diff --git a/cmake/OpenCVPackaging.cmake b/cmake/OpenCVPackaging.cmake
index 732eae9e4e..3d63c09fa9 100644
--- a/cmake/OpenCVPackaging.cmake
+++ b/cmake/OpenCVPackaging.cmake
@@ -64,7 +64,7 @@ set(CPACK_DEBIAN_PACKAGE_HOMEPAGE "http://opencv.org")
 
 #depencencies
 set(CPACK_DEBIAN_PACKAGE_SHLIBDEPS TRUE)
-set(CPACK_COMPONENT_samples_DEPENDS libs)
+set(CPACK_COMPONENT_samples_DEPENDS libs dev)
 set(CPACK_COMPONENT_dev_DEPENDS libs)
 set(CPACK_COMPONENT_docs_DEPENDS libs)
 set(CPACK_COMPONENT_java_DEPENDS libs)
diff --git a/samples/CMakeLists.txt b/samples/CMakeLists.txt
index 48a419d75c..659ab058c1 100644
--- a/samples/CMakeLists.txt
+++ b/samples/CMakeLists.txt
@@ -19,6 +19,9 @@ if(ANDROID AND BUILD_ANDROID_EXAMPLES)
   add_subdirectory(android)
 endif()
 
+install(FILES "CMakeLists.txt"
+        DESTINATION ${OPENCV_SAMPLES_SRC_INSTALL_PATH}
+        PERMISSIONS OWNER_READ GROUP_READ WORLD_READ COMPONENT samples)
 
 #
 # END OF BUILD CASE 1: Build samples with library sources
@@ -60,7 +63,7 @@ endif()
 add_subdirectory(c)
 add_subdirectory(cpp)
 add_subdirectory(ocl)
-# FIXIT: can't use cvconfig.h in samples: add_subdirectory(gpu)
+add_subdirectory(gpu)
 
 #
 # END OF BUILD CASE 2: Build samples with library binaries
diff --git a/samples/c/CMakeLists.txt b/samples/c/CMakeLists.txt
index 6d374e7443..61f23666f3 100644
--- a/samples/c/CMakeLists.txt
+++ b/samples/c/CMakeLists.txt
@@ -51,8 +51,8 @@ if(BUILD_EXAMPLES AND OCV_DEPENDENCIES_FOUND)
   endforeach()
 endif()
 
-if (INSTALL_C_EXAMPLES AND NOT WIN32)
-  file(GLOB C_SAMPLES *.c *.cpp *.jpg *.png *.data makefile.* build_all.sh *.dsp *.cmd )
+if (OCV_DEPENDENCIES_FOUND AND INSTALL_C_EXAMPLES AND NOT WIN32)
+  file(GLOB C_SAMPLES *.c *.cpp *.jpg *.png *.data makefile.* build_all.sh *.dsp *.cmd *.txt)
   install(FILES ${C_SAMPLES}
           DESTINATION ${OPENCV_SAMPLES_SRC_INSTALL_PATH}/c
           PERMISSIONS OWNER_READ GROUP_READ WORLD_READ COMPONENT samples)
diff --git a/samples/c/latentsvmdetect.cpp b/samples/c/latentsvmdetect.cpp
index e74b227a62..92e237a24a 100644
--- a/samples/c/latentsvmdetect.cpp
+++ b/samples/c/latentsvmdetect.cpp
@@ -2,13 +2,6 @@
 #include "opencv2/highgui/highgui.hpp"
 #include <stdio.h>
 
