High end-to-end latency on non-B200 GPUs

I found that when running on GPUs other than B200, the actual network/display latency becomes significantly higher.

Is there a way to optimize this for lower-end GPUs? Also, would it be possible to lower the FPS or reduce the number of frames processed per step to improve real-time responsiveness and reduce accumulated latency?

The issue names no files, tests, or entry points. First reproduce the latency difference on a non-B200 GPU and compare network/display timing with B200; then locate the controls for FPS and frames processed per step. Done means lower-end GPUs show reduced accumulated latency without losing the intended real-time behavior.

Written by the indexing model from the issue text.