kernel-workflow

Name: kernel-workflow
Availability: InStock
Author: MindSpore

Automate end-to-end AI kernel generation and optimization workflows across Triton, CUDA C, C++, and TileLang.

258|48|Updated Jun 22, 2020

Search Tags:#ai #task-orchestration #multi-backend #kernel-optimization #kernel-generation #ml-compiler

Ready to run

npx skills add https://github.com/mindspore-ai/akg --skill kernel-workflow-mindspore-ai

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: kernel-workflow
Source: https://github.com/mindspore-ai/akg/tree/main/akg_agents/python/akg_agents/op/resources/skills/kernel-workflow
Command: npx skills add https://github.com/mindspore-ai/akg --skill kernel-workflow-mindspore-ai

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

AI kernel development often requires coordinating generation, verification, and optimization across multiple backends. This skill provides end-to-end orchestration to streamline kernel engineering tasks with Triton, CUDA C, C++, and TileLang.

Core Features & Use Cases

Workflow orchestration: coordinates task flows, verification steps, and optimization strategies across multiple agents and references.
Multi-backend support: enables kernel generation and optimization for Triton, CUDA C, C++, TileLang.
Practical scenarios: user submits a kernel request, the skill orchestrates task building, verification, and deployment with deterministic outputs.

Quick Start

Automates end-to-end kernel generation and optimization workflows.

Frequently Asked Questions about kernel-workflow

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

FAQPage Schema

How do I automate AI kernel generation and optimization across multiple backends?▼

You can automate AI kernel generation and optimization by orchestrating end-to-end workflows across Triton, CUDA C, C++, and TileLang, coordinating task building, verification, and deployment with deterministic outputs.

What is the best way to coordinate verification and optimization for CUDA C and Triton kernels?▼

The best way to coordinate kernel verification and optimization is using a workflow orchestration skill that manages task flows and tool choices across multiple agents and references for deterministic execution.

Does this kernel workflow support generating and validating TileLang and C++ kernels?▼

Yes, this kernel workflow supports generating and validating TileLang and C++ kernels, applying task orchestration and tool selection to satisfy input handling and deterministic execution requirements.

How do I ensure deterministic execution when generating AI kernels across different backends?▼

You ensure deterministic execution during AI kernel generation by using frontmatter metadata, references, and optional scripts to coordinate task flows and tool choices across backends like Triton and CUDA C.

Can I use this workflow to streamline kernel engineering tasks without writing custom orchestration scripts?▼

Yes, you can streamline kernel engineering tasks without custom orchestration scripts by relying on built-in workflow coordination that automates task building, verification, and deployment across supported backends.