Benchmark #1219

Download
Teardown | BORE vs SCX

This was a static CPU limited scene with some background load composed of a Discord call + 1080p60fps CPU encoded streaming. 200 second frame recording with 0ms interval to get the most accurate results; single run only but duration should account for some amount of load variation. Performance Governor/EPP | BORE default settings | BPFLAND --performance | LAVD --performance | LAVD --autopilot | Flash default settings

Submitted 1 week ago by summedkibbles36

Specifications
Label OS GPU CPU RAM OS specific
Teardown BORE Freedesktop SDK 24.08 (Flatpak runtime) AMD Radeon RX 7900 GRE (RADV NAVI31) AMD Ryzen 7 7800X3D 8-Core Processor 33 GB 6.14.0-cachyos3.lto.fc43.x86_64 performance
Teardown BPFLAND Freedesktop SDK 24.08 (Flatpak runtime) AMD Radeon RX 7900 GRE (RADV NAVI31) AMD Ryzen 7 7800X3D 8-Core Processor 33 GB 6.14.0-cachyos3.lto.fc43.x86_64 performance
Teardown FLASH Freedesktop SDK 24.08 (Flatpak runtime) AMD Radeon RX 7900 GRE (RADV NAVI31) AMD Ryzen 7 7800X3D 8-Core Processor 33 GB 6.14.0-cachyos3.lto.fc43.x86_64 performance
Teardown LAVD Freedesktop SDK 24.08 (Flatpak runtime) AMD Radeon RX 7900 GRE (RADV NAVI31) AMD Ryzen 7 7800X3D 8-Core Processor 33 GB 6.14.0-cachyos3.lto.fc43.x86_64 performance
Teardown LAVD Autopilot Freedesktop SDK 24.08 (Flatpak runtime) AMD Radeon RX 7900 GRE (RADV NAVI31) AMD Ryzen 7 7800X3D 8-Core Processor 33 GB 6.14.0-cachyos3.lto.fc43.x86_64 performance
# Top runs: * **Highest FPS**: `Teardown FLASH` with significantly higher average FPS compared to other runs. * **Smoothest FPS**: `Teardown LAVD Autopilot` with the lowest standard deviation and variance in FPS, indicating more consistent frame rates. * **Best overall**: `Teardown BPFLAND` as it achieves a high average FPS while maintaining a relatively low variance, balancing performance and stability effectively. # Issues: * The `Teardown FLASH` run shows drastic variation in FPS, with a notably high standard deviation and variance, indicating inconsistency in performance. * All runs were carried out on identical hardware and software configurations, ensuring a fair comparison across different settings. # Summary This benchmark comparison of different performance settings for a CPU-limited scene analysis in *Teardown* points towards the utility of the `BPFLAND --performance` setting as a balanced option for achieving a high frame rate with moderate FPS consistency. While the `FLASH default` setting delivers the highest average FPS, its performance fluctuates significantly. Conversely, the `LAVD --autopilot` setting achieves the smoothest delivery of frames at a lower average FPS. Across the board, the tests were conducted using the same hardware and software environment, guaranteeing the validity of the observed performance results.