generic-library/opencv
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

fixed ocl::sum, ocl::sqrSum, ocl::absSum

Browse Source
This commit is contained in:
Ilya Lavrenov 2013-09-27 17:56:30 +04:00
parent b54228fb83
commit b864f48274
4 changed files with 215 additions and 423 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

110
modules/ocl/src/arithm.cpp
View File

@ -66,7 +66,6 @@ namespace cv
extern const char *arithm_nonzero;
extern const char *arithm_sum;
extern const char *arithm_sum_3;
extern const char *arithm_minMax;
extern const char *arithm_minMaxLoc;
extern const char *arithm_minMaxLoc_mask;
@ -317,21 +316,28 @@ void cv::ocl::compare(const oclMat &src1, const oclMat &src2, oclMat &dst , int
////////////////////////////////// sum //////////////////////////////////////
//////////////////////////////////////////////////////////////////////////////
//type = 0 sum,type = 1 absSum,type = 2 sqrSum
static void arithmetic_sum_buffer_run(const oclMat &src, cl_mem &dst, int vlen , int groupnum, int type = 0)
enum { SUM = 0, ABS_SUM, SQR_SUM };
static void arithmetic_sum_buffer_run(const oclMat &src, cl_mem &dst, int groupnum, int type, int ddepth)
{
vector<pair<size_t , const void *> > args;
int all_cols = src.step / (vlen * src.elemSize1());
int pre_cols = (src.offset % src.step) / (vlen * src.elemSize1());
int sec_cols = all_cols - (src.offset % src.step + src.cols * src.elemSize() - 1) / (vlen * src.elemSize1()) - 1;
int ochannels = src.oclchannels();
int all_cols = src.step / src.elemSize();
int pre_cols = (src.offset % src.step) / src.elemSize();
int sec_cols = all_cols - (src.offset % src.step + src.cols * src.elemSize() - 1) / src.elemSize() - 1;
int invalid_cols = pre_cols + sec_cols;
int cols = all_cols - invalid_cols , elemnum = cols * src.rows;;
int offset = src.offset / (vlen * src.elemSize1());
int repeat_s = src.offset / src.elemSize1() - offset * vlen;
int repeat_e = (offset + cols) * vlen - src.offset / src.elemSize1() - src.cols * src.oclchannels();
char build_options[512];
CV_Assert(type == 0 || type == 1 || type == 2);
sprintf(build_options, "-D DEPTH_%d -D REPEAT_S%d -D REPEAT_E%d -D FUNC_TYPE_%d", src.depth(), repeat_s, repeat_e, type);
int offset = src.offset / src.elemSize();
const char * const typeMap[] = { "uchar", "char", "ushort", "short", "int", "float", "double" };
const char * const funcMap[] = { "FUNC_SUM", "FUNC_ABS_SUM", "FUNC_SQR_SUM" };
const char * const channelMap[] = { " ", " ", "2", "4", "4" };
string buildOptions = format("-D srcT=%s%s -D dstT=%s%s -D convertToDstT=convert_%s%s -D %s",
typeMap[src.depth()], channelMap[ochannels],
typeMap[ddepth], channelMap[ochannels],
typeMap[ddepth], channelMap[ochannels],
funcMap[type]);
vector<pair<size_t , const void *> > args;
args.push_back( make_pair( sizeof(cl_int) , (void *)&cols ));
args.push_back( make_pair( sizeof(cl_int) , (void *)&invalid_cols ));
args.push_back( make_pair( sizeof(cl_int) , (void *)&offset));
@ -339,55 +345,63 @@ static void arithmetic_sum_buffer_run(const oclMat &src, cl_mem &dst, int vlen ,
args.push_back( make_pair( sizeof(cl_int) , (void *)&groupnum));
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)
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);
size_t globalThreads[3] = { groupnum * 256, 1, 1 };
size_t localThreads[3] = { 256, 1, 1 };
