that generate little heat themselves. I suspect running these tools together could cause issues after extended use.
I didn’t check whether running Gigabyte software and MSI simultaneously would conflict with each other.
I would choose MSI Afterburner, because it offers all the same features as GigaByte software but with extra capabilities. If you use both tools, any adjustments made in one will appear in the other. You can set up the default and OC modes using profiles in Afterburner. In short, default stays default, and to return to default just press the Reset button. Then configure your OC settings in Afterburner and save them as a profile.
techgeek:
I would stick with MSI Afterburner, it not only offers the same features as GigaByte software but also includes additional capabilities. If you use both, any adjustments made in one should appear in the other. The settings for default and OC modes can be created using a profile in Afterburner. Generally, default settings remain unchanged, while reverting to default simply requires using the Reset button in Afterburner. To switch to OC mode, save your changes as a Profile.
I chose to disable Gigabyte's software temporarily because I won't be using it and am currently running Heaven with a +125 on the core clock (2113MHz). It keeps looping without exceeding 65°C. So far everything is working well. I previously tried +135/150 and experienced some artifacts occasionally. I also switched to MSI's default fan curve instead of my previous setting.