MindStudio Profiler
Ascend Performance Data Collection Tool
✨ What's New
- [2025.12.30]: Initial release of the MindStudio Profiler (msProf) project.
ℹ️ Overview
MindStudio Profiler (msProf) is a performance analysis tool for AI training and inference scenarios. It supports the collection and parsing of software and hardware performance data from the CANN layer and Ascend AI Processor NPU hardware layer, helping users identify performance bottlenecks during model training or inference. msProf also serves as the foundational capability for other profiling collection interfaces — many upper-layer performance collection and analysis capabilities ultimately rely on msProf for underlying data collection. For a complete picture of Ascend performance tuning tools, see MindStudio Profiler Documentation Overview.
⚙️ Functions
| Function | Description | Document | Source Code Repository |
|---|---|---|---|
| Performance Data Collection | Collects software and hardware performance data from the CANN platform and Ascend AI Processors using msprof commands. |
Performance Data Collection | msprof |
| Performance Data Parsing | Parses the collected performance data using the msProf tool to generate readable analysis results. |
Performance Data Parsing | analysis |
🚀 Quick Start
The msProf tool is invoked via the command line. The general collection command format is as follows:
msprof --output=<output_dir> --application="<application> <arguments>"
Examples:
# Example 1: Collect data for a Python task
msprof --output=./output --application="python3 train.py"
# Example 2: Collect data for an AI task launched by a shell script
msprof --output=./output --application="./run_standalone_train.sh"
Taking the ResNet-50 model training task as an example, Quick Start guides you through the full performance tuning workflow, helping you quickly experience the core functions of the msProf tool — including data collection, parsing and export, and performance analysis — within 10 minutes.
📦 Installation Guide
The msProf tool is included in the CANN Toolkit development suite. It is recommended to download the CANN package directly for installation. For details, see CANN Quick Installation.
To build and install from source code, see msProf Installation Guide.
📘 User Guide
For detailed usage instructions, see msProf User Guide.
💡 Typical Cases
Explore typical problem scenarios to understand and master the tool. See msProf Typical Cases.
❓ FAQ
For common questions and solutions, see msProf FAQ.
🌌 Smart Search
To improve documentation lookup efficiency, we offer several efficient search methods:
🔹 AI Q&A (DeepWiki): Natural language Q&A for quickly understanding project architecture and module relationships.
🔹 AI Q&A (ZRead): Optimized Chinese Q&A experience for precisely locating function usage and details.
🔹 Exact Search (ReadTheDocs): Full-text keyword search for direct access to APIs, parameters, and error messages.
🛠️ Contributing Guide
We welcome contributions to the project. See Contributing Guide.
⚖️ Legal Notices
🤝 Suggestions and Feedback
We welcome everyone to contribute to the community. If you have any questions or suggestions, please submit Issues. We will reply as soon as possible. Thank you for your support.
| Instant Messaging (WeChat Group) | Official News (Official Account) | In-Depth Support (Assistant / Forum) |
|---|---|---|
 Scan to join the technical discussion group |
 Scan to follow the official account |
Scan the QR code to join the group and follow the official account for the fastest way to connect with the MindStudio user and developer community: Quick Questions: Discuss technical issues with community members in real time Stay Updated: Get notified about version releases and feature updates firsthand Share Experience: Exchange best practices and practical insights with fellow developers More Support Channels: 👉 Ascend Assistant:  👉 Ascend Forum:  |
🙏 Acknowledgments
This tool is developed by the following Huawei department:
- Ascend Computing MindStudio Development Department
Thank you to everyone in the community for your PRs. We warmly welcome your contributions.