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

add some clFinish for better profile timings

Browse Source 
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)
This commit is contained in:
erwincoumans
2013-07-15 10:22:40 -07:00
parent 05ed1fdbcf
commit a5d00c8219
4 changed files with 28 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
src/Bullet3OpenCL/RigidBody/b3GpuBatchingPgsSolver.cpp
View File

@@ -373,7 +373,7 @@ void b3GpuBatchingPgsSolver::solveContactConstraint( const b3OpenCLArray<b3Rigi
}
}
//clFinish(m_data->m_queue);
clFinish(m_data->m_queue);
}
@@ -416,7 +416,7 @@ void b3GpuBatchingPgsSolver::solveContactConstraint( const b3OpenCLArray<b3Rigi
launcher.launch1D( 64*nn/B3_SOLVER_N_BATCHES, 64 );
}
}
//clFinish(m_data->m_queue);
clFinish(m_data->m_queue);
}
#ifdef DEBUG_ME
@@ -838,7 +838,8 @@ void b3GpuBatchingPgsSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
launcher.setConst( cdata );
launcher.launch1D( nContacts, 64 );
//clFinish(m_data->m_queue);
//we use the clFinish for proper benchmark/profile
clFinish(m_data->m_queue);
}
@@ -900,7 +901,9 @@ void b3GpuBatchingPgsSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
b3Printf("maxNumBatches = %d\n",maxNumBatches);
}
//clFinish(m_data->m_queue);
//we use the clFinish for proper benchmark/profile
clFinish(m_data->m_queue);
}
}
@@ -925,7 +928,7 @@ void b3GpuBatchingPgsSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
contactConstraintOut,
additionalData, nContacts,
(b3SolverBase::ConstraintCfg&) csCfg );
//clFinish(m_data->m_queue);
clFinish(m_data->m_queue);
}
@@ -968,7 +971,7 @@ void b3GpuBatchingPgsSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
m_data->m_solverGPU->solveContactConstraintHost(m_data->m_bodyBufferGPU, m_data->m_inertiaBufferGPU, m_data->m_contactCGPU,0, nContactOut ,maxNumBatches);
}
//clFinish(m_data->m_queue);
}