Ignore Userbenchmark altogether. it's just not good or reliable data. Frankly it's garbage.
Timespy provides several scores, including an overall score and the graphics score.
If your graphics score was 9650 on Timespy that's actually alright:
The teaser was no joke, NVIDIA today launches the first two cards in their new Super series. We review the GeForce RTX 2060 and 2070 SUPER and check out if they are actually that, super that is.... DX12: 3DMark Time Spy (DX12) and Port Royale (DXR)
If a Founders Edition (twin fan version of the 2070 Super) gets 9890, your stock-clocked, blower-cooled card getting only 2-3% lower is actually pretty decent.
Are you benchmarking for self-satisfaction purposes, or because you are concerned there is a specific problem?