Bart/bullet3
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Some performance improvements and fixes related to btVector3 being aligned on SPU.

Browse Source 
btQuantizedBvh has a version number, memory layout might be different now (due to aligned btVector3)
reorganized some data members of some classes, to reduce memory footprint
This commit is contained in:
erwin.coumans
2008-10-29 02:44:50 +00:00
parent ce5df7cf47
commit 55b29e2355
10 changed files with 241 additions and 110 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
src/BulletMultiThreaded/SpuGatheringCollisionDispatcher.h
View File

@@ -21,8 +21,13 @@ subject to the following restrictions:
///Tuning value to optimized SPU utilization
///Too small value means Task overhead is large compared to computation (too fine granularity)
///Too big value might render some SPUs are idle, while a few other SPUs are doing all work.
#define SPU_BATCHSIZE_BROADPHASE_PAIRS 16
//#define SPU_BATCHSIZE_BROADPHASE_PAIRS 8
//#define SPU_BATCHSIZE_BROADPHASE_PAIRS 16
#define SPU_BATCHSIZE_BROADPHASE_PAIRS 64
//#define SPU_BATCHSIZE_BROADPHASE_PAIRS 128
//#define SPU_BATCHSIZE_BROADPHASE_PAIRS 256
//#define SPU_BATCHSIZE_BROADPHASE_PAIRS 1024
class SpuCollisionTaskProcess;