Select your cookie preferences

We use essential cookies and similar tools that are necessary to provide our site and services. We use performance cookies to collect anonymous statistics, so we can understand how customers use our site and make improvements. Essential cookies cannot be deactivated, but you can choose “Customize” or “Decline” to decline performance cookies.

If you agree, AWS and approved third parties will also use cookies to provide useful site features, remember your preferences, and display relevant content, including relevant advertising. To accept or decline all non-essential cookies, choose “Accept” or “Decline.” To make more detailed choices, choose “Customize.”

Preferences
Contact Us
Feedback
AWS Documentation
Create an AWS Account

Profiling Unreal Engine performance

PDF
Focus mode
Profiling Unreal Engine performance - Amazon GameLift Streams

In this section, learn how to analyze your Unreal Engine game or application performance. This can help you identify area's to optimize, leading to smoother streaming in Amazon GameLift Streams.

You can use Unreal Engine's console and its built-in stat commands to get a detailed look at your game's performance. You can access the console in a non-shippable build or the Editor. A non-shippable build refers to a project that was built using a debug or development configuration.

To access the console

In non-shippable builds and the Play In Editor mode, press the tilde (~) key to open the console. Double-press the tilde key to expand the console.

Here are some tips for using the console:

  • Type a keyword to list all possible commands containing that keyword. Scroll through the list using the arrow keys.

  • Scroll through the history by using the arrow keys or Page up and Page down keys.

  • Logs are saved in a .txt file in your project's Saved/Logs directory

To profile your game's performance

  1. Start by running the stat fps and stat unit commands. This will give you an overview of where your game struggles with performance.

    • stat fps: Shows the current frames per second.

    • stat unit: Breaks down the frame into several subsections.

      • Frame: Total wall-clock time starting from when the simulation of the frame starts to when the presentation of the frame is on the screen.

      • Game: Total CPU time taken by the game simulation thread per frame.

      • Draw: Total CPU time for the rendering threads to translate the scene to commands for the GPU and submit them to the GPU.

      • GPU: Total time for the GPU to process all commands.

      • Draws: Total number of draws submitted for the frame.

      • Prims: Total number of triangles drawn.

  2. Play through the game and identify areas with low performance, indicated by decreased FPS and increased time in Game, Draw, or GPU.

  3. Run stat game to see how time is spent for the various gameplay groups.

  4. Refine the stats for specific gameplay factors like AI, animation, physics, gameplay, scripting, and so on. Here are a few examples:

    • stat ai: Time to compute AI behavior.

    • stat anim: Time to compute skinned meshes.

    • stat physics: Time to compute physics simulations.

  5. Run stat drawcount to see which render areas generate the most draws. The list shows the render passes that emit draws, and the number of draws emitted each frame. You can get more information by analyzing the GPU stats in the next step.

  6. Run stat gpu to see which render types consume the most GPU time.

  7. Refine the rendering types into broad groups, such as lights, shadows, lumen (lighting), hair, post processing, and so on. Here are a few common examples:

    • stat lightrendering: GPU time to render lights and shadows.

    • stat shadowrendering: GPU time to update the various shadows.

    • stat scenerendering: GPU time to render the scene.

This section covers only a subset of available commands. Depending on your game's features, look into stats for areas such as asset streaming, virtual texturing, CPU task workload distribution, threading, sound, particles, and so on. For more information, refer to Stat commands.

PrivacySite termsCookie preferences
© 2025, Amazon Web Services, Inc. or its affiliates. All rights reserved.