Installing AMD uProf - Installing AMD uProf - 5.2 English - 57368

uProf User Guide (57368)

Document ID: 57368
Release Date: 2026-03-11
Version: 5.2 English

uProf User Guide
About AMD uProf
1. About AMD uProf
1.1. User Interfaces
1.2. Hardware Sources
1.2.1. Performance Monitor Counters (PMC)
1.2.2. Instruction-Based Sampling (IBS)
1.2.3. L3 Cache Performance Monitor Counters (L3PMC)
1.2.4. Data Fabric Performance Monitor Counters (DFPMC)
1.3. Support
Hardware and Software Requirements
2. Hardware and Software Requirements
2.1. Processors
2.2. Operating Systems
2.3. Compilers and Application Environment
2.4. Virtualization Support
2.5. Container Support
Installing AMD uProf
3. Installing AMD uProf
3.1. Installing Using Windows
3.1.1. Sample Programs
3.2. Installing Using Linux
3.2.1. Installing Using TAR File
3.2.2. Installing Using a RPM Package (RHEL)
3.2.3. Installing Using a Debian Package (Ubuntu)
3.2.4. Installing Power Profiling Driver on Linux
3.2.5. Linux Power Profiling Driver Support for DKMS
3.2.6. Sample Programs
3.3. Installing Using FreeBSD
3.3.1. Sample Programs
Performance Characterization Using AMDuProfPcm
4. Performance Characterization Using AMDuProfPcm
4.1. Overview
4.2. Performance Monitoring Counters (PMC)
4.3. Key Features
4.3.1. Multiplexing
4.3.2. Profiling Data Collection and Reporting
4.3.2.1. Perf Mode (Only available on Linux)
4.3.2.2. MSR Mode
4.3.3. Data Collection and Reporting
4.3.4. Process Tracking (Only available in Perf mode)
4.4. Prerequisite(s)
4.4.1. Windows
4.4.2. Linux
4.4.3. FreeBSD
4.4.3.1. Synopsis
4.4.3.2. Common Usages
4.5. AMDuProfPcm Command Line Options
4.5.1. Commands
4.5.2. Examples
4.5.2.1. Linux and FreeBSD
4.5.2.2. Windows Commands
4.6. Metrics
4.6.1. Performance Metrics for AMD EPYC™ Zen 2 Core Architecture Processors
4.6.2. Performance Metrics for AMD EPYC™ Zen 3 Core Architecture Processors
4.6.3. Performance Metrics for AMD EPYC™ Zen 4 and later Core Architecture Processors
4.7. Interpreting Profile Data
4.7.1. Pipeline Utilization
4.8. Virtualization Support
4.9. Known Behavior - Issues Due to BIOS Settings
4.10. Constraints and Limitations
4.11. Monitoring Without Root Privileges
Performance Modelling using AMDuProfPcm
5. Performance Modelling using AMDuProfPcm
5.1. Classic Roofline Model
5.2. Command Line Options
Application Analysis - Getting Started
6. Application Analysis - Getting Started
6.1. Profiling Concepts
6.1.1. Sampling
6.1.2. Sampling Interval
6.1.3. Sampling-point
6.1.4. Event-Counter Multiplexing
6.2. Profile Types
6.2.1. Time-Based Profile (TBP)
6.2.2. Event-Based Profile (EBP)
6.2.3. Instruction-Based Sampling (IBS)
6.2.4. Custom Profile
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
6.5.1. Generating Debug Information on Windows
6.5.2. Generating Debug Information with Inline functions on Windows
6.5.3. Generating Debug Information on Linux
6.6. Workflow
6.6.1. Collect Phase
