MindStudio Kernel Performance Prediction

✨ Latest News

🔹 [2025.12.31]: MindStudio Kernel Performance Prediction is fully open-sourced.

️ ℹ️ Overview

MindStudio Kernel Performance Prediction (msKPP) is a performance simulation tool that supports quick prediction of the peak performance of an operator based on the given algorithm implementation. The execution time is estimated based on the input/output scale, without actual computation. The result can be returned in seconds, and the simulation speed is several orders of magnitude faster than that of the cycle-level simulator.

⚙️ Features

Function	Description
Operator Feature Modeling	Simulates the operator time consumption based on the APIs provided by msKPP.
Operator Computing and Transferring Specification Analysis	Generates the transfer pipeline statistics file and instruction information statistics file to view the msKPP modeling result.
Peak Performance Analysis	Generates the instruction pipeline diagram and instruction proportion pie chart to view the msKPP modeling result.
Preliminary Design of Operator Tiling	Quickly filters out optimal tiling policies.

🚀 Quick Start

For details, see msKPP Quick Start.

📦 Installation Guide

This section describes the environment dependencies and installation methods of the msKPP tool. For details, see msKPP Installation Guide.

📘 User Guide

For details about how to use the tool, see msKPP User Guide.

📚 API Reference

The msKPP tool provides two types of APIs: basic APIs and instruction APIs. For details, see msKPP API Reference.

🌌 Smart Search

To improve the efficiency of document retrieval, we provide the following efficient search methods:
🔹 AI Q&A (DeepWiki): Natural language Q&A to quickly grasp the project architecture and module relationships.
🔹 AI Q&A (ZRead): Better Chinese Q&A experience, precisely locating feature usage and details.
🔹 Precise Search (ReadTheDocs): Full-text keyword search, directly accessing APIs, parameters, and error messages.

🛠️ Contribution Guide

You are welcome to contribute to the project. For details, see Contribution Guide.

⚖️ Related Notes

🔹 Release Notes 🔹 License Notice 🔹 Security Statement 🔹 Disclaimer

🤝 Suggestions and Communication

You are welcome to contribute to the community. If you have any questions or suggestions, please submit an Issues. We will reply as soon as possible. Thank you for your support.

Live Chat (WeChat Group)	Official Info (WeChat Official Account)	In-Depth Support (Assistant/Forum)
Scan to join the tech group	Scan to follow the official account	Scan to join the group and follow the official account for the fastest way to reach the MindStudio user and developer community: Quick Questions: Discuss technical issues with community members in real time Stay Updated: Get version release and feature update notifications first Share Experience: Exchange best practices and practical insights with developers More Support Channels: 👉 Ascend Assistant: 👉 Ascend Forum:

🙏 Acknowledgements

This tool is jointly developed by the following Huawei departments:
🔹 Ascend Computing MindStudio Development Department
🔹 Ascend Computing Ecosystem Enablement Department
🔹 Huawei Cloud AI Compute Service
🔹 Compiler Technologies Lab, 2012 Labs
🔹 Markov Lab, 2012 Labs
Thank you to everyone in the community for your PRs. We warmly welcome your contributions.