Chiebot-Mirror
/
opencv
8
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

fixed minMaxLoc kernel (removed compilation errors)

Browse Source
pull/1503/head
Ilya Lavrenov 12 years ago
parent 544c02407e
commit 0faac595a8
2 changed files with 102 additions and 230 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 126
      modules/ocl/src/arithm.cpp
  2. 206
      modules/ocl/src/opencl/arithm_minMaxLoc.cl

126
modules/ocl/src/arithm.cpp
Unescape View File

@ -341,7 +341,7 @@ static void arithmetic_sum_buffer_run(const oclMat &src, cl_mem &dst, int vlen ,
args.push_back( make_pair( sizeof(cl_mem) , (void *)&src.data));
args.push_back( make_pair( sizeof(cl_mem) , (void *)&dst ));
size_t gt[3] = {groupnum * 256, 1, 1}, lt[3] = {256, 1, 1};
if(src.oclchannels() != 3)
if (src.oclchannels() != 3)
openCLExecuteKernel(src.clCxt, &arithm_sum, "arithm_op_sum", gt, lt, args, -1, -1, build_options);
else
openCLExecuteKernel(src.clCxt, &arithm_sum_3, "arithm_op_sum_3", gt, lt, args, -1, -1, build_options);
@ -365,9 +365,9 @@ Scalar arithmetic_sum(const oclMat &src, int type = 0)
memset(p, 0, dbsize * sizeof(T));
openCLReadBuffer(clCxt, dstBuffer, (void *)p, dbsize * sizeof(T));
for(int i = 0; i < dbsize;)
for (int i = 0; i < dbsize;)
{
for(int j = 0; j < src.oclchannels(); j++, i++)
for (int j = 0; j < src.oclchannels(); j++, i++)
s.val[j] += p[i];
}
delete[] p;
@ -378,9 +378,9 @@ Scalar arithmetic_sum(const oclMat &src, int type = 0)
typedef Scalar (*sumFunc)(const oclMat &src, int type);
Scalar cv::ocl::sum(const oclMat &src)
{
if(!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.depth() == CV_64F)
if (!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.depth() == CV_64F)
{
CV_Error(CV_GpuNotSupported, "select device don't support double");
CV_Error(CV_GpuNotSupported, "Selected device doesn't support double");
}
static sumFunc functab[2] =
{
@ -395,9 +395,9 @@ Scalar cv::ocl::sum(const oclMat &src)
Scalar cv::ocl::absSum(const oclMat &src)
{
if(!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.depth() == CV_64F)
if (!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.depth() == CV_64F)
{
CV_Error(CV_GpuNotSupported, "select device don't support double");
CV_Error(CV_GpuNotSupported, "Selected device doesn't support double");
}
static sumFunc functab[2] =
{
@ -412,9 +412,9 @@ Scalar cv::ocl::absSum(const oclMat &src)
Scalar cv::ocl::sqrSum(const oclMat &src)
{
if(!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.depth() == CV_64F)
if (!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.depth() == CV_64F)
{
CV_Error(CV_GpuNotSupported, "select device don't support double");
CV_Error(CV_GpuNotSupported, "Selected device doesn't support double");
}
static sumFunc functab[2] =
{
@ -446,7 +446,7 @@ void cv::ocl::meanStdDev(const oclMat &src, Scalar &mean, Scalar &stddev)
m1 = (Mat)dst1;
m2 = (Mat)dst2;
int i = 0, *p = (int *)m1.data, *q = (int *)m2.data;
for(; i < channels; i++)
for (; i < channels; i++)
{
mean.val[i] = (double)p[i] / (src.cols * src.rows);
stddev.val[i] = std::sqrt(std::max((double) q[i] / (src.cols * src.rows) - mean.val[i] * mean.val[i] , 0.));
@ -476,7 +476,7 @@ static void arithmetic_minMax_run(const oclMat &src, const oclMat &mask, cl_mem
args.push_back( make_pair( sizeof(cl_int) , (void *)&elemnum));