openCLExecuteKernel(src.clCxt, &arithm_sum, "arithm_op_sum", globalThreads, localThreads,
args, -1, -1, buildOptions.c_str());
}
template <typename T>
Scalar arithmetic_sum(const oclMat &src, int type = 0)
Scalar arithmetic_sum(const oclMat &src, int type, int ddepth)
{
CV_Assert(src.step % src.elemSize() == 0);
size_t groupnum = src.clCxt->computeUnits();
CV_Assert(groupnum != 0);
int vlen = src.oclchannels() == 3 ? 12 : 8, dbsize = groupnum * vlen;
int dbsize = groupnum * src.oclchannels();
Context *clCxt = src.clCxt;
AutoBuffer<T> _buf(dbsize);
T *p = (T*)_buf;
cl_mem dstBuffer = openCLCreateBuffer(clCxt, CL_MEM_WRITE_ONLY, dbsize * sizeof(T));
Scalar s = Scalar::all(0.0);
arithmetic_sum_buffer_run(src, dstBuffer, vlen, groupnum, type);
memset(p, 0, dbsize * sizeof(T));
openCLReadBuffer(clCxt, dstBuffer, (void *)p, dbsize * sizeof(T));
for (int i = 0; i < dbsize;)
{
for (int j = 0; j < src.oclchannels(); j++, i++)
s.val[j] += p[i];
}
cl_mem dstBuffer = openCLCreateBuffer(clCxt, CL_MEM_WRITE_ONLY, dbsize * sizeof(T));
arithmetic_sum_buffer_run(src, dstBuffer, groupnum, type, ddepth);
openCLReadBuffer(clCxt, dstBuffer, (void *)p, dbsize * sizeof(T));
openCLFree(dstBuffer);
Scalar s = Scalar::all(0.0);
for (int i = 0; i < dbsize;)
for (int j = 0; j < src.oclchannels(); j++, i++)
s.val[j] += p[i];
return s;
}
typedef Scalar (*sumFunc)(const oclMat &src, int type);
typedef Scalar (*sumFunc)(const oclMat &src, int type, int ddepth);
Scalar cv::ocl::sum(const oclMat &src)
{
if (!src.clCxt->supportsFeature(Context::CL_DOUBLE) && src.depth() == CV_64F)
{
CV_Error(CV_GpuNotSupported, "Selected device doesn't support double");
}
static sumFunc functab[2] =
static sumFunc functab[3] =
{
arithmetic_sum<int>,
arithmetic_sum<float>,
arithmetic_sum<double>
};
sumFunc func;
func = functab[(int)src.clCxt->supportsFeature(Context::CL_DOUBLE)];
return func(src, 0);
bool hasDouble = src.clCxt->supportsFeature(Context::CL_DOUBLE);
int ddepth = std::max(src.depth(), CV_32S);
if (!hasDouble && ddepth == CV_64F)
ddepth = CV_32F;
sumFunc func = functab[ddepth - CV_32S];
return func(src, SUM, ddepth);
}
Scalar cv::ocl::absSum(const oclMat &src)
@ -396,15 +410,20 @@ Scalar cv::ocl::absSum(const oclMat &src)
{
CV_Error(CV_GpuNotSupported, "Selected device doesn't support double");
}
static sumFunc functab[2] =
static sumFunc functab[3] =
{
arithmetic_sum<int>,
arithmetic_sum<float>,
arithmetic_sum<double>
};
sumFunc func;
func = functab[(int)src.clCxt->supportsFeature(Context::CL_DOUBLE)];
return func(src, 1);
bool hasDouble = src.clCxt->supportsFeature(Context::CL_DOUBLE);
int ddepth = std::max(src.depth(), CV_32S);
if (!hasDouble && ddepth == CV_64F)
ddepth = CV_32F;
sumFunc func = functab[ddepth - CV_32S];
return func(src, ABS_SUM, ddepth);
}
Scalar cv::ocl::sqrSum(const oclMat &src)
@ -413,15 +432,20 @@ Scalar cv::ocl::sqrSum(const oclMat &src)
{
CV_Error(CV_GpuNotSupported, "Selected device doesn't support double");
}
static sumFunc functab[2] =
static sumFunc functab[3] =
{
arithmetic_sum<int>,
arithmetic_sum<float>,
arithmetic_sum<double>
};
sumFunc func;
func = functab[(int)src.clCxt->supportsFeature(Context::CL_DOUBLE)];
return func(src, 2);
bool hasDouble = src.clCxt->supportsFeature(Context::CL_DOUBLE);
int ddepth = std::max(src.depth(), CV_32S);
if (!hasDouble && ddepth == CV_64F)
ddepth = CV_32F;
