Benchmark #661

Download
BORE 5.7.1 vs SCX schedulers in Rocket League / kernel 6.11.5

Rocket League first two minutes of match replay no additional workload Ryzen 5700X RTX 3060Ti EPP powersave / balance_performance

Submitted 4 weeks ago by yubysowhat

Specifications
Label OS GPU CPU RAM OS specific
BORE 5.7.1 CachyOS NVIDIA GeForce RTX 3060 Ti AMD Ryzen 7 5700X 8-Core Processor 17 GB 6.11.5-2611B571 powersave
Bpfland CachyOS NVIDIA GeForce RTX 3060 Ti AMD Ryzen 7 5700X 8-Core Processor 17 GB 6.11.5-2611B571 powersave
Bpfland-next CachyOS NVIDIA GeForce RTX 3060 Ti AMD Ryzen 7 5700X 8-Core Processor 17 GB 6.11.5-2611B571 powersave
Fair CachyOS NVIDIA GeForce RTX 3060 Ti AMD Ryzen 7 5700X 8-Core Processor 17 GB 6.11.5-2611B571 powersave
LAVD CachyOS NVIDIA GeForce RTX 3060 Ti AMD Ryzen 7 5700X 8-Core Processor 17 GB 6.11.5-2611B571 powersave
Rusty CachyOS NVIDIA GeForce RTX 3060 Ti AMD Ryzen 7 5700X 8-Core Processor 17 GB 6.11.5-2611B571 powersave
# Top runs: * **Highest FPS**: `BORE 5.7.1` with an approximately 2% higher mean FPS than the other runs. * **Smoothest FPS**: `Bpfland` showing 3% lower variance and standard deviation, indicating more consistent performance. * **Best overall**: `BORE 5.7.1` offers both the highest FPS and competitive stability with a variance only about 6% higher than the smoothest, striking a strong balance between raw performance and stability. # Issues: * The run using the `LAVD` scheduler has approximately a 3% lower mean FPS compared to others, coupled with noticeably higher variance and standard deviation, making it less stable. * Additionally, both the GPU load and mean GPU temp in the `LAVD` scheduler run were around 2-3% lower than other runs, possibly indicating underutilization. # Summary This benchmark evaluated Rocket League performance using different schedulers on `CachyOS` with the same hardware specifications, focusing primarily on the `BORE` scheduler versus implementations under `sched_ext`, including `Bpfland`, `Fair`, `LAVD`, and `Rusty`. The `BORE` scheduler showcased the highest average FPS, maintaining an optimal balance between raw performance and stability. On the other hand, the `Bpfland` scheduler proved to be the most stable, with minimal frame time fluctuation. The `LAVD` scheduler displayed noticeably lower performance in terms of FPS and consistency, which might be linked to sub-optimal resource utilization. Overall, these results highlight the impact of scheduler choice on gaming performance, emphasizing the trade-offs between high FPS and smooth gameplay experience.