args.push_back( make_pair( sizeof(cl_int) , (void *)&groupnum));
args.push_back( make_pair( sizeof(cl_mem) , (void *)&src.data));
if(!mask.empty())
if (!mask.empty())
{
int mall_cols = mask.step / (vlen * mask.elemSize1());
int mpre_cols = (mask.offset % mask.step) / (vlen * mask.elemSize1());
@ -499,7 +499,7 @@ static void arithmetic_minMax_mask_run(const oclMat &src, const oclMat &mask, cl
vector<pair<size_t , const void *> > args;
size_t gt[3] = {groupnum * 256, 1, 1}, lt[3] = {256, 1, 1};
char build_options[50];
if(src.oclchannels() == 1)
if (src.oclchannels() == 1)
{
int cols = (src.cols - 1) / vlen + 1;
int invalid_cols = src.step / (vlen * src.elemSize1()) - cols;
@ -519,8 +519,6 @@ static void arithmetic_minMax_mask_run(const oclMat &src, const oclMat &mask, cl
args.push_back( make_pair( sizeof(cl_int) , (void *)&moffset ));
args.push_back( make_pair( sizeof(cl_mem) , (void *)&mask.data ));
args.push_back( make_pair( sizeof(cl_mem) , (void *)&dst ));
// printf("elemnum:%d,cols:%d,invalid_cols:%d,offset:%d,minvalid_cols:%d,moffset:%d,repeat_e:%d\r\n",
// elemnum,cols,invalid_cols,offset,minvalid_cols,moffset,repeat_me);
openCLExecuteKernel(src.clCxt, &arithm_minMax_mask, kernelName, gt, lt, args, -1, -1, build_options);
}
}
@ -549,18 +547,18 @@ template <typename T> void arithmetic_minMax(const oclMat &src, double *minVal,
Mat matbuf = Mat(buf);
T *p = matbuf.ptr<T>();
if(minVal != NULL)
if (minVal != NULL)
{
*minVal = std::numeric_limits<double>::max();
for(int i = 0; i < vlen * (int)groupnum; i++)
for (int i = 0; i < vlen * (int)groupnum; i++)
{
*minVal = *minVal < p[i] ? *minVal : p[i];
}
}
if(maxVal != NULL)
if (maxVal != NULL)
{
*maxVal = -std::numeric_limits<double>::max();
for(int i = vlen * (int)groupnum; i < 2 * vlen * (int)groupnum; i++)
for (int i = vlen * (int)groupnum; i < 2 * vlen * (int)groupnum; i++)
{
*maxVal = *maxVal > p[i] ? *maxVal : p[i];
}
@ -577,9 +575,9 @@ void cv::ocl::minMax(const oclMat &src, double *minVal, double *maxVal, const oc
void cv::ocl::minMax_buf(const oclMat &src, double *minVal, double *maxVal, const oclMat &mask, oclMat &buf)
{
CV_Assert(src.oclchannels() == 1);
if(!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.depth() == CV_64F)
if (!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.depth() == CV_64F)
{
CV_Error(CV_GpuNotSupported, "select device don't support double");
CV_Error(CV_GpuNotSupported, "Selected device doesn't support double");
}
static minMaxFunc functab[8] =
{
@ -625,7 +623,7 @@ double cv::ocl::norm(const oclMat &src1, const oclMat &src2, int normType)
m = (gm2);
p = (int *)m.data;
r = -std::numeric_limits<double>::max();
for(i = 0; i < channels; i++)
for (i = 0; i < channels; i++)
{
r = std::max(r, (double)p[i]);
}
@ -635,7 +633,7 @@ double cv::ocl::norm(const oclMat &src1, const oclMat &src2, int normType)
//arithmetic_sum_run(gm1, gm2,"arithm_op_sum");
m = (gm2);
p = (int *)m.data;
for(i = 0; i < channels; i++)
for (i = 0; i < channels; i++)
{
r = r + (double)p[i];
}
@ -645,14 +643,14 @@ double cv::ocl::norm(const oclMat &src1, const oclMat &src2, int normType)
//arithmetic_sum_run(gm1, gm2,"arithm_op_squares_sum");
m = (gm2);
p = (int *)m.data;
for(i = 0; i < channels; i++)
for (i = 0; i < channels; i++)
