RTX 5090 score below expected

[David (PassMark)]

The 5090 didn't do so well (compared to 4090) in less complex scenes at lower resolutions (like our older DirectX9 & 10 tests). These older tests are more CPU bound as so the video card doesn't have so much work to do.

No point buying a 5090 is you are going to play an older game at 1080p I guess.

For the next major release of PerformanceTest (version 12) we are going to drop DirectX9 (and maybe 10). Plus also optimize DX12 to use less CPU time.

[a_good_person]

Cool, thanks for investigating! It's interesting how little the 5090 above the 5080, since the 5090 is just two 5080s on one die. What do you think the bottleneck could be?

[David (PassMark)]

We've got 4 samples now. One of them was very anomalous, with a frame rate of 17fps for the DirectX10 test. The expected result was more around 220fps.
This one,
https://www.passmark.com/baselines/V...d=238215916898

So something was very wrong with that machine or that test run. We've excluded it now.

Results look plausible now.


But it is early days. Result will jump around a bit until we get a dozen or more examples.

Wish I had the hardware to troubleshoot!

Indeed. We don't have one either. Seems like there is only a few thousand made worldwide.

[a_good_person]

With 3 samples, it's scoring at 82% of the RTX 5080 and RTX 4090. Something not right? Wish I had the hardware to troubleshoot!