FIXED FOR ME: Disabled Studio Telemetry plugin, UDP Messaging / TCP Messaging. That improves editor lag, then for game performance, use the settings below.
[/Script/Engine.RendererSettings]
;Transluceny has big cost when raining reducing here to half res is acceptable
r.SeparateTranslucencyScreenPercentage=66
r.ReflectionMethod=2
r.GenerateMeshDistanceFields=True
r.DynamicGlobalIlluminationMethod=1
r.Shadow.Virtual.Enable=1
r.SkinCache.CompileShaders=True
r.RayTracing=False
r.Shadow.UnbuiltPreviewInGame=False
r.AllowStaticLighting=False
r.VirtualTextures=True
r.VirtualTexturedLightmaps=False
r.Lumen.TraceMeshSDFs=0
r.Shadow.CSMCaching=True
r.ReflectionCaptureResolution=64
r.AntiAliasingMethod=4
r.MSAACount=4
r.Mobile.EnableStaticAndCSMShadowReceivers=False
r.Mobile.EnableStaticAndCSMShadowReceivers=False
r.CustomDepth=3
r.DefaultFeature.AutoExposure.ExtendDefaultLuminanceRange=True
r.GPUSkin.Support16BitBoneIndex=True
r.GPUSkin.UnlimitedBoneInfluences=True
r.SkinCache.DefaultBehavior=0
SkeletalMesh.UseExperimentalChunking=1
r.Nanite.ProjectEnabled=True
bEnableVirtualTextureOpacityMask=True
r.Water.SingleLayer.DepthPrepass=True
r.Water.SingleLayer.ShadersSupportVSMFiltering=True
r.Water.SingleLayer.VSMFiltering=True
r.Lumen.SampleFog=0
r.Lumen.Reflections.Contrast=0.75
r.VolumetricFog.Emissive=0
r.optimizedWPO=1
r.Nanite.MaxVisibleClusters=10485760
r.Nanite.MaxCandidateClusters=16777216
r.Nanite.MaxNodes=10485760
r.Nanite.RasterIndirectionMultiplier=1.1
r.Shadow.Virtual.TranslucentQuality=1
r.Lumen.HardwareRayTracing=False
r.RayTracing.Culling=3
r.RayTracing.Culling.Radius=10000
r.RayTracing.Culling.Angle=2
r.RayTracing.Geometry.NiagaraRibbons=0
r.RayTracing.Geometry.NiagaraSprites=0
r.RayTracing.Geometry.NiagaraMeshes=0
r.TSR.History.SampleCount=32
r.TSR.ShadingRejection.Flickering.FrameRateCap=30
r.TSR.ShadingRejection.Flickering.MaxParralaxVelocity=2
r.TSR.ShadingRejection.Flickering.Period=3
r.SupportStationarySkylight=False
r.SupportLowQualityLightmaps=False
r.Mobile.AllowDistanceFieldShadows=False
r.DefaultBackBufferPixelFormat=4
r.GBufferFormat=3
r.Substrate.AccurateSRGB=1
r.Substrate=False
r.Substrate.BytesPerPixel=170;
r.Nanite.AllowTessellation=1
r.Nanite.Tessellation=1
r.PathTracing=False
r.Shadow.Virtual.UseFarShadowCulling=0
r.Shadow.Virtual.NonNanite.IncludeInCoarsePages=0
r.Lumen.Reflections.HardwareRayTracing.Translucent.Refraction.EnableForProject=False
r.DefaultFeature.MotionBlur=False
r.DefaultFeature.AmbientOcclusionStaticFraction=False
r.SupportLocalFogVolumes=False
r.SupportSkyAtmosphereAffectsHeightFog=False
r.SeparateTranslucency=False
Hi @IntXThOrn
What is your CPU? M1 / M2 or M3?
I tried the settings you pasted and it does not help much for me
Well since they mention running unreal on windows I doubt it’s either of those…
Aside that, latest (5.4.1) has nanite fixed for visual glitches it seems but performance is still bad unless you use “medium” quality. I tried M1 and M3pro
M2 here. Severe degradation in performance.
Interesting empirical titbit: upon project load it is sluggish but usable, but after “playing” it or simulating it degrades into single fps digits for me.
Even my m3 max 40c GPU with 64 GB RAM struggles quite a lot. Nanite tessellation is still glitchy.
5.3 was far superior performance-wise. We need fixes for MacOS ASAP.
Yes, it’s super sad that instead of finaly utilizing HWRT, mesh shading and the dynamic memory stuff we just got worse performance and no new features. Epic must be really mad at apple.
mesh shading is supposed to be solved by Nanite (at least the gains behind it, not the actual principle)
dynamic memory was a solution to a problem that Apple created by having unified memory, this has (virtually/mostly) nothing to do with UE or anything else - oversimplified explanation is now hardware/software stack of unified memory does not blast memory to the full with GPU workloads, but does it smarter. it is how it was supposed to be in the first place, but wasn’t
do not be that dismissive of Epic engineers, especially when you apparently mix apples with oranges (pun intended). at this rate you might as well say “why doesn’t Epic make 8gb RAM run like 16gb RAM, Apple said this innovation comes for free with each 8gb RAM config!”
What?
And you forgot about my complain about lack of hwrt support. Please also give me a very wrong but confident answer. Like it’s a solution to a problem apple has created or that it’s got nothing to do with epic 
ok
I have the same issue with 5.4.1 on a M1 Pro MacBook Pro with 32GB RAM.
Major decrease in performance compared to the 5.2 I was using before, maybe ~50%. No single render setting or AA change seems to be responsible as I’ve tried everything including the console commands above. The only thing that has an outsized impact is the screen resolution.
I do appreciate that Niagara NeighborGrid seems to compile on Mac/metal on my M1 now and the new Xcode stuff is a big improvement so thanks for that.
On a new top down template project the most expensive item in statGPU by far is “[unaccounted]”. And profileGPU gives wildly different and obviously incorrect results every time I run it. But profileGPU has done that on Mac as long as I can remember.
5.4.2
same, unusable on my end
Same on my MBP M1 Max. Same project, same settings, about 10% slower from 5.3.2 to 5.4.2. And the packaged app is about 10% bigger. This wasn´t a step forward for Mac users… :-/
It might not sound much to you, but it is in fact an abundant amount of RAM for Unreal Editor on any system. I have UE 5.4.4 running on my system right now and it uses whole 344 MB of RAM and on the whole system only 12 GB is used out of 32 GB on the system and I have loads of other stuff running: multiple Chromium-based browsers running with 200 tabs in them and 50 other programs.
In fact, UE can run on a system with 4 GB of memory total if you just run Unreal itself, because the system doesn’t need much to run a single program.