{
r = r + (double)p[i];
}
r = std::sqrt(r);
break;
}
if(isRelative)
if (isRelative)
r = r / norm(src2, normType);
return r;
}
@ -663,9 +661,9 @@ double cv::ocl::norm(const oclMat &src1, const oclMat &src2, int normType)
static void arithmetic_flip_rows_run(const oclMat &src, oclMat &dst, string kernelName)
{
if(!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.type() == CV_64F)
if (!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.type() == CV_64F)
{
CV_Error(CV_GpuNotSupported, "Selected device don't support double\r\n");
CV_Error(CV_GpuNotSupported, "Selected device doesn't support double\r\n");
return;
}
@ -710,9 +708,9 @@ static void arithmetic_flip_rows_run(const oclMat &src, oclMat &dst, string kern
static void arithmetic_flip_cols_run(const oclMat &src, oclMat &dst, string kernelName, bool isVertical)
{
if(!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.type() == CV_64F)
if (!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.type() == CV_64F)
{
CV_Error(CV_GpuNotSupported, "Selected device don't support double\r\n");
CV_Error(CV_GpuNotSupported, "Selected device doesn't support double\r\n");
return;
}
@ -749,7 +747,7 @@ static void arithmetic_flip_cols_run(const oclMat &src, oclMat &dst, string kern
args.push_back( make_pair( sizeof(cl_int), (void *)&dst.rows ));
args.push_back( make_pair( sizeof(cl_int), (void *)&dst.cols ));
if(isVertical)
if (isVertical)
args.push_back( make_pair( sizeof(cl_int), (void *)&rows ));
else
args.push_back( make_pair( sizeof(cl_int), (void *)&cols ));
@ -764,11 +762,11 @@ static void arithmetic_flip_cols_run(const oclMat &src, oclMat &dst, string kern
void cv::ocl::flip(const oclMat &src, oclMat &dst, int flipCode)
{
dst.create(src.size(), src.type());
if(flipCode == 0)
if (flipCode == 0)
{
arithmetic_flip_rows_run(src, dst, "arithm_flip_rows");
}
else if(flipCode > 0)
else if (flipCode > 0)
arithmetic_flip_cols_run(src, dst, "arithm_flip_cols", false);
else
arithmetic_flip_cols_run(src, dst, "arithm_flip_rc", true);
@ -877,9 +875,9 @@ void cv::ocl::log(const oclMat &src, oclMat &dst)
static void arithmetic_magnitude_phase_run(const oclMat &src1, const oclMat &src2, oclMat &dst, string kernelName)
{
if(!src1.clCxt->supportsFeature(Context::CL_DOUBLE) && src1.type() == CV_64F)
if (!src1.clCxt->supportsFeature(Context::CL_DOUBLE) && src1.type() == CV_64F)
{
CV_Error(CV_GpuNotSupported, "Selected device don't support double\r\n");
CV_Error(CV_GpuNotSupported, "Selected device doesn't support double\r\n");
return;
}
@ -921,9 +919,9 @@ void cv::ocl::magnitude(const oclMat &src1, const oclMat &src2, oclMat &dst)
static void arithmetic_phase_run(const oclMat &src1, const oclMat &src2, oclMat &dst, string kernelName, const char **kernelString)
{
if(!src1.clCxt->supportsFeature(Context::CL_DOUBLE) && src1.type() == CV_64F)
if (!src1.clCxt->supportsFeature(Context::CL_DOUBLE) && src1.type() == CV_64F)
{
CV_Error(CV_GpuNotSupported, "Selected device don't support double\r\n");
CV_Error(CV_GpuNotSupported, "Selected device doesn't support double\r\n");
return;
}
@ -964,7 +962,7 @@ void cv::ocl::phase(const oclMat &x, const oclMat &y, oclMat &Angle , bool angle
CV_Assert(x.type() == y.type() && x.size() == y.size() && (x.depth() == CV_32F || x.depth() == CV_64F));
