Unreal Engine Performance Profiling: A Guide to Stat Commands and Insights

When a performance issue arises in an Unreal Engine project, the first step is to measure where it occurs. Epic Games’ official documentation explains how to view basic performance data by entering stat commands in the console while the project is running, then examine recorded traces in greater detail with Unreal Insights. This guide covers a workflow from measurement through analysis, particularly for real-time architectural visualization and interactive scenes.

1. Identify the overall bottleneck first

Run the project in Play In Editor (PIE) mode and enter the following command in the console:

stat unit

This command displays performance information under the Frame, Game, Draw, GPU, RHIT and DynRes headings. According to the documentation, Frame is the total time taken to produce a frame. If Frame time is close to Game time, the issue may lie on the Game thread; if it is close to Draw time, the render thread may be the source. GPU time shows how long the graphics card takes to render the scene. Treat these values as indicators that help guide further investigation, not as a definitive diagnosis.

Another useful starting point for an overview is stat sceneRendering. Epic describes it as a good place to begin investigating general areas of slowdowns in the rendering process. You can then choose a command based on what the measurements indicate: stat memory shows memory usage, stat streaming provides basic information about streamed assets, and stat lightRendering displays the time spent processing lighting and shadows.

stat sceneRendering

If you need a list of commands, use stat help in the console. The documentation also notes that you can find commands in the Editor’s Stat menu by clicking the dropdown arrow next to the Viewport Setting button.

2. Save PIE measurements to a file

If real-time stats aren’t enough to capture the issue, save the statistics samples to a file. Start recording in the PIE console with:

stat startfile

The recording typically creates a .uestats file under Saved/Profiling/UnrealStats in the project folder. Be sure to stop the recording when you’re done:

stat stopfile

This step matters: according to the documentation, StartFile can continue recording in the background even after PIE is closed. Leaving a recording running can result in large .uestats files. To inspect the file, go to Tools > Session Frontend in the Unreal Editor menu, open the Profiler tab and select the file using Load. The captured data will then appear in Profiler.

3. Inspect traces with Unreal Insights

Unreal Insights is a collection of tools for capturing events from a project at high data rates, then analyzing and visualizing them. You can launch it from the Trace/Insights Status Bar Widget in the Editor. Trace sessions are stored in .utrace files; associated data may be found in .ucache files in the same location.

To start an analysis, double-click a saved session in the Session Browser, or select it and click Open Trace. You can locate a .utrace file elsewhere using the Open Trace menu; dragging and dropping the file into the Unreal Insights window is another option described in the documentation. Live sessions appear in the list with a LIVE status and update in real time as you inspect them.

The Timing Insights window lets you view CPU and GPU traces and see how long tasks take. Use Memory Insights to investigate memory usage and call stacks, and Asset Loading Insights to examine how long assets take to load and the order in which they load. Choosing the tool that matches the issue helps keep the investigation focused.

4. Adapt the workflow for architectural visualization

If you notice stuttering, slow responses or loading delays in an interactive architectural scene, start by measuring with stat unit. Note whether the results point to Game, Draw or GPU; rather than making an optimization decision based on a single snapshot, measure again in the same scene under similar conditions. Then capture a stats recording or an Unreal Insights trace, depending on how much detail you need.

For scene transitions and level loading status, use stat levels. The command shows whether levels are active, visible, loading, unloading or in the preloading stage, as well as the time from the loading request to completion. stat streaming can help assess memory usage for streamed content, while stat lightRendering can help investigate lighting and shadow processing. These indicators help architectural teams choose which scene or rendering stage to investigate; on their own, they don’t require changing any particular setting.

Unreal Insights documentation states that Trace Server is enabled on all desktop platforms starting with UE 5.3. You can set a different folder for storing traces and add additional trace folders. Before sharing analysis files, teams should check which folder is in use and whether the recording has finished. The steps here are based on commands and tools in Epic Games’ Unreal Engine documentation; since the documentation doesn’t specify hardware requirements or licensing conditions for this workflow, no specific recommendation can be made about them.

Common mistakes

• Forgetting to stop a recording after stat startfile: recording may continue even after PIE closes. Enter stat stopfile when you’re done.

• Assuming .uestats and .utrace are the same type of recording: the official documentation associates these extensions with different workflows. Open .uestats recordings in the Session Frontend Profiler and .utrace traces in Unreal Insights.

• Treating the first value you see as the definitive cause: stat unit indicates potential bottlenecks; use the relevant recording and analysis tools for a closer investigation.