Benchmark #1901

Download
Using LLVM-Objdump To Optimize Performance

Testing some optimization where llvm-objdump is used to make some refactors to the scheduler performance.

Submitted 1 week ago by ritzdacat

Specifications
Label OS GPU CPU RAM OS specific
scx_gamer_llvm_objdumprefactor1 CachyOS NVIDIA GeForce RTX 4090 AMD Ryzen 7 9800X3D 8-Core Processor 101 GB 6.17.7-5-cachyos performance
scx_gamer_llvm_objdumprefactor2 CachyOS NVIDIA GeForce RTX 4090 AMD Ryzen 7 9800X3D 8-Core Processor 101 GB 6.17.7-5-cachyos performance
scx_gamer_llvm_objdumprefactor3 CachyOS NVIDIA GeForce RTX 4090 AMD Ryzen 7 9800X3D 8-Core Processor 101 GB 6.17.7-5-cachyos performance
scx_gamer_llvm_objdumprefactor4 CachyOS NVIDIA GeForce RTX 4090 AMD Ryzen 7 9800X3D 8-Core Processor 101 GB 6.17.7-5-cachyos performance
scx_gamer_llvm_objdumprefactor5 CachyOS NVIDIA GeForce RTX 4090 AMD Ryzen 7 9800X3D 8-Core Processor 101 GB 6.17.7-5-cachyos performance
scx_gamer_1115B1 CachyOS NVIDIA GeForce RTX 4090 AMD Ryzen 7 9800X3D 8-Core Processor 101 GB 6.17.7-5-cachyos performance
scx_gamer_1115B2 CachyOS NVIDIA GeForce RTX 4090 AMD Ryzen 7 9800X3D 8-Core Processor 101 GB 6.17.7-5-cachyos performance
scx_gamer_1115B3 CachyOS NVIDIA GeForce RTX 4090 AMD Ryzen 7 9800X3D 8-Core Processor 101 GB 6.17.7-5-cachyos performance
scx_gamer_ltofat_cpunative_200_2000OC1 CachyOS NVIDIA GeForce RTX 4090 AMD Ryzen 7 9800X3D 8-Core Processor 101 GB 6.17.7-5-cachyos performance
scx_gamer_ltofat_cpunative_200_2000OC2 CachyOS NVIDIA GeForce RTX 4090 AMD Ryzen 7 9800X3D 8-Core Processor 101 GB 6.17.7-5-cachyos performance
scx_gamer_ltofat_cpunative_200_2000OC3 CachyOS NVIDIA GeForce RTX 4090 AMD Ryzen 7 9800X3D 8-Core Processor 101 GB 6.17.7-5-cachyos performance
scx_gamer_ltofat_cpunative_bugfixes111520251 CachyOS NVIDIA GeForce RTX 4090 AMD Ryzen 7 9800X3D 8-Core Processor 101 GB 6.17.7-5-cachyos performance
scx_gamer_ltofat_cpunative_bugfixes111520252 CachyOS NVIDIA GeForce RTX 4090 AMD Ryzen 7 9800X3D 8-Core Processor 101 GB 6.17.7-5-cachyos performance
scx_gamer_ltofat_cpunative_bugfixes111520253 CachyOS NVIDIA GeForce RTX 4090 AMD Ryzen 7 9800X3D 8-Core Processor 101 GB 6.17.7-5-cachyos performance
# Top runs: * **Highest FPS**: `scx_gamer_ltofat_cpunative_200_2000OC3` with approximately 7% higher average FPS than the next best run. * **Smoothest FPS**: `scx_gamer_llvm_objdumprefactor3` with about 60% lower standard deviation compared to the next best option. * **Best overall**: `scx_gamer_llvm_objdumprefactor3` as it maintains a balance of higher average FPS with about 70% smoother FPS than the average across all runs. # Issues: * The configurations using `scx_gamer_llvm_objdumprefactor4` and `scx_gamer_llvm_objdumprefactor5` report roughly 47% and 33% higher variance in FPS respectively compared to the average among the objdump refactor benchmarks, indicating inconsistent and less stable performance. * Similarly, the last set of configurations (`scx_gamer_ltofat_cpunative_bugfixes111520251`, `scx_gamer_ltofat_cpunative_bugfixes111520252`, `scx_gamer_ltofat_cpunative_bugfixes111520253`) encounter around 120% greater variance in FPS than some of the other streamlined configurations like `scx_gamer_ltofat_cpunative_200_2000OC1` and `scx_gamer_ltofat_cpunative_200_2000OC2`. # Summary This benchmark focuses on testing the performance improvements achieved using `llvm-objdump` to refactor schedulers on `CachyOS` with a focus on balancing FPS stability and overall performance. Key improvements are noted in the CPU-native configurations highlighted by `scx_gamer_ltofat_cpunative_200_2000OC3`, achieving the highest FPS. The configurations employing `llvm-objdump` refactor, specifically `scx_gamer_llvm_objdumprefactor3`, attain the smoothest FPS, showcasing more stability even with slightly less average FPS than the leading run. Among the lower-performing configurations, the last bugfix updates displayed inconsistent frame outputs, suggesting further refinement in balancing stability and resource management is needed.