If you have the option to adjust the voltage, do it in small increments while testing and watching the temps. Good luck, let us know how you get on and give us a shout if you get stuck I have vram used, load %, cpu % etc (settings - monitoring tab, select the readings you want to see and tick the box below that says 'show in on-screen display'.) It's worth setting up the on-screen display to watch temperatures. If you go too far you might notice glitches/artifacts, or the graphics driver will just crash, you'll get a black screen for a while and when it comes back it will have reset the clocks on the card.
If that's all fine, start applying an offset to the gpu and memory, do them one at a time - find the max for gpu, drop it back to 0 and move to the memory (I'd go up in 100mhz increments at first, when it crashes start going up in smaller increments), then once you found the edge of stability for both, set them a little below maximum to avoid crashes. Then adjust your power limit if it's a modern card with gpu boost, apply and play games/run benchmark to check thermals are still okay, try to stay <80c if you can. Generally it's just a case of setting a custom fan curve, or manually apply a fan %. You need something to stress the card to to check stability, you can use a game with vsync off, or download a benchmark like Unigine Heaven.
0 kommentar(er)
