Code-style consistency improvement:
Apply clang-format-all.sh using the _clang-format file through all the cpp/.h files. make sure not to apply it to certain serialization structures, since some parser expects the * as part of the name, instead of type. This commit contains no other changes aside from adding and applying clang-format-all.sh
This commit is contained in:
erwincoumans
@@ -1,199 +1,198 @@
|
||||
//this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
|
||||
static const char* gridBroadphaseCL= \
|
||||
"int getPosHash(int4 gridPos, __global float4* pParams)\n"
|
||||
"{\n"
|
||||
" int4 gridDim = *((__global int4*)(pParams + 1));\n"
|
||||
" gridPos.x &= gridDim.x - 1;\n"
|
||||
" gridPos.y &= gridDim.y - 1;\n"
|
||||
" gridPos.z &= gridDim.z - 1;\n"
|
||||
" int hash = gridPos.z * gridDim.y * gridDim.x + gridPos.y * gridDim.x + gridPos.x;\n"
|
||||
" return hash;\n"
|
||||
"} \n"
|
||||
"int4 getGridPos(float4 worldPos, __global float4* pParams)\n"
|
||||
"{\n"
|
||||
" int4 gridPos;\n"
|
||||
" int4 gridDim = *((__global int4*)(pParams + 1));\n"
|
||||
" gridPos.x = (int)floor(worldPos.x * pParams[0].x) & (gridDim.x - 1);\n"
|
||||
" gridPos.y = (int)floor(worldPos.y * pParams[0].y) & (gridDim.y - 1);\n"
|
||||
" gridPos.z = (int)floor(worldPos.z * pParams[0].z) & (gridDim.z - 1);\n"
|
||||
" return gridPos;\n"
|
||||
"}\n"
|
||||
"// calculate grid hash value for each body using its AABB\n"
|
||||
"__kernel void kCalcHashAABB(int numObjects, __global float4* allpAABB, __global const int* smallAabbMapping, __global int2* pHash, __global float4* pParams )\n"
|
||||
"{\n"
|
||||
" int index = get_global_id(0);\n"
|
||||
" if(index >= numObjects)\n"
|
||||
" {\n"
|
||||
" return;\n"
|
||||
" }\n"
|
||||
" float4 bbMin = allpAABB[smallAabbMapping[index]*2];\n"
|
||||
" float4 bbMax = allpAABB[smallAabbMapping[index]*2 + 1];\n"
|
||||
" float4 pos;\n"
|
||||
" pos.x = (bbMin.x + bbMax.x) * 0.5f;\n"
|
||||
" pos.y = (bbMin.y + bbMax.y) * 0.5f;\n"
|
||||
" pos.z = (bbMin.z + bbMax.z) * 0.5f;\n"
|
||||
" pos.w = 0.f;\n"
|
||||
" // get address in grid\n"
|
||||
" int4 gridPos = getGridPos(pos, pParams);\n"
|
||||
" int gridHash = getPosHash(gridPos, pParams);\n"
|
||||
" // store grid hash and body index\n"
|
||||
" int2 hashVal;\n"
|
||||
" hashVal.x = gridHash;\n"
|
||||
" hashVal.y = index;\n"
|
||||
" pHash[index] = hashVal;\n"
|
||||
"}\n"
|
||||
"__kernel void kClearCellStart( int numCells, \n"
|
||||
" __global int* pCellStart )\n"
|
||||
"{\n"
|
||||
" int index = get_global_id(0);\n"
|
||||
" if(index >= numCells)\n"
|
||||
" {\n"
|
||||
" return;\n"
|
||||
" }\n"
|
||||
" pCellStart[index] = -1;\n"
|
||||
"}\n"
|
||||
"__kernel void kFindCellStart(int numObjects, __global int2* pHash, __global int* cellStart )\n"
|
||||
"{\n"
|
||||
" __local int sharedHash[513];\n"
|
||||
" int index = get_global_id(0);\n"
|
||||
" int2 sortedData;\n"
|
||||
" if(index < numObjects)\n"
|
||||
" {\n"
|
||||
" sortedData = pHash[index];\n"
|
||||
" // Load hash data into shared memory so that we can look \n"
|
||||
" // at neighboring body's hash value without loading\n"
|
||||
" // two hash values per thread\n"
|
||||
" sharedHash[get_local_id(0) + 1] = sortedData.x;\n"
|
||||
" if((index > 0) && (get_local_id(0) == 0))\n"
|
||||
" {\n"
|
||||
" // first thread in block must load neighbor body hash\n"
|
||||
" sharedHash[0] = pHash[index-1].x;\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" if(index < numObjects)\n"
|
||||
" {\n"
|
||||
" if((index == 0) || (sortedData.x != sharedHash[get_local_id(0)]))\n"
|
||||
" {\n"
|
||||
" cellStart[sortedData.x] = index;\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
"}\n"
|
||||
"int testAABBOverlap(float4 min0, float4 max0, float4 min1, float4 max1)\n"
|
||||
"{\n"
|
||||
" return (min0.x <= max1.x)&& (min1.x <= max0.x) && \n"
|
||||
" (min0.y <= max1.y)&& (min1.y <= max0.y) && \n"
|
||||
" (min0.z <= max1.z)&& (min1.z <= max0.z); \n"
|
||||
"}\n"
|
||||
"//search for AABB 'index' against other AABBs' in this cell\n"
|
||||
"void findPairsInCell( int numObjects,\n"
|
||||
" int4 gridPos,\n"
|
||||
" int index,\n"
|
||||
" __global int2* pHash,\n"
|
||||
" __global int* pCellStart,\n"
|
||||
" __global float4* allpAABB, \n"
|
||||
" __global const int* smallAabbMapping,\n"
|
||||
" __global float4* pParams,\n"
|
||||
" volatile __global int* pairCount,\n"
|
||||
" __global int4* pPairBuff2,\n"
|
||||
" int maxPairs\n"
|
||||
" )\n"
|
||||
"{\n"
|
||||
" int4 pGridDim = *((__global int4*)(pParams + 1));\n"
|
||||
" int maxBodiesPerCell = pGridDim.w;\n"
|
||||
" int gridHash = getPosHash(gridPos, pParams);\n"
|
||||
" // get start of bucket for this cell\n"
|
||||
" int bucketStart = pCellStart[gridHash];\n"
|
||||
" if (bucketStart == -1)\n"
|
||||
" {\n"
|
||||
" return; // cell empty\n"
|
||||
" }\n"
|
||||
" // iterate over bodies in this cell\n"
|
||||
" int2 sortedData = pHash[index];\n"