Angle.create(x.size(), x.type());
string kernelName = angleInDegrees ? "arithm_phase_indegrees" : "arithm_phase_inradians";
if(angleInDegrees)
if (angleInDegrees)
arithmetic_phase_run(x, y, Angle, kernelName, &arithm_phase);
else
arithmetic_phase_run(x, y, Angle, kernelName, &arithm_phase);
@ -977,9 +975,9 @@ void cv::ocl::phase(const oclMat &x, const oclMat &y, oclMat &Angle , bool angle
static void arithmetic_cartToPolar_run(const oclMat &src1, const oclMat &src2, oclMat &dst_mag, oclMat &dst_cart,
string kernelName, bool angleInDegrees)
{
if(!src1.clCxt->supportsFeature(Context::CL_DOUBLE) && src1.type() == CV_64F)
if (!src1.clCxt->supportsFeature(Context::CL_DOUBLE) && src1.type() == CV_64F)
{
CV_Error(CV_GpuNotSupported, "Selected device don't support double\r\n");
CV_Error(CV_GpuNotSupported, "Selected device doesn't support double\r\n");
return;
}
@ -1030,9 +1028,9 @@ void cv::ocl::cartToPolar(const oclMat &x, const oclMat &y, oclMat &mag, oclMat
static void arithmetic_ptc_run(const oclMat &src1, const oclMat &src2, oclMat &dst1, oclMat &dst2, bool angleInDegrees,
string kernelName)
{
if(!src1.clCxt->supportsFeature(Context::CL_DOUBLE) && src1.type() == CV_64F)
if (!src1.clCxt->supportsFeature(Context::CL_DOUBLE) && src1.type() == CV_64F)
{
CV_Error(CV_GpuNotSupported, "Selected device don't support double\r\n");
CV_Error(CV_GpuNotSupported, "Selected device doesn't support double\r\n");
return;
}
@ -1048,7 +1046,7 @@ static void arithmetic_ptc_run(const oclMat &src1, const oclMat &src2, oclMat &d
int tmp = angleInDegrees ? 1 : 0;
vector<pair<size_t , const void *> > args;
if(src1.data)
if (src1.data)
{
args.push_back( make_pair( sizeof(cl_mem), (void *)&src1.data ));
args.push_back( make_pair( sizeof(cl_int), (void *)&src1.step ));
@ -1077,7 +1075,7 @@ void cv::ocl::polarToCart(const oclMat &magnitude, const oclMat &angle, oclMat &
x.create(angle.size(), angle.type());
y.create(angle.size(), angle.type());
if( magnitude.data )
if ( magnitude.data )
{
CV_Assert( magnitude.size() == angle.size() && magnitude.type() == angle.type() );
arithmetic_ptc_run(magnitude, angle, x, y, angleInDegrees, "arithm_polarToCart_mag");
@ -1119,7 +1117,7 @@ static void arithmetic_minMaxLoc_mask_run(const oclMat &src, const oclMat &mask,
vector<pair<size_t , const void *> > args;
size_t gt[3] = {groupnum * 256, 1, 1}, lt[3] = {256, 1, 1};
char build_options[50];
if(src.oclchannels() == 1)
if (src.oclchannels() == 1)
{
int cols = (src.cols - 1) / vlen + 1;
int invalid_cols = src.step / (vlen * src.elemSize1()) - cols;
@ -1143,7 +1141,8 @@ static void arithmetic_minMaxLoc_mask_run(const oclMat &src, const oclMat &mask,
openCLExecuteKernel(src.clCxt, &arithm_minMaxLoc_mask, "arithm_op_minMaxLoc_mask", gt, lt, args, -1, -1, build_options);
}
}
template<typename T>
template <typename T>
void arithmetic_minMaxLoc(const oclMat &src, double *minVal, double *maxVal,
Point *minLoc, Point *maxLoc, const oclMat &mask)
{
@ -1164,12 +1163,12 @@ void arithmetic_minMaxLoc(const oclMat &src, double *minVal, double *maxVal,
T *p = new T[groupnum * vlen * 4];
memset(p, 0, dbsize);
openCLReadBuffer(clCxt, dstBuffer, (void *)p, dbsize);
for(int i = 0; i < vlen * (int)groupnum; i++)
for (int i = 0; i < vlen * (int)groupnum; i++)
{