6.6.1.1. Profile Target
6.6.1.2. Profile Type
6.6.1.3. Profile Configuration
6.6.1.4. Profile Session (or Profile Run)
6.6.2. Translate and Report Phases
6.6.3. Analyze Phase
6.6.3.1. View Configuration
6.6.4. Export Session
6.6.4.1. Common Usage
6.6.5. Import Session
6.6.6. Remote Profile
6.6.6.1. Overview
6.6.6.2. Setting up Authorization
6.6.6.3. Launching AMDProfilerService
6.6.6.4. Launching AMDProfilerService with IPv6
6.6.6.5. Connecting to a Remote Target
6.6.6.6. Limitations
Investigate Performance Issues
7. Investigate Performance Issues
7.1. Profiling Support on Linux for perf_event_paranoid Values
7.2. Configure Profile
7.2.1. Select Profile Target
7.2.2. Select Profile Type
7.2.3. Advanced Options
7.2.4. Start Profile
7.3. AMDuProf Overhead Estimation
7.4. Analyze Profile Data
7.4.1. Overview of Performance Hotspots
7.4.1.1. Summary Overview
7.4.1.2. OS Trace Screen
7.4.1.3. GPU Trace Screen
7.4.1.4. MPI Trace Screen
7.4.1.5. CPU Profile
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
7.4.9.1. Timeline Analysis GUI in Linux
7.4.9.2. Region Selection
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
7.5.1. User Mode Sampling
7.5.2. Callstack Stitching
7.5.3. Data Collection
7.5.3.1. Data Collection Using GUI
7.5.3.2. Data Collection Using CLI
7.5.4. Analyze the Data
7.5.4.1. CLI Report
7.5.4.2. Identify the Hottest Function
7.5.4.3. Identify the Hot Code Paths
7.5.5. Troubleshoot
7.5.6. Limitations
7.6. Threading Analysis
7.6.1. User Mode Tracing
7.6.2. Data Collection
7.6.2.1. Data Collection Using GUI
7.6.2.2. Data Collection Using CLI
7.6.3. Analyze the Data
7.6.3.1. CLI Report
7.6.3.2. Threading Summary
7.6.3.3. Wait Time Analysis
7.6.3.4. Timeline
7.6.4. Troubleshoot
7.6.5. Limitations
7.6.6. Reference
7.6.6.1. Posix Thread APIs
7.6.6.2. libc System Call Wrapper APIs
7.7. Overview Analysis
7.7.1. Prerequisites
7.7.2. Data Collection
7.7.2.1. Data Collection Using GUI
7.7.2.2. Data Collection Using CLI
7.7.3. Analyze the Data
7.7.3.1. CLI Report
7.7.3.2. GPU Offload Analysis
7.7.3.3. Function Trace Analysis
7.7.3.4. Timeline
7.7.3.5. Limitations
7.8. Time-based Profiling
7.8.1. Configuring and Starting Profile
7.8.2. Analyzing Profile Data
7.9. Micro Architecture Analysis
7.9.1. Overview
7.9.1.1. Predefined Sampling Configuration
7.9.1.2. Predefined Core PMC Events
7.9.2. Analysis with Event-based Profiling
7.9.2.1. Data Collection
7.9.2.1.1. Data Collection Using GUI
7.9.2.1.2. Data Collection Using CLI
7.9.2.1.3. Microarchitecture Analysis Options
7.9.2.2. Analyze Data
7.9.2.2.1. CLI Report
7.9.2.2.2. Identify the Hot Code Paths
7.9.2.2.3. Advisory
7.9.2.3. Limitations
7.9.3. Analysis with Instruction Based Sampling
7.9.3.1. Data Collection
7.9.3.1.1. Data Collection Using GUI
7.9.3.1.2. Data Collection Using CLI
7.9.3.2. Analyze Data
7.9.3.2.1. CLI Report
7.9.3.2.2. Identify the Hot Code Paths
7.9.3.2.3. ASCII Dump of IBS Samples
7.9.3.2.4. IBS Derived Events