|
||||
" int unsorted_indx = sortedData.y;\n"
|
||||
" float4 min0 = allpAABB[smallAabbMapping[unsorted_indx]*2 + 0]; \n"
|
||||
" float4 max0 = allpAABB[smallAabbMapping[unsorted_indx]*2 + 1];\n"
|
||||
" int handleIndex = as_int(min0.w);\n"
|
||||
" \n"
|
||||
" int bucketEnd = bucketStart + maxBodiesPerCell;\n"
|
||||
" bucketEnd = (bucketEnd > numObjects) ? numObjects : bucketEnd;\n"
|
||||
" for(int index2 = bucketStart; index2 < bucketEnd; index2++) \n"
|
||||
" {\n"
|
||||
" int2 cellData = pHash[index2];\n"
|
||||
" if (cellData.x != gridHash)\n"
|
||||
" {\n"
|
||||
" break; // no longer in same bucket\n"
|
||||
" }\n"
|
||||
" int unsorted_indx2 = cellData.y;\n"
|
||||
" //if (unsorted_indx2 < unsorted_indx) // check not colliding with self\n"
|
||||
" if (unsorted_indx2 != unsorted_indx) // check not colliding with self\n"
|
||||
" { \n"
|
||||
" float4 min1 = allpAABB[smallAabbMapping[unsorted_indx2]*2 + 0];\n"
|
||||
" float4 max1 = allpAABB[smallAabbMapping[unsorted_indx2]*2 + 1];\n"
|
||||
" if(testAABBOverlap(min0, max0, min1, max1))\n"
|
||||
" {\n"
|
||||
" if (pairCount)\n"
|
||||
" {\n"
|
||||
" int handleIndex2 = as_int(min1.w);\n"
|
||||
" if (handleIndex<handleIndex2)\n"
|
||||
" {\n"
|
||||
" int curPair = atomic_add(pairCount,1);\n"
|
||||
" if (curPair<maxPairs)\n"
|
||||
" {\n"
|
||||
" int4 newpair;\n"
|
||||
" newpair.x = handleIndex;\n"
|
||||
" newpair.y = handleIndex2;\n"
|
||||
" newpair.z = -1;\n"
|
||||
" newpair.w = -1;\n"
|
||||
" pPairBuff2[curPair] = newpair;\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" \n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
"}\n"
|
||||
"__kernel void kFindOverlappingPairs( int numObjects,\n"
|
||||
" __global float4* allpAABB, \n"
|
||||
" __global const int* smallAabbMapping,\n"
|
||||
" __global int2* pHash, \n"
|
||||
" __global int* pCellStart, \n"
|
||||
" __global float4* pParams ,\n"
|
||||
" volatile __global int* pairCount,\n"
|
||||
" __global int4* pPairBuff2,\n"
|
||||
" int maxPairs\n"
|
||||
" )\n"
|
||||
"{\n"
|
||||
" int index = get_global_id(0);\n"
|
||||
" if(index >= numObjects)\n"
|
||||
" {\n"
|
||||
" return;\n"
|
||||
" }\n"
|
||||
" int2 sortedData = pHash[index];\n"
|
||||
" int unsorted_indx = sortedData.y;\n"
|
||||
" float4 bbMin = allpAABB[smallAabbMapping[unsorted_indx]*2 + 0];\n"
|
||||
" float4 bbMax = allpAABB[smallAabbMapping[unsorted_indx]*2 + 1];\n"
|
||||
" float4 pos;\n"
|
||||
" pos.x = (bbMin.x + bbMax.x) * 0.5f;\n"
|
||||
" pos.y = (bbMin.y + bbMax.y) * 0.5f;\n"
|
||||
" pos.z = (bbMin.z + bbMax.z) * 0.5f;\n"
|
||||
" // get address in grid\n"
|
||||
" int4 gridPosA = getGridPos(pos, pParams);\n"
|
||||
" int4 gridPosB; \n"
|
||||
" // examine only neighbouring cells\n"
|
||||
" for(int z=-1; z<=1; z++) \n"
|
||||
" {\n"
|
||||
" gridPosB.z = gridPosA.z + z;\n"
|
||||
" for(int y=-1; y<=1; y++) \n"
|
||||
" {\n"
|
||||
" gridPosB.y = gridPosA.y + y;\n"
|
||||
" for(int x=-1; x<=1; x++) \n"
|
||||
" {\n"
|
||||
" gridPosB.x = gridPosA.x + x;\n"
|
||||
" findPairsInCell(numObjects, gridPosB, index, pHash, pCellStart, allpAABB,smallAabbMapping, pParams, pairCount,pPairBuff2, maxPairs);\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
"}\n"
|
||||
;
|
||||
static const char* gridBroadphaseCL =
|
||||
"int getPosHash(int4 gridPos, __global float4* pParams)\n"
|
||||
"{\n"
|
||||
" int4 gridDim = *((__global int4*)(pParams + 1));\n"
|
||||
" gridPos.x &= gridDim.x - 1;\n"
|
||||
" gridPos.y &= gridDim.y - 1;\n"
|
||||
" gridPos.z &= gridDim.z - 1;\n"
|
||||
" int hash = gridPos.z * gridDim.y * gridDim.x + gridPos.y * gridDim.x + gridPos.x;\n"
|
||||
" return hash;\n"
|
||||
"} \n"
|
||||
"int4 getGridPos(float4 worldPos, __global float4* pParams)\n"
|
||||
"{\n"
|
||||
" int4 gridPos;\n"
|
||||
" int4 gridDim = *((__global int4*)(pParams + 1));\n"
|
||||
" gridPos.x = (int)floor(worldPos.x * pParams[0].x) & (gridDim.x - 1);\n"
|
||||
" gridPos.y = (int)floor(worldPos.y * pParams[0].y) & (gridDim.y - 1);\n"
|
||||
" gridPos.z = (int)floor(worldPos.z * pParams[0].z) & (gridDim.z - 1);\n"
|
||||
" return gridPos;\n"
|
||||
"}\n"
|
||||
"// calculate grid hash value for each body using its AABB\n"
|
||||
"__kernel void kCalcHashAABB(int numObjects, __global float4* allpAABB, __global const int* smallAabbMapping, __global int2* pHash, __global float4* pParams )\n"
|
||||
"{\n"
|
||||
" int index = get_global_id(0);\n"
|
||||
" if(index >= numObjects)\n"
|
||||
" {\n"
|
||||
" return;\n"
|
||||
" }\n"
|
||||
" float4 bbMin = allpAABB[smallAabbMapping[index]*2];\n"
|
||||
" float4 bbMax = allpAABB[smallAabbMapping[index]*2 + 1];\n"
|
||||
" float4 pos;\n"
|
||||
" pos.x = (bbMin.x + bbMax.x) * 0.5f;\n"
|
||||
" pos.y = (bbMin.y + bbMax.y) * 0.5f;\n"
|
||||
" pos.z = (bbMin.z + bbMax.z) * 0.5f;\n"
|
||||
" pos.w = 0.f;\n"
|
||||
" // get address in grid\n"
|
||||
" int4 gridPos = getGridPos(pos, pParams);\n"
|
||||
" int gridHash = getPosHash(gridPos, pParams);\n"
|
||||
" // store grid hash and body index\n"
|
||||
" int2 hashVal;\n"
|
||||
" hashVal.x = gridHash;\n"
|
||||
" hashVal.y = index;\n"
|
||||
" pHash[index] = hashVal;\n"
|
||||
"}\n"
|
||||
"__kernel void kClearCellStart( int numCells, \n"
|
||||
" __global int* pCellStart )\n"
|
||||
"{\n"
|
||||
" int index = get_global_id(0);\n"
|
||||
" if(index >= numCells)\n"
|
||||
" {\n"
|
||||
" return;\n"
|
||||
" }\n"
|
||||
" pCellStart[index] = -1;\n"
|
||||
"}\n"
|
||||
"__kernel void kFindCellStart(int numObjects, __global int2* pHash, __global int* cellStart )\n"
|
||||
"{\n"
|
||||
" __local int sharedHash[513];\n"
|
||||
" int index = get_global_id(0);\n"
|
||||
" int2 sortedData;\n"
|
||||
" if(index < numObjects)\n"
|
||||
" {\n"
|
||||
" sortedData = pHash[index];\n"
|
||||
" // Load hash data into shared memory so that we can look \n"
|
||||
" // at neighboring body's hash value without loading\n"
|
||||
" // two hash values per thread\n"
|
||||
" sharedHash[get_local_id(0) + 1] = sortedData.x;\n"
|
||||
" if((index > 0) && (get_local_id(0) == 0))\n"
|
||||
" {\n"
|
||||
" // first thread in block must load neighbor body hash\n"
|
||||
" sharedHash[0] = pHash[index-1].x;\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" if(index < numObjects)\n"
|
||||
" {\n"
|
||||
" if((index == 0) || (sortedData.x != sharedHash[get_local_id(0)]))\n"
|
||||
" {\n"
|
||||
" cellStart[sortedData.x] = index;\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
"}\n"
|
||||
"int testAABBOverlap(float4 min0, float4 max0, float4 min1, float4 max1)\n"
|
||||
"{\n"
|
||||
" return (min0.x <= max1.x)&& (min1.x <= max0.x) && \n"
|
||||
" (min0.y <= max1.y)&& (min1.y <= max0.y) && \n"
|
||||
" (min0.z <= max1.z)&& (min1.z <= max0.z); \n"
|
||||
"}\n"
|
||||
"//search for AABB 'index' against other AABBs' in this cell\n"
|
||||
"void findPairsInCell( int numObjects,\n"
|
||||
" int4 gridPos,\n"
|
||||
" int index,\n"
|
||||
" __global int2* pHash,\n"
|
||||
" __global int* pCellStart,\n"
|
||||
" __global float4* allpAABB, \n"
|
||||
" __global const int* smallAabbMapping,\n"
|
||||
" __global float4* pParams,\n"
|
||||
" volatile __global int* pairCount,\n"
|
||||
" __global int4* pPairBuff2,\n"
|
||||
" int maxPairs\n"
|
||||
" )\n"
|
||||
"{\n"
|
||||
" int4 pGridDim = *((__global int4*)(pParams + 1));\n"
|
||||
" int maxBodiesPerCell = pGridDim.w;\n"
|
||||
" int gridHash = getPosHash(gridPos, pParams);\n"
|
||||
" // get start of bucket for this cell\n"
|
||||
" int bucketStart = pCellStart[gridHash];\n"
|
||||
" if (bucketStart == -1)\n"
|
||||
" {\n"
|
||||
" return; // cell empty\n"
|
||||
" }\n"
|
||||
" // iterate over bodies in this cell\n"
|
||||
" int2 sortedData = pHash[index];\n"
|
||||
" int unsorted_indx = sortedData.y;\n"
|
||||
" float4 min0 = allpAABB[smallAabbMapping[unsorted_indx]*2 + 0]; \n"
|
||||
" float4 max0 = allpAABB[smallAabbMapping[unsorted_indx]*2 + 1];\n"
|
||||
" int handleIndex = as_int(min0.w);\n"
|
||||
" \n"
|
||||
" int bucketEnd = bucketStart + maxBodiesPerCell;\n"
|
||||
" bucketEnd = (bucketEnd > numObjects) ? numObjects : bucketEnd;\n"
|
||||
" for(int index2 = bucketStart; index2 < bucketEnd; index2++) \n"
|
||||
" {\n"
|
||||
" int2 cellData = pHash[index2];\n"
|
||||
" if (cellData.x != gridHash)\n"
|
||||
" {\n"
|
||||
" break; // no longer in same bucket\n"
|
||||
" }\n"
|
||||
" int unsorted_indx2 = cellData.y;\n"
|
||||
" //if (unsorted_indx2 < unsorted_indx) // check not colliding with self\n"
|
||||
" if (unsorted_indx2 != unsorted_indx) // check not colliding with self\n"
|
||||
" { \n"
|
||||
" float4 min1 = allpAABB[smallAabbMapping[unsorted_indx2]*2 + 0];\n"
|
||||
" float4 max1 = allpAABB[smallAabbMapping[unsorted_indx2]*2 + 1];\n"
|
||||
" if(testAABBOverlap(min0, max0, min1, max1))\n"
|
||||
" {\n"
|
||||
" if (pairCount)\n"
|
||||
" {\n"
|
||||
" int handleIndex2 = as_int(min1.w);\n"
|
||||
" if (handleIndex<handleIndex2)\n"
|
||||
" {\n"
|
||||
" int curPair = atomic_add(pairCount,1);\n"
|
||||
" if (curPair<maxPairs)\n"
|
||||
" {\n"
|
||||
" int4 newpair;\n"
|
||||
" newpair.x = handleIndex;\n"
|
||||
" newpair.y = handleIndex2;\n"
|
||||
" newpair.z = -1;\n"
|
||||
" newpair.w = -1;\n"
|
||||
" pPairBuff2[curPair] = newpair;\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" \n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
"}\n"
|
||||
"__kernel void kFindOverlappingPairs( int numObjects,\n"
|
||||
" __global float4* allpAABB, \n"
|
||||
" __global const int* smallAabbMapping,\n"
|
||||
" __global int2* pHash, \n"
|
||||
" __global int* pCellStart, \n"
|
||||
" __global float4* pParams ,\n"
|
||||
" volatile __global int* pairCount,\n"
|
||||
" __global int4* pPairBuff2,\n"
|
||||
" int maxPairs\n"
|
||||
" )\n"
|
||||
"{\n"
|
||||
" int index = get_global_id(0);\n"
|
||||
" if(index >= numObjects)\n"
|
||||
" {\n"
|