*minVal = (*minVal < p[i] || p[i + 2 * vlen * groupnum] == -1) ? *minVal : p[i];
minloc = (*minVal < p[i] || p[i + 2 * vlen * groupnum] == -1) ? minloc : cvRound(p[i + 2 * vlen * groupnum]);
}
for(int i = vlen * (int)groupnum; i < 2 * vlen * (int)groupnum; i++)
for (int i = vlen * (int)groupnum; i < 2 * vlen * (int)groupnum; i++)
{
*maxVal = (*maxVal > p[i] || p[i + 2 * vlen * groupnum] == -1) ? *maxVal : p[i];
maxloc = (*maxVal > p[i] || p[i + 2 * vlen * groupnum] == -1) ? maxloc : cvRound(p[i + 2 * vlen * groupnum]);
@ -1178,9 +1177,9 @@ void arithmetic_minMaxLoc(const oclMat &src, double *minVal, double *maxVal,
int pre_rows = src.offset / src.step;
int pre_cols = (src.offset % src.step) / src.elemSize1();
int wholecols = src.step / src.elemSize1();
if( minLoc )
if ( minLoc )
{
if( minloc >= 0 )
if ( minloc >= 0 )
{
minLoc->y = minloc / wholecols - pre_rows;
minLoc->x = minloc % wholecols - pre_cols;
@ -1188,9 +1187,9 @@ void arithmetic_minMaxLoc(const oclMat &src, double *minVal, double *maxVal,
else
minLoc->x = minLoc->y = -1;
}
if( maxLoc )
if ( maxLoc )
{
if( maxloc >= 0 )
if ( maxloc >= 0 )
{
maxLoc->y = maxloc / wholecols - pre_rows;
maxLoc->x = maxloc % wholecols - pre_cols;
@ -1209,9 +1208,9 @@ typedef void (*minMaxLocFunc)(const oclMat &src, double *minVal, double *maxVal,
void cv::ocl::minMaxLoc(const oclMat &src, double *minVal, double *maxVal,
Point *minLoc, Point *maxLoc, const oclMat &mask)
{
if(!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.depth() == CV_64F)
if (!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.depth() == CV_64F)
{
CV_Error(CV_GpuNotSupported, "select device don't support double");
CV_Error(CV_GpuNotSupported, "Selected device doesn't support double");
return;
}
@ -1259,12 +1258,11 @@ static void arithmetic_countNonZero_run(const oclMat &src, cl_mem &dst, int vlen
int cv::ocl::countNonZero(const oclMat &src)
{
size_t groupnum = src.clCxt->computeUnits();
if(!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.depth() == CV_64F)
if (!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.depth() == CV_64F)
{
CV_Error(CV_GpuNotSupported, "select device don't support double");
CV_Error(CV_GpuNotSupported, "selected device doesn't support double");
}
CV_Assert(groupnum != 0);
// groupnum = groupnum * 2;
int vlen = 8 , dbsize = groupnum * vlen;
Context *clCxt = src.clCxt;
string kernelName = "arithm_op_nonzero";
@ -1274,7 +1272,7 @@ int cv::ocl::countNonZero(const oclMat &src)
memset(p, 0, dbsize * sizeof(int));
openCLReadBuffer(clCxt, dstBuffer, (void *)p, dbsize * sizeof(int));
for(int i = 0; i < dbsize; i++)
for (int i = 0; i < dbsize; i++)
nonzero += p[i];
delete[] p;
@ -1677,7 +1675,7 @@ static void arithmetic_pow_run(const oclMat &src1, double p, oclMat &dst, string
args.push_back( make_pair( sizeof(cl_int), (void *)&dst_step1 ));
float pf = static_cast<float>(p);
if(!src1.clCxt->supportsFeature(Context::CL_DOUBLE))
if (!src1.clCxt->supportsFeature(Context::CL_DOUBLE))
args.push_back( make_pair( sizeof(cl_float), (void *)&pf ));
else
args.push_back( make_pair( sizeof(cl_double), (void *)&p ));
@ -1687,7 +1685,7 @@ static void arithmetic_pow_run(const oclMat &src1, double p, oclMat &dst, string
void cv::ocl::pow(const oclMat &x, double p, oclMat &y)
{
