added more assertion on device features to gpu functions and tests

moved TargerArchs and DeviceInfo to core
fixed bug in GpuMat::copy with mask (incorrect index in function tab)
This commit is contained in:
Vladislav Vinogradov
2012-03-27 10:34:30 +00:00
parent e8fab91d51
commit eaea6782d5
11 changed files with 1501 additions and 1619 deletions

View File

@@ -54,94 +54,6 @@
namespace cv { namespace gpu {
//////////////////////////////// Initialization & Info ////////////////////////
//! This is the only function that do not throw exceptions if the library is compiled without Cuda.
CV_EXPORTS int getCudaEnabledDeviceCount();
//! Functions below throw cv::Expception if the library is compiled without Cuda.
CV_EXPORTS void setDevice(int device);
CV_EXPORTS int getDevice();
//! Explicitly destroys and cleans up all resources associated with the current device in the current process.
//! Any subsequent API call to this device will reinitialize the device.
CV_EXPORTS void resetDevice();
enum FeatureSet
{
FEATURE_SET_COMPUTE_10 = 10,
FEATURE_SET_COMPUTE_11 = 11,
FEATURE_SET_COMPUTE_12 = 12,
FEATURE_SET_COMPUTE_13 = 13,
FEATURE_SET_COMPUTE_20 = 20,
FEATURE_SET_COMPUTE_21 = 21,
GLOBAL_ATOMICS = FEATURE_SET_COMPUTE_11,
SHARED_ATOMICS = FEATURE_SET_COMPUTE_12,
NATIVE_DOUBLE = FEATURE_SET_COMPUTE_13
};
// Gives information about what GPU archs this OpenCV GPU module was
// compiled for
class CV_EXPORTS TargetArchs
{
public:
static bool builtWith(FeatureSet feature_set);
static bool has(int major, int minor);
static bool hasPtx(int major, int minor);
static bool hasBin(int major, int minor);
static bool hasEqualOrLessPtx(int major, int minor);
static bool hasEqualOrGreater(int major, int minor);
static bool hasEqualOrGreaterPtx(int major, int minor);
static bool hasEqualOrGreaterBin(int major, int minor);
private:
TargetArