Extras/RigidBodyGpuPipeline/opencl/primitives/AdlPrimitives/Search/BoundSearchKernelsCL.h

   1 static const char* boundSearchKernelsCL= \\r
   2 "/*\n"\r
   3 "               2011 Takahiro Harada\n"\r
   4 "*/\n"\r
   5 "\n"\r
   6 "typedef unsigned int u32;\n"\r
   7 "#define GET_GROUP_IDX get_group_id(0)\n"\r
   8 "#define GET_LOCAL_IDX get_local_id(0)\n"\r
   9 "#define GET_GLOBAL_IDX get_global_id(0)\n"\r
  10 "#define GET_GROUP_SIZE get_local_size(0)\n"\r
  11 "#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"\r
  12 "\n"\r
  13 "typedef struct\n"\r
  14 "{\n"\r
  15 "       u32 m_key; \n"\r
  16 "       u32 m_value;\n"\r
  17 "}SortData;\n"\r
  18 "\n"\r
  19 "\n"\r
  20 "\n"\r
  21 "typedef struct\n"\r
  22 "{\n"\r
  23 "       u32 m_nSrc;\n"\r
  24 "       u32 m_nDst;\n"\r
  25 "       u32 m_padding[2];\n"\r
  26 "} ConstBuffer;\n"\r
  27 "\n"\r
  28 "\n"\r
  29 "\n"\r
  30 "__attribute__((reqd_work_group_size(64,1,1)))\n"\r
  31 "__kernel\n"\r
  32 "void SearchSortDataLowerKernel(__global SortData* src, __global u32 *dst, \n"\r
  33 "                                       ConstBuffer cb)\n"\r
  34 "{\n"\r
  35 "       int gIdx = GET_GLOBAL_IDX;\n"\r
  36 "       u32 nSrc = cb.m_nSrc;\n"\r
  37 "       u32 nDst = cb.m_nDst;\n"\r
  38 "\n"\r
  39 "       if( gIdx < nSrc )\n"\r
  40 "       {\n"\r
  41 "               SortData first; first.m_key = (u32)(-1); first.m_value = (u32)(-1);\n"\r
  42 "               SortData end; end.m_key = nDst; end.m_value = nDst;\n"\r
  43 "\n"\r
  44 "               SortData iData = (gIdx==0)? first: src[gIdx-1];\n"\r
  45 "               SortData jData = (gIdx==nSrc)? end: src[gIdx];\n"\r
  46 "\n"\r
  47 "               if( iData.m_key != jData.m_key )\n"\r
  48 "               {\n"\r
  49 "//                     for(u32 k=iData.m_key+1; k<=min(jData.m_key, nDst-1); k++)\n"\r
  50 "                       u32 k = jData.m_key;\n"\r
  51 "                       {\n"\r
  52 "                               dst[k] = gIdx;\n"\r
  53 "                       }\n"\r
  54 "               }\n"\r
  55 "       }\n"\r
  56 "}\n"\r
  57 "\n"\r
  58 "\n"\r
  59 "__attribute__((reqd_work_group_size(64,1,1)))\n"\r
  60 "__kernel\n"\r
  61 "void SearchSortDataUpperKernel(__global SortData* src, __global u32 *dst, \n"\r
  62 "                                       ConstBuffer cb)\n"\r
  63 "{\n"\r
  64 "       int gIdx = GET_GLOBAL_IDX;\n"\r
  65 "       u32 nSrc = cb.m_nSrc;\n"\r
  66 "       u32 nDst = cb.m_nDst;\n"\r
  67 "\n"\r
  68 "       if( gIdx < nSrc+1 )\n"\r
  69 "       {\n"\r
  70 "               SortData first; first.m_key = 0; first.m_value = 0;\n"\r
  71 "               SortData end; end.m_key = nDst; end.m_value = nDst;\n"\r
  72 "\n"\r
  73 "               SortData iData = (gIdx==0)? first: src[gIdx-1];\n"\r
  74 "               SortData jData = (gIdx==nSrc)? end: src[gIdx];\n"\r
  75 "\n"\r
  76 "               if( iData.m_key != jData.m_key )\n"\r
  77 "               {\n"\r
  78 "//                     for(u32 k=iData.m_key; k<min(jData.m_key, nDst); k++)\n"\r
  79 "                       u32 k = iData.m_key;\n"\r
  80 "                       {\n"\r
  81 "                               dst[k] = gIdx;\n"\r
  82 "                       }\n"\r
  83 "               }\n"\r
  84 "       }\n"\r
  85 "}\n"\r
  86 "\n"\r
  87 "__attribute__((reqd_work_group_size(64,1,1)))\n"\r
  88 "__kernel\n"\r
  89 "void SubtractKernel(__global u32* A, __global u32 *B, __global u32 *C, \n"\r
  90 "                                       ConstBuffer cb)\n"\r
  91 "{\n"\r
  92 "       int gIdx = GET_GLOBAL_IDX;\n"\r
  93 "       u32 nSrc = cb.m_nSrc;\n"\r
  94 "       u32 nDst = cb.m_nDst;\n"\r
  95 "\n"\r
  96 "       if( gIdx < nDst )\n"\r
  97 "       {\n"\r
  98 "               C[gIdx] = A[gIdx] - B[gIdx];\n"\r
  99 "       }\n"\r
 100 "}\n"\r
 101 "\n"\r
 102 ;\r