The process of overclocking resulted in a lower firestrike rating.
The process of overclocking resulted in a lower firestrike rating.
Defined tK :
Vellinious :
On the top one, 9553....the GPU is running 1200 core? The physics score seems acceptable for the CPU clock. The GPU score appears poor. Something must have gone wrong during that run.
Middle score, 14298: That GPU score aligns more with what I'd anticipate from a 980ti at near stock clocks. Your CPU clock seems to be somewhat inconsistent, or it would have achieved higher scores in the physics section of the test.
Third score, 12103: The GPU score is disappointing. The physics score matches what I expect at that CPU clock. Now, your GPU clock seems unstable at 1527.
First step I'd take is to run DDU and uninstall all GPU drivers. Then perform a clean install. Trial and error... that's the essence of benchmarking. Discovering what works best for you and sticking with it.
You're clearly new to overclocking, so I recommend looking for a tutorial and going through it.
I'm not entirely sure this is related to overclocking.
The 14000 score came from this summer. I used an air cooler and ran everything stock. Now, when I attempt the stock benchmark, I get 9000. The 9000 uses stock default BIOS, fan curves, and afterburner settings.
To reach close to that 14000 again, I need to heavily overclock both CPU and GPU. Drivers didn't help. I plan to do a clean driver install; it might be the solution. After that, I'll format my hard drive and try once more.
I feel like the issue lies in the software—if my GPU were a brick, it would likely either function or fail completely. It would be odd if it got damaged and now performance drops significantly.
The Maswell architecture was sensitive to heat, but... not too critical. I'd probably have to consider the drivers next.
RAM speed has minimal impact on Firestrike... certainly not enough to explain the huge score differences.
The Maswell architecture was responsive to heat, though...not too severe. It seems the drivers might be the main factor.
RAM speed has minimal impact on Firestrike—it definitely isn't enough to explain the big score differences.
I share your thoughts in typical conditions, but excessive overclocking with RAM can significantly change outcomes, or even cause memory controller issues.
I've never noticed memory clock influence on firestrike ratings...actually, not firestrike.
I don't think painting the memory heatsinks is a good choice. They aren't painted—they're anodized aluminum, which is just bare metal designed for heat transfer and corrosion resistance. The color is just an extra feature. Your CPU and GPU both depend on consistent memory performance. You've set up a special situation with a specific challenge.