-#ifdef HAVE_CVCONFIG_H
-#include <cvconfig.h>
-#endif
-#ifdef HAVE_TBB
-#include "tbb/task_scheduler_init.h"
-#endif
-
 using namespace cv;
 
 static void help()
@@ -31,28 +24,13 @@ static void detect_and_draw_objects( IplImage* image, CvLatentSvmDetector* detec
     CvSeq* detections = 0;
     int i = 0;
     int64 start = 0, finish = 0;
-#ifdef HAVE_TBB
-    tbb::task_scheduler_init init(tbb::task_scheduler_init::deferred);
-    if (numThreads > 0)
-    {
-        init.initialize(numThreads);
-        printf("Number of threads %i\n", numThreads);
-    }
-    else
-    {
-        printf("Number of threads is not correct for TBB version");
-        return;
-    }
-#endif
+    cv::setNumThreads(numThreads);
 
     start = cvGetTickCount();
     detections = cvLatentSvmDetectObjects(image, detector, storage, 0.5f, numThreads);
     finish = cvGetTickCount();
     printf("detection time = %.3f\n", (float)(finish - start) / (float)(cvGetTickFrequency() * 1000000.0));
 
-#ifdef HAVE_TBB
-    init.terminate();
-#endif
     for( i = 0; i < detections->total; i++ )
     {
         CvObjectDetection detection = *(CvObjectDetection*)cvGetSeqElem( detections, i );
diff --git a/samples/cpp/CMakeLists.txt b/samples/cpp/CMakeLists.txt
index 1f55ed90f3..65e402bc13 100644
--- a/samples/cpp/CMakeLists.txt
+++ b/samples/cpp/CMakeLists.txt
@@ -90,8 +90,8 @@ if(BUILD_EXAMPLES AND OCV_DEPENDENCIES_FOUND)
   endforeach()
 endif()
 
-if (INSTALL_C_EXAMPLES AND NOT WIN32)
-  file(GLOB C_SAMPLES *.c *.cpp *.jpg *.png *.data makefile.* build_all.sh *.dsp *.cmd )
+if (OCV_DEPENDENCIES_FOUND AND INSTALL_C_EXAMPLES AND NOT WIN32)
+  file(GLOB C_SAMPLES *.c *.cpp *.jpg *.png *.data makefile.* build_all.sh *.dsp *.cmd *.txt)
   install(FILES ${C_SAMPLES}
           DESTINATION ${OPENCV_SAMPLES_SRC_INSTALL_PATH}/cpp
           PERMISSIONS OWNER_READ GROUP_READ WORLD_READ COMPONENT samples)
diff --git a/samples/cpp/latentsvm_multidetect.cpp b/samples/cpp/latentsvm_multidetect.cpp
index 37297e23de..de368eab5d 100644
--- a/samples/cpp/latentsvm_multidetect.cpp
+++ b/samples/cpp/latentsvm_multidetect.cpp
@@ -9,14 +9,6 @@
 #include <dirent.h>
 #endif
 
-#ifdef HAVE_CVCONFIG_H
-#include <cvconfig.h>
-#endif
-
-#ifdef HAVE_TBB
-#include "tbb/task_scheduler_init.h"
-#endif
-
 using namespace std;
 using namespace cv;
 
diff --git a/samples/gpu/CMakeLists.txt b/samples/gpu/CMakeLists.txt
index d25c3a6b5d..dc7464f942 100644
--- a/samples/gpu/CMakeLists.txt
+++ b/samples/gpu/CMakeLists.txt
@@ -69,8 +69,28 @@ if(BUILD_EXAMPLES AND OCV_DEPENDENCIES_FOUND)
     endif()
   ENDMACRO()
 
+  # remove all matching elements from the list
+  MACRO(list_filterout lst regex)
+    foreach(item ${${lst}})
+      if(item MATCHES "${regex}")
+        list(REMOVE_ITEM ${lst} "${item}")
+      endif()
+    endforeach()
+  ENDMACRO()
+
   file(GLOB all_samples RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} *.cpp)
 
