flashinfer-bench
GPU kernel benchmarking, workload tracing, and model onboarding
All Skills in This Repository (3)
Pure Emerald Level Indicators
flashinfer-aiclone-repos
Clone and update SGLang and FlashInfer repositories with submodules from GitHub.
flashinfer-aiextract-kernel-definitions
Extract GPU kernel schemas from SGLang models into deduplicated Definition JSON files.
flashinfer-aiadd-reference-tests
Generate and validate pytest tests for flashinfer_trace reference implementations against FlashInfer or SGLang ground truth.
Frequently Asked Questions
FAQPage SchemaHow to install flashinfer-bench?▼
Run `npx skills add flashinfer-ai/flashinfer-bench --all -g -y` in your terminal to install all skills in this suite globally.
What does flashinfer-bench do?▼
It benchmarks GPU kernels for large language models against real inference workloads, checking both speed and correctness. It also manages a shared dataset of kernel definitions, workloads, and performance traces.
How to onboard a new LLM for kernel benchmarking?▼
Use the onboard-model skill, which runs a five-phase pipeline: clone repos, discover the model's kernels, generate definitions, collect workloads, and submit pull requests automatically.
Does flashinfer-bench work with Claude Code?▼
Yes. All skills follow the universal SKILL.md standard and run in Claude Code, Cursor, and other compatible coding agents.
Do I need a GPU to use flashinfer-bench?▼
A CUDA GPU is required for workload collection and benchmark runs, but dataset validation and definition editing can run with GPU checks disabled.
Related Repositories in Software Engineering
View All in Software Engineering→
openclawopenclaw
Run a personal AI assistant across your devices and chat apps
obrasuperpowers
Gives coding agents a disciplined workflow from idea to merged code
facebookreact
AI agent skills for building, testing, and porting React core