sumFunc func = functab[ddepth - CV_32S];
return func(src, SQR_SUM, ddepth);
}
//////////////////////////////////////////////////////////////////////////////

161
modules/ocl/src/opencl/arithm_sum.cl
View File

@ -43,163 +43,62 @@
//
//M*/
/**************************************PUBLICFUNC*************************************/
#if defined (DOUBLE_SUPPORT)
#ifdef cl_khr_fp64
#pragma OPENCL EXTENSION cl_khr_fp64:enable
#define RES_TYPE double8
#define CONVERT_RES_TYPE convert_double8
#else
#define RES_TYPE float8
#define CONVERT_RES_TYPE convert_float8
#elif defined (cl_amd_fp64)
#pragma OPENCL EXTENSION cl_amd_fp64:enable
#endif
#endif
#if defined (DEPTH_0)
#define VEC_TYPE uchar8
#if defined (FUNC_SUM)
#define FUNC(a, b) b += a;
#endif
#if defined (DEPTH_1)
#define VEC_TYPE char8
#if defined (FUNC_ABS_SUM)
#define FUNC(a, b) b += a >= 0 ? a : -a;
#endif
#if defined (DEPTH_2)
#define VEC_TYPE ushort8
#if defined (FUNC_SQR_SUM)
#define FUNC(a, b) b += a * a;
#endif
#if defined (DEPTH_3)
#define VEC_TYPE short8
#endif
#if defined (DEPTH_4)
#define VEC_TYPE int8
#endif
#if defined (DEPTH_5)
#define VEC_TYPE float8
#endif
#if defined (DEPTH_6)
#define VEC_TYPE double8
#endif
#if defined (FUNC_TYPE_0)
#define FUNC(a,b) b += a;
#endif
#if defined (FUNC_TYPE_1)
#define FUNC(a,b) b = b + (a >= 0 ? a : -a);
#endif
#if defined (FUNC_TYPE_2)
#define FUNC(a,b) b = b + a * a;
#endif
#if defined (REPEAT_S0)
#define repeat_s(a) a = a;
#endif
#if defined (REPEAT_S1)
#define repeat_s(a) a.s0 = 0;
#endif
#if defined (REPEAT_S2)
#define repeat_s(a) a.s0 = 0;a.s1 = 0;
#endif
#if defined (REPEAT_S3)
#define repeat_s(a) a.s0 = 0;a.s1 = 0;a.s2 = 0;
#endif
#if defined (REPEAT_S4)
#define repeat_s(a) a.s0 = 0;a.s1 = 0;a.s2 = 0;a.s3 = 0;
#endif
#if defined (REPEAT_S5)
#define repeat_s(a) a.s0 = 0;a.s1 = 0;a.s2 = 0;a.s3 = 0;a.s4 = 0;
#endif
#if defined (REPEAT_S6)
#define repeat_s(a) a.s0 = 0;a.s1 = 0;a.s2 = 0;a.s3 = 0;a.s4 = 0;a.s5 = 0;
#endif
#if defined (REPEAT_S7)
#define repeat_s(a) a.s0 = 0;a.s1 = 0;a.s2 = 0;a.s3 = 0;a.s4 = 0;a.s5 = 0;a.s6 = 0;
#endif
#if defined (REPEAT_E0)
#define repeat_e(a) a = a;
#endif
#if defined (REPEAT_E1)
#define repeat_e(a) a.s7 = 0;
#endif
#if defined (REPEAT_E2)
#define repeat_e(a) a.s7 = 0;a.s6 = 0;
#endif
#if defined (REPEAT_E3)
#define repeat_e(a) a.s7 = 0;a.s6 = 0;a.s5 = 0;
#endif
#if defined (REPEAT_E4)
#define repeat_e(a) a.s7 = 0;a.s6 = 0;a.s5 = 0;a.s4 = 0;
#endif
#if defined (REPEAT_E5)
#define repeat_e(a) a.s7 = 0;a.s6 = 0;a.s5 = 0;a.s4 = 0;a.s3 = 0;
#endif
#if defined (REPEAT_E6)
#define repeat_e(a) a.s7 = 0;a.s6 = 0;a.s5 = 0;a.s4 = 0;a.s3 = 0;a.s2 = 0;
#endif
#if defined (REPEAT_E7)
#define repeat_e(a) a.s7 = 0;a.s6 = 0;a.s5 = 0;a.s4 = 0;a.s3 = 0;a.s2 = 0;a.s1 = 0;
#endif
#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics:enable
#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics:enable
/**************************************Array buffer SUM**************************************/
__kernel void arithm_op_sum (int cols,int invalid_cols,int offset,int elemnum,int groupnum,
__global VEC_TYPE *src, __global RES_TYPE *dst)
__kernel void arithm_op_sum(int cols,int invalid_cols,int offset,int elemnum,int groupnum,
__global srcT *src, __global dstT *dst)
{
unsigned int lid = get_local_id(0);
unsigned int gid = get_group_id(0);
unsigned int id = get_global_id(0);
unsigned int id = get_global_id(0);