7.9.3.2.5. Limitations
7.9.3.3. Cache Analysis
7.9.3.3.1. Supported Metrics
7.9.3.3.2. Cache Analysis Using GUI
7.9.3.3.3. Cache Analysis Using CLI
7.9.3.4. Branch Analysis
7.9.3.4.1. Prerequisites
7.9.3.4.2. Configuration
7.9.3.4.3. Analyze the Data
7.9.3.4.4. Limitations
7.9.3.5. Virtualization Support
7.9.3.5.1. Profiling of Guest VM from Guest VM
7.9.3.5.2. Profiling of Guest VM from Host System (KVM Hypervisor)
7.9.3.5.3. Preparing Host system to Profile Guest Kernel Modules
7.9.3.5.4. AMD uProf CLI with Profiling Options
7.10. Parallelism - OpenMP Analysis
7.10.1. Data Collection Using GUI
7.10.2. Data Collection Using CLI
7.10.3. Analyze the Data
7.10.4. Generate Profile Report
7.10.5. Environment Variables
7.10.6. Limitations
7.11. Parallelism - MPI Trace Analysis
7.11.1. MPI Light-weight Tracing Using CLI
7.11.2. MPI Full Tracing Using CLI
7.11.3. MPI Full Tracing Report Visualization Using GUI
7.11.4. MPI Runtime Library Mismatch (Undefined Symbol Errors)
7.12. Parallel Strong Scaling Metrics (MPI + OpenMP)
7.12.1. Overview
7.12.2. Hierarchy
7.12.3. Metric Definitions
7.12.4. Interpretation Guidelines
7.12.5. Preconditions & Reporting Behavior
7.12.6. Concepts & Attribution Model
7.12.7. Usage Workflow
7.12.8. Optimization Focus Examples
7.12.8.1. Examples
7.12.9. Limitations
7.12.10. Best Practices
7.13. Accelerators
7.13.1. GPU Profiling
7.13.1.1. Prerequisites
7.13.1.2. Optional Settings
7.13.1.3. Supported Options
7.13.1.4. Supported Events and Metrics
7.13.1.5. GPU Profiling Using CLI
7.13.1.6. GPU Profiling Using GUI
7.13.1.7. Analyze Data
7.13.1.7.1. Using CLI
7.13.1.7.2. Using GUI
7.13.1.8. Identify the Hottest GPU Kernel
7.13.1.9. Limitations
7.13.2. GPU Offloading Analysis (GPU Tracing)
7.13.3. Prerequisites
7.13.4. Optional Settings
7.13.5. Supported Events
7.13.6. Data Collection Using CLI
7.13.7. Data Collection Using GUI
7.13.8. Analyze the Data
7.13.8.1. Using CLI
7.13.8.2. Using GUI
7.13.8.3. Limitations
7.14. Other Analysis
7.14.1. Function Tracing
7.14.1.1. Prerequisites
7.14.1.2. Data Collection Using CLI
7.14.1.3. Data Collection Using GUI
7.14.1.4. Function Tracing Options
7.14.1.5. Analyze the Data Using CLI
7.14.1.6. Analyzing Data Using GUI
7.14.1.7. Limitations
7.14.2. Memory Tracing Options
7.14.2.1. Data Collection using CLI
7.14.2.2. Analyze the Data Using CLI
7.14.2.3. Limitations
7.14.3. Pagefault Tracing
7.14.3.1. Prerequisites
7.14.3.2. Data Collection using CLI
7.14.3.3. Analyze the Data Using CLI
7.14.3.4. Limitations
7.14.4. Kernel Block I/O Analysis
7.14.4.1. Prerequisites
7.14.4.2. Data Collection Using GUI
7.14.4.3. Data Collection Using CLI
7.14.4.4. Analyze the Data Using CLI
7.14.4.5. Analyze the Data Using GUI
7.14.4.6. Limitations
7.15. Custom Profile
7.15.1. Configuring and Starting Profile
7.15.2. Analyzing Profile Data
7.15.3. Limitations
Code Profiling
8. Code Profiling
8.1. Profiling a .NET/CLR Application