||||
" return;\n"
|
||||
" }\n"
|
||||
" int2 sortedData = pHash[index];\n"
|
||||
" int unsorted_indx = sortedData.y;\n"
|
||||
" float4 bbMin = allpAABB[smallAabbMapping[unsorted_indx]*2 + 0];\n"
|
||||
" float4 bbMax = allpAABB[smallAabbMapping[unsorted_indx]*2 + 1];\n"
|
||||
" float4 pos;\n"
|
||||
" pos.x = (bbMin.x + bbMax.x) * 0.5f;\n"
|
||||
" pos.y = (bbMin.y + bbMax.y) * 0.5f;\n"
|
||||
" pos.z = (bbMin.z + bbMax.z) * 0.5f;\n"
|
||||
" // get address in grid\n"
|
||||
" int4 gridPosA = getGridPos(pos, pParams);\n"
|
||||
" int4 gridPosB; \n"
|
||||
" // examine only neighbouring cells\n"
|
||||
" for(int z=-1; z<=1; z++) \n"
|
||||
" {\n"
|
||||
" gridPosB.z = gridPosA.z + z;\n"
|
||||
" for(int y=-1; y<=1; y++) \n"
|
||||
" {\n"
|
||||
" gridPosB.y = gridPosA.y + y;\n"
|
||||
" for(int x=-1; x<=1; x++) \n"
|
||||
" {\n"
|
||||
" gridPosB.x = gridPosA.x + x;\n"
|
||||
" findPairsInCell(numObjects, gridPosB, index, pHash, pCellStart, allpAABB,smallAabbMapping, pParams, pairCount,pPairBuff2, maxPairs);\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
"}\n";
|
||||
|
||||
File diff suppressed because it is too large
Load Diff
@@ -1,342 +1,341 @@
|
||||
//this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
|
||||
static const char* sapCL= \
|
||||
"/*\n"
|
||||
"Copyright (c) 2012 Advanced Micro Devices, Inc. \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"
|
||||
"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"
|
||||
"#define NEW_PAIR_MARKER -1\n"
|
||||
"typedef struct \n"
|
||||
"{\n"
|
||||
" union\n"
|
||||
" {\n"
|
||||
" float4 m_min;\n"
|
||||
" float m_minElems[4];\n"
|
||||
" int m_minIndices[4];\n"
|
||||
" };\n"
|
||||
" union\n"
|
||||
" {\n"
|
||||
" float4 m_max;\n"
|
||||
" float m_maxElems[4];\n"
|
||||
" int m_maxIndices[4];\n"
|
||||
" };\n"
|
||||
"} btAabbCL;\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"
|
||||
"{\n"
|
||||
" bool overlap = true;\n"
|
||||
" overlap = (aabb1->m_min.x > aabb2->m_max.x || aabb1->m_max.x < aabb2->m_min.x) ? false : overlap;\n"
|
||||
" overlap = (aabb1->m_min.z > aabb2->m_max.z || aabb1->m_max.z < aabb2->m_min.z) ? false : overlap;\n"
|
||||
" overlap = (aabb1->m_min.y > aabb2->m_max.y || aabb1->m_max.y < aabb2->m_min.y) ? false : overlap;\n"
|
||||
" return overlap;\n"
|
||||
"}\n"
|
||||
"bool TestAabbAgainstAabb2GlobalGlobal(__global const btAabbCL* aabb1, __global const btAabbCL* aabb2);\n"
|
||||
"bool TestAabbAgainstAabb2GlobalGlobal(__global const btAabbCL* aabb1, __global const btAabbCL* aabb2)\n"
|
||||
"{\n"
|
||||
" bool overlap = true;\n"
|
||||
" overlap = (aabb1->m_min.x > aabb2->m_max.x || aabb1->m_max.x < aabb2->m_min.x) ? false : overlap;\n"
|
||||
" overlap = (aabb1->m_min.z > aabb2->m_max.z || aabb1->m_max.z < aabb2->m_min.z) ? false : overlap;\n"
|
||||
" overlap = (aabb1->m_min.y > aabb2->m_max.y || aabb1->m_max.y < aabb2->m_min.y) ? false : overlap;\n"
|
||||
" return overlap;\n"
|
||||
"}\n"
|
||||
"bool TestAabbAgainstAabb2Global(const btAabbCL* aabb1, __global const btAabbCL* aabb2);\n"
|
||||
"bool TestAabbAgainstAabb2Global(const btAabbCL* aabb1, __global const btAabbCL* aabb2)\n"
|
||||
"{\n"
|
||||
" bool overlap = true;\n"
|
||||
" overlap = (aabb1->m_min.x > aabb2->m_max.x || aabb1->m_max.x < aabb2->m_min.x) ? false : overlap;\n"
|
||||
" overlap = (aabb1->m_min.z > aabb2->m_max.z || aabb1->m_max.z < aabb2->m_min.z) ? false : overlap;\n"
|
||||
" overlap = (aabb1->m_min.y > aabb2->m_max.y || aabb1->m_max.y < aabb2->m_min.y) ? false : overlap;\n"
|
||||
" return overlap;\n"
|
||||
"}\n"
|
||||
"__kernel void computePairsKernelTwoArrays( __global const btAabbCL* unsortedAabbs, __global const int* unsortedAabbMapping, __global const int* unsortedAabbMapping2, volatile __global int4* pairsOut,volatile __global int* pairCount, int numUnsortedAabbs, int numUnSortedAabbs2, int axis, int maxPairs)\n"
|
||||
"{\n"
|
||||
" int i = get_global_id(0);\n"
|
||||
" if (i>=numUnsortedAabbs)\n"
|
||||
" return;\n"
|
||||
" int j = get_global_id(1);\n"
|
||||
" if (j>=numUnSortedAabbs2)\n"
|
||||
" return;\n"
|
||||
" __global const btAabbCL* unsortedAabbPtr = &unsortedAabbs[unsortedAabbMapping[i]];\n"
|
||||
" __global const btAabbCL* unsortedAabbPtr2 = &unsortedAabbs[unsortedAabbMapping2[j]];\n"
|
||||
" if (TestAabbAgainstAabb2GlobalGlobal(unsortedAabbPtr,unsortedAabbPtr2))\n"
|
||||
" {\n"
|
||||
" int4 myPair;\n"
|
||||
" \n"
|
||||
" int xIndex = unsortedAabbPtr[0].m_minIndices[3];\n"
|
||||
" int yIndex = unsortedAabbPtr2[0].m_minIndices[3];\n"
|
||||
" if (xIndex>yIndex)\n"
|
||||
" {\n"
|
||||
" int tmp = xIndex;\n"
|
||||
" xIndex=yIndex;\n"
|
||||
" yIndex=tmp;\n"
|
||||
" }\n"
|
||||
" \n"
|
||||
" myPair.x = xIndex;\n"
|
||||
" myPair.y = yIndex;\n"
|
||||
" myPair.z = NEW_PAIR_MARKER;\n"
|
||||
" myPair.w = NEW_PAIR_MARKER;\n"
|
||||
" int curPair = atomic_inc (pairCount);\n"
|
||||
" if (curPair<maxPairs)\n"
|
||||
