uProf User Guide
Advanced Micro Devices
Contents
- 1. About AMD uProf
- 2. Hardware and Software Requirements
- 3. Installing AMD uProf
- 4. Performance Characterization Using AMDuProfPcm
- 4.1. Overview
- 4.2. Performance Monitoring Counters (PMC)
- 4.3. Key Features
- 4.4. Prerequisite(s)
- 4.5. AMDuProfPcm Command Line Options
- 4.6. Metrics
- 4.7. Interpreting Profile Data
- 4.8. Virtualization Support
- 4.9. Known Behavior - Issues Due to BIOS Settings
- 4.10. Constraints and Limitations
- 4.11. Monitoring Without Root Privileges
- 5. Performance Modelling using AMDuProfPcm
- 6. Application Analysis - Getting Started
- 6.1. Profiling Concepts
- 6.2. Profile Types
- 6.3. Predefined Sampling Configuration
- 6.4. Predefined View Configuration
- 6.4.1. Assess Performance Configurations
- 6.4.2. Threading Configuration
- 6.4.3. Overview Configuration
- 6.4.4. Investigate Data Access Configurations
- 6.4.5. Investigate Branch Configurations
- 6.4.6. Assess Performance (Extended) Configurations
- 6.4.7. Investigate Instruction Access Configurations
- 6.4.8. Investigate CPI Configurations
- 6.4.9. Instruction Based Sampling Configurations
- 6.5. Preparing an Application for Profiling
- 6.6. Workflow
- 7. Investigate Performance Issues
- 7.1. Profiling Support on Linux for perf_event_paranoid Values
- 7.2. Configure Profile
- 7.3. AMDuProf Overhead Estimation
- 7.4. Analyze Profile Data
- 7.4.1. Overview of Performance Hotspots
- 7.4.2. Thread Concurrency Graph
- 7.4.3. Function HotSpots
- 7.4.4. Process and Functions
- 7.4.5. Source and Assembly
- 7.4.6. Top-down Callstack
- 7.4.7. Flame Graph
- 7.4.8. Call Graph
- 7.4.9. All Thread Timeline
- 7.4.10. Per Thread Timeline
- 7.4.11. Notes on GPU Acceleration
- 7.4.12. IMIX View
- 7.4.13. Wait Object Hotspots
- 7.5. Hotspots Analysis
- 7.6. Threading Analysis
- 7.7. Overview Analysis
- 7.8. Time-based Profiling
- 7.9. Micro Architecture Analysis
- 7.10. Parallelism - OpenMP Analysis
- 7.11. Parallelism - MPI Trace Analysis
- 7.12. Parallel Strong Scaling Metrics (MPI + OpenMP)
- 7.13. Accelerators
- 7.14. Other Analysis
- 7.15. Custom Profile
- 8. Code Profiling
- 9. AMD uProf CLI Options
- 10. AMD uProf GUI Reference
- 11. AMD uProf Application Programming Interfaces
- 11.1. CPU Profile Control APIs
- 11.2. Instrument APIs - Domain and Task
- 11.3. Instrument APIs - Event
- 11.3.1. Definition and Concepts
- 11.3.2. Overall workflow with small example
- 11.3.3. How to Run and Get Task/Event detail
- 11.3.4. Compiling Instrumented Target Application
- 11.3.5. Profiling Instrumented Target Application
- 11.3.6. Disabling Instrumentation APIs
- 11.3.7. Example Steps to Attach Instrumented Process
- 11.3.8. Unmatched Task/Event
- 11.3.9. Depth Within Domain
- 11.3.10. Pedantic Mode support
- 11.3.11. Output
- 11.3.12. Limitations
- 11.4. OneAPI support in AMDuProf
- 12. Live Power Profile
- 13. Performance Debug using AMDuProfSys
- 14. Reference
- 15. Appendix - AMDSystemCheck Utility
- 16. Additional Resources and Legal Notices
Legal Notice
Trademarks
AMD, the AMD Arrow logo, and combinations thereof are trademarks of Advanced Micro Devices, Inc. Dolby is a trademark of Dolby Laboratories.
ENERGY STAR is a registered trademark of the U.S. Environmental Protection Agency. HDMI is a trademark of HDMI Licensing, LLC.
HyperTransport is a licensed trademark of the HyperTransport Technology Consortium.
Microsoft, Windows, Windows Vista, Windows Server, Visual Studio,and DirectX are registered trademarks of Microsoft Corporation.
MMX is a trademark of Intel Corporation.
OpenCL is a trademark of Apple Inc. used by permission by Khronos. PCIe is a registered trademark of PCI-Special Interest Group (PCI-SIG).
Other product names used in this publication are for identification purposes only and may be trademarks of their respective companies.
Dolby Laboratories, Inc.
Manufactured under license from Dolby Laboratories.
Rovi Corporation
This device is protected by U.S. patents and other intellectual property rights. The use of Rovi Corporation’s copy protection technology in the device must be authorized by Rovi Corporation and is intended for home and other limited pay-per-view uses only, unless otherwise authorized in writing by Rovi Corporation.
Reverse engineering or disassembly is prohibited.
USE OF THIS PRODUCT IN ANY MANNER THAT COMPLIES WITH THE MPEG-2 STANDARD IS EXPRESSLY PROHIBITED WITHOUT A LICENSE UNDER APPLICABLE PATENTS IN THE MPEG-2 PATENT PORTFOLIO, WHICH LICENSE IS AVAILABLE FROM MPEG LA, L.L.C., 6312 S. FIDDLERS GREEN CIRCLE, SUITE 400E, GREENWOOD VILLAGE, COLORADO 80111.
Revision History
Date |
Revision |
Description |
|---|---|---|
May 2026 |
5.3 |
Documented 5.3 features. |
December 2025 |
5.2 |
Documented 5.2 features. |
June 2025 |
5.1 |
Documented 5.1 features. |
October 2024 |
5.0 |
Documented 5.0 features. |
January 2024 |
4.2 |
Minor edits and updates. |
August 2023 |
4.1 |
Documented AMD uProf 4.1 features. |
November 2022 |
4.0 |
Included AMD uProf 4.0 features. |
July 2022 |
3.6 |
|
January 2022 |
3.5 |
Documented AMD uProf 3.5 features. |
April 2021 |
Initial |
Documented AMD uProf 3.4 features. |