ejcoumans
eeb78b8d45
free memory for btShapeHulls, keep track of it in GL_ShapeDrawer.
...
move btShapeHull and btConvexHull into its own library in Extras/ConvexHull (it allocates memory using mem/delete and refactoring into using btAlignedAlloc/Free takes too much time)
fix heightfield / btOptimizedBvh for quantization, so that raycast can use quantized aabb (clamp up for maxima and down for minima)
work-in-progress (update projectfiles etc)
2008-02-13 07:14:19 +00:00
ejcoumans
5399a9ae8c
fixes to compile/share more code between SPU and PPU
2008-02-01 03:35:54 +00:00
ejcoumans
1cdd5bf67f
add support for 16bit and 32bit indices in SPU/BulletMultiThreaded version of btBvhTriangleMeshShape
2008-02-01 02:13:24 +00:00
johnmccutchan
be0beaf7bd
Refactored SpuGatheringCollisionTask to use code in SpuCollisionShapes.
...
More work on SpuBatchRaycaster. It is working now on the PS3 and Windows.
2008-01-14 23:44:07 +00:00
ejcoumans
6ba6805b43
bugfix: need to use returned address of cellDmaSmallGetReadOnly
...
Thanks John McCutchan for pointing this out. Affects non-Cell BulletMultiThreaded.
2008-01-12 00:07:39 +00:00
ejcoumans
61ffa6a1b5
Updated cmake files, added header files, Thanks John McCutchan
...
re-autogenerated msvc projectfiles with new HelloWorld/BulletMultiThreaded Demo
Minor fixes
2007-12-15 02:48:24 +00:00
ejcoumans
6dff743d68
Cleanup some #defines in parallel collision detector, add conditional freeglut support to return from glutMainLoop.
2007-12-15 00:40:34 +00:00
ejcoumans
c5e6044e53
added MultiThreadedDemo
2007-12-14 07:17:35 +00:00
ejcoumans
d7354e5387
get SequentialThreadSupport to compile on the Mac
2007-12-14 06:21:51 +00:00
ejcoumans
8d10a6ba1d
Fix friction and restitution in BulletMultiThreaded (collision detection and parallel solver)
...
Thanks to lsgmasa for reporting, see http://www.bulletphysics.com/Bullet/phpBB3/viewtopic.php?f=9&t=1758
2007-12-12 03:41:35 +00:00
johnmccutchan
66cdbb9659
Fixes for IBM Cell SDK 3.0 by Jochen Roth
2007-12-11 19:49:09 +00:00
ejcoumans
830c096e10
added a cellDmaLargeGetReadOnly/cellDmaGetReadOnly, which avoid memcpy on non-SPU platforms.
...
enabled it for triangle mesh aabb DMA transfers. Todo: check if it is worth also modifying other parts of the code to avoid the memcpy into the 256kb Task memory.
2007-10-16 23:58:26 +00:00
ejcoumans
6303d98844
- use mfc directly, instead of cellDmaSmallGet (allow sizes under 32 bytes, instead of 16)
...
- align btBroadphaseProxy to 16 bytes
2007-10-16 00:52:23 +00:00
ejcoumans
9e530a005c
make parallel solver compile on MSVC 6
2007-10-12 06:59:59 +00:00
ejcoumans
29dd9681f4
Added libspe2 parallel Cell SPE support contribution by IBM Germany 'Extreme Blue' project:
...
Thanks to Benjamin Hoeferlin, Minh Cuong Tran,Martina Huellmann,Frederick Roth.
2007-09-26 23:43:51 +00:00
ejcoumans
08d036afc7
multi-threading improvements: optionally use software caching (IBM/PS3), move some code from cpp to header to inline.
2007-09-20 22:42:56 +00:00
ejcoumans
b43b8f0f75
- added 'init' function for DoubleBuffer, in case constructor is not called.
...
- use proper transform for concave case
thanks Marten for the patch
2007-08-15 04:16:25 +00:00
ejcoumans
5279f9e129
BulletMultiThreaded (SPU/multi-core): added compound shape support and concave-convex (swapped case). Thanks to Marten Svanfeldt
2007-08-02 20:16:58 +00:00
ejcoumans
38b7f474c3
Added better support for btUniformScalingShape, by moving some data that is not shared from btConvexShape to btConvexInternalShape. This reduces the sizeof btUniformScalingShape to 16 bytes (from 64).
...
This is good when having lots of re-used shapes with different sizes.
Convex shapes will need to derive from btConvexInternalShape (which is a subclass of btConvexShape). We could have renamed btConvexShape to 'btConvexShapeInterface' (can still do that later)
2007-07-28 21:10:21 +00:00
ejcoumans
272a7dfac4
More work on multi-threaded version, fixed alignment issues in DMA.
...
BulletMultiThreaded shared implementation works now on both Win32 Threads and PS3 Cell SPU.
2007-06-16 02:06:24 +00:00
ejcoumans
ea4dfc15d5
updated multi threading shared code on multiple platforms.
2007-06-15 02:06:56 +00:00
ejcoumans
fc6cf5dbef
DMA / memcpy the actual size, rather then over-estimated (can cause crashes)
...
added alternative to memcpy, for debugging
2007-06-14 00:21:15 +00:00
ejcoumans
f4d31a2dd8
unknown issue with dma size, need to investigate deeper (had to increase target buffer size to avoid crashes)
2007-06-13 23:18:04 +00:00
ejcoumans
0fb0b6056c
bug in dma size calculation
2007-06-13 23:05:37 +00:00
ejcoumans
05cbd96972
updated multithreaded version for concave trimeshes
2007-06-13 22:25:37 +00:00
ejcoumans
4f9d719493
add snippets for convex/concave multithreaded
2007-06-13 06:44:47 +00:00
ejcoumans
2b39be94a3
enable convex polyhedra (btConvexHullShape) for multi-threaded version
2007-06-13 06:01:56 +00:00
ejcoumans
6f26aa1591
Added multi-threaded collision detection. Original code is written for Cell SPU, but wrappers are provided to run on multi-core using Win32 Threads.
...
SpuLibspe2Support is on the todo list, so it can run on Cell Blade & PS3 Linux.
2007-06-13 01:04:43 +00:00