" {\n"
|
||||
" pairsOut[curPair] = myPair; //flush to main memory\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
"}\n"
|
||||
"__kernel void computePairsKernelBruteForce( __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"
|
||||
" if (i>=numObjects)\n"
|
||||
" return;\n"
|
||||
" for (int j=i+1;j<numObjects;j++)\n"
|
||||
" {\n"
|
||||
" if (TestAabbAgainstAabb2GlobalGlobal(&aabbs[i],&aabbs[j]))\n"
|
||||
" {\n"
|
||||
" int4 myPair;\n"
|
||||
" myPair.x = aabbs[i].m_minIndices[3];\n"
|
||||
" myPair.y = aabbs[j].m_minIndices[3];\n"
|
||||
" myPair.z = NEW_PAIR_MARKER;\n"
|
||||
" myPair.w = NEW_PAIR_MARKER;\n"
|
||||
" int curPair = atomic_inc (pairCount);\n"
|
||||
" if (curPair<maxPairs)\n"
|
||||
" {\n"
|
||||
" pairsOut[curPair] = myPair; //flush to main memory\n"
|
||||
" }\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"
|
||||
" if (i>=numObjects)\n"
|
||||
" return;\n"
|
||||
" for (int j=i+1;j<numObjects;j++)\n"
|
||||
" {\n"
|
||||
" if(aabbs[i].m_maxElems[axis] < (aabbs[j].m_minElems[axis])) \n"
|
||||
" {\n"
|
||||
" break;\n"
|
||||
" }\n"
|
||||
" if (TestAabbAgainstAabb2GlobalGlobal(&aabbs[i],&aabbs[j]))\n"
|
||||
" {\n"
|
||||
" int4 myPair;\n"
|
||||
" myPair.x = aabbs[i].m_minIndices[3];\n"
|
||||
" myPair.y = aabbs[j].m_minIndices[3];\n"
|
||||
" myPair.z = NEW_PAIR_MARKER;\n"
|
||||
" myPair.w = NEW_PAIR_MARKER;\n"
|
||||
" int curPair = atomic_inc (pairCount);\n"
|
||||
" if (curPair<maxPairs)\n"
|
||||
" {\n"
|
||||
" pairsOut[curPair] = myPair; //flush to main memory\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"
|
||||
" __local int numActiveWgItems[1];\n"
|
||||
" __local int breakRequest[1];\n"
|
||||
" if (localId==0)\n"
|
||||
" {\n"
|
||||
" numActiveWgItems[0] = 0;\n"
|
||||
" breakRequest[0] = 0;\n"
|
||||
" }\n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" atomic_inc(numActiveWgItems);\n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" int localBreak = 0;\n"
|
||||
" int j=i+1;\n"
|
||||
" do\n"
|
||||
" {\n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" \n"
|
||||
" if (j<numObjects)\n"
|
||||
" {\n"
|
||||
" if(aabbs[i].m_maxElems[axis] < (aabbs[j].m_minElems[axis])) \n"
|
||||
" {\n"
|
||||
" if (!localBreak)\n"
|
||||
" {\n"
|
||||
" atomic_inc(breakRequest);\n"
|
||||
" localBreak = 1;\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" \n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" \n"
|
||||
" if (j>=numObjects && !localBreak)\n"
|
||||
" {\n"
|
||||
" atomic_inc(breakRequest);\n"
|
||||
" localBreak = 1;\n"
|
||||
" }\n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" \n"
|
||||
" if (!localBreak)\n"
|
||||
" {\n"
|
||||
" if (TestAabbAgainstAabb2GlobalGlobal(&aabbs[i],&aabbs[j]))\n"
|
||||
" {\n"
|
||||
" int4 myPair;\n"
|
||||
" myPair.x = aabbs[i].m_minIndices[3];\n"
|
||||
" myPair.y = aabbs[j].m_minIndices[3];\n"
|
||||
" myPair.z = NEW_PAIR_MARKER;\n"
|
||||
" myPair.w = NEW_PAIR_MARKER;\n"
|
||||
" int curPair = atomic_inc (pairCount);\n"
|
||||
" if (curPair<maxPairs)\n"
|
||||
" {\n"
|
||||
" pairsOut[curPair] = myPair; //flush to main memory\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" j++;\n"
|
||||
" } while (breakRequest[0]<numActiveWgItems[0]);\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"
|
||||
" __local int numActiveWgItems[1];\n"
|
||||
" __local int breakRequest[1];\n"
|
||||
" __local btAabbCL localAabbs[128];// = aabbs[i];\n"
|
||||
" \n"
|
||||
" btAabbCL myAabb;\n"
|
||||
" \n"
|
||||
" myAabb = (i<numObjects)? aabbs[i]:aabbs[0];\n"
|
||||
" float testValue = myAabb.m_maxElems[axis];\n"
|
||||
" \n"
|
||||
" if (localId==0)\n"
|
||||
" {\n"
|
||||
" numActiveWgItems[0] = 0;\n"
|
||||
" breakRequest[0] = 0;\n"
|
||||
" }\n"
|
||||
" int localCount=0;\n"
|
||||
" int block=0;\n"
|
||||
" localAabbs[localId] = (i+block)<numObjects? aabbs[i+block] : aabbs[0];\n"
|
||||
" localAabbs[localId+64] = (i+block+64)<numObjects? aabbs[i+block+64]: aabbs[0];\n"
|
||||
" \n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" atomic_inc(numActiveWgItems);\n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" int localBreak = 0;\n"
|
||||
" \n"
|
||||
" int j=i+1;\n"
|
||||
" do\n"
|
||||
" {\n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" \n"
|
||||
" if (j<numObjects)\n"
|
||||
" {\n"
|
||||
" if(testValue < (localAabbs[localCount+localId+1].m_minElems[axis])) \n"
|
||||
" {\n"
|
||||
" if (!localBreak)\n"
|
||||
" {\n"
|
||||
" atomic_inc(breakRequest);\n"
|
||||
" localBreak = 1;\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" \n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" \n"
|
||||
" if (j>=numObjects && !localBreak)\n"
|
||||
" {\n"
|
||||
" atomic_inc(breakRequest);\n"
|
||||
" localBreak = 1;\n"
|
||||
" }\n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" \n"
|
||||
" if (!localBreak)\n"
|
||||
" {\n"
|
||||
" if (TestAabbAgainstAabb2(&myAabb,&localAabbs[localCount+localId+1]))\n"