unsigned int idx = offset + id + (id / cols) * invalid_cols;
__local RES_TYPE localmem_sum[128];
RES_TYPE sum = 0,temp;
if(id < elemnum)
{
temp = CONVERT_RES_TYPE(src[idx]);
if(id % cols == 0 )
{
repeat_s(temp);
}
if(id % cols == cols - 1)
{
repeat_e(temp);
}
FUNC(temp,sum);
}
else
{
sum = 0;
}
for(id=id + (groupnum << 8); id < elemnum;id = id + (groupnum << 8))
__local dstT localmem_sum[128];
dstT sum = (dstT)(0), temp;
for (int grainSize = groupnum << 8; id < elemnum; id += grainSize)
{
idx = offset + id + (id / cols) * invalid_cols;
temp = CONVERT_RES_TYPE(src[idx]);
if(id % cols == 0 )
{
repeat_s(temp);
}
if(id % cols == cols - 1)
{
repeat_e(temp);
}
FUNC(temp,sum);
temp = convertToDstT(src[idx]);
FUNC(temp, sum);
}
if(lid > 127)
{
if (lid > 127)
localmem_sum[lid - 128] = sum;
}
barrier(CLK_LOCAL_MEM_FENCE);
if(lid < 128)
{
if (lid < 128)
localmem_sum[lid] = sum + localmem_sum[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_sum[lid] = localmem_sum[lid] + localmem_sum[lid2];
}
barrier(CLK_LOCAL_MEM_FENCE);
}
if( lid == 0)
{
if (lid == 0)
dst[gid] = localmem_sum[0];
}
}

247
modules/ocl/src/opencl/arithm_sum_3.cl
View File

@ -1,247 +0,0 @@
/*M///////////////////////////////////////////////////////////////////////////////////////
//
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
//
// By downloading, copying, installing or using the software you agree to this license.
// If you do not agree to this license, do not download, install,
// copy or use the software.
//
//
// License Agreement
// For Open Source Computer Vision Library
//
// Copyright (C) 2010-2012, Institute Of Software Chinese Academy Of Science, all rights reserved.
// Copyright (C) 2010-2012, Advanced Micro Devices, Inc., all rights reserved.
// Third party copyrights are property of their respective owners.
//
// @Authors
// Shengen Yan,yanshengen@gmail.com
//
// Redistribution and use in source and binary forms, with or without modification,
// are permitted provided that the following conditions are met:
//
// * Redistribution's of source code must retain the above copyright notice,
// this list of conditions and the following disclaimer.
//
// * Redistribution's in binary form must reproduce the above copyright notice,
// this list of conditions and the following disclaimer in the documentation
// and/or other oclMaterials provided with the distribution.
//
// * The name of the copyright holders may not be used to endorse or promote products
// derived from this software without specific prior written permission.
//
// This software is provided by the copyright holders and contributors as is and
// any express or implied warranties, including, but not limited to, the implied
// warranties of merchantability and fitness for a particular purpose are disclaimed.
// In no event shall the Intel Corporation or contributors be liable for any direct,
// indirect, incidental, special, exemplary, or consequential damages
// (including, but not limited to, procurement of substitute goods or services;
// loss of use, data, or profits; or business interruption) however caused
// and on any theory of liability, whether in contract, strict liability,
// or tort (including negligence or otherwise) arising in any way out of
// the use of this software, even if advised of the possibility of such damage.
