Overclock SLI 970s?
Overclock SLI 970s?
I adjusted the settings a bit more this morning before deciding to stop. Without raising the voltage, I managed to keep my cards running smoothly at +80 core clock and +375 memory clock. This improved my FPS in the Heaven Benchmark at 1440p by 7 frames per second.
This result isn't as impressive as what others with identical cards could accomplish, but it's still better than anything stable in my system. It might be related to the attempt to overclock two cards together. Either way, it's a small boost without a major jump.
The change in performance was likely much more noticeable in gaming compared to previous benchmarks like Heaven or Star Wars Battlefront. At high resolutions and full settings, I typically achieved 95-105 FPS, while playing for about two hours I maintained around 120-130 FPS. My turbo clock was set to 1440MHz and memory clock to 1940MHz. It seems unlikely that the modest overclocking added just a 25 FPS improvement. I plan to try Fallout 4 later today, as I’ve been playing it more frequently recently. If I no longer experience drops into the low 40s with shadow distance at maximum, it would confirm the impact of this overclock.
High settings at high resolution could be responsible for the frame rate increase. Evga cards are typically clocked quite high from the start. It doesn't surprise me that you didn't notice a significant GPU speed boost. Perhaps. I didn't observe much variation in Fallout 4. With shadow distance set to maximum, I still experience frame rate drops into the 40s in cities and around 60 fps elsewhere. From what I've learned, this pattern occurs with any hardware level. I'm curious to see similar results in other games.
I made a few adjustments earlier this morning before deciding to stop working on it permanently. Without raising the voltage, I managed to keep my cards stable at +80 core clock and +375 memory clock. My FPS improved in the Heaven Benchmark at 1440p by 7 frames per second. This performance is better than what some users with identical cards could achieve, though anything above that wasn't consistent in my setup. It might be related to the attempt to overclock two cards together. Regardless, it still provides a noticeable improvement without being a major jump.
Before you give up, I have another rig using an i7-4790k. I’m realizing now that to get a stable overclock, I’ll likely need a power supply rated for around 1000 watts. In testing, one card reached 1404 stock while the other hit 1342, so I adjusted the settings in MSI Afterburner to manage core and memory frequencies. This should help with the display driver issues.
Jayhawker32 shared his experience with some improvements, noting stable performance at certain settings without needing higher voltage. He mentioned a slight gain in FPS and observed that other users with similar cards achieved better results. He also discussed power requirements for overclocking and noted differences in performance between his 970s and i7-4790k systems.
King3pj shared some observations about their system performance after adjusting settings. They noted stable card operation at certain clock speeds and a slight improvement in FPS during benchmarking. They mentioned challenges with overclocking two cards simultaneously and the need for higher power supplies. Additionally, they discussed differences in performance when using SLI versus single GPU and how boosting one card's speed affected overall results.
It seems you're questioning whether it's worth pushing the better card beyond its stable limit. If the slower card handles most tasks, boosting the faster one further might not significantly improve performance during games.
I believe you now grasp what I'm referring to. You're suggesting I should focus on overclocking the slower card instead of the faster one.
It makes sense in this context. But with my +80MHz boost, both cards can reach around 1440MHz. This exceeds the 1392MHz boost clock I get on my faster card without any overclocking.
I suppose if I only overclocked the slower card, it would cap at 1392MHz boost, making my faster card the bottleneck. Then its performance would drop noticeably—almost 50MHz slower than what I can achieve by boosting both to +80MHz.
I'm still unclear on the reasoning. I haven't tried overclocking much because the gains in most games didn't justify the extra heat. If I couldn't sustain 60FPS at 1440p, I'd likely limit the overclock to specific titles.
It seems that if I only boosted the slower card to match the 1392MHz boost speed, my faster card would keep up with its original performance, giving me a 26MHz boost in SLI without significant temperature rise.
I believe you're grasping the idea now. You're suggesting I should focus on boosting the slower card instead of the faster one.
It makes sense in this context. But with my +80MHz boost, both cards can reach around 1440MHz. This is higher than the 1392MHz boost clock I get on my faster card without overclocking.
I think if I only boosted the slower card, they'd both cap at 1392MHz because that's what my faster card achieves at normal speeds.
Then my faster card would become the bottleneck, and I'd be running nearly 50MHz slower than my current performance when both are overclocked to the same stable +80MHz setting.
I'm still unclear on the reasoning. I haven't tried overclocking recently because I didn't notice enough performance gains in most games to justify the extra heat. If I couldn't consistently hit 60FPS at 1440p, I'd probably limit the boost to match my faster card's stock speed.
That would let my faster card keep up with its normal pace while still benefiting from the boost.
In short, what I did was run each card separately so both could operate at their optimal speeds without one trying to outpace the other. If I hadn't done that, the system would have been unstable and drivers would often crash.