skills.rest
mindspore-ai
mindspore-aiOFFICIAL
Explore all skills by this author →
WebsiteGitHub

kernel-verifier

Validate generated kernel implementations via static analysis and accuracy verification.

258|48|Updated Jun 22, 2020
Search Tags:#frameworks#verification#kernel#AST#backends#triton#static-check
One-click install
npx skills add https://github.com/mindspore-ai/akg --skill kernel-verifier-mindspore-ai
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: kernel-verifier
Source: https://github.com/mindspore-ai/akg/tree/main/akg_agents/workspace/.opencode/skills/kernel-verifier
Command: npx skills add https://github.com/mindspore-ai/akg --skill kernel-verifier-mindspore-ai

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This Skill provides a two-stage verification workflow to ensure kernel implementations are correct by performing static checks and an accuracy comparison between generated and framework outputs.

Core Features & Use Cases

  • Static analysis: syntax, import validity, and code quality checks.
  • DSL compliance: verifies Triton kernel definitions and correct kernel launches.
  • Cross-backend validation: supports cuda/ascend/cpu with torch/mindspore backends for end-to-end kernel verification.

Quick Start

Run code_check.py to statically inspect your generated kernel code, then run verify.py to compare outputs against the framework results.

Frequently Asked Questions about kernel-verifier

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I statically check kernel code before execution?▼

To statically check kernel code, you run code_check.py to inspect syntax, validate imports, and verify AST-based code quality for your generated kernel implementations.

What is the best way to verify kernel accuracy across torch and mindspore frameworks?▼

The best way to verify kernel accuracy across torch and mindspore frameworks is running verify.py to compare generated outputs against framework results across cuda, ascend, and cpu backends.

Does the verification workflow support DSL compliance checks for Triton kernels?▼

Yes, the verification workflow supports DSL compliance by verifying Triton kernel definitions and ensuring correct kernel launches during the static analysis stage.

Can I use this for cross-backend validation on cuda, ascend, and cpu?▼

Yes, you can use this for cross-backend validation on cuda, ascend, and cpu, enabling end-to-end kernel verification and precision tests across multiple supported environments.

What are the two stages of kernel validation?▼

The two stages of kernel validation are static analysis for syntax and import checks, followed by accuracy verification comparing generated outputs against framework results.

Why do I need to create a validation project for kernel verification?▼

You need to create a validation project to establish the environment for executing precision tests and comparing end-to-end results between your generated kernel and framework outputs.

Related Skills

Explore More Software Engineering Skills→
mindspore-aimindspore-ai
258

triton-ascend-examples-mindspore

Register Triton Ascend custom operators with MindSpore primitives and tensor patterns.

Official
Intermediate
ali5terali5ter
4

obsidian-project-assistant

Generate and update project notes in an Obsidian vault from templates.

Community
Intermediate
hey-konghey-kong

add-sgl-kernel

Implement and integrate CUDA/C++ kernels with tests and benchmarks.

Community
Advanced
namninhndnamninhnd

lint-and-validate

Lint and statically analyze Node.js and Python projects with eslint, tsc, ruff, bandit, and mypy.

Community
Intermediate