KernelWiki

KernelWiki — Blackwell & Hopper Kernel Optimization Wiki

Query a structured, cross-referenced knowledge base of GPU kernel optimization for NVIDIA Blackwell (SM100) and Hopper (SM90). The repository update date is recorded in README.md; run python3 scripts/repo_status.py for current corpus counts.

When To Use This Skill

Trigger this skill when the user asks about:

• Blackwell/SM100 kernel programming — tcgen05.mma, TMEM, CLC, 2-SM cooperative, NVFP4, FP8/FP4 block scaling, PDL/GDC
• Kernel implementations — FlashAttention-4, DeepGEMM, FlashMLA, NSA, GatedDeltaNet, NVFP4 GEMM/GEMV, fused MoE, gated dual GEMM
• Performance patterns — low SM utilization, memory-bound, register pressure, compute-bound, tail effects, pipeline stalls
• DSLs for Blackwell — CuTe DSL, CUDA C++ with PTX inline, Triton on Blackwell
• Hopper → Blackwell migration — wgmma → tcgen05, register → TMEM accumulators
• PR references — "how did vLLM/SGLang/FlashInfer/CUTLASS/PyTorch implement X for SM100?"
• Competition context — GPU Mode NVFP4 hackathon and FlashInfer MLSys 2026 task definitions, speed-of-light data, and dated results snapshots; participant write-ups are separate blog sources

Do NOT use this skill for:

• Generic CUDA questions unrelated to Blackwell/Hopper tensor cores
• Host-side framework integration (model loading, request routing, scheduling policy)
• Distributed systems topics — DeepEP, EPLB, DualPipe are out of scope

How To Query

All commands below run from the skill directory (the clone root — the directory this SKILL.md lives in). The scripts auto-resolve the wiki root; no environment variable required.

Runtime dependencies

The query and maintenance scripts are self-contained. They use the host PyYAML package when present and fall back to the bundled pure-Python loader and dumper when it is missing. Invoke them directly with python3; do not assume that pip, a virtualenv, network access, or a pre-installed PyYAML package is available. requirements.txt is optional and only installs the faster host implementation.

Path 1: Unified search (preferred for natural language)

python3 scripts/query.py "how to fuse gate-up dual GEMM on Blackwell"
python3 scripts/query.py --tag nvfp4 --type kernel
python3 scripts/query.py --repo cutlass --limit 20
python3 scripts/query.py --symptom tail-effect --compact

Filters: --type, --tag, --repo, --language, --architecture, --symptom, --confidence, --limit, --compact, --paths-only. --tag and --architecture accept aliases — --tag UMMA matches tcgen05, --architecture B200 matches sm100, etc.

Path 2: Fetch a specific page by id or path

python3 scripts/get_page.py kernel-flash-attention-4
python3 scripts/get_page.py pr-cutlass-2472
python3 scripts/get_page.py kernel-flash-attention-4 --follow-sources
python3 scripts/get_page.py kernel-flash-attention-4 --body-only

Path 3: Regex text search across wiki bodies and PR pages

python3 scripts/grep_wiki.py "tcgen05" --only wiki
python3 scripts/grep_wiki.py "two-CTA" --only wiki
python3 scripts/grep_wiki.py "nvfp4" "block_scale" --any

Path 4: Pre-built cross-reference indices

Auto-generated under queries/:

• queries/by-architecture.md — exact SM, family-only, and validated-unknown architecture evidence
• queries/by-problem.md — symptom → pattern page → candidate techniques
• queries/by-technique.md — 17 techniques with architectures, confidence, reproducibility, source count
• queries/by-hardware-feature.md — tcgen05/tmem/clc/tma/nvfp4/etc. → related wiki + PR pages
• queries/by-kernel-type.md — gemm/attention/moe/mla/gated-delta-net → pages
• queries/by-language.md — cute-dsl/cuda-cpp/ptx/triton → guide page + related kernels/sources
• queries/by-repo.md — PR pages grouped by source repository

Path 5: Primer, schema, examples

Companion docs under references/:

• references/primer.md — topic map: hardware features, techniques, symptoms, canonical page IDs. Read this first when the question is broad.
• references/schema.md — condensed frontmatter schema, confidence rules, reproducibility ladder, controlled vocabulary, canonical aliases.
• references/examples.md — 10 worked query patterns mapping user questions → command sequences → synthesis.

Output Pattern

When answering from this KB:

1. Cite specific pages with paths (e.g., wiki/kernels/flash-attention-4.md) and IDs (kernel-flash-attention-4).
2. Follow sources: fields to trace claims back to PRs/blogs/docs.
3. Respect confidence levels — verified > source-reported > inferred > experimental. Call out when a claim is experimental or inferred.
4. Include code snippets from wiki pages when they exist — technique/kernel/language pages are guaranteed snippet-reproducibility (validator-enforced).
5. Report performance claims with all six fields — gpu, dtype, shape, metric, value, source_id.

Knowledge Base Contents

• Source PR pages, synthesized wiki pages, blog/doc/contest summaries, candidate ledgers, query indices, and artifact bundles.
• Verbatim upstream asset bundles in artifacts/ (PR patches and complete kernel files or excerpts) — pinned to upstream SHAs via PROVENANCE.yaml
• Auto-generated query indices in queries/
• Controlled vocabulary (80+ tags) in data/tags.yaml, alias map in data/aliases.yaml
• Hybrid version-claim registry — per-page version_sensitive: <id> pointers + data/version-claims.yaml central registry, validated for bidirectional consistency
• Status script scripts/repo_status.py — current corpus counts
• Validator scripts/validate.py — schema, link, artifact, and ledger checks
• Blackwell-first — Hopper-only wiki pages carry explicit blackwell_relevance; source pages preserve upstream evidence and are exempt

To refresh the corpus: run scripts/refresh_candidate_ledger.py, regenerate PR pages and query indices, then validate.

Quality Guarantees

• Every verified page has official-doc + upstream-code evidence
• Every technique/kernel/language page has a compilable snippet
• Every PR page has inclusion_reason and an evidence-backed status; current distribution: 942 merged, 2 closed without merge
• All Hopper-only wiki pages have explicit blackwell_relevance; source pages are exempt