Reduce the number of drawcalls by generating a mesh per 8 blocks (2x2x2). Only blocks with even coordinates (lowest bit set to 0) will get a mesh.
Note: This also removes the old 'loops' algorithm for building the draw list, because it produces visual artifacts and cannot be made compatible with the approach of having a mesh for every 8th block without hurting performance.
Co-authored-by: Jude Melton-Houghton <jwmhjwmh@gmail.com>
Co-authored-by: Lars <larsh@apache.org>
Co-authored-by: sfan5 <sfan5@live.de>
* refactoring(StaticObjectList): don't expose m_active and m_stored anymore
This prevents our old crap code where anyone can access to StaticObjectList. use proper modifiers. It also permits to do a short cleanup on MapBlock using a helper
* refactoring(MapBlock): reduce a bit exposed m_active_blocks variable
* refactoring: MapBlock::m_node_timers is now private
We already had various helpers to perform this privatization, just use it. Also factorize the MapBlock stepping code for timers using already existing code and importing them from ServerEnvironment to MapBlock.
It's currently done pretty straight forward without any inheritance as MapBlock is just used everywhere, maybe in a future we'll have ServerMapBlock over MapBlock. Currently for a simple function let's just use proper objects and add a comment warning
* refactoring(Server): fix duplicated function for add/remove node
* refactoring(guiFormSpecMenu): add removeAll function to prevent duplicated code
* refactoring(ShadowRenderer) + perf: code quality + increase performance
* All callers are already using the point and we should never test a function with nullptr node, it's a bug. Removed workaround which was hacky and fix the bug
* Drop clientmap lookup from shadowrendered, just use directly its
pointer and forbid to push it in the generic list
* Reduce memory pressure on the renderShadowObject by preventing
deallocating and reallocating multiple vectors on each node
* refactoring(MapBlock): reduce exposure of MapBlock::m_static_objects
It's not complete as some parts of the code are pretty nested, but it's better than before :)
* fix: better working on new functions & drop unwanted 2 lines
Co-authored-by: Jude Melton-Houghton <jwmhjwmh@gmail.com>
Co-authored-by: Jude Melton-Houghton <jwmhjwmh@gmail.com>
Reduces stutter and freezes when playing.
* Maintains double SM and SM Color textures
* Light frustum update triggers incremental generation of shadow map into secondary 'future' textures.
* Every incremental update renders a portion of the shadow draw list (split equally).
* After defined number of frames (currently, 4), 'future' and 'current' textures are swapped, and DirectionalLight 'commits' the new frustum to use when rendering shadows on screen.
Co-authored-by: sfan5 <sfan5@live.de>
Update the profiler names to make more sense of what they actually represent
Move the profiler code from header to its source file
Use monospace font to align lines
Format the statistics line to align better with surrounding values
Refresh the profiler each 3 seconds (roughly)
lhofhansl
x2048
Loïc Blot
sfan5
paradust7
Jude Melton-Houghton
Liso
Loic Blot
hecks
SmallJoker
Quentin Bazin