if(!x.clCxt->supportsFeature(Context::CL_DOUBLE) && x.type() == CV_64F)
if (!x.clCxt->supportsFeature(Context::CL_DOUBLE) && x.type() == CV_64F)
{
cout << "Selected device do not support double" << endl;
return;
@ -1714,14 +1712,14 @@ void cv::ocl::setIdentity(oclMat& src, double scalar)
size_t global_threads[] = {src.cols, src.rows, 1};
string kernelName = "setIdentityKernel";
if(src.type() == CV_32FC1)
if (src.type() == CV_32FC1)
kernelName += "_F1";
else if(src.type() == CV_32SC1)
else if (src.type() == CV_32SC1)
kernelName += "_I1";
else
{
kernelName += "_D1";
if(!(clCxt->supportsFeature(Context::CL_DOUBLE)))
if (!(clCxt->supportsFeature(Context::CL_DOUBLE)))
{
oclMat temp;
src.convertTo(temp, CV_32FC1);
@ -1738,9 +1736,9 @@ void cv::ocl::setIdentity(oclMat& src, double scalar)
int scalar_i = 0;
float scalar_f = 0.0f;
if(clCxt->supportsFeature(Context::CL_DOUBLE))
if (clCxt->supportsFeature(Context::CL_DOUBLE))
{
if(src.type() == CV_32SC1)
if (src.type() == CV_32SC1)
{
scalar_i = (int)scalar;
args.push_back(make_pair(sizeof(cl_int), (void*)&scalar_i));
@ -1750,7 +1748,7 @@ void cv::ocl::setIdentity(oclMat& src, double scalar)
}
else
{
if(src.type() == CV_32SC1)
if (src.type() == CV_32SC1)
{
scalar_i = (int)scalar;
args.push_back(make_pair(sizeof(cl_int), (void*)&scalar_i));

206
modules/ocl/src/opencl/arithm_minMaxLoc.cl
Unescape View File

@ -142,29 +142,35 @@
#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics:enable
/**************************************Array minMax**************************************/
__kernel void arithm_op_minMaxLoc (int cols,int invalid_cols,int offset,int elemnum,int groupnum,
__kernel void arithm_op_minMaxLoc(int cols, int invalid_cols, int offset, int elemnum, int groupnum,
__global VEC_TYPE *src, __global RES_TYPE *dst)
{
unsigned int lid = get_local_id(0);
unsigned int gid = get_group_id(0);
unsigned int id = get_global_id(0);
unsigned int idx = offset + id + (id / cols) * invalid_cols;
__local VEC_TYPE localmem_max[128],localmem_min[128];
VEC_TYPE minval,maxval,temp;
__local VEC_TYPE_LOC localmem_maxloc[128],localmem_minloc[128];
VEC_TYPE_LOC minloc,maxloc,temploc,negative = -1;
__local VEC_TYPE localmem_max[128], localmem_min[128];
VEC_TYPE minval, maxval, temp;
__local VEC_TYPE_LOC localmem_maxloc[128], localmem_minloc[128];
VEC_TYPE_LOC minloc, maxloc, temploc, negative = -1;
int idx_c;
if(id < elemnum)
if (id < elemnum)
{
temp = src[idx];
idx_c = idx << 2;
temploc = (VEC_TYPE_LOC)(idx_c,idx_c+1,idx_c+2,idx_c+3);
if(id % cols == 0 )
temploc = (VEC_TYPE_LOC)(idx_c, idx_c + 1, idx_c + 2, idx_c + 3);
if (id % cols == 0 )
{
repeat_s(temp);
repeat_s(temploc);
}
if(id % cols == cols - 1)
if (id % cols == cols - 1)
{
repeat_e(temp);
repeat_e(temploc);
@ -181,164 +187,33 @@ __kernel void arithm_op_minMaxLoc (int cols,int invalid_cols,int offset,int elem
minloc = negative;
maxloc = negative;
}
float4 aaa;
for(id=id + (groupnum << 8); id < elemnum;id = id + (groupnum << 8))
int grainSize = (groupnum << 8);
for (id = id + grainSize; id < elemnum; id = id + grainSize)
{
idx = offset + id + (id / cols) * invalid_cols;
temp = src[idx];
idx_c = idx << 2;
temploc = (VEC_TYPE_LOC)(idx_c,idx_c+1,idx_c+2,idx_c+3);
if(id % cols == 0 )
temploc = (VEC_TYPE_LOC)(idx_c, idx_c+1, idx_c+2, idx_c+3);
if (id % cols == 0 )
