CL 4871955
Fixed TInlineAllocator::GetAllocatedSize reporting SecondaryData AllocatedSize even when there wasn't any SecondaryData
CL 4871994
MeshDrawCommand memory optimizations
* When GPUScene is enabled, cached FMeshDrawCommands are only stored in FScene::CachedMeshDrawCommandStateBuckets, so we only have one command per state bucket.
* Removed FGraphicsMinimalPipelineStateInitializer, now relying on index into the global table
* Union IndirectArgsBuffer with draw params
* Trimmed debug data
* Removed duplicated space for SRV values, now using a bitmask to indicate SRV or TextureRHI
* Deferred renderer GPUPerfTest cached MeshDrawCommands 28Mb -> 2.1Mb
* Mobile renderer GPUPerfTest cached MeshDrawCommands 30.7Mb -> 20.5Mb
CL 4892243
Resupported r.DoLazyStaticMeshUpdate with mesh draw commands. Now enabled when in game mode, no longer requires cooked build.
* GPUPerfTest Total cached MeshDrawCommands 2.324Mb -> 1.691Mb and Primitive StaticMeshCommandInfos 3529.2Kb -> 563.7Kb
CL 4925071
Allocate MeshDrawCommands at lowest free index instead of last freed index
Shrink MeshDrawCommands periodically when r.DoLazyStaticMeshUpdate is enabled
Measured in replay ES31: Cached MeshDrawCommands 5.236Mb -> 2.216Mb
CL 4937280
Removed r.DeferUniformBufferUpdatesUntilVisible since it is no longer supported with MeshDrawCommands, which cache the uniform buffer pointer at AddToScene time. Removed dead code referencing it.
CL 4937591
Fixed FMeshDrawCommandStateCache always resetting resources at index 0. Found by Rolando.
CL 4951781
Attempt to silence static analysis warning
#jira none
#rb none
#rnx
#ROBOMERGE-SOURCE: CL 4993090 in //UE4/Release-4.22/...
#ROBOMERGE-BOT: RELEASE (Release-4.22 -> Main)
[CL 4993091 by krzysztof narkowicz in Main branch]
This fixes a memory leak resulting from UShapeComponent::CreateShapeBodySetupIfNeeded sometimes being called as part of async navigation collision gathering.
#jira UE-69914
#rb Robert.Manuszewski
#ROBOMERGE-SOURCE: CL 4992792 in //UE4/Release-4.22/...
#ROBOMERGE-BOT: RELEASE (Release-4.22 -> Main)
[CL 4992793 by mieszko zielinski in Main branch]
#jira UE-69962
#rb none
#ROBOMERGE-SOURCE: CL 4992138 in //UE4/Release-4.22/...
#ROBOMERGE-BOT: RELEASE (Release-4.22 -> Main)
[CL 4992139 by max chen in Main branch]
fix HTML5 build errors
#jira UE-68340 HTML5 Servers crash on connection cleanup due to malformed connection sources
#rnx
#rb none
#ROBOMERGE-SOURCE: CL 4992075 in //UE4/Release-4.22/...
#ROBOMERGE-BOT: RELEASE (Release-4.22 -> Main)
[CL 4992079 by nick shin in Main branch]
#jira UE-68340 HTML5 Servers crash on connection cleanup due to malformed connection sources
#rnx
#rb none
#ROBOMERGE-SOURCE: CL 4991534 in //UE4/Release-4.22/...
#ROBOMERGE-BOT: RELEASE (Release-4.22 -> Main)
[CL 4991535 by nick shin in Main branch]
#jira none
#rb ryan.gerleve
#ROBOMERGE-SOURCE: CL 4991241 in //UE4/Release-4.22/...
#ROBOMERGE-BOT: RELEASE (Release-4.22 -> Main)
[CL 4991242 by brian bekich in Main branch]
#rb none
#jira UE-70042
#ROBOMERGE-SOURCE: CL 4991095 in //UE4/Release-4.22/...
#ROBOMERGE-BOT: RELEASE (Release-4.22 -> Main)
[CL 4991098 by max chen in Main branch]
Added 2 new texture filter modes : TF_FirstMipNearest and TF_FirstMipBilinear. Those always sample the first (highest) mip.
When running with cooked content, TF_FirstMipNearest and TF_FirstMipBilinear will drop the all mips but the first (freeing mip tail memory).
In uncooked game, the visuals will be valid but memory of the mip trail is still allocated.
In the editor, the mip tail can still be visualized.
Fixed the texture editor streaming method description to actually reflect the state of the texture.
#rb nick.darnell
#ROBOMERGE-SOURCE: CL 4985873 via CL 4987451
[CL 4990180 by uriel doyon in Main branch]
Merging CL#4987059 from//Fortnite/Dev-EngineMerge
Dynamic Memory tracking for Rep Layout. This still doesn't track Dynamic Memory for NetSerialize or Custom Delta Serialize items.
#review-4978647 @Brian.Bekich @Ryan.Gerleve @Bart.Hawthorne @LouisPhilippe.Seguin @Mattias.Hornlund @Peter.Engstrom
#jira FORT-141917
#rb Peter.Engstrom
[CL 4989716 by Bob Tellez in Main branch]
Also, make sure more code is properly tracked as LLM Networking.
[at]Ryan.Gerleve [at]Brian.Bekich [at]Bart.Hawthorne
#rb Ryan.Gerleve, Brian.Bekich
#ROBOMERGE-SOURCE: CL 4982995 via CL 4982996 via CL 4986892
[CL 4989186 by jon nabozny in Main branch]
#rb aaron.mcleran
#jira none
#ROBOMERGE-SOURCE: CL 4989031 in //UE4/Release-4.22/...
#ROBOMERGE-BOT: RELEASE (Release-4.22 -> Main)
[CL 4989032 by rob gay in Main branch]
#rb mark.satterthwaite
#ROBOMERGE-OWNER: peter.sauerbrei
#ROBOMERGE-AUTHOR: peter.sauerbrei
#ROBOMERGE-SOURCE: CL 4982155 via CL 4982162 via CL 4982613
[CL 4988648 by peter sauerbrei in Main branch]
#rb arciel.rekman,brandon.schaefer
#jira UE-66130
#ROBOMERGE-SOURCE: CL 4988608 in //UE4/Release-4.22/...
#ROBOMERGE-BOT: RELEASE (Release-4.22 -> Main)
[CL 4988613 by marcin undak in Main branch]
#rb none
#jira
#ROBOMERGE-SOURCE: CL 4988259 in //UE4/Release-4.22/...
#ROBOMERGE-BOT: RELEASE (Release-4.22 -> Main)
[CL 4988260 by ben marsh in Main branch]