//
//M*/
/**************************************PUBLICFUNC*************************************/
#if defined (DOUBLE_SUPPORT)
#pragma OPENCL EXTENSION cl_khr_fp64:enable
#define RES_TYPE double4
#define CONVERT_RES_TYPE convert_double4
#else
#define RES_TYPE float4
#define CONVERT_RES_TYPE convert_float4
#endif
#if defined (DEPTH_0)
#define VEC_TYPE uchar4
#endif
#if defined (DEPTH_1)
#define VEC_TYPE char4
#endif
#if defined (DEPTH_2)
#define VEC_TYPE ushort4
#endif
#if defined (DEPTH_3)
#define VEC_TYPE short4
#endif
#if defined (DEPTH_4)
#define VEC_TYPE int4
#endif
#if defined (DEPTH_5)
#define VEC_TYPE float4
#endif
#if defined (DEPTH_6)
#define VEC_TYPE double4
#endif
#if defined (FUNC_TYPE_0)
#define FUNC(a,b) b += a;
#endif
#if defined (FUNC_TYPE_1)
#define FUNC(a,b) b = b + (a >= 0 ? a : -a);
#endif
#if defined (FUNC_TYPE_2)
#define FUNC(a,b) b = b + a * a;
#endif
#if defined (REPEAT_S0)
#define repeat_s(a,b,c) a=a; b =b; c=c;
#endif
#if defined (REPEAT_S1)
#define repeat_s(a,b,c) a.s0=0; b=b; c=c;
#endif
#if defined (REPEAT_S2)
#define repeat_s(a,b,c) a.s0=0; a.s1=0; b=b; c=c;
#endif
#if defined (REPEAT_S3)
#define repeat_s(a,b,c) a.s0=0; a.s1=0; a.s2=0; b=b; c=c;
#endif
#if defined (REPEAT_S4)
#define repeat_s(a,b,c) a=0;b=b; c=c;
#endif
#if defined (REPEAT_S5)
#define repeat_s(a,b,c) a=0; b.s0=0;c=c;
#endif
#if defined (REPEAT_S6)
#define repeat_s(a,b,c) a=0; b.s0=0; b.s1=0; c=c;
#endif
#if defined (REPEAT_S7)
#define repeat_s(a,b,c) a=0; b.s0=0; b.s1=0; b.s2=0; c=c;
#endif
#if defined (REPEAT_S8)
#define repeat_s(a,b,c) a=0; b=0; c=c;
#endif
#if defined (REPEAT_S9)
#define repeat_s(a,b,c) a=0; b=0; c.s0=0;
#endif
#if defined (REPEAT_S10)
#define repeat_s(a,b,c) a=0; b=0; c.s0=0; c.s1=0;
#endif
#if defined (REPEAT_S11)
#define repeat_s(a,b,c) a=0; b=0; c.s0=0; c.s1=0; c.s2=0;
#endif
#if defined (REPEAT_E0)
#define repeat_e(a,b,c) a=a; b =b; c=c;
#endif
#if defined (REPEAT_E1)
#define repeat_e(a,b,c) a=a; b=b; c.s3=0;
#endif
#if defined (REPEAT_E2)
#define repeat_e(a,b,c) a=a; b=b; c.s3=0; c.s2=0;
#endif
#if defined (REPEAT_E3)
#define repeat_e(a,b,c) a=a; b=b; c.s3=0; c.s2=0; c.s1=0;
#endif
#if defined (REPEAT_E4)
#define repeat_e(a,b,c) a=a; b=b; c=0;
#endif
#if defined (REPEAT_E5)
#define repeat_e(a,b,c) a=a; b.s3=0; c=0;
#endif
#if defined (REPEAT_E6)
#define repeat_e(a,b,c) a=a; b.s3=0; b.s2=0; c=0;
#endif
#if defined (REPEAT_E7)
#define repeat_e(a,b,c) a=a; b.s3=0; b.s2=0; b.s1=0; c=0;
#endif
#if defined (REPEAT_E8)
#define repeat_e(a,b,c) a=a; b=0; c=0;
#endif
#if defined (REPEAT_E9)
#define repeat_e(a,b,c) a.s3=0; b=0; c=0;
#endif
#if defined (REPEAT_E10)
#define repeat_e(a,b,c) a.s3=0; a.s2=0; b=0; c=0;
#endif
#if defined (REPEAT_E11)
#define repeat_e(a,b,c) a.s3=0; a.s2=0; a.s1=0; b=0; c=0;
