Bart/bullet3
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
205 Commits 10 Branches 0 Tags
733f9027fb3de9cac9b58afe502cfe5fc46c621b
Commit Graph

47 Commits

Author SHA1 Message Date
erwincoumans
733f9027fb Deterministic simulation for concave and compound collision shapes: added childShapeA/B to b3Contact4 + sort on them 
Use tetrahedra instead of barrel for convex demo (until performance for edge-edge is improved)
Increased #overlapping pair capacity from 12 to 16 / objec
 2013-07-17 22:42:50 -07:00
erwincoumans
ab125fbb6d implement GPU breakable constraints 
add GPU fixed constraint
fix performance issue with concave meshes (didn't clear the number of concave-convex pairs, so it increased every frame)
 2013-07-17 16:11:54 -07:00
erwincoumans
3beae80a73 prepare for GPU fixed constraint (not done yet) 2013-07-16 19:05:07 -07:00
erwincoumans
20f9e41ff0 remove some debug printf's from kernel (stringify doesn't handle it properly) 2013-07-15 10:36:02 -07:00
erwincoumans
a5d00c8219 add some clFinish for better profile timings 
improved parallel batching, don't try to write for static objects,
this fixed a bug, when the hash of a static object was identical with hash of dynamic objects, causing it to be assigned a bogus 100+i batching number
The parallel batching is still not enabled, because we need to measure the batching size (todo)
 2013-07-15 10:22:40 -07:00
erwincoumans
05ed1fdbcf revert to useNewBatchingKernel, the other one is broken (need to figure out why) 2013-07-15 01:06:04 -07:00
erwin coumans
1e31073f4b made the simulation deterministic 
disable 'simdwidth' optimization for determinism (need to double-check)
made the spatial batching 3D
 2013-07-14 19:16:33 -07:00
erwin coumans
4b293130bd Merge branch 'master' of https://github.com/erwincoumans/bullet3 2013-07-12 20:46:49 -07:00
erwin coumans
061f7173b8 compute best axis on host and OpenCL for 1-axis SAP, based on best variance 2013-07-12 20:46:43 -07:00
erwin coumans
0fcdd290f5 remove dynamic OpenCL allocations at run-time 2013-07-12 16:00:25 -07:00
erwin coumans
2aad8419b7 add support for picking, using point 2 point constraint 
allow to remove constraints by unique id
added tiny wavefront loader, plan to use this instead of existing slow wavefront loader
 2013-07-10 00:21:23 -07:00
erwin coumans
9610b369a3 fix constraint issue with b3GpuDynamicsWorld 2013-07-09 19:32:14 -07:00
erwin coumans
a3c2456c65 introduce b3GpuSolverConstraint and fix Windows build 2013-07-09 18:03:51 -07:00
erwincoumans
282e19171f fix Linux version of ConstraintDemo 2013-07-09 14:41:56 -07:00
erwin coumans
4a0d2847c4 allow to run each part of the GPU solver on CPU, for debugging purposes 2013-07-09 13:07:34 -07:00
erwin coumans
38bfcc859b fix linux build 2013-07-09 11:36:21 -07:00
erwin coumans
bd2cd7b2a7 remove clFinish and copyToHost from GPU joint solver, performance is looking better now. 2013-07-09 11:19:16 -07:00
erwin coumans
c4375a09e4 added GPU joint solver for non-contact constraints. Only point 2 point version for now, will add some other constraints soon (changes are very local) 2013-07-09 10:46:47 -07:00
erwin coumans
b8d5cecfe3 joint (non-contact constraint) solver iterations is now working on GPU, but overall slower because of data copy. 
Will move joint setup to GPU, and then some benefit should be visible.
Don't use 64 alignment, it causes data structures size mismatch between cpu and gpu
 2013-07-06 14:11:32 -07:00
erwin coumans
3ca4d68f3e added jointSolver.cl/h (not working yet) 2013-07-06 13:20:34 -07:00
erwin coumans
3a6e1ccff4 work towards GPU joint solver (non-contact constraints), not working yet. 2013-07-06 13:08:58 -07:00
erwin coumans
28f6e0fbd3 towards GPU joint (non-contact) constraint solving. The ConstraintDemo shows that CPU performance is very good, so not much performance win is to be expected. 2013-07-02 21:13:42 -07:00
erwin coumans
6d6ef5d3dc fixed many memory leaks 
added working shadows, using basic shadow mapping
 2013-06-30 23:19:14 -07:00
erwin coumans
330bf3ea09 start implementing ray-convex on GPU (work-in-progress) 2013-06-20 11:33:00 -07:00
erwin coumans
531af1b394 add ray-convex CPU implementation, make CPU raytest default for now 
tweak/increase the b3Config values again, so it works better on the Macbook Retina GPU.