|
||||
" {\n"
|
||||
" int4 myPair;\n"
|
||||
" myPair.x = myAabb.m_minIndices[3];\n"
|
||||
" myPair.y = localAabbs[localCount+localId+1].m_minIndices[3];\n"
|
||||
" myPair.z = NEW_PAIR_MARKER;\n"
|
||||
" myPair.w = NEW_PAIR_MARKER;\n"
|
||||
" int curPair = atomic_inc (pairCount);\n"
|
||||
" if (curPair<maxPairs)\n"
|
||||
" {\n"
|
||||
" pairsOut[curPair] = myPair; //flush to main memory\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" \n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" localCount++;\n"
|
||||
" if (localCount==64)\n"
|
||||
" {\n"
|
||||
" localCount = 0;\n"
|
||||
" block+=64; \n"
|
||||
" localAabbs[localId] = ((i+block)<numObjects) ? aabbs[i+block] : aabbs[0];\n"
|
||||
" localAabbs[localId+64] = ((i+64+block)<numObjects) ? aabbs[i+block+64] : aabbs[0];\n"
|
||||
" }\n"
|
||||
" j++;\n"
|
||||
" \n"
|
||||
" } while (breakRequest[0]<numActiveWgItems[0]);\n"
|
||||
" \n"
|
||||
"}\n"
|
||||
"//http://stereopsis.com/radix.html\n"
|
||||
"unsigned int FloatFlip(float fl);\n"
|
||||
"unsigned int FloatFlip(float fl)\n"
|
||||
"{\n"
|
||||
" unsigned int f = *(unsigned int*)&fl;\n"
|
||||
" unsigned int mask = -(int)(f >> 31) | 0x80000000;\n"
|
||||
" return f ^ mask;\n"
|
||||
"}\n"
|
||||
"float IFloatFlip(unsigned int f);\n"
|
||||
"float IFloatFlip(unsigned int f)\n"
|
||||
"{\n"
|
||||
" unsigned int mask = ((f >> 31) - 1) | 0x80000000;\n"
|
||||
" unsigned int fl = f ^ mask;\n"
|
||||
" return *(float*)&fl;\n"
|
||||
"}\n"
|
||||
"__kernel void copyAabbsKernel( __global const btAabbCL* allAabbs, __global btAabbCL* destAabbs, int numObjects)\n"
|
||||
"{\n"
|
||||
" int i = get_global_id(0);\n"
|
||||
" if (i>=numObjects)\n"
|
||||
" return;\n"
|
||||
" int src = destAabbs[i].m_maxIndices[3];\n"
|
||||
" destAabbs[i] = allAabbs[src];\n"
|
||||
" destAabbs[i].m_maxIndices[3] = src;\n"
|
||||
"}\n"
|
||||
"__kernel void flipFloatKernel( __global const btAabbCL* allAabbs, __global const int* smallAabbMapping, __global int2* sortData, int numObjects, int axis)\n"
|
||||
"{\n"
|
||||
" int i = get_global_id(0);\n"
|
||||
" if (i>=numObjects)\n"
|
||||
" return;\n"
|
||||
" \n"
|
||||
" \n"
|
||||
" sortData[i].x = FloatFlip(allAabbs[smallAabbMapping[i]].m_minElems[axis]);\n"
|
||||
" sortData[i].y = i;\n"
|
||||
" \n"
|
||||
"}\n"
|
||||
"__kernel void scatterKernel( __global const btAabbCL* allAabbs, __global const int* smallAabbMapping, 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] = allAabbs[smallAabbMapping[sortData[i].y]];\n"
|
||||
"}\n"
|
||||
"__kernel void prepareSumVarianceKernel( __global const btAabbCL* allAabbs, __global const int* smallAabbMapping, __global float4* sum, __global float4* sum2,int numAabbs)\n"
|
||||
"{\n"
|
||||
" int i = get_global_id(0);\n"
|
||||
" if (i>=numAabbs)\n"
|
||||
" return;\n"
|
||||
" \n"
|
||||
" btAabbCL smallAabb = allAabbs[smallAabbMapping[i]];\n"
|
||||
" \n"
|
||||
" float4 s;\n"
|
||||
" s = (smallAabb.m_max+smallAabb.m_min)*0.5f;\n"
|
||||
" sum[i]=s;\n"
|
||||
" sum2[i]=s*s; \n"
|
||||
"}\n"
|
||||
;
|
||||
static const char* sapCL =
|
||||
"/*\n"
|
||||
"Copyright (c) 2012 Advanced Micro Devices, Inc. \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"
|
||||
"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"
|
||||
"#define NEW_PAIR_MARKER -1\n"
|
||||
"typedef struct \n"
|
||||
"{\n"
|
||||
" union\n"
|
||||
" {\n"
|
||||
" float4 m_min;\n"
|
||||
" float m_minElems[4];\n"
|
||||
" int m_minIndices[4];\n"
|
||||
" };\n"
|
||||
" union\n"
|
||||
" {\n"
|
||||
" float4 m_max;\n"
|
||||
" float m_maxElems[4];\n"
|
||||
" int m_maxIndices[4];\n"
|
||||
" };\n"
|
||||
"} btAabbCL;\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"
|
||||
"{\n"
|
||||
" bool overlap = true;\n"
|
||||
" overlap = (aabb1->m_min.x > aabb2->m_max.x || aabb1->m_max.x < aabb2->m_min.x) ? false : overlap;\n"
|
||||
" overlap = (aabb1->m_min.z > aabb2->m_max.z || aabb1->m_max.z < aabb2->m_min.z) ? false : overlap;\n"
|
||||
" overlap = (aabb1->m_min.y > aabb2->m_max.y || aabb1->m_max.y < aabb2->m_min.y) ? false : overlap;\n"
|
||||
" return overlap;\n"
|
||||
"}\n"
|
||||
"bool TestAabbAgainstAabb2GlobalGlobal(__global const btAabbCL* aabb1, __global const btAabbCL* aabb2);\n"
|
||||
"bool TestAabbAgainstAabb2GlobalGlobal(__global const btAabbCL* aabb1, __global const btAabbCL* aabb2)\n"
|
||||
"{\n"
|
||||
" bool overlap = true;\n"
|
||||
" overlap = (aabb1->m_min.x > aabb2->m_max.x || aabb1->m_max.x < aabb2->m_min.x) ? false : overlap;\n"
|
||||
" overlap = (aabb1->m_min.z > aabb2->m_max.z || aabb1->m_max.z < aabb2->m_min.z) ? false : overlap;\n"
|
||||
" overlap = (aabb1->m_min.y > aabb2->m_max.y || aabb1->m_max.y < aabb2->m_min.y) ? false : overlap;\n"
|
||||
" return overlap;\n"
|
||||
"}\n"
|
||||
"bool TestAabbAgainstAabb2Global(const btAabbCL* aabb1, __global const btAabbCL* aabb2);\n"
|
||||
