Add add-flashinfer-solution skill (end-to-end Solution authoring + benchmarking + visualization + docker image)
Summary
Adds a new agent skill add-flashinfer-solution under .claude/skills/
covering the Solution authoring side of the flashinfer-bench workflow:
pick a Definition → wrap a kernel → write Solution JSON → run benchmark
→ inspect traces → visualize
The existing skill suite covers the Definition / Workload / model side
(extract-kernel-definitions, collect-workloads, onboard-model,add-reference-tests, track-models, clone-repos,collect-workloads-bench). Solution authoring had no skill-level
documentation; this PR fills that gap.
Motivation
Three real solutions were authored end-to-end during this work:
| Solution | Definition | Verified on |
|---|---|---|
sdpa_paged_decode_v1 |
gqa_paged_decode_h32_kv8_d128_ps1 |
H100 PCIe |
fla_gdn_decode_v1 |
gdn_decode_qk4_v8_d128_k_last |
H100 PCIe |
sglang_mla_decode_v1 |
mla_paged_decode_h16_ckv512_kpe64_ps1 |
H100 PCIe + B200 |
Recurring patterns surfaced — KV-cache layout conversion, LSE base-2 vs
base-e, GQA H_kv → H_q expansion, FLA's transpose_state_layout flag,
vendored-Triton sys.path, NGC PyTorch + CuTe DSL ABI mismatch — were
costly to re-derive each time. This skill captures them as a reusable
playbook so subsequent Solution contributors don't relive the debug loops.
What's added
.claude/skills/add-flashinfer-solution/
├── SKILL.md 7-step workflow + gotchas
├── reference/
│ ├── definition_schema.md Definition JSON full schema
│ ├── solution_schema.md Solution JSON full schema
│ ├── wrapper_gotchas.md 17+ failure modes with diagnostics
│ └── visualization.md Web UI / public site / SSH forwarding
├── templates/
│ ├── dense_baseline_main.py + .json PyTorch SDPA-style baseline
│ ├── linear_attention_main.py + .json External lib wrapper (FLA)
│ └── vendored_kernel_main.py + .json Vendored Triton kernel (SGLang-style)
└── docker/
├── Dockerfile Pinned reproducible runtime
├── README.md Build / run / volume-mount guide
├── build.sh, run.sh Helper scripts
Docker environment
Pinned exact stack used to verify the three reference Solutions:
| Component | Version |
|---|---|
| Base | nvcr.io/nvidia/pytorch:24.10-py3 |
flashinfer-python |
==0.6.9 (pinned for CuTe DSL ABI compat) |
flashinfer-bench |
>=0.6.0,<0.7 |
flash-linear-attention |
>=0.5.0 |
nvidia-cutlass-dsl[cu13], cuda-python |
latest |
safetensors, huggingface-hub, git-lfs |
latest |
Build + smoke test:
cd .claude/skills/add-flashinfer-solution/docker
./build.sh # ~5-10 min
./run.sh --smoke # imports verified
Layout note for reviewers
Existing skills are single-file SKILL.md. This one uses subdirectories
(reference/, templates/, docker/) because:
- Schema reference docs (~200 lines each) are more useful as standalone
linkable artifacts than as inline sections of a longer SKILL.md - Wrapper templates need to be runnable as
.py/.json, not as
inline code blocks - The Dockerfile is consumed by
docker build, not read by Claude
Total content is ~1600 lines but distributed across 11 purposeful files.
Happy to flatten if maintainers prefer strict one-file convention; in
that case the templates and Dockerfile would still need to live as
separate executable artifacts somewhere.
How to test
1. Skill metadata sanity
head -4 .claude/skills/add-flashinfer-solution/SKILL.md
# Expect a valid YAML frontmatter with name + description
2. Docker build
cd .claude/skills/add-flashinfer-solution/docker
./build.sh
./run.sh --smoke
Smoke output should show: torch >= 2.7, CUDA: True, FLA: OK,fbench: present.
3. End-to-end Solution run inside the container
HOST_TRACE=/path/to/your/flashinfer-trace ./run.sh
# Inside the container:
cd /workspace/flashinfer-trace
flashinfer-bench run --local . \
--definitions gqa_paged_decode_h32_kv8_d128_ps1 \
--solutions <baseline_or_your_solution> \
--num-trials 1 --iterations 5 --timeout 300
4. Lint / format
No code changes; only markdown + Dockerfile + shell + json.
Related
- Existing skills suite: see
.claude/skills/and CLAUDE.md - Last skill-related PR: #384 (docs: add skill summaries and
cross-references) - Prior skill-restructure PR: #224 (referenced in #384)
---
## How to actually open the PR (recap of the procedure)
```bash
# 0. set git identity if not yet (replace with yours)
cd /home/scratch.yuny_wwfo/kernel_arena/flashinfer-bench
git config user.name "Your Name"
git config user.email "your@email"
# 1. fork on GitHub, then add as an additional remote
git remote add fork https://github.com/<your-gh-handle>/flashinfer-bench.git
# 2. create a clean branch from current main
git fetch origin
git checkout -b yyang/add-solution-skill origin/main
# 3. copy the skill in place
mkdir -p .claude/skills/add-flashinfer-solution
cp -r /home/yuny/kernel_arena/skills/add-flashinfer-solution/* \
.claude/skills/add-flashinfer-solution/
# 4. confirm only the skill files are staged (the working tree currently has
# unrelated 'deleted' entries from the flashinfer_trace symlink — DO NOT
# stage those)
git add .claude/skills/add-flashinfer-solution
git status
# 5. commit (paste the suggested commit message above)
git commit -F /home/yuny/kernel_arena/skills/add-flashinfer-solution/PR_DESCRIPTION.md
# (or interactively edit, taking just the commit-message section)
# 6. push to your fork
git push -u fork yyang/add-solution-skill
# 7. open PR via gh CLI (or GitHub web)
gh pr create \
--repo flashinfer-ai/flashinfer-bench \
--base main \
--head <your-gh-handle>:yyang/add-solution-skill \
--title "feat: add add-flashinfer-solution skill (end-to-end Solution authoring + benchmarking + visualization)" \
--body-file <(sed -n '/^## PR description/,/^---$/p' /home/yuny/kernel_arena/skills/add-flashinfer-solution/PR_DESCRIPTION.md | sed -n '/```markdown/,/^```$/p' | grep -v '^```')