Windows Development Documentation

Home > Tools > Profilers > CPU Usage Tool

CPU Usage Tool

The CPU Usage tool helps you identify performance bottlenecks in your application by showing you the amount of time spent in functions. You can use it to diagnose issues like high CPU usage, slow performance, and unexpected behavior.

On this page

Introduction

This tool is an indispensable part of the Windows development toolkit, providing deep insights into your application's CPU consumption. By capturing and analyzing CPU activity, you can pinpoint exactly which parts of your code are consuming the most processing power.

Note: This tool is designed to work with applications built for Windows, including native C++, C#, and .NET applications.

Key Features

  • Real-time CPU Monitoring: Observe CPU usage as your application runs.
  • Function-level Profiling: Drill down into the performance of individual functions.
  • Call Tree Visualization: Understand the relationships between functions and how they contribute to CPU load.
  • Flame Graph Support: Visualize CPU usage across function call stacks in an intuitive flame graph.
  • Sampling and Instrumentation: Choose between different profiling methods to suit your needs.
  • Exportable Reports: Save profiling data for offline analysis and sharing.

Getting Started

To begin using the CPU Usage tool:

  1. Ensure you have the appropriate version of Visual Studio or the Windows SDK installed.
  2. Launch your application.
  3. Open the Performance Profiler (usually under the "Analyze" menu in Visual Studio).
  4. Select the "CPU Usage" option and start profiling.

For command-line usage, you can utilize tools like VsPerfCmd.exe or the built-in profiling capabilities of .NET Core.

Usage Scenarios

Identifying Hot Spots

One of the primary uses of the CPU Usage tool is to identify "hot spots" – sections of code that consume a disproportionate amount of CPU time. By looking at the list of functions sorted by exclusive or inclusive time, you can quickly focus your optimization efforts.


// Example of a potentially "hot" function
void ProcessData(const std::vector& data) {
    for (int value : data) {
        // Complex calculation here
        // ...
    }
}

Analyzing Function Call Stacks

Understanding how functions call each other is crucial. The call tree view allows you to see the entire chain of calls leading to a particular function, helping you understand the context of CPU usage.

Information: A high inclusive time for a function, with low exclusive time, often indicates that the function itself is not computationally expensive, but it calls many other functions that are.

Advanced Analysis

Beyond basic hot spot identification, the tool offers advanced features:

  • Thread Analysis: Examine CPU usage per thread to identify thread contention or imbalances.
  • Marks and Events: Insert custom markers in your code to correlate performance spikes with specific application events.
  • Symbol Loading: Ensure that debugging symbols are loaded correctly for accurate function names.
Tip: Use the "View" options to switch between different representations like "Call Tree," "Flat List," and "Functions" to find the most insightful view for your analysis.

Integration

The CPU Usage tool integrates seamlessly with other development tools:

  • Visual Studio: Built-in integration for live profiling and analysis.
  • Azure DevOps: Include profiling steps in your build and release pipelines to catch performance regressions early.
  • Command-Line Tools: Automate profiling tasks and integrate them into CI/CD workflows.
Important: Always profile your application under realistic load conditions to get the most accurate performance data.