"bool TestAabbAgainstAabb2Global(const btAabbCL* aabb1, __global const btAabbCL* aabb2)\n"
|
||||
"{\n"
|
||||
" bool overlap = true;\n"
|
||||
" overlap = (aabb1->m_min.x > aabb2->m_max.x || aabb1->m_max.x < aabb2->m_min.x) ? false : overlap;\n"
|
||||
" overlap = (aabb1->m_min.z > aabb2->m_max.z || aabb1->m_max.z < aabb2->m_min.z) ? false : overlap;\n"
|
||||
" overlap = (aabb1->m_min.y > aabb2->m_max.y || aabb1->m_max.y < aabb2->m_min.y) ? false : overlap;\n"
|
||||
" return overlap;\n"
|
||||
"}\n"
|
||||
"__kernel void computePairsKernelTwoArrays( __global const btAabbCL* unsortedAabbs, __global const int* unsortedAabbMapping, __global const int* unsortedAabbMapping2, volatile __global int4* pairsOut,volatile __global int* pairCount, int numUnsortedAabbs, int numUnSortedAabbs2, int axis, int maxPairs)\n"
|
||||
"{\n"
|
||||
" int i = get_global_id(0);\n"
|
||||
" if (i>=numUnsortedAabbs)\n"
|
||||
" return;\n"
|
||||
" int j = get_global_id(1);\n"
|
||||
" if (j>=numUnSortedAabbs2)\n"
|
||||
" return;\n"
|
||||
" __global const btAabbCL* unsortedAabbPtr = &unsortedAabbs[unsortedAabbMapping[i]];\n"
|
||||
" __global const btAabbCL* unsortedAabbPtr2 = &unsortedAabbs[unsortedAabbMapping2[j]];\n"
|
||||
" if (TestAabbAgainstAabb2GlobalGlobal(unsortedAabbPtr,unsortedAabbPtr2))\n"
|
||||
" {\n"
|
||||
" int4 myPair;\n"
|
||||
" \n"
|
||||
" int xIndex = unsortedAabbPtr[0].m_minIndices[3];\n"
|
||||
" int yIndex = unsortedAabbPtr2[0].m_minIndices[3];\n"
|
||||
" if (xIndex>yIndex)\n"
|
||||
" {\n"
|
||||
" int tmp = xIndex;\n"
|
||||
" xIndex=yIndex;\n"
|
||||
" yIndex=tmp;\n"
|
||||
" }\n"
|
||||
" \n"
|
||||
" myPair.x = xIndex;\n"
|
||||
" myPair.y = yIndex;\n"
|
||||
" myPair.z = NEW_PAIR_MARKER;\n"
|
||||
" myPair.w = NEW_PAIR_MARKER;\n"
|
||||
" int curPair = atomic_inc (pairCount);\n"
|
||||
" if (curPair<maxPairs)\n"
|
||||
" {\n"
|
||||
" pairsOut[curPair] = myPair; //flush to main memory\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
"}\n"
|
||||
"__kernel void computePairsKernelBruteForce( __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"
|
||||
" if (i>=numObjects)\n"
|
||||
" return;\n"
|
||||
" for (int j=i+1;j<numObjects;j++)\n"
|
||||
" {\n"
|
||||
" if (TestAabbAgainstAabb2GlobalGlobal(&aabbs[i],&aabbs[j]))\n"
|
||||
" {\n"
|
||||
" int4 myPair;\n"
|
||||
" myPair.x = aabbs[i].m_minIndices[3];\n"
|
||||
" myPair.y = aabbs[j].m_minIndices[3];\n"
|
||||
" myPair.z = NEW_PAIR_MARKER;\n"
|
||||
" myPair.w = NEW_PAIR_MARKER;\n"
|
||||
" int curPair = atomic_inc (pairCount);\n"
|
||||
" if (curPair<maxPairs)\n"
|
||||
" {\n"
|
||||
" pairsOut[curPair] = myPair; //flush to main memory\n"
|
||||
" }\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"
|
||||
" if (i>=numObjects)\n"
|
||||
" return;\n"
|
||||
" for (int j=i+1;j<numObjects;j++)\n"
|
||||
" {\n"
|
||||
" if(aabbs[i].m_maxElems[axis] < (aabbs[j].m_minElems[axis])) \n"
|
||||
" {\n"
|
||||
" break;\n"
|
||||
" }\n"
|
||||
" if (TestAabbAgainstAabb2GlobalGlobal(&aabbs[i],&aabbs[j]))\n"
|
||||
" {\n"
|
||||
" int4 myPair;\n"
|
||||
" myPair.x = aabbs[i].m_minIndices[3];\n"
|
||||
" myPair.y = aabbs[j].m_minIndices[3];\n"
|
||||
" myPair.z = NEW_PAIR_MARKER;\n"
|
||||
" myPair.w = NEW_PAIR_MARKER;\n"
|
||||
" int curPair = atomic_inc (pairCount);\n"
|
||||
" if (curPair<maxPairs)\n"
|
||||
" {\n"
|
||||
" pairsOut[curPair] = myPair; //flush to main memory\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"
|
||||
" __local int numActiveWgItems[1];\n"
|
||||
" __local int breakRequest[1];\n"
|
||||
" if (localId==0)\n"
|
||||
" {\n"
|
||||
" numActiveWgItems[0] = 0;\n"
|
||||
" breakRequest[0] = 0;\n"
|
||||
" }\n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" atomic_inc(numActiveWgItems);\n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" int localBreak = 0;\n"
|
||||
" int j=i+1;\n"
|
||||
" do\n"
|
||||
" {\n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" \n"
|
||||
" if (j<numObjects)\n"
|
||||
" {\n"
|
||||
" if(aabbs[i].m_maxElems[axis] < (aabbs[j].m_minElems[axis])) \n"
|
||||
" {\n"
|
||||
" if (!localBreak)\n"
|
||||
" {\n"
|
||||
" atomic_inc(breakRequest);\n"
|
||||
" localBreak = 1;\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" \n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" \n"
|
||||
" if (j>=numObjects && !localBreak)\n"
|
||||
" {\n"
|
||||
" atomic_inc(breakRequest);\n"
|
||||
" localBreak = 1;\n"
|
||||
" }\n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" \n"
|
||||
" if (!localBreak)\n"
|
||||
" {\n"
|
||||
" if (TestAabbAgainstAabb2GlobalGlobal(&aabbs[i],&aabbs[j]))\n"
|
||||
" {\n"
|
||||
" int4 myPair;\n"
|
||||
" myPair.x = aabbs[i].m_minIndices[3];\n"
|
||||
" myPair.y = aabbs[j].m_minIndices[3];\n"
|
||||
" myPair.z = NEW_PAIR_MARKER;\n"
|
||||
" myPair.w = NEW_PAIR_MARKER;\n"
|
||||
" int curPair = atomic_inc (pairCount);\n"
|
||||
" if (curPair<maxPairs)\n"
|
||||
" {\n"
|
||||