{
repeat_s(temp);
repeat_s(temploc);
}
if(id % cols == cols - 1)
if (id % cols == cols - 1)
{
repeat_e(temp);
repeat_e(temploc);
}
minval = min(minval,temp);
maxval = max(maxval,temp);
minloc = CONDITION_FUNC(minval == temp, temploc , minloc);
maxloc = CONDITION_FUNC(maxval == temp, temploc , maxloc);
aaa= convert_float4(maxval == temp);
maxloc = convert_int4(aaa) ? temploc : maxloc;
}
if(lid > 127)
{
localmem_min[lid - 128] = minval;
localmem_max[lid - 128] = maxval;
localmem_minloc[lid - 128] = minloc;
localmem_maxloc[lid - 128] = maxloc;
}
barrier(CLK_LOCAL_MEM_FENCE);
if(lid < 128)
{
localmem_min[lid] = min(minval,localmem_min[lid]);
localmem_max[lid] = max(maxval,localmem_max[lid]);
localmem_minloc[lid] = CONDITION_FUNC(localmem_min[lid] == minval, minloc , localmem_minloc[lid]);
localmem_maxloc[lid] = CONDITION_FUNC(localmem_max[lid] == maxval, maxloc , localmem_maxloc[lid]);
}
barrier(CLK_LOCAL_MEM_FENCE);
for(int lsize = 64; lsize > 0; lsize >>= 1)
{
if(lid < lsize)
{
int lid2 = lsize + lid;
localmem_min[lid] = min(localmem_min[lid] , localmem_min[lid2]);
localmem_max[lid] = max(localmem_max[lid] , localmem_max[lid2]);
localmem_minloc[lid] =
CONDITION_FUNC(localmem_min[lid] == localmem_min[lid2], localmem_minloc[lid2] , localmem_minloc[lid]);
localmem_maxloc[lid] =
CONDITION_FUNC(localmem_max[lid] == localmem_max[lid2], localmem_maxloc[lid2] , localmem_maxloc[lid]);
}
barrier(CLK_LOCAL_MEM_FENCE);
}
if( lid == 0)
{
dst[gid] = CONVERT_RES_TYPE(localmem_min[0]);
dst[gid + groupnum] = CONVERT_RES_TYPE(localmem_max[0]);
dst[gid + 2 * groupnum] = CONVERT_RES_TYPE(localmem_minloc[0]);
dst[gid + 3 * groupnum] = CONVERT_RES_TYPE(localmem_maxloc[0]);
}
}
#if defined (REPEAT_S0)
#define repeat_ms(a) a = a;
#endif
#if defined (REPEAT_S1)
#define repeat_ms(a) a.s0 = 0;
#endif
#if defined (REPEAT_S2)
#define repeat_ms(a) a.s0 = 0;a.s1 = 0;
#endif
#if defined (REPEAT_S3)
#define repeat_ms(a) a.s0 = 0;a.s1 = 0;a.s2 = 0;
#endif
#if defined (REPEAT_E0)
#define repeat_me(a) a = a;
#endif
#if defined (REPEAT_E1)
#define repeat_me(a) a.s3 = 0;
#endif
#if defined (REPEAT_E2)
#define repeat_me(a) a.s3 = 0;a.s2 = 0;
#endif
#if defined (REPEAT_E3)
#define repeat_me(a) a.s3 = 0;a.s2 = 0;a.s1 = 0;
#endif
/**************************************Array minMaxLoc mask**************************************/
/*
__kernel void arithm_op_minMaxLoc_mask (int cols,int invalid_cols,int offset,int elemnum,int groupnum,__global VEC_TYPE *src,
int minvalid_cols,int moffset,__global uchar4 *mask,__global RES_TYPE *dst)
{
unsigned int lid = get_local_id(0);
unsigned int gid = get_group_id(0);
unsigned int id = get_global_id(0);
unsigned int idx = offset + id + (id / cols) * invalid_cols;
unsigned int midx = moffset + id + (id / cols) * minvalid_cols;
__local VEC_TYPE localmem_max[128],localmem_min[128];
VEC_TYPE minval,maxval,temp,max_val = MAX_VAL,min_val = MIN_VAL,zero = 0,m_temp;
__local VEC_TYPE_LOC localmem_maxloc[128],localmem_minloc[128];
VEC_TYPE_LOC minloc,maxloc,temploc,negative = -1;
if(id < elemnum)
{
temp = src[idx];
m_temp = CONVERT_TYPE(mask[midx]);
int idx_c = idx << 2;
temploc = (VEC_TYPE_LOC)(idx_c,idx_c+1,idx_c+2,idx_c+3);
if(id % cols == 0 )
{
repeat_ms(m_temp);
repeat_s(temploc);
}
if(id % cols == cols - 1)
{
repeat_me(m_temp);