#endif
__kernel void arithm_op_sum_3 (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;
idx = idx * 3;
__local RES_TYPE localmem_sum1[128];
__local RES_TYPE localmem_sum2[128];
__local RES_TYPE localmem_sum3[128];
RES_TYPE sum1 = 0,sum2 = 0,sum3 = 0,temp1,temp2,temp3;
if(id < elemnum)
{
temp1 = CONVERT_RES_TYPE(src[idx]);
temp2 = CONVERT_RES_TYPE(src[idx+1]);
temp3 = CONVERT_RES_TYPE(src[idx+2]);
if(id % cols == 0 )
{
repeat_s(temp1,temp2,temp3);
}
if(id % cols == cols - 1)
{
repeat_e(temp1,temp2,temp3);
}
FUNC(temp1,sum1);
FUNC(temp2,sum2);
FUNC(temp3,sum3);
}
else
{
sum1 = 0;
sum2 = 0;
sum3 = 0;
}
for(id=id + (groupnum << 8); id < elemnum;id = id + (groupnum << 8))
{
idx = offset + id + (id / cols) * invalid_cols;
idx = idx * 3;
temp1 = CONVERT_RES_TYPE(src[idx]);
temp2 = CONVERT_RES_TYPE(src[idx+1]);
temp3 = CONVERT_RES_TYPE(src[idx+2]);
if(id % cols == 0 )
{
repeat_s(temp1,temp2,temp3);
}
if(id % cols == cols - 1)
{
repeat_e(temp1,temp2,temp3);
}
FUNC(temp1,sum1);
FUNC(temp2,sum2);
FUNC(temp3,sum3);
}
if(lid > 127)
{
localmem_sum1[lid - 128] = sum1;
localmem_sum2[lid - 128] = sum2;
localmem_sum3[lid - 128] = sum3;
}
barrier(CLK_LOCAL_MEM_FENCE);
if(lid < 128)
{
localmem_sum1[lid] = sum1 + localmem_sum1[lid];
localmem_sum2[lid] = sum2 + localmem_sum2[lid];
localmem_sum3[lid] = sum3 + localmem_sum3[lid];
}
barrier(CLK_LOCAL_MEM_FENCE);
for(int lsize = 64; lsize > 0; lsize >>= 1)
{
if(lid < lsize)
{
int lid2 = lsize + lid;
localmem_sum1[lid] = localmem_sum1[lid] + localmem_sum1[lid2];
localmem_sum2[lid] = localmem_sum2[lid] + localmem_sum2[lid2];
localmem_sum3[lid] = localmem_sum3[lid] + localmem_sum3[lid2];
}
barrier(CLK_LOCAL_MEM_FENCE);
}
if( lid == 0)
{
dst[gid*3] = localmem_sum1[0];
dst[gid*3+1] = localmem_sum2[0];
dst[gid*3+2] = localmem_sum3[0];
}
}

120
modules/ocl/test/test_arithm.cpp
View File

@ -1022,7 +1022,7 @@ TEST_P(MinMaxLoc, MASK)
typedef ArithmTestBase Sum;
TEST_P(Sum, DISABLED_MAT)
TEST_P(Sum, MAT)
{
for (int j = 0; j < LOOP_TIMES; j++)
{
@ -1031,7 +1031,121 @@ TEST_P(Sum, DISABLED_MAT)
Scalar cpures = cv::sum(src1_roi);
Scalar gpures = cv::ocl::sum(gsrc1);
//check results
// check results
EXPECT_NEAR(cpures[0], gpures[0], 0.1);
EXPECT_NEAR(cpures[1], gpures[1], 0.1);
EXPECT_NEAR(cpures[2], gpures[2], 0.1);
EXPECT_NEAR(cpures[3], gpures[3], 0.1);
}
}
typedef ArithmTestBase SqrSum;
template <typename T, typename WT>
static Scalar sqrSum(const Mat & src)
{
Scalar sum = Scalar::all(0);
int cn = src.channels();
WT data[4] = { 0, 0, 0, 0 };
int cols = src.cols * cn;
for (int y = 0; y < src.rows; ++y)
{
const T * const sdata = src.ptr<T>(y);
for (int x = 0; x < cols; )
for (int i = 0; i < cn; ++i, ++x)
{
WT t = static_cast<WT>(sdata[x]);
data[i] += t * t;
