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

change lcpp Lua preprocessor, to keep #defines and comments, remove empty lines

Browse Source 
remove duplicate data in b3Contact4 (now in btContact4Data shared between CPU/C++ and OpenCL)
OpenCL kernels use #include "Bullet3Collision/NarrowPhaseCollision/shared/b3Contact4Data.h"
Increase number of batches back to 250 (from 50), need to fix this hard coded number (see https://github.com/erwincoumans/bullet3/issues/12)
Work towards GJK/EPA, in addition to SAT/clipping (early on)
This commit is contained in:
erwincoumans
2013-08-08 12:24:09 -07:00
parent 46a08e3282
commit 3bf003ace1
50 changed files with 920 additions and 2731 deletions
Download Patch File Download Diff File Expand all files Collapse all files

45
src/Bullet3OpenCL/BroadphaseCollision/kernels/sapKernels.h
View File

@@ -2,21 +2,17 @@
static const char* sapCL= \
"/*\n"
"Copyright (c) 2012 Advanced Micro Devices, Inc. \n"
"\n"
"This software is provided 'as-is', without any express or implied warranty.\n"
"In no event will the authors be held liable for any damages arising from the use of this software.\n"
"Permission is granted to anyone to use this software for any purpose, \n"
"including commercial applications, and to alter it and redistribute it freely, \n"
"subject to the following restrictions:\n"
"\n"
"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
"3. This notice may not be removed or altered from any source distribution.\n"
"*/\n"
"//Originally written by Erwin Coumans\n"
"\n"
"#define NEW_PAIR_MARKER -1\n"
"\n"
"typedef struct \n"
"{\n"
" union\n"
@@ -32,8 +28,6 @@ static const char* sapCL= \
" int m_maxIndices[4];\n"
" };\n"
"} btAabbCL;\n"
"\n"
"\n"
"/// conservative test for overlap between two aabbs\n"
"bool TestAabbAgainstAabb2(const btAabbCL* aabb1, __local const btAabbCL* aabb2);\n"
"bool TestAabbAgainstAabb2(const btAabbCL* aabb1, __local const btAabbCL* aabb2)\n"
@@ -53,7 +47,6 @@ static const char* sapCL= \
" overlap = (aabb1->m_min.y > aabb2->m_max.y || aabb1->m_max.y < aabb2->m_min.y) ? false : overlap;\n"
" return overlap;\n"
"}\n"
"\n"
"bool TestAabbAgainstAabb2Global(const btAabbCL* aabb1, __global const btAabbCL* aabb2);\n"
"bool TestAabbAgainstAabb2Global(const btAabbCL* aabb1, __global const btAabbCL* aabb2)\n"
"{\n"
@@ -63,18 +56,14 @@ static const char* sapCL= \
" overlap = (aabb1->m_min.y > aabb2->m_max.y || aabb1->m_max.y < aabb2->m_min.y) ? false : overlap;\n"
" return overlap;\n"
"}\n"
"\n"
"\n"
"__kernel void computePairsKernelTwoArrays( __global const btAabbCL* unsortedAabbs, __global const btAabbCL* sortedAabbs, volatile __global int4* pairsOut,volatile __global int* pairCount, int numUnsortedAabbs, int numSortedAabbs, int axis, int maxPairs)\n"
"{\n"
" int i = get_global_id(0);\n"
" if (i>=numUnsortedAabbs)\n"
" return;\n"
"\n"
" int j = get_global_id(1);\n"
" if (j>=numSortedAabbs)\n"
" return;\n"
"\n"
" if (TestAabbAgainstAabb2GlobalGlobal(&unsortedAabbs[i],&sortedAabbs[j]))\n"
" {\n"
" int4 myPair;\n"
@@ -92,8 +81,6 @@ static const char* sapCL= \
" myPair.y = yIndex;\n"
" myPair.z = NEW_PAIR_MARKER;\n"
" myPair.w = NEW_PAIR_MARKER;\n"
"\n"
"\n"
" int curPair = atomic_inc (pairCount);\n"
" if (curPair<maxPairs)\n"
" {\n"
@@ -101,7 +88,6 @@ static const char* sapCL= \
" }\n"
" }\n"
"}\n"
"\n"
"__kernel void computePairsKernelOriginal( __global const btAabbCL* aabbs, volatile __global int4* pairsOut,volatile __global int* pairCount, int numObjects, int axis, int maxPairs)\n"
"{\n"
