Soldato
- Joined
- 5 Feb 2009
- Posts
- 3,962
I've been trying to see how far I can push my 980 Ti at 3440x1440, but I've been a frustrated with the experience compared to how I used to overclock graphics cards in the past.
At stock (Asus Strix 980 Ti) I get core boosts of 1228-1240MHz.
Overclocking (with max, +110% power limit) seems to widen the range of boost speeds I'm seeing - and even more so with added core voltage.
So, at +240 Core speed, I see boosts from 1467 to 1505MHz in Superposition. Adding voltage I see lower low boosts and higher high boosts.
Adding memory speed seems to lower the boosts a little (possibly running into the power limit).
So I get it's variable, BUT:
Increasing the settings in Rise of the Tomb Raider gave me lower boosts in the benchmark!?! (1452-1490 at default graphics settings; 1414-1478 with higher textures, AA and HBAO, with more time spent at lower clocks).
This is not only confusing and frustrating, it makes it harder to test for stability. I had a driver crash in RoTR but it happened when the core clocks were boosting quite low (around 1414MHz) when they'd been sat over 50MHz higher for a while beforehand.
How do people manage this? And what's the best way a) to wring out the most consistent performance and b) to test to stability reliably?
At stock (Asus Strix 980 Ti) I get core boosts of 1228-1240MHz.
Overclocking (with max, +110% power limit) seems to widen the range of boost speeds I'm seeing - and even more so with added core voltage.
So, at +240 Core speed, I see boosts from 1467 to 1505MHz in Superposition. Adding voltage I see lower low boosts and higher high boosts.
Adding memory speed seems to lower the boosts a little (possibly running into the power limit).
So I get it's variable, BUT:
Increasing the settings in Rise of the Tomb Raider gave me lower boosts in the benchmark!?! (1452-1490 at default graphics settings; 1414-1478 with higher textures, AA and HBAO, with more time spent at lower clocks).
This is not only confusing and frustrating, it makes it harder to test for stability. I had a driver crash in RoTR but it happened when the core clocks were boosting quite low (around 1414MHz) when they'd been sat over 50MHz higher for a while beforehand.
How do people manage this? And what's the best way a) to wring out the most consistent performance and b) to test to stability reliably?