- Defines will still be there if we're dumping the shader code for debug.
- This makes the code less unique in general case and also avoids unnecessary works that no one sees.
#rb Lukas.Hermanns
[FYI] Lukas.Hermanns, Rolando.Caloca
#ROBOMERGE-SOURCE: CL 11311362 via CL 11311438
#ROBOMERGE-BOT: (v653-11302973)
[CL 11311455 by arciel rekman in Main branch]
#rnx
#rb none
#ROBOMERGE-SOURCE: CL 10869240 via CL 10869516 via CL 10869902
#ROBOMERGE-BOT: (v613-10869866)
[CL 10870584 by ryan durand in Main branch]
#jira UE-80424
#rb None
[FYI] Lukas.Hermanns
#ROBOMERGE-SOURCE: CL 10108886 in //UE4/Release-4.24/...
#ROBOMERGE-BOT: RELEASE (Release-4.24 -> Main) (v586-10095595)
[CL 10108946 by jack porter in Main branch]
- Added error reporting to prevent future silent mcpp failures
- Merged CL 9919664 from //UE4/Dev-Rendering to Release-4.24, Fixes compiler warnings in MCPP and made main arguments a dynamic array when MAX_OPTIONS is exceeded.
#jira UE-80424
#rb Dmitriy.Dyomin
[FYI] Lukas.Hermanns
#ROBOMERGE-SOURCE: CL 10092550 in //UE4/Release-4.24/...
#ROBOMERGE-BOT: RELEASE (Release-4.24 -> Main) (v574-10069753)
[CL 10092553 by jack porter in Main branch]
Replicated from CL# 7924370.
#rb none
#ROBOMERGE-OWNER: steve.robb
#ROBOMERGE-AUTHOR: steve.robb
#ROBOMERGE-SOURCE: CL 9279060 via CL 9279063
#ROBOMERGE-BOT: (v443-9013191)
[CL 9279836 by steve robb in Main branch]
Speeds up ShaderCompilerWorker almost 3x in test case (Fortnite shaders, all local compilation). In basic benchmarking the Preprocess step went from some 65% of wall time to less than 10%.
For XGE the difference is likely smaller in practice since XGE seems to introduce massive stalls during process startup for the first few jobs starting on a remote worker.
#rb rolando.caloca
[CL 4668598 by Stefan Boberg in Dev-Core branch]