+  if(NOT HAVE_OPENGL)
+    list(REMOVE_ITEM all_samples "opengl.cpp")
+  endif()
+
+  if(NOT HAVE_CUDA)
+    list(REMOVE_ITEM all_samples "opticalflow_nvidia_api.cpp")
+    list(REMOVE_ITEM all_samples "cascadeclassifier_nvidia_api.cpp")
+    list(REMOVE_ITEM all_samples "driver_api_multi.cpp")
+    list(REMOVE_ITEM all_samples "driver_api_stereo_multi.cpp")
+  endif()
+
   foreach(sample_filename ${all_samples})
     get_filename_component(sample ${sample_filename} NAME_WE)
     file(GLOB sample_srcs RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} ${sample}.*)
@@ -80,8 +100,17 @@ if(BUILD_EXAMPLES AND OCV_DEPENDENCIES_FOUND)
   include("performance/CMakeLists.txt")
 endif()
 
-if (INSTALL_C_EXAMPLES AND NOT WIN32)
-  file(GLOB install_list *.c *.cpp *.jpg *.png *.data makefile.* build_all.sh *.dsp *.cmd )
+if (OCV_DEPENDENCIES_FOUND AND INSTALL_C_EXAMPLES AND NOT WIN32)
+  file(GLOB install_list *.c *.cpp *.jpg *.png *.data makefile.* build_all.sh *.dsp *.cmd *.txt)
+  if(NOT HAVE_OPENGL)
+    list_filterout(install_list ".*opengl.cpp")
+  endif()
+  if(NOT HAVE_CUDA)
+    list_filterout(install_list ".*opticalflow_nvidia_api.cpp")
+    list_filterout(install_list ".*cascadeclassifier_nvidia_api.cpp")
+    list_filterout(install_list ".*driver_api_multi.cpp")
+    list_filterout(install_list ".*driver_api_stereo_multi.cpp")
+  endif()
   install(FILES ${install_list}
           DESTINATION ${OPENCV_SAMPLES_SRC_INSTALL_PATH}/gpu
           PERMISSIONS OWNER_READ GROUP_READ WORLD_READ COMPONENT samples)
diff --git a/samples/gpu/brox_optical_flow.cpp b/samples/gpu/brox_optical_flow.cpp
index 7cd5089b41..ecbd4d9fa2 100644
--- a/samples/gpu/brox_optical_flow.cpp
+++ b/samples/gpu/brox_optical_flow.cpp
@@ -3,7 +3,6 @@
 #include <string>
 #include <ctype.h>
 
-#include "cvconfig.h"
 #include "opencv2/core/core.hpp"
 #include "opencv2/highgui/highgui.hpp"
 #include "opencv2/gpu/gpu.hpp"
diff --git a/samples/gpu/cascadeclassifier_nvidia_api.cpp b/samples/gpu/cascadeclassifier_nvidia_api.cpp
index 98195b35c2..87258b9a05 100644
--- a/samples/gpu/cascadeclassifier_nvidia_api.cpp
+++ b/samples/gpu/cascadeclassifier_nvidia_api.cpp
@@ -2,39 +2,24 @@
 #pragma warning( disable : 4201 4408 4127 4100)
 #endif
 
-#include "cvconfig.h"
 #include <iostream>
 #include <iomanip>
 #include <cstdio>
 #include "opencv2/gpu/gpu.hpp"
 #include "opencv2/highgui/highgui.hpp"
-
-#ifdef HAVE_CUDA
 #include "NCVHaarObjectDetection.hpp"
-#endif
 
 using namespace std;
 using namespace cv;
 
-
-#if !defined(HAVE_CUDA) || defined(__arm__)
-
-int main( int, const char** )
-{
-#if !defined(HAVE_CUDA)
-    std::cout << "CUDA support is required (CMake key 'WITH_CUDA' must be true)." << std::endl;
-#endif
-
 #if defined(__arm__)
+int main()
+{
     std::cout << "Unsupported for ARM CUDA library." << std::endl;
-#endif
-
     return 0;
 }
-
 #else
 
