A Look Ahead: Improving UEFN Stability and Iteration Times

Authors: Aurel Cordonnier @aurel.cordonnier and Kavita Chodavarapu @CKnew234

Hello Fortnite developers,

We made a commitment last year to improve the stability and iteration speed of UEFN. Our goal is a 50% reduction in iteration time by June (or more—we hope to underpromise and overdeliver here). This post outlines changes we’ve made to our internal testing and QA process to address release stability and the initial suite of updates we’re introducing in the coming months to improve iteration times.

Release Stability and Testing

We’ve changed our testing approach to focus on what matters most to you—high-engagement and persistence-heavy islands. Our testing coverage now includes additional developer content more prone to regressions, and as a result, we’re catching critical defects earlier, before they impact the developer experience.

We’ve also implemented a 24-hour triage escalation process, which means when we find issues, we’re able to fix them faster than ever before. Additionally, we’re now testing six complex developer islands twice a week to track key performance benchmarks like boot time, load time, and minigame operations—escalating any regressions immediately.

The results so far are encouraging. In release 39.30, we prevented 57% of defects from going live, and in releases 39.40 and 39.50, we had zero critical defects tied to developer content or persistence—a direct result of faster triage and expanded coverage. Our escape rate (defects that slip through to live releases) has also dropped significantly: from 18.5% in 39.11 to just 4.7% in 39.40.

We know internal metrics only tell part of the story, and we’re partnering with the community team to launch monthly stability surveys so we can hear directly from developers and ensure that they’re seeing a noticeable difference. This feedback will help us know where to focus our stability efforts.

Stability and Asset Management Tools

Alongside an expanded internal QA process, we’re also adding new tooling and workflows to help you identify and fix errors in your projects before they cause crashes or stalls.

  • Asset Health Tooling: Resolve asset validation issues and ‘out of memory’ errors as you iterate using new post-import monitoring, reporting, and per-platform memory thresholds.

  • Spatial Profiler Updates: With expanded profiling capabilities in UEFN, you’ll be able to spot heavy assets quickly across platforms. This includes new direct memory calculation results and a new Live Asset View with UObject analysis.

Cooker Optimizations

To speed up the time it takes to prep your island for a play session and improve iteration times, we are refactoring the UEFN cooking pipeline to be more efficient. “Cooking” is the process of converting raw assets into an optimized, client-ready format so content loads and renders quickly.

These will be under-the-hood improvements to UEFN’s backend infrastructure, so you won’t need to change anything about your workflow.

  • Zen DDC (Derived Data Cache): We’re deploying the latest Zen Server systems to the UEFN backend and introducing cooked snapshots to improve the efficiency of the cooking process, reducing the ‘play on client’ iteration loop.

  • Incremental Cooking: We’re adding a system that only cooks the required asset deltas (changes) pushed from your project, rather than reprocessing unchanged data.

  • Distributed Shader Compilation: Instead of your local machine bearing the full load, we will distribute shader compilation to the cloud, speeding up compile time.

Live Edit Improvements

We’re expanding Live Edit, providing more ways to iterate during a Live Edit session with support for additional asset types and new in-editor workflows for editing a running session.

  • Expanded Asset Type Support: You will be able to modify more asset types directly in a Live Edit session, reducing how often you need to Push Changes.Ex: Sound Bus, Level Sequence.

  • Live Edit During Game Mode: While the game mode is in progress during Live Edit, a subset of edits in the UEFN Editor will be applied without pushing changes and stopping the game. For example: If you’re building an obstacle course, you will be able to transform objects and tweak their position.

  • Edit History Panel: Shows you which changes in your session have been implemented, and the edits that still require you to Push Changes. This panel will work during Edit Mode and Game Mode while the session is running.

You can keep track of these updates on the Fortnite Developer Roadmap. We will also use this thread when sharing updates on our ongoing work to improve stability and iteration times in UEFN.

    • Zen DDC (Derived Data Cache): We’re deploying the latest Zen Server systems to the UEFN backend and introducing cooked snapshots to improve the efficiency of the cooking process, reducing the ‘play on client’ iteration loop.

    • Incremental Cooking: We’re adding a system that only cooks the required asset deltas (changes) pushed from your project, rather than reprocessing unchanged data.

This! Yes, these are the platform feature QOL updates I dream about! :100:

1 Like

This is GOATED! Really really good change. Thank you guys​:folded_hands:

2 Likes

Ahhh incremental cooking lets goo, goated update for large projects :fire:

Also this is huge!!! Live Edit During Game Mode

Does this mean we’re getting 1.0 support for the level sequencer? If so, that’s pretty cool. If not, still very cool!

I love all of the changes here. Anything to make UEFN iteration times faster is something I want to see!