Benchmark #641

Download
osu!stable + stress-ng - EEVDF

Submitted 1 month ago by .dnaim

Specifications
Label OS GPU CPU RAM OS specific
eevdf-1.2ms-0.5ms-16 AMD Radeon Graphics powersave
eevdf-2ms-0.5ms-16-5ms AMD Radeon Graphics powersave
eevdf-2ms-0.5ms-16 AMD Radeon Graphics powersave
eevdf-2ms-0.3ms-16 AMD Radeon Graphics powersave
eevdf-2ms-0.5ms-8 AMD Radeon Graphics powersave
eevdf-cachy-2ms AMD Radeon Graphics powersave
eevdf-cachy AMD Radeon Graphics powersave
# Top runs: * **Highest FPS**: `eevdf-cachy` with an average FPS roughly 6% higher than the next best, showing the highest performance metrics across the runs. * **Smoothest FPS**: `eevdf-2ms-0.5ms-16` demonstrates the lowest FPS variance, making it the smoothest experience, with a variability about 5% lower compared to others. * **Best overall**: `eevdf-2ms-0.5ms-16` is considered the best overall due to its balance of stability and performance, offering slightly higher stability and moderately good FPS in relation to its peers. # Issues: * The run with the 2ms and 0.3ms scheduler configuration, though similar to other metrics, shows a consistently lower FPS performance, about 7% less than the `eevdf-cachy` configuration. * An outlier is detected with the `eevdf-2ms-0.5ms-16-5ms` run, exhibiting a 10% lower FPS than the `eevdf-cachy` despite similar testing conditions, suggesting non-optimal scheduler settings or hardware behavior under stress. # Summary This benchmark, focusing on the impact of different `eevdf` scheduler configurations on Linux with consistent hardware properties, reveals insightful distinctions between setups. The `powersave` scheduler shows varied FPS outcomes based on slightly different input settings, with the `eevdf-cachy` setup reaching peak performance concerning FPS. At the same time, the `eevdf-2ms-0.5ms-16` configuration offers notable stability, making it favorable for stable performance scenarios. Some configuration adjustments fall short in execution speed, likely due to nuanced scheduler interactions under high load and the test scenarios' specifics.