Peter M. Groen / openbr

Commit aaabb3cbe5738cd9f9418c4b6b2ddb78c5937559

Authored by DepthDeluxe
1 parent bdfd554c

fixed copyto and copyfrom functions to be thread safe

Inline Side-by-side
Showing 4 changed files with 61 additions and 38 deletions
openbr/plugins/cuda/copyfrom.cpp
  View file @aaabb3c
1 1 #include <iostream>
2 2  
3 3 #include <opencv2/opencv.hpp>
4   -#include <opencv2/gpu/gpu.hpp>
5 4  
6 5 #include <openbr/plugins/openbr_internal.h>
7 6  
8 7 using namespace std;
9 8  
10 9 using namespace cv;
11   -using namespace cv::gpu;
12 10  
  11 +// extern CUDA declaration
  12 +namespace br { namespace cuda { namespace cudacopyfrom {
  13 + //template <typename T> void wrapper(void* src, T* out, int rows, int cols) {
  14 + void wrapper(void* src, unsigned char* out, const int rows, const int cols);
  15 +}}}
13 16  
14 17 namespace br
15 18 {
... ... @@ -20,20 +23,16 @@ namespace br
20 23 private:
21 24 void project(const Template &src, Template &dst) const
22 25 {
23   - // reassemble the integer and then build pointer to it
24   - uint64_t gpuMatInt = (((uint64_t)src.m().at<int>(1,0)) << (uint64_t)32) + ((uint64_t)src.m().at<int>(0,0));
25   - GpuMat* gpuMat = (GpuMat*)gpuMatInt;
  26 + // pull the data back out of the Mat
  27 + void* const* dataPtr = src.m().ptr<void*>();
  28 + void* cudaMemPtr = dataPtr[0];
  29 + int rows = *((int*)dataPtr[1]);
  30 + int cols = *((int*)dataPtr[2]);
  31 + int type = *((int*)dataPtr[3]);
26 32  
27   - printf("gpuMatInt: %li\n", gpuMatInt);
28   - printf("m.at(0,0): %i\nm.at(1,0): %i\n", src.m().at<int>(0,0), src.m().at<int>(1,0));
  33 + dst = Mat(rows, cols, type);
29 34  
30   - // download the data back into the destination
31   - Size size = gpuMat->size();
32   - Mat out = Mat(size.height, size.width, gpuMat->depth());
33   -
34   - gpuMat->download(out);
35   -
36   - dst = out;
  35 + br::cuda::cudacopyfrom::wrapper(cudaMemPtr, dst.m().ptr<unsigned char>(), rows, cols);
37 36 }
38 37 };
39 38  
... ...
openbr/plugins/cuda/copyfrom.cu 0 → 100644
  View file @aaabb3c
  1 +namespace br { namespace cuda { namespace cudacopyfrom {
  2 + //template <typename T> void wrapper(void* src, T* out, int rows, int cols) {
  3 + void wrapper(void* src, unsigned char* out, const int rows, const int cols) {
  4 + cudaMemcpy(out, src, rows*cols*sizeof(unsigned char), cudaMemcpyDeviceToHost);
  5 + cudaFree(src);
  6 + }
  7 +}}}
... ...
openbr/plugins/cuda/copyto.cpp
  View file @aaabb3c
1 1 #include <iostream>
2 2  
3 3 #include <opencv2/opencv.hpp>
4   -#include <opencv2/gpu/gpu.hpp>
5 4  
6 5 #include <openbr/plugins/openbr_internal.h>
7 6  
8 7 using namespace std;
9 8  
10 9 using namespace cv;
11   -using namespace cv::gpu;
  10 +
  11 +extern string type2str(int type);
  12 +
  13 +namespace br { namespace cuda { namespace cudacopyto {
  14 + //template<typename T>
  15 + //void wrapper(const T* in, void** out, const int rows, const int cols);
  16 + void wrapper(const unsigned char* in, void** out, const int rows, const int cols);
  17 +}}}
12 18  
13 19 namespace br
14 20 {
... ... @@ -19,32 +25,35 @@ namespace br
19 25 private:
20 26 void project(const Template &src, Template &dst) const
21 27 {
22   - // get the mat to send to the GPU
23   - GpuMat* gpuMat = new GpuMat;
24   -
25   - try
26   - {
27   - // copy the contents to the GPU
28   - gpuMat->upload(src.m());
  28 + const Mat& srcMat = src.m();
  29 + const int rows = srcMat.rows;
  30 + const int cols = srcMat.cols;
  31 +
  32 + void* cudaMemPtr;
  33 + switch(srcMat.type()) {
  34 + //case CV_32FC1:
  35 + // br::cuda::cudacopyfrom::wrapper<float>(srcMat.ptr<float>(), &cudaMemPtr, rows, cols);
  36 + // break;
  37 + case CV_8UC1:
  38 + //br::cuda::cudacopyfrom::wrapper<unsigned char>(srcMat.ptr<unsigned char>(), &cudaMemPtr, rows, cols);
  39 + br::cuda::cudacopyto::wrapper(srcMat.ptr<unsigned char>(), &cudaMemPtr, rows, cols);
  40 + break;
  41 + default:
  42 + cout << "ERR: Invalid image type! " << type2str(srcMat.type()) << endl;
  43 + return;
29 44 }
30   - catch(const cv::Exception& ex)
31   - {
32   - cout << "Error: " << ex.what() << endl;
33   - }
34   -
35   - // now create a new Mat that contains the 64-bit pointer
36   - Mat m = Mat(2, 1, CV_32S);
37 45  
38   - // pointer magic
39   - uint64_t gpuMatInt = (uint64_t)gpuMat;
40   - m.at<int>(0,0) = (int32_t)(gpuMatInt & 0x00000000FFFFFFFF);
41   - m.at<int>(1,0) = (int32_t)((gpuMatInt & 0xFFFFFFFF00000000) >> (uint64_t)32);
  46 + // output will be a single pointer to graphics card memory
  47 + Mat dstMat = Mat(4, 1, DataType<void*>::type);
  48 + void** dstMatData = dstMat.ptr<void*>();
42 49  
43   - printf("gpuMatInt: %li\n", gpuMatInt);
44   - printf("m.at(0,0): %i\nm.at(1,0): %i\n", m.at<int>(0,0), m.at<int>(1,0));
  50 + // save cuda ptr, rows, cols, then type
  51 + dstMatData[0] = cudaMemPtr;
  52 + dstMatData[1] = new int; *((int*)dstMatData[1]) = rows;
  53 + dstMatData[2] = new int; *((int*)dstMatData[2]) = cols;
  54 + dstMatData[3] = new int; *((int*)dstMatData[3]) = srcMat.type();
45 55  
46   - // save away in the destination mat
47   - dst += m;
  56 + dst = dstMat;
48 57 }
49 58 };
50 59  
... ...
openbr/plugins/cuda/copyto.cu 0 → 100644
  View file @aaabb3c
  1 +namespace br { namespace cuda { namespace cudacopyto {
  2 + //template<typename T>
  3 + //void wrapper(const T* in, void** out, const int rows, const int cols) {
  4 + void wrapper(const unsigned char* in, void** out, const int rows, const int cols) {
  5 + cudaMalloc(out, rows*cols*sizeof(unsigned char));
  6 + cudaMemcpy(*out, in, rows*cols*sizeof(unsigned char), cudaMemcpyHostToDevice);
  7 + }
  8 +}}}
... ...