dist-op-analysis

Name: dist-op-analysis
Availability: InStock
Author: MindSpore

Extract interface signatures, Primitive/ATen mappings, and sharding strategies for MindSpore mint and PyTorch ops.

7|3|Updated Feb 27, 2026

Search Tags:#sharding #dtensor #distributed-ops #op-analysis #interface-signatures #layout-inference

Ready to run

npx skills add https://github.com/mindspore-ai/hyper-parallel --skill dist-op-analysis

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: dist-op-analysis
Source: https://github.com/mindspore-ai/hyper-parallel/tree/main/.claude/skills/dist-op-analysis
Command: npx skills add https://github.com/mindspore-ai/hyper-parallel --skill dist-op-analysis

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This SKILL provides read-only analysis for HyperParallel distributed operator development. Given a MindSpore mint or PyTorch op name, it explores framework source code to extract interface specifications, Primitive/ATen mappings and HyperParallel layout derivation logic to support dist-op-dev workflows. It is internal and not intended for direct user invocation.

Core Features & Use Cases

Read actual source to extract full interface signatures, including parameter names, defaults and constraints for MindSpore mint and PyTorch ops.
Trace distributed implementation details and sharding strategies via YAML configs, mapping to HyperParallel's layout derivation logic.
Support workflow automation by feeding interface data, layout data, and expand logic into dist-op-dev pipelines for automated analysis.

Quick Start

Invoke the dist-op-dev workflow with a framework op name to generate its interface and layout analysis.

Frequently Asked Questions about dist-op-analysis

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

FAQPage Schema

How do I extract interface signatures and sharding strategies for distributed ops?▼

To extract interface signatures and sharding strategies for distributed ops, read the framework source code and HyperParallel YAML mapping files to compute infer_layout and get_expand_impl logic for MindSpore mint and PyTorch ops.

How does DTensor layout inference work for PyTorch distributed operators?▼

DTensor layout inference for PyTorch distributed operators works by tracing Primitive/ATen mappings and sharding strategies via YAML configs to map to HyperParallel's layout derivation logic and derive compatible layouts.

Can I analyze MindSpore mint ops for HyperParallel distributed development?▼

Yes, you can analyze MindSpore mint ops for HyperParallel distributed development by exploring the framework source to extract full interface specifications, parameter defaults, and layout derivation logic.

What do I need to compute infer_layout and get_expand_impl logic for a framework op?▼

To compute infer_layout and get_expand_impl logic for a framework op, you need access to the framework source code and the HyperParallel YAML mapping files to extract interface data and trace sharding strategies.

Why do I need YAML mapping files for distributed operator analysis?▼

You need YAML mapping files for distributed operator analysis because they contain the distributed implementation details and sharding strategies required to map framework ops to HyperParallel's layout derivation logic.