Sculpt/dyntopo: Make the omp threads configurable to overcome performance issues
authorJens Verwiebe <info@jensverwiebe.de>
Mon, 31 Mar 2014 11:51:40 +0000 (13:51 +0200)
committerJens Verwiebe <info@jensverwiebe.de>
Mon, 31 Mar 2014 11:51:49 +0000 (13:51 +0200)
commit277fb1a31fc4b0c9691b3bbab43fd1a970d3e575
tree14fdf30a1783d0d4b17ca83f4de8b4e0c4276c66
parente05d35bfaffec69ed4d990f6a90a1b9244970aa4
Sculpt/dyntopo: Make the omp threads configurable to overcome performance issues
- autodetect optimal default, which typically avoids HT threads
- can store setting in .blend per scene
- this does not touch general omp max threads, due i found other areas where the calculations are fitting for huge corecount
- Intel notes, some of the older generation processors with HyperThreading would not provide significant performance boost for FPU intensive applications. On those systems you might want to set OMP_NUM_THREADS = total number of cores (not total number of hardware theads).
release/scripts/startup/bl_ui/space_view3d_toolbar.py
source/blender/blenkernel/BKE_scene.h
source/blender/blenkernel/intern/scene.c
source/blender/blenlib/BLI_threads.h
source/blender/blenlib/intern/threads.c
source/blender/editors/sculpt_paint/sculpt.c
source/blender/makesdna/DNA_scene_types.h
source/blender/makesrna/intern/rna_scene.c