opencv/samples/gpu/driver_api_stereo_multi.cpp

/* This sample demonstrates working on one piece of data using two GPUs.
   It splits input into two parts and processes them separately on different
   GPUs. */

// Disable some warnings which are caused with CUDA headers
#if defined(_MSC_VER)
#pragma warning(disable: 4201 4408 4100)
#endif

#include <iostream>
#include "cvconfig.h"
#include "opencv2/core/core.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/gpu/gpu.hpp"

#if !defined(HAVE_CUDA) || !defined(HAVE_TBB)

int main()
{
#if !defined(HAVE_CUDA)
    std::cout << "CUDA support is required (CMake key 'WITH_CUDA' must be true).\n";
#endif

#if !defined(HAVE_TBB)
    std::cout << "TBB support is required (CMake key 'WITH_TBB' must be true).\n";
#endif

    return 0;
}

#else

#include <cuda.h>
#include <cuda_runtime.h>
#include <GL/gl.h>
#include <cudaGL.h>
#include "opencv2/core/internal.hpp" // For TBB wrappers
#include "tbb/tbb.h"
#include "tbb/mutex.h"

using namespace std;
using namespace cv;
using namespace cv::gpu;

struct Worker { void operator()(int device_id) const; };
void destroyContexts();

#define safeCall(expr) safeCall_(expr, #expr, __FILE__, __LINE__)
inline void safeCall_(int code, const char* expr, const char* file, int line)
{
    if (code != CUDA_SUCCESS)
    {
        std::cout << "CUDA driver API error: code " << code << ", expr " << expr
            << ", file " << file << ", line " << line << endl;
        destroyContexts();
        exit(-1);
    }
}

// Each GPU is associated with its own context
CUcontext contexts[/*2*/1];

void inline contextOn(int id) 
{
    safeCall(cuCtxPushCurrent(contexts[id]));
}

void inline contextOff() 
{
    CUcontext prev_context;
    safeCall(cuCtxPopCurrent(&prev_context));
}

// GPUs data
GpuMat d_left[2];
GpuMat d_right[2];
StereoBM_GPU* bm[2];
GpuMat d_result[2];

// CPU result
Mat result;

int some[2];

tbb::mutex mutex;

int main(int argc, char** argv)
{
    if (argc < 3)
    {
        std::cout << "Usage: stereo_multi_gpu <left_image> <right_image>\n";
        return -1;
    }

    int num_devices = getCudaEnabledDeviceCount();
//    if (num_devices < 2)
//    {
//        std::cout << "Two or more GPUs are required\n";
//        return -1;
//    }

    for (int i = 0; i < num_devices; ++i)
    {
        DeviceInfo dev_info(i);
        if (!dev_info.isCompatible())
        {
            std::cout << "GPU module isn't built for GPU #" << i << " ("
                 << dev_info.name() << ", CC " << dev_info.majorVersion()
                 << dev_info.minorVersion() << "\n";
            return -1;
        }
    }

    // Load input data
    Mat left = imread(argv[1], CV_LOAD_IMAGE_GRAYSCALE);
    Mat right = imread(argv[2], CV_LOAD_IMAGE_GRAYSCALE);
    if (left.empty())
    {
        std::cout << "Cannot open '" << argv[1] << "'\n";
        return -1;
    }
    if (right.empty())
    {
        std::cout << "Cannot open '" << argv[2] << "'\n";
        return -1;
    }

    // Init CUDA Driver API
    safeCall(cuInit(0));

    // Create context for GPU #0
    CUdevice device;
    safeCall(cuDeviceGet(&device, 0));
    safeCall(cuGLCtxCreate(&contexts[0], 0, device));
    //safeCall(cuCtxCreate(&contexts[0], 0, device));
    contextOff();

//    // Create context for GPU #1
//    safeCall(cuDeviceGet(&device, 0));
//    safeCall(cuCtxCreate(&contexts[1], 0, device));
//    contextOff();

    // Split source images for processing on GPU #0
    contextOn(0);
    d_left[0].upload(left.rowRange(0, left.rows / 2));
    d_right[0].upload(right.rowRange(0, right.rows / 2));
    bm[0] = new StereoBM_GPU();
    contextOff();

    // Split source images for processing on the GPU #1
    contextOn(0);
    d_left[1].upload(left.rowRange(left.rows / 2, left.rows));
    d_right[1].upload(right.rowRange(right.rows / 2, right.rows));
    bm[1] = new StereoBM_GPU();
    contextOff();

    some[0] = some[1] = 0;
    // Execute calculation in two threads using two GPUs
    vector<int> devices;
    for (int i = 0; i < 4; ++i)
        devices.push_back(rand()%2);
    tbb::parallel_do(&devices[0], &devices[devices.size() - 1], Worker());

    cout << some[0] << " " << some[1] << endl;

    // Release the first GPU resources
    contextOn(0);
    imshow("GPU #0 result", Mat(d_result[0]));
    d_left[0].release();
    d_right[0].release();
    d_result[0].release();
    delete bm[0];
    contextOff();

    // Release the second GPU resources
    contextOn(0);
    imshow("GPU #1 result", Mat(d_result[1]));
    d_left[1].release();
    d_right[1].release();
    d_result[1].release();
    delete bm[1];
    contextOff();

    waitKey();
    destroyContexts();
    return 0;
}


void Worker::operator()(int device_id) const
{
    mutex.lock();

    contextOn(0);

    bm[device_id]->operator()(d_left[device_id], d_right[device_id],
                              d_result[device_id]);

    std::cout << "GPU #" << device_id << " (" << DeviceInfo().name()
        << "): finished\n";

    contextOff();

    mutex.unlock();
}


void destroyContexts()
{
    safeCall(cuCtxDestroy(contexts[0]));
    //safeCall(cuCtxDestroy(contexts[1]));
}

#endif