" pairsOut[curPair] = myPair; //flush to main memory\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" j++;\n"
|
||||
" } while (breakRequest[0]<numActiveWgItems[0]);\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"
|
||||
" __local int numActiveWgItems[1];\n"
|
||||
" __local int breakRequest[1];\n"
|
||||
" __local btAabbCL localAabbs[128];// = aabbs[i];\n"
|
||||
" \n"
|
||||
" btAabbCL myAabb;\n"
|
||||
" \n"
|
||||
" myAabb = (i<numObjects)? aabbs[i]:aabbs[0];\n"
|
||||
" float testValue = myAabb.m_maxElems[axis];\n"
|
||||
" \n"
|
||||
" if (localId==0)\n"
|
||||
" {\n"
|
||||
" numActiveWgItems[0] = 0;\n"
|
||||
" breakRequest[0] = 0;\n"
|
||||
" }\n"
|
||||
" int localCount=0;\n"
|
||||
" int block=0;\n"
|
||||
" localAabbs[localId] = (i+block)<numObjects? aabbs[i+block] : aabbs[0];\n"
|
||||
" localAabbs[localId+64] = (i+block+64)<numObjects? aabbs[i+block+64]: aabbs[0];\n"
|
||||
" \n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" atomic_inc(numActiveWgItems);\n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" int localBreak = 0;\n"
|
||||
" \n"
|
||||
" int j=i+1;\n"
|
||||
" do\n"
|
||||
" {\n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" \n"
|
||||
" if (j<numObjects)\n"
|
||||
" {\n"
|
||||
" if(testValue < (localAabbs[localCount+localId+1].m_minElems[axis])) \n"
|
||||
" {\n"
|
||||
" if (!localBreak)\n"
|
||||
" {\n"
|
||||
" atomic_inc(breakRequest);\n"
|
||||
" localBreak = 1;\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" \n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" \n"
|
||||
" if (j>=numObjects && !localBreak)\n"
|
||||
" {\n"
|
||||
" atomic_inc(breakRequest);\n"
|
||||
" localBreak = 1;\n"
|
||||
" }\n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" \n"
|
||||
" if (!localBreak)\n"
|
||||
" {\n"
|
||||
" if (TestAabbAgainstAabb2(&myAabb,&localAabbs[localCount+localId+1]))\n"
|
||||
" {\n"
|
||||
" int4 myPair;\n"
|
||||
" myPair.x = myAabb.m_minIndices[3];\n"
|
||||
" myPair.y = localAabbs[localCount+localId+1].m_minIndices[3];\n"
|
||||
" myPair.z = NEW_PAIR_MARKER;\n"
|
||||
" myPair.w = NEW_PAIR_MARKER;\n"
|
||||
" int curPair = atomic_inc (pairCount);\n"
|
||||
" if (curPair<maxPairs)\n"
|
||||
" {\n"
|
||||
" pairsOut[curPair] = myPair; //flush to main memory\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
" \n"
|
||||
" barrier(CLK_LOCAL_MEM_FENCE);\n"
|
||||
" localCount++;\n"
|
||||
" if (localCount==64)\n"
|
||||
" {\n"
|
||||
" localCount = 0;\n"
|
||||
" block+=64; \n"
|
||||
" localAabbs[localId] = ((i+block)<numObjects) ? aabbs[i+block] : aabbs[0];\n"
|
||||
" localAabbs[localId+64] = ((i+64+block)<numObjects) ? aabbs[i+block+64] : aabbs[0];\n"
|
||||
" }\n"
|
||||
" j++;\n"
|
||||
" \n"
|
||||
" } while (breakRequest[0]<numActiveWgItems[0]);\n"
|
||||
" \n"
|
||||
"}\n"
|
||||
"//http://stereopsis.com/radix.html\n"
|
||||
"unsigned int FloatFlip(float fl);\n"
|
||||
"unsigned int FloatFlip(float fl)\n"
|
||||
"{\n"
|
||||
" unsigned int f = *(unsigned int*)&fl;\n"
|
||||
" unsigned int mask = -(int)(f >> 31) | 0x80000000;\n"
|
||||
" return f ^ mask;\n"
|
||||
"}\n"
|
||||
"float IFloatFlip(unsigned int f);\n"
|
||||
"float IFloatFlip(unsigned int f)\n"
|
||||
"{\n"
|
||||
" unsigned int mask = ((f >> 31) - 1) | 0x80000000;\n"
|
||||
" unsigned int fl = f ^ mask;\n"
|
||||
" return *(float*)&fl;\n"
|
||||
"}\n"
|
||||
"__kernel void copyAabbsKernel( __global const btAabbCL* allAabbs, __global btAabbCL* destAabbs, int numObjects)\n"
|
||||
"{\n"
|
||||
" int i = get_global_id(0);\n"
|
||||
" if (i>=numObjects)\n"
|
||||
" return;\n"
|
||||
" int src = destAabbs[i].m_maxIndices[3];\n"
|
||||
" destAabbs[i] = allAabbs[src];\n"
|
||||
" destAabbs[i].m_maxIndices[3] = src;\n"
|
||||
"}\n"
|
||||
"__kernel void flipFloatKernel( __global const btAabbCL* allAabbs, __global const int* smallAabbMapping, __global int2* sortData, int numObjects, int axis)\n"
|
||||
"{\n"
|
||||
" int i = get_global_id(0);\n"
|
||||
" if (i>=numObjects)\n"
|
||||
" return;\n"
|
||||
" \n"
|
||||
" \n"
|
||||
" sortData[i].x = FloatFlip(allAabbs[smallAabbMapping[i]].m_minElems[axis]);\n"
|
||||
" sortData[i].y = i;\n"
|
||||
" \n"
|
||||
"}\n"
|
||||
"__kernel void scatterKernel( __global const btAabbCL* allAabbs, __global const int* smallAabbMapping, 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] = allAabbs[smallAabbMapping[sortData[i].y]];\n"
|
||||
"}\n"
|
||||
"__kernel void prepareSumVarianceKernel( __global const btAabbCL* allAabbs, __global const int* smallAabbMapping, __global float4* sum, __global float4* sum2,int numAabbs)\n"
|
||||
"{\n"
|
||||
" int i = get_global_id(0);\n"
|
||||
" if (i>=numAabbs)\n"
|
||||
" return;\n"
|
||||
" \n"
|
||||
" btAabbCL smallAabb = allAabbs[smallAabbMapping[i]];\n"
|
||||
" \n"
|
||||
" float4 s;\n"
|
||||
" s = (smallAabb.m_max+smallAabb.m_min)*0.5f;\n"
|
||||
" sum[i]=s;\n"
|
||||
" sum2[i]=s*s; \n"
|
||||
"}\n";
|
||||
|
||||
Reference in New Issue
Block a user