" int i = get_global_id(0);\n"
@@ -120,7 +106,6 @@ static const char* sapCL= \
" myPair.y = aabbs[j].m_minIndices[3];\n"
" myPair.z = NEW_PAIR_MARKER;\n"
" myPair.w = NEW_PAIR_MARKER;\n"
"\n"
" int curPair = atomic_inc (pairCount);\n"
" if (curPair<maxPairs)\n"
" {\n"
@@ -129,18 +114,12 @@ static const char* sapCL= \
" }\n"
" }\n"
"}\n"
"\n"
"\n"
"\n"
"\n"
"__kernel void computePairsKernelBarrier( __global const btAabbCL* aabbs, volatile __global int4* pairsOut,volatile __global int* pairCount, int numObjects, int axis, int maxPairs)\n"
"{\n"
" int i = get_global_id(0);\n"
" int localId = get_local_id(0);\n"
"\n"
" __local int numActiveWgItems[1];\n"
" __local int breakRequest[1];\n"
"\n"
" if (localId==0)\n"
" {\n"
" numActiveWgItems[0] = 0;\n"
@@ -150,7 +129,6 @@ static const char* sapCL= \
" atomic_inc(numActiveWgItems);\n"
" barrier(CLK_LOCAL_MEM_FENCE);\n"
" int localBreak = 0;\n"
"\n"
" int j=i+1;\n"
" do\n"
" {\n"
@@ -186,7 +164,6 @@ static const char* sapCL= \
" myPair.y = aabbs[j].m_minIndices[3];\n"
" myPair.z = NEW_PAIR_MARKER;\n"
" myPair.w = NEW_PAIR_MARKER;\n"
"\n"
" int curPair = atomic_inc (pairCount);\n"
" if (curPair<maxPairs)\n"
" {\n"
@@ -195,16 +172,12 @@ static const char* sapCL= \
" }\n"
" }\n"
" j++;\n"
"\n"
" } while (breakRequest[0]<numActiveWgItems[0]);\n"
"}\n"
"\n"
"\n"
"__kernel void computePairsKernelLocalSharedMemory( __global const btAabbCL* aabbs, volatile __global int4* pairsOut,volatile __global int* pairCount, int numObjects, int axis, int maxPairs)\n"
"{\n"
" int i = get_global_id(0);\n"
" int localId = get_local_id(0);\n"
"\n"
" __local int numActiveWgItems[1];\n"
" __local int breakRequest[1];\n"
" __local btAabbCL localAabbs[128];// = aabbs[i];\n"
@@ -264,7 +237,6 @@ static const char* sapCL= \
" myPair.y = localAabbs[localCount+localId+1].m_minIndices[3];\n"
" myPair.z = NEW_PAIR_MARKER;\n"
" myPair.w = NEW_PAIR_MARKER;\n"
"\n"
" int curPair = atomic_inc (pairCount);\n"
" if (curPair<maxPairs)\n"
" {\n"
@@ -274,7 +246,6 @@ static const char* sapCL= \
" }\n"
" \n"
" barrier(CLK_LOCAL_MEM_FENCE);\n"
"\n"
" localCount++;\n"
" if (localCount==64)\n"
" {\n"
@@ -288,10 +259,6 @@ static const char* sapCL= \
" } while (breakRequest[0]<numActiveWgItems[0]);\n"
" \n"
"}\n"
"\n"
"\n"
"\n"
"\n"
"//http://stereopsis.com/radix.html\n"
"unsigned int FloatFlip(float fl);\n"
"unsigned int FloatFlip(float fl)\n"
@@ -307,10 +274,6 @@ static const char* sapCL= \
" unsigned int fl = f ^ mask;\n"
" return *(float*)&fl;\n"
"}\n"
"\n"
"\n"
"\n"
"\n"
"__kernel void copyAabbsKernel( __global const btAabbCL* allAabbs, __global btAabbCL* destAabbs, int numObjects)\n"
"{\n"
" int i = get_global_id(0);\n"
@@ -320,8 +283,6 @@ static const char* sapCL= \
" destAabbs[i] = allAabbs[src];\n"
" destAabbs[i].m_maxIndices[3] = src;\n"
"}\n"
"\n"
"\n"
"__kernel void flipFloatKernel( __global const btAabbCL* aabbs, volatile __global int2* sortData, int numObjects, int axis)\n"
"{\n"
" int i = get_global_id(0);\n"
@@ -332,19 +293,13 @@ static const char* sapCL= \
" sortData[i].y = i;\n"
" \n"
"}\n"
"\n"
"\n"
"__kernel void scatterKernel( __global const btAabbCL* aabbs, volatile __global const int2* sortData, __global btAabbCL* sortedAabbs, int numObjects)\n"
"{\n"
" int i = get_global_id(0);\n"
" if (i>=numObjects)\n"
" return;\n"
"\n"
" sortedAabbs[i] = aabbs[sortData[i].y];\n"
"}\n"
"\n"
"\n"
"\n"
"__kernel void prepareSumVarianceKernel( __global const btAabbCL* aabbs, __global float4* sum, __global float4* sum2,int numAabbs)\n"
"{\n"
" int i = get_global_id(0);\n"