-
 const Size2i preferredVideoFrameSize(640, 480);
 const string wndTitle = "NVIDIA Computer Vision :: Haar Classifiers Cascade";
 
@@ -386,4 +371,4 @@ int main(int argc, const char** argv)
     return 0;
 }
 
-#endif //!defined(HAVE_CUDA)
+#endif
diff --git a/samples/gpu/driver_api_multi.cpp b/samples/gpu/driver_api_multi.cpp
index c829830e72..4f5231c32d 100644
--- a/samples/gpu/driver_api_multi.cpp
+++ b/samples/gpu/driver_api_multi.cpp
@@ -7,56 +7,101 @@
 #endif
 
 #include <iostream>
-#include "cvconfig.h"
 #include "opencv2/core/core.hpp"
 #include "opencv2/gpu/gpu.hpp"
 
-#if !defined(HAVE_CUDA) || !defined(HAVE_TBB) || defined(__arm__)
-
+#if defined(__arm__)
 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
-
-#if defined(__arm__)
     std::cout << "Unsupported for ARM CUDA library." << std::endl;
-#endif
-
     return 0;
 }
-
 #else
 
 #include <cuda.h>
 #include <cuda_runtime.h>
-#include "opencv2/core/internal.hpp" // For TBB wrappers
 
 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();
+        << ", file " << file << ", line " << line << endl;
         exit(-1);
     }
 }
 
-// Each GPU is associated with its own context
-CUcontext contexts[2];
+struct Worker: public ParallelLoopBody
+{
+    Worker(int num_devices)
+    {
+        count = num_devices;
+        contexts = new CUcontext[num_devices];
+        for (int device_id = 0; device_id < num_devices; device_id++)
+        {
+            CUdevice device;
+            safeCall(cuDeviceGet(&device, device_id));
+            safeCall(cuCtxCreate(&contexts[device_id], 0, device));
+        }
+    }
+
+    virtual void operator() (const Range& range) const
+    {
+        for (int device_id = range.start; device_id != range.end; ++device_id)
+        {
+            // Set the proper context
+            safeCall(cuCtxPushCurrent(contexts[device_id]));
+
+            Mat src(1000, 1000, CV_32F);
+            Mat dst;
+
+            RNG rng(0);
+            rng.fill(src, RNG::UNIFORM, 0, 1);
+
+            // CPU works
+            transpose(src, dst);
+
+            // GPU works
+            GpuMat d_src(src);
+            GpuMat d_dst;
+            transpose(d_src, d_dst);
+
+            // Check results
+            bool passed = norm(dst - Mat(d_dst), NORM_INF) < 1e-3;
+            std::cout << "GPU #" << device_id << " (" << DeviceInfo().name() << "): "
+            << (passed ? "passed" : "FAILED") << endl;
+
+            // Deallocate data here, otherwise deallocation will be performed
+            // after context is extracted from the stack
+            d_src.release();
+            d_dst.release();
+
+            CUcontext prev_context;
+            safeCall(cuCtxPopCurrent(&prev_context));
+        }
+    }
+
+    ~Worker()
+    {
+        if ((contexts != NULL) && count != 0)
+        {
+            for (int device_id = 0; device_id < count; device_id++)
+            {
+                safeCall(cuCtxDestroy(contexts[device_id]));
+            }
+
+            delete[] contexts;
+        }
+    }
+
+    CUcontext* contexts;
+    int count;
+};
 
 int main()
 {
@@ -84,67 +129,10 @@ int main()
     // Init CUDA Driver API
     safeCall(cuInit(0));
 
-    // Create context for GPU #0
-    CUdevice device;
-    safeCall(cuDeviceGet(&device, 0));
-    safeCall(cuCtxCreate(&contexts[0], 0, device));
-
-    CUcontext prev_context;
-    safeCall(cuCtxPopCurrent(&prev_context));
-
-    // Create context for GPU #1
-    safeCall(cuDeviceGet(&device, 1));
-    safeCall(cuCtxCreate(&contexts[1], 0, device));
-
-    safeCall(cuCtxPopCurrent(&prev_context));
+    // Execute calculation
+    parallel_for_(cv::Range(0, num_devices), Worker(num_devices));
 
