While optimizing a separate custom project using the Spatial Profiler, I noticed that no matter how much I aggressively cut down textures and static assets, the baseline memory usage would not drop. The profiler was also tracking references to native Fortnite assets (like vehicles) that were not present anywhere in my map.
Out of curiosity, I created a completely blank project using the default template just to check the baseline.
The results of the blank template:
Memory Usage: 6.5 GiB â 6.7 GiB (Triggers Orange âWarningâ status)
Frame Time: 33.33 ms â 35.33 ms (Triggers Red âCriticalâ status)
This is on a project that literally only has two player spawners and a few default grid planes. Nothing else is on the map.
Is this actually normal, expected UEFN performance out of the gate, and something creators should just ignore? Itâs a very weird performance curve to be handed a critical warning before placing a single prop.
My system: RTX 3090, 128gb of ram, ryzen 9 (12core)
Please select what you are reporting on:
Creative
What Type of Bug are you experiencing?
Memory Calculation
Steps to Reproduce
Create an island on blank template
Expected Result
Every stat on spatial profiler should be green since it is a completely blank map
Observed Result
Spatial Profiler showing 6.6GB memory & red frame times on a completely blank map
I was checking it in 3 of my maps after update to v41.00 and it was showing critical warning because of memory usage around 7 GB in all of them although they are very different regarding overall size and usage of unique actors in them, so I would not expect memory usage at the same level like this.
I watched the whole FNâs stream about it few days ago and it was said that Memory Snapshot should take into consideration loaded actors from the actual running game session and not from the editor. But when I searched for individual actors in Memory Snapshot I found there actors from the whole map even if there is big distance between them so all of them are never loaded at the same time in the actual game when the player is just at the edge of the map, while streaming is enabled and everything is set up for streaming and working correctly in edit session and also in live versions.
Then I checked the map that I am working on right now that has locations separated from each other by big areas of nothing (with streaming enabled of course) to optimize memory usage and it looks like it doesnât take it into consideration, because it also showed memory usage around 7 GB again right from the start area that is on the edge of a long line of individual locations with free spaces between them. When I am in edit session, other locations are correctly not visible, because they are not loaded, but it looks like these memory tools doesnât think so. Or there is just some issue why it shows in very different cases I tested memory usage right around 7 GB.
So I am just ignoring it for now, because it doesnât make sense.
I tested, and it seems that the Spatial Profiler is reporting data based on the combined usage of Fortnite and UEFN.
These are the Frame Times I get when:
UEFN is visible (covering Fortnite): 33.2 ms
Split the screen between Fortnite and the Spatial Profiler (the two covering UEFN): 16.7 ms
The same as 2, but setting Vsync to Off and Frame Rate Limit to Unlimited: 6.0 ms
Memory Usage is unaffected by the setups listed above.
The Spatial Profiler window gets minimized with UEFN, so the best I could do was split the screen as mentioned above. Obviously, the Spatial Profiler should only report data from Fortnite to be useful, but since it records in the background, one can ignore the Editor data (at least for Frame Time).
Iâm glad Iâm not the only one having this issue. Thank you for going out of your way to research the frametime issue, I think itâll be helpful for anyone reading this.
Here are a few new screenshots after the 41.10 update.
My blank projectâs memory usage grew to 7.0 GB out of the 7.2 GB budget, even though nothing was added to the scene. This completely blows past the 1.5 GB budget for mobile.
I can replicate this with all my other projects, which are all above the critical limit on desktop with no ability to reduce their memory footprint in any significant way.
I assume the memory and spatial profilers are not working as intended. Could a dev from Epic weigh in to confirm whether this is expected behavior or a bug?
Iâm having these exact same issues. My project started spiking above the 7.2 threshold constantly since the 41.00 update, and I was testing deleting massive chunks of my game (geometry, devices, etc) and the differences were almost non-existent with the memory usage in my map/island.
I then decided to open up a blank island and the GiB number was sitting at an average of 6.8 on a blank map.