repeat_e(temploc);
}
minval = m_temp > zero ? temp : max_val;
maxval = m_temp > zero ? temp : min_val;
minloc = CONDITION_FUNC(m_temp > zero, temploc , negative);
maxloc = minloc;
minval = min(minval, temp);
maxval = max(maxval, temp);
minloc = CONDITION_FUNC(minval == temp, temploc, minloc);
maxloc = CONDITION_FUNC(maxval == temp, temploc, maxloc);
}
else
{
minval = MAX_VAL;
maxval = MIN_VAL;
minloc = negative;
maxloc = negative;
}
for(id=id + (groupnum << 8); id < elemnum;id = id + (groupnum << 8))
{
idx = offset + id + (id / cols) * invalid_cols;
midx = moffset + id + (id / cols) * minvalid_cols;
temp = src[idx];
m_temp = CONVERT_TYPE(mask[midx]);
int idx_c = idx << 2;
temploc = (VEC_TYPE_LOC)(idx_c,idx_c+1,idx_c+2,idx_c+3);
if(id % cols == 0 )
{
repeat_ms(m_temp);
repeat_s(temploc);
}
if(id % cols == cols - 1)
{
repeat_me(m_temp);
repeat_e(temploc);
}
minval = min(minval,m_temp > zero ? temp : max_val);
maxval = max(maxval,m_temp > zero ? temp : min_val);
temploc = CONDITION_FUNC(m_temp > zero, temploc , negative);
minloc = CONDITION_FUNC(minval == temp, temploc , minloc);
maxloc = CONDITION_FUNC(maxval == temp, temploc , maxloc);
}
if(lid > 127)
if (lid > 127)
{
localmem_min[lid - 128] = minval;
localmem_max[lid - 128] = maxval;
@ -346,29 +221,30 @@ __kernel void arithm_op_minMaxLoc_mask (int cols,int invalid_cols,int offset,int
localmem_maxloc[lid - 128] = maxloc;
}
barrier(CLK_LOCAL_MEM_FENCE);
if(lid < 128)
if (lid < 128)
{
localmem_min[lid] = min(minval,localmem_min[lid]);
localmem_max[lid] = max(maxval,localmem_max[lid]);
localmem_minloc[lid] = CONDITION_FUNC(localmem_min[lid] == minval, minloc , localmem_minloc[lid]);
localmem_maxloc[lid] = CONDITION_FUNC(localmem_max[lid] == maxval, maxloc , localmem_maxloc[lid]);
localmem_minloc[lid] = CONDITION_FUNC(localmem_min[lid] == minval, minloc, localmem_minloc[lid]);
localmem_maxloc[lid] = CONDITION_FUNC(localmem_max[lid] == maxval, maxloc, localmem_maxloc[lid]);
}
barrier(CLK_LOCAL_MEM_FENCE);
for(int lsize = 64; lsize > 0; lsize >>= 1)
for (int lsize = 64; lsize > 0; lsize >>= 1)
{
if(lid < lsize)
if (lid < lsize)
{
int lid2 = lsize + lid;
localmem_min[lid] = min(localmem_min[lid] , localmem_min[lid2]);
localmem_max[lid] = max(localmem_max[lid] , localmem_max[lid2]);
localmem_minloc[lid] =
CONDITION_FUNC(localmem_min[lid] == localmem_min[lid2], localmem_minloc[lid2] , localmem_minloc[lid]);
localmem_maxloc[lid] =
CONDITION_FUNC(localmem_max[lid] == localmem_max[lid2], localmem_maxloc[lid2] , localmem_maxloc[lid]);
localmem_min[lid] = min(localmem_min[lid], localmem_min[lid2]);
localmem_max[lid] = max(localmem_max[lid], localmem_max[lid2]);
localmem_minloc[lid] = CONDITION_FUNC(localmem_min[lid] == localmem_min[lid2], localmem_minloc[lid2], localmem_minloc[lid]);
localmem_maxloc[lid] = CONDITION_FUNC(localmem_max[lid] == localmem_max[lid2], localmem_maxloc[lid2], localmem_maxloc[lid]);
}
barrier(CLK_LOCAL_MEM_FENCE);
}
if( lid == 0)
if ( lid == 0)
{
dst[gid] = CONVERT_RES_TYPE(localmem_min[0]);
dst[gid + groupnum] = CONVERT_RES_TYPE(localmem_max[0]);
@ -376,5 +252,3 @@ __kernel void arithm_op_minMaxLoc_mask (int cols,int invalid_cols,int offset,int
dst[gid + 3 * groupnum] = CONVERT_RES_TYPE(localmem_maxloc[0]);
}
}
*/

Write Preview
Loading…
Cancel
Save