-    // Execute calculation in two threads using two GPUs
-    int devices[] = {0, 1};
-    parallel_do(devices, devices + 2, Worker());
-
-    destroyContexts();
     return 0;
 }
 
-
-void Worker::operator()(int device_id) const
-{
-    // Set the proper context
-    safeCall(cuCtxPushCurrent(contexts[device_id]));
-
-    Mat src(1000, 1000, CV_32F);
-    Mat dst;
-
-    RNG rng(0);
-    rng.fill(src, RNG::UNIFORM, 0, 1);
-
-    // CPU works
-    transpose(src, dst);
-
-    // GPU works
-    GpuMat d_src(src);
-    GpuMat d_dst;
-    transpose(d_src, d_dst);
-
-    // Check results
-    bool passed = norm(dst - Mat(d_dst), NORM_INF) < 1e-3;
-    std::cout << "GPU #" << device_id << " (" << DeviceInfo().name() << "): "
-        << (passed ? "passed" : "FAILED") << endl;
-
-    // Deallocate data here, otherwise deallocation will be performed
-    // after context is extracted from the stack
-    d_src.release();
-    d_dst.release();
-
-    CUcontext prev_context;
-    safeCall(cuCtxPopCurrent(&prev_context));
-}
-
-
-void destroyContexts()
-{
-    safeCall(cuCtxDestroy(contexts[0]));
-    safeCall(cuCtxDestroy(contexts[1]));
-}
-
 #endif
diff --git a/samples/gpu/driver_api_stereo_multi.cpp b/samples/gpu/driver_api_stereo_multi.cpp
index d4d0af451c..507203f858 100644
--- a/samples/gpu/driver_api_stereo_multi.cpp
+++ b/samples/gpu/driver_api_stereo_multi.cpp
@@ -8,30 +8,16 @@
 #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) || defined(__arm__)
-
+#if defined(__arm__)
 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
-
-#if defined(__arm__)
     std::cout << "Unsupported for ARM CUDA library." << std::endl;
-#endif
-
     return 0;
 }
-
 #else
 
 #include <cuda.h>
@@ -42,7 +28,6 @@ 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__)
@@ -77,6 +62,27 @@ GpuMat d_right[2];
 StereoBM_GPU* bm[2];
 GpuMat d_result[2];
 
+
+struct Worker: public ParallelLoopBody
+{
+    virtual void operator() (const Range& range) const
+    {
+        for (int device_id = range.start; device_id != range.end; ++device_id)
+        {
+            contextOn(device_id);
+
+            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();
+
+        }
+    }
+};
+
 static void printHelp()
 {
     std::cout << "Usage: driver_api_stereo_multi_gpu --left <left_image> --right <right_image>\n";
@@ -162,8 +168,7 @@ int main(int argc, char** argv)
     contextOff();
 
     // Execute calculation in two threads using two GPUs
-    int devices[] = {0, 1};
-    parallel_do(devices, devices + 2, Worker());
+    parallel_for_(cv::Range(0, 2), Worker());
 
     // Release the first GPU resources
     contextOn(0);
@@ -188,21 +193,6 @@ int main(int argc, char** argv)
     return 0;
 }
 
-
-void Worker::operator()(int device_id) const
-{
-    contextOn(device_id);
-
-    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();
-}
-
-
 void destroyContexts()
 {
     safeCall(cuCtxDestroy(contexts[0]));
diff --git a/samples/gpu/multi.cpp b/samples/gpu/multi.cpp
index c3925fc442..ddb512f6fc 100644
--- a/samples/gpu/multi.cpp
+++ b/samples/gpu/multi.cpp
@@ -7,34 +7,47 @@
 #endif
 