}
}
for (int i = 0; i < cn; ++i)
sum[i] = static_cast<double>(data[i]);
return sum;
}
typedef Scalar (*sumFunc)(const Mat &);
TEST_P(SqrSum, MAT)
{
for (int j = 0; j < LOOP_TIMES; j++)
{
random_roi();
static sumFunc funcs[] = { sqrSum<uchar, int>,
sqrSum<char, int>,
sqrSum<ushort, int>,
sqrSum<short, int>,
sqrSum<int, int>,
sqrSum<float, double>,
sqrSum<double, double>,
0 };
sumFunc func = funcs[src1_roi.depth()];
CV_Assert(func != 0);
Scalar cpures = func(src1_roi);
Scalar gpures = cv::ocl::sqrSum(gsrc1);
// check results
EXPECT_NEAR(cpures[0], gpures[0], 1.0);
EXPECT_NEAR(cpures[1], gpures[1], 1.0);
EXPECT_NEAR(cpures[2], gpures[2], 1.0);
EXPECT_NEAR(cpures[3], gpures[3], 1.0);
}
}
typedef ArithmTestBase AbsSum;
template <typename T, typename WT>
static Scalar absSum(const Mat & src)
{
Scalar sum = Scalar::all(0);
int cn = src.channels();
WT data[4] = { 0, 0, 0, 0 };
int cols = src.cols * cn;
for (int y = 0; y < src.rows; ++y)
{
const T * const sdata = src.ptr<T>(y);
for (int x = 0; x < cols; )
for (int i = 0; i < cn; ++i, ++x)
{
WT t = static_cast<WT>(sdata[x]);
data[i] += t >= 0 ? t : -t;
}
}
for (int i = 0; i < cn; ++i)
sum[i] = static_cast<double>(data[i]);
return sum;
}
TEST_P(AbsSum, MAT)
{
for (int j = 0; j < LOOP_TIMES; j++)
{
random_roi();
static sumFunc funcs[] = { absSum<uchar, int>,
absSum<char, int>,
absSum<ushort, int>,
absSum<short, int>,
absSum<int, int>,
absSum<float, double>,
absSum<double, double>,
0 };
sumFunc func = funcs[src1_roi.depth()];
CV_Assert(func != 0);
Scalar cpures = func(src1_roi);
Scalar gpures = cv::ocl::absSum(gsrc1);
// check results
EXPECT_NEAR(cpures[0], gpures[0], 0.1);
EXPECT_NEAR(cpures[1], gpures[1], 0.1);
EXPECT_NEAR(cpures[2], gpures[2], 0.1);
@ -1319,6 +1433,8 @@ INSTANTIATE_TEST_CASE_P(Arithm, Flip, Combine(testing::Range(CV_8U, CV_USRTYPE1)
INSTANTIATE_TEST_CASE_P(Arithm, MinMax, Combine(testing::Range(CV_8U, CV_USRTYPE1), Values(1), Bool()));
INSTANTIATE_TEST_CASE_P(Arithm, MinMaxLoc, Combine(testing::Range(CV_8U, CV_USRTYPE1), Values(1), Bool())); // +
INSTANTIATE_TEST_CASE_P(Arithm, Sum, Combine(testing::Range(CV_8U, CV_USRTYPE1), testing::Range(1, 5), Bool()));
INSTANTIATE_TEST_CASE_P(Arithm, SqrSum, Combine(testing::Range(CV_8U, CV_USRTYPE1), testing::Range(1, 5), Bool()));
INSTANTIATE_TEST_CASE_P(Arithm, AbsSum, Combine(testing::Range(CV_8U, CV_USRTYPE1), testing::Range(1, 5), Bool()));
INSTANTIATE_TEST_CASE_P(Arithm, CountNonZero, Combine(testing::Range(CV_8U, CV_USRTYPE1), Values(1), Bool())); // +
INSTANTIATE_TEST_CASE_P(Arithm, Phase, Combine(Values(CV_32F, CV_64F), testing::Range(1, 5), Bool())); // +
INSTANTIATE_TEST_CASE_P(Arithm, Bitwise_and, Combine(testing::Range(CV_8U, CV_USRTYPE1), testing::Range(1, 5), Bool())); // +