8.1.1. Prerequisites
8.1.2. Launching a .NET Application
8.1.2.1. Using GUI
8.1.2.2. Using CLI
8.1.3. Analyzing Profiled .NET Data
8.1.3.1. .NET Source View
8.1.4. Limitations
8.2. Profiling a Java Application
8.2.1. Prerequisites
8.2.2. Configuration
8.2.2.1. Launching a Java Application
8.2.3. Attaching a Java Process to Profile
8.2.3.1. On Linux Platforms
8.2.3.1.1. Using GUI
8.2.3.1.2. Using CLI
8.2.3.1.3. On Windows and FreeBSD Platforms
8.2.3.1.4. Using GUI
8.2.3.2. Java Source View
8.2.3.3. Java Call Stack and Flame Graph
8.2.4. Limitations
8.3. Profiling a Python Application
8.3.1. eBPF Sampling Mode
8.3.1.1. Key Characteristics
8.3.1.2. Prerequisites
8.3.1.3. Configuration
8.3.1.3.1. Launching a Python Application
8.3.1.3.1.1. Using GUI
8.3.1.3.1.2. Using CLI
8.3.1.3.2. Attaching to a Running Python Process
8.3.1.3.2.1. Using GUI
8.3.1.3.2.2. Using CLI
8.3.1.4. Analyze the Data
8.3.1.5. Identify the hot code paths
8.3.1.6. Python Source View
8.3.1.7. Limitations
8.3.2. Tracing Mode
8.3.2.1. Key Characteristics
8.3.2.2. Prerequisites
8.3.2.3. Configuration
8.3.2.3.1. Launching a Python Application
8.3.2.3.1.1. Using GUI
8.3.2.3.1.2. Using CLI
8.3.2.4. Analyze the Data
8.3.2.5. Identify the hottest function
8.3.2.6. Identify the hot code paths
8.3.2.7. Limitations
8.4. MPI Code Profiling
8.4.1. Collecting Data Using CLI
8.4.2. Analyzing the Data with CLI
8.4.3. Analyze the Data with GUI
8.4.4. Limitations
8.5. Profiling Linux System Modules
8.6. Profiling Linux Kernel
8.6.1. Enabling Kernel Symbol Resolution
8.6.2. Downloading and Installing Kernel Debug Symbol Packages
8.6.3. Build Linux Kernel with Debug Symbols
8.6.4. Analyzing Hotspots in Kernel Functions
8.6.5. Linux Kernel Callstack Sampling
8.6.6. Limitations
8.7. Profiling FreeBSD Kernel Modules (Pre-release Builds)
AMD uProf CLI Options
9. AMD uProf CLI Options
9.1. Overview
9.2. Starting a CPU Profile
9.2.1. List of Predefined Sample Configurations
9.3. Starting a Power Profile
9.3.1. System-wide Power Profiling (Live)
9.4. Collect Command
9.4.1. Synopsis
9.4.2. Common Usages
9.4.3. Options
9.4.4. Examples
9.4.4.1. Windows
9.4.4.2. Linux
9.5. Report Command
9.5.1. Synopsis
9.5.2. Common Usages
9.5.3. Examples
9.5.3.1. Windows
9.5.3.2. Linux
9.6. Translate Command
9.6.1. Synopsis
9.6.2. Common Usages
9.6.3. Options
9.6.4. Examples
9.6.4.1. Windows
9.6.4.2. Linux
9.7. Timechart Command
9.7.1. Synopsis
9.7.2. Common Usages
9.7.3. Examples
9.7.3.1. Windows
9.7.3.2. Linux
9.8. Diff Command
9.8.1. Synopsis
9.8.2. Common Usages
9.8.3. Profile Comparison Eligibility Criteria
9.8.4. Options
9.8.5. Examples
9.8.5.1. Windows
9.8.5.2. Linux
9.9. Profile Command
9.9.1. Synopsis
9.9.2. Common Usages
9.9.3. Options
9.9.4. Examples
9.9.4.1. Windows
9.9.4.2. Linux
9.10. Info Command
9.10.1. Synopsis
9.10.2. Common Usages
9.10.3. Options
9.10.4. Examples
AMD uProf GUI Reference
10. AMD uProf GUI Reference
10.1. User Interface