 #include <iostream>
-#include "cvconfig.h"
 #include "opencv2/core/core.hpp"
 #include "opencv2/gpu/gpu.hpp"
 
-#if !defined(HAVE_CUDA) || !defined(HAVE_TBB)
+using namespace std;
+using namespace cv;
+using namespace cv::gpu;
 
-int main()
+struct Worker: public ParallelLoopBody
 {
-#if !defined(HAVE_CUDA)
-    std::cout << "CUDA support is required (CMake key 'WITH_CUDA' must be true).\n";
-#endif
+    virtual void operator() (const Range& range) const
+    {
+        for (int device_id = range.start; device_id != range.end; ++device_id)
+        {
+            setDevice(device_id);
 
-#if !defined(HAVE_TBB)
-    std::cout << "TBB support is required (CMake key 'WITH_TBB' must be true).\n";
-#endif
+            Mat src(1000, 1000, CV_32F);
+            Mat dst;
 
-    return 0;
-}
+            RNG rng(0);
+            rng.fill(src, RNG::UNIFORM, 0, 1);
 
-#else
+            // CPU works
+            transpose(src, dst);
 
-#include "opencv2/core/internal.hpp" // For TBB wrappers
+            // GPU works
+            GpuMat d_src(src);
+            GpuMat d_dst;
+            transpose(d_src, d_dst);
 
-using namespace std;
-using namespace cv;
-using namespace cv::gpu;
+            // Check results
+            bool passed = norm(dst - Mat(d_dst), NORM_INF) < 1e-3;
+            std::cout << "GPU #" << device_id << " (" << DeviceInfo().name() << "): "
+            << (passed ? "passed" : "FAILED") << endl;
 
-struct Worker { void operator()(int device_id) const; };
+            // Deallocate data here, otherwise deallocation will be performed
+            // after context is extracted from the stack
+            d_src.release();
+            d_dst.release();
+        }
+    }
+};
 
 int main()
 {
@@ -58,41 +71,8 @@ int main()
         }
     }
 
-    // Execute calculation in two threads using two GPUs
-    int devices[] = {0, 1};
-    parallel_do(devices, devices + 2, Worker());
+    // Execute calculation in several threads, 1 GPU per thread
+    parallel_for_(cv::Range(0, num_devices), Worker());
 
     return 0;
 }
-
-
-void Worker::operator()(int device_id) const
-{
-    setDevice(device_id);
-
-    Mat src(1000, 1000, CV_32F);
-    Mat dst;
-
-    RNG rng(0);
-    rng.fill(src, RNG::UNIFORM, 0, 1);
-
-    // CPU works
-    transpose(src, dst);
-
-    // GPU works
-    GpuMat d_src(src);
-    GpuMat d_dst;
-    transpose(d_src, d_dst);
-
-    // Check results
-    bool passed = norm(dst - Mat(d_dst), NORM_INF) < 1e-3;
-    std::cout << "GPU #" << device_id << " (" << DeviceInfo().name() << "): "
-        << (passed ? "passed" : "FAILED") << endl;
-
-    // Deallocate data here, otherwise deallocation will be performed
-    // after context is extracted from the stack
-    d_src.release();
-    d_dst.release();
-}
-
-#endif
diff --git a/samples/gpu/opengl.cpp b/samples/gpu/opengl.cpp
index 5b84a5bef5..d90a6658ec 100644
--- a/samples/gpu/opengl.cpp
+++ b/samples/gpu/opengl.cpp
@@ -1,13 +1,4 @@
 #include <iostream>
-#include "cvconfig.h"
-
-#ifndef HAVE_OPENGL
-int main()
-{
-    std::cerr << "Library was built without OpenGL support" << std::endl;
-    return -1;
-}
-#else
 
