Benchmark #1513

Download
Clone Drone in the Danger Zone: BORE vs SCX

CPU Limited static scene with no background load from a custom map. 30 second frame recording with 0ms interval. Performance Governor/EPP. SCX commit: 847e767 | Patched LAVD so that Preferred Cores are still identified correctly but not split into two groups since it is detrimental to performance when the difference is minor (due to binning).

Submitted 5 days ago by summedkibbles36

Specifications
Label OS GPU CPU RAM OS specific
BORE Freedesktop SDK 24.08 (Flatpak runtime) AMD Radeon RX 7900 GRE (RADV NAVI31) AMD Ryzen 7 7800X3D 8-Core Processor 33 GB 6.15.1-cachyos1.lto.fc43.x86_64 performance
BPFLAND -m all Freedesktop SDK 24.08 (Flatpak runtime) AMD Radeon RX 7900 GRE (RADV NAVI31) AMD Ryzen 7 7800X3D 8-Core Processor 33 GB 6.15.1-cachyos1.lto.fc43.x86_64 performance
FLASH -m all Freedesktop SDK 24.08 (Flatpak runtime) AMD Radeon RX 7900 GRE (RADV NAVI31) AMD Ryzen 7 7800X3D 8-Core Processor 33 GB 6.15.1-cachyos1.lto.fc43.x86_64 performance
P2DQ Freedesktop SDK 24.08 (Flatpak runtime) AMD Radeon RX 7900 GRE (RADV NAVI31) AMD Ryzen 7 7800X3D 8-Core Processor 33 GB 6.15.1-cachyos1.lto.fc43.x86_64 performance
Patched LAVD Freedesktop SDK 24.08 (Flatpak runtime) AMD Radeon RX 7900 GRE (RADV NAVI31) AMD Ryzen 7 7800X3D 8-Core Processor 33 GB 6.15.1-cachyos1.lto.fc43.x86_64 performance
Patched LAVD --performance Freedesktop SDK 24.08 (Flatpak runtime) AMD Radeon RX 7900 GRE (RADV NAVI31) AMD Ryzen 7 7800X3D 8-Core Processor 33 GB 6.15.1-cachyos1.lto.fc43.x86_64 performance
# Top runs: * **Highest FPS**: `BORE` with highest average FPS, significantly higher than other runs, indicating this setup offered superior overall performance. * **Smoothest FPS**: `Patched LAVD` exhibited the lowest standard deviation, providing the most consistent frame times amongst all runs. * **Best overall**: `BORE` achieved the best balance of high average FPS and reasonable stability, making it the best choice overall for this benchmarking test. # Issues: * The frame rate in the `BPFLAND -m all`, `FLASH -m all`, `Patched LAVD`, and `Patched LAVD --performance` configurations were significantly lower, around 10% to 15% below the higher-performing configurations. * `FLASH -m all` and `Patched LAVD --performance` exhibited a standard deviation in FPS that could indicate more inconsistency in performance compared to others, diminishing the experience with noticeable frame dips. * The CPU and GPU temperatures remained consistent across all benchmarks, showing no signs of thermal throttling impacting any specific configuration. This suggests the issue lies primarily with FPS performance. * Performance drops in certain configurations did not correlate with any change in hardware or software, indicating the intrinsic performance of the scheduler settings. # Summary This comparison evaluated the performance of different scheduler implementations in a CPU-limited, static scene within the game "Clone Drone in the Danger Zone", focusing on `BORE`, `BPFLAND -m all`, `FLASH -m all`, `P2DQ`, and `Patched LAVD` configurations, among others. The key focus was on maintaining high FPS and consistent frame times to ensure smooth gameplay. While `BORE` showcased superior FPS and `Patched LAVD` offered the most stable frame times, other configurations like `BPFLAND -m all` and `FLASH -m all` faltered in both categories, suggesting a lack of optimization for peak performance. Despite using a consistent hardware and software environment, certain scheduler configurations did not fully leverage the system's capabilities, highlighting the impact of scheduler choice on gaming experience.