10.2. Launching GUI
10.3. Configure Callstack Samples Collection
10.4. Translation Progress
10.5. Importing Profile Database
10.6. Analyzing Saved Profile Session
10.7. Using Saved Profile Configuration
10.8. Settings
10.8.1. Preferences
10.8.2. Symbols
10.8.3. Source Data
10.8.4. Profile Data
AMD uProf Application Programming Interfaces
11. AMD uProf Application Programming Interfaces
11.1. CPU Profile Control APIs
11.1.1. Difference between Group1 and Group2 APIs
11.1.2. C/C++ API Descriptions
11.1.3. Static Library
11.1.3.1. Windows
11.1.3.2. Linux
11.1.4. C/C++ API Descriptions
11.1.4.1. Group1
11.1.4.2. Group2
11.1.5. Python API Descriptions
11.1.5.1. Group2
11.1.5.2. Unmatched API Calls
11.2. Instrument APIs - Domain and Task
11.2.1. Definition and Concepts
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
11.3.3.1. Description
11.3.3.1.1. Using the Task APIs
11.3.3.2. C++ Task APIs
11.3.3.3. Python APIs
11.3.3.3.1. Using the Event APIs
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
11.3.8.1. Definition
11.3.8.2. Reason for Unmatched Task/Event
11.3.8.3. Handling the Unmatched Task
11.3.9. Depth Within Domain
11.3.9.1. Definition
11.3.10. Output
11.3.10.1. GUI Task Hotspot Summary
11.3.11. Limitations
11.4. OneAPI support in AMDuProf
11.4.1. Overview
11.4.2. Supported APIs
11.4.3. How to Run
11.4.4. Compiling an Instrumented Target Application
11.4.5. Profiling Instrumented Target Application
11.4.6. Example Steps to Attach Instrumented Process
11.4.7. Output
11.4.7.1. GUI Task Hotspot Summary
11.4.8. APIs Usage Example
11.4.9. Limitations
Live Power Profile
12. Live Power Profile
12.1. Features
12.2. Metrics
12.3. Using Profile through GUI
12.3.1. System-wide Power Profile (Live)
12.3.2. Configuring a Profile
12.3.3. Analyzing a Profile
12.4. Using CLI to Profile
12.4.1. Examples
12.4.1.1. Windows
12.4.1.2. Linux
12.4.2. AMDPowerProfileAPI Library
12.4.2.1. Using the APIs
12.4.2.1.1. Windows
12.4.2.1.2. Linux
12.5. Limitations
Performance Debug using AMDuProfSys
13. Performance Debug using AMDuProfSys
13.1. Overview
13.2. Supported Platforms
13.3. Supported Hardware Counters
13.4. Supported Operating Systems
13.4.1. Prerequisites
13.4.1.1. Windows
13.4.2. Set up
13.4.2.1. Linux
13.4.2.2. Windows
13.5. Options
13.5.1. Synopsis
13.5.2. Generic
13.5.3. Collect Command
13.5.4. Report Command
13.6. Configuration Option
13.6.1. Default configuration
13.6.2. Predefined Configurations
13.6.3. Obfuscation Support
13.6.3.1. Feature Workflow
13.6.4. Custom Metrics
13.7. Limitations
13.8. Virtualization Support
Reference
14. Reference
14.1. CPU Profiling
14.2. Virtualization Support
14.2.1. Feature Support Matrix on Various Hypervisors
14.3. Troubleshooting
14.3.1. Enable Logging
14.3.2. Running the AMDuProf Installer with Verbose Logging
14.4. Useful URLs
Additional Resources and Legal Notices
15. Additional Resources and Legal Notices
15.1. Abbreviations

Back to home page