 #ifdef WIN32
     #define WIN32_LEAN_AND_MEAN 1
diff --git a/samples/gpu/opticalflow_nvidia_api.cpp b/samples/gpu/opticalflow_nvidia_api.cpp
index 31ee569788..d0b071f64c 100644
--- a/samples/gpu/opticalflow_nvidia_api.cpp
+++ b/samples/gpu/opticalflow_nvidia_api.cpp
@@ -9,26 +9,14 @@
 #include <ctime>
 #include <ctype.h>
 
-#include "cvconfig.h"
 #include <iostream>
 #include <iomanip>
 #include "opencv2/gpu/gpu.hpp"
 #include "opencv2/highgui/highgui.hpp"
 
-#ifdef HAVE_CUDA
 #include "NPP_staging/NPP_staging.hpp"
 #include "NCVBroxOpticalFlow.hpp"
-#endif
-
-#if !defined(HAVE_CUDA)
-int main( int, const char** )
-{
-    std::cout << "Please compile the library with CUDA support" << std::endl;
-    return -1;
-}
-#else
 
-//using std::tr1::shared_ptr;
 using cv::Ptr;
 
 #define PARAM_LEFT  "--left"
@@ -649,5 +637,3 @@ int main(int argc, char **argv)
 
     return 0;
 }
-
-#endif
diff --git a/samples/gpu/performance/CMakeLists.txt b/samples/gpu/performance/CMakeLists.txt
index de0feadd26..4a1603a916 100644
--- a/samples/gpu/performance/CMakeLists.txt
+++ b/samples/gpu/performance/CMakeLists.txt
@@ -27,7 +27,7 @@ if(WIN32)
 endif()
 
 if(INSTALL_C_EXAMPLES AND NOT WIN32)
-  file(GLOB GPU_FILES performance/*.cpp performance/*.h)
+  file(GLOB GPU_FILES performance/*.cpp performance/*.h performance/CMakeLists.txt)
   install(FILES ${GPU_FILES}
           DESTINATION ${OPENCV_SAMPLES_SRC_INSTALL_PATH}/gpu/performance
           PERMISSIONS OWNER_READ GROUP_READ WORLD_READ
diff --git a/samples/gpu/pyrlk_optical_flow.cpp b/samples/gpu/pyrlk_optical_flow.cpp
index 5ccd66009b..1c3eb20a95 100644
--- a/samples/gpu/pyrlk_optical_flow.cpp
+++ b/samples/gpu/pyrlk_optical_flow.cpp
@@ -1,7 +1,6 @@
 #include <iostream>
 #include <vector>
 
-#include "cvconfig.h"
 #include "opencv2/core/core.hpp"
 #include "opencv2/imgproc/imgproc.hpp"
 #include "opencv2/highgui/highgui.hpp"
diff --git a/samples/ocl/CMakeLists.txt b/samples/ocl/CMakeLists.txt
index 7fc20fd356..10db9b10fb 100644
--- a/samples/ocl/CMakeLists.txt
+++ b/samples/ocl/CMakeLists.txt
@@ -51,8 +51,8 @@ if(BUILD_EXAMPLES AND OCV_DEPENDENCIES_FOUND)
   endforeach()
 endif()
 
-if (INSTALL_C_EXAMPLES AND NOT WIN32)
-  file(GLOB install_list *.c *.cpp *.jpg *.png *.data makefile.* build_all.sh *.dsp *.cmd )
+if (OCV_DEPENDENCIES_FOUND AND INSTALL_C_EXAMPLES AND NOT WIN32)
+  file(GLOB install_list *.c *.cpp *.jpg *.png *.data makefile.* build_all.sh *.dsp *.cmd *.txt)
   install(FILES ${install_list}
           DESTINATION ${OPENCV_SAMPLES_SRC_INSTALL_PATH}/ocl
           PERMISSIONS OWNER_READ GROUP_READ WORLD_READ COMPONENT samples)