Benchmark #611

Download
osu!stable + stress-ng - BORE vs bpfland vs lavd

Submitted 1 month ago by .dnaim

Specifications
Label OS GPU CPU RAM OS specific
BORE CachyOS AMD Radeon Graphics AMD Ryzen 5 6600HS Creator Edition 14 GB 6.11.4-1-cachyos-naim powersave
lavd CachyOS AMD Radeon Graphics AMD Ryzen 5 6600HS Creator Edition 14 GB 6.11.4-1-cachyos-naim powersave
lavd-tuning CachyOS AMD Radeon Graphics AMD Ryzen 5 6600HS Creator Edition 14 GB 6.11.4-1-cachyos-naim powersave
bpfland-k CachyOS AMD Radeon Graphics AMD Ryzen 5 6600HS Creator Edition 14 GB 6.11.4-1-cachyos-naim powersave
bpfland-k-L CachyOS AMD Radeon Graphics AMD Ryzen 5 6600HS Creator Edition 14 GB 6.11.4-1-cachyos-naim powersave
# Top runs: * **Highest FPS**: `BORE` with approximately 20% higher average FPS compared to others. * **Smoothest FPS**: `bpfland-k` has the lowest FPS variance and standard deviation, indicating greater stability. * **Best overall**: `bpfland-k` for its decent FPS and the best stability balance, with much smoother performance compared to `BORE`. # Issues: * The `lavd` run demonstrates a significant drop in FPS by nearly 40% compared to others, resulting in less smooth performance. * The `lavd-tuning` has a substantially larger frame time variability, with a frame time variance significantly higher compared to other runs, undermining performance consistency. * Both `lavd` and `lavd-tuning` configurations are inferior in GPU utilization, showing lower efficiency, especially in the `lavd-tuning` variant. # Summary This benchmark evaluates the performance of different Linux schedulers (`bpf land` and `lavd`) using the `powersave` governor on `CachyOS` with the AMD Ryzen CPU and Radeon GPU. The `bpf land` scheduler variants generally provided a better balance of FPS and stability, with the `bpfland-k` being particularly noteworthy due to its smooth and consistent performance. In contrast, the `lavd` variants struggled to maintain higher FPS and exhibited significant instability, indicating potential inefficiencies in system resource utilization.