 2013-06-20 00:29:09 -07:00
erwin coumans
a69ba48de4 Move b3Quickprof.* from Bullet 3.x src folder to btgui/Timing 
The Bullet 3.x B3_PROFILE can be customized using b3SetCustomEnterProfileZoneFunc/b3SetCustomLeaveProfileZoneFunc defined in Bullet3Common/b3Logging,
so you can hook Bullet 3.x up to your profiler of choice.
The Demos3/BasicGpuDemo will show the Bullet 3.x timings inside the Bullet 2.x btQuickprof profiler.
 2013-06-19 22:08:03 -07:00
erwin coumans
f10eb86f55 enable the GPU ray test in BasicGpuDemo, only for spheres at the moment 2013-06-19 00:12:06 -07:00
erwin coumans
481d54967f Fixes in raycast (against sphere), point 2 point constraint 
Picking test works, holding shift in App_BasicGpuDemo_*
 2013-06-18 19:33:45 -07:00
erwincoumans
9a92eecf10 allow larger amount of rigid bodies, dynamically increase b3Config limits 
avoid crashes in instancing renderer if instance maximum is exceeded.
 2013-06-17 17:05:01 -07:00
erwincoumans
561a44e5d8 Try to avoid crashes if memory allocations fail 
Deal with cases exceeding allocations of bodies/shapes above the limits set in b3Config
(work-in-progress )
 2013-06-17 15:55:41 -07:00
erwincoumans
e179d5882e Merge branch 'master' of https://github.com/erwincoumans/bullet3 2013-06-17 13:47:41 -07:00
erwincoumans
161c48d331 work-in-progress Bullet 3.x GPU raytest 
work-in-progress P2P constraint for b3GpuDynamicsWorld
 2013-06-17 13:47:35 -07:00
erwin coumans
31282ab85e don't crash if the maximum number of pairs is exceeded, but report an error using b3Error 2013-06-17 13:23:41 -07:00
erwin coumans
f73d11d896 tweak default values of b3Config.h so some laptop GPUs can run 128k scenes 
return the result of resize
use size_t instead of int for btOpenCLArray
BasicInitialize demo will test how much memory can be allocated, using the btOpenCLArray
 2013-06-14 12:10:16 -07:00
erwin coumans
100449d76f faster scene construction when not using instancing (avoid copy individual vertices/shapes to GPU, but do it afterwards in a batch), copy GPU data in 'writeAllBodiesToGpu' method 
add option --no_instanced_collision_shapes, this only applies to the 2 benchmark screnes at the moment
always use the user specified b3Config file, so the settings can be set at runtime, not just at compile time
adjust default constants in b3Config (needs more tweaking, ideally at run-time)
 2013-06-08 11:08:44 -07:00
erwin coumans
3f10082aa3 add missing rayCast kernel 2013-06-07 20:47:47 -07:00
erwincoumans
9fe48d887d b3GpuDynamicsWorld improvements: 
apply forces
copy linear/angular velocity every frame at the CPU side, initial velocity works now
hook up setGravity
Note: the 'stepSimulation' for GPU only simulates a single simulation frame.
 2013-06-06 22:01:06 -07:00
Erwin Coumans
a9d4184e2d fix Mac version of cloth rendering 2013-06-03 17:35:31 -07:00
erwin coumans
a886a978b7 add preliminary GPU ray cast (brute-force slow, only against sphere, allocates OpenCL buffers at every cast) 2013-05-28 18:39:53 -07:00
Erwin Coumans
a92223890f improve Mac OSX/Unix benchmark 2013-05-23 20:38:27 -07:00
erwin coumans
4877a2af8a enable CPU joint (non-contact) constraint solver in GPU rigid body pipeline 
small change to BasicGpuDemo
 2013-05-08 15:55:02 -07:00
erwin coumans
26873328cf gridtable needs to be in the __constant memory space 2013-05-03 11:38:23 -07:00
erwin coumans
1185de51d5 experiment with first-level batching using object index instead of spatial hash in uniform grid (to avoid tuning average object size for uniform grid) 2013-05-03 01:14:34 -07:00
erwin coumans
6ee9eb9bb5 fix bug in assignment of contact constraints to solver grid (always use dynamic body to determine constraint assignment, otherwise write conflicts can occur) 
implement CPU version of narrowphase convex collision, for comparison/debug purposes
start towards cpu/gpu sync, for adding/removing bodies (work in progress)
 2013-05-02 09:49:16 -07:00
erwincoumans
e999903c15 resolve some naming conflicts between Bullet 3.x and Bullet 2.x 
add BasicGpuDemo that bridges Bullet 2.x and Bullet 3.x (still preliminary/crashing/instable)
 2013-04-30 18:17:44 -07:00
erwincoumans
92f0938af3 add bitonic sort, as comparison. 
fix stringify.bat for Windows (need to fix Mac/Linux version too)
 2013-04-30 11:40:09 -07:00
erwincoumans
3ac332f3a7 reorder files, in preparation for Bullet 3 -> Bullet 2 merge 2013-04-29 19:04:08 -07:00