Significant Performance Gap between Benchmark and Actual Export on RTX 5070 Ti

GPU: NVIDIA GeForce RTX 5070 Ti (Driver: Latest Studio Driver)

CPU: AMD Ryzen 9 9900X (12-Core)

RAM: 32GB DDR5 6400MHz

The Issue: I am experiencing a massive performance discrepancy between the in-app Benchmark and actual rendering tasks.

Benchmark Results: Proteus 2X (1080P to 4K) shows 19.5 FPS.

Actual Export: The same model/settings only achieve 9.5 - 11 FPS.

Observed Bottlenecks:

CPU Usage: Stays at 60% during export.

GPU Usage: Hovers around only 40%, with temps remaining low (49°C).

Bus Interface Load: Reported at 100% in GPU-Z (potential Blackwell architecture reporting bug?).

It has been brought up before that the benchmark, being a synthetic test, isn’t a very accurate indicator of real world performance in many instances.

Have you looked at the individual core performance, to see if there are any threads that are close to maxing out?

Codec settings can have an impact too. In the past, changing over to ProRes export gave a speed boost (don’t know whether that still has a significant affect).