| benchmark | General CPU, memory, and compression benchmarks: Geekbench, OpenSSL speed, lmbench memory bandwidth (bw_mem), and compression of the Silesia corpus. |
| benchmark-ffmpeg | FFmpeg transcoding-capacity benchmark of audio and H.264/H.265 video encoding. |
| benchmark-llm | Low-concurrency LLM inference speed benchmark with llama.cpp's llama-bench, measuring prompt processing and text generation across models and token counts. |
| benchmark-passmark | PassMark PerformanceTest CPU and memory benchmark suite. |
| benchmark-pgbench-postgres | PostgreSQL database benchmark focusing on CPU and memory performance, measuring transactions per second and latency against a local or remote server. |
| benchmark-redis | Redis 7.4 in-memory key-value store benchmark. |
| benchmark-storage | Raw block device benchmark with fio, testing each distinct disk type of the host and multi-disk groups concurrently. |
| benchmark-vllm-cpu | High-concurrency LLM inference benchmark with vLLM and GuideLLM, targeted CPUs with AVX-512. |
| benchmark-vllm-cpu-avx2 | High-concurrency LLM inference benchmark with vLLM and GuideLLM, targeted CPUs with AVX2 but no AVX-512. |
| benchmark-vllm-gpu | High-concurrency LLM inference benchmark with vLLM and GuideLLM, using tensor parallelism across all available GPUs. |
| benchmark-web | Static web server benchmark, serving files with binserve and measuring throughput and latency with wrk. |
| nvbandwidth | NVIDIA GPU memory bandwidth benchmark with nvbandwidth, measuring host-to-device and device-to-device copy speeds. |
| stress-ng | CPU benchmark running stress-ng stressors and counting their operations per second. |
| stress-ng-longrun | Pinned stress-ng version for the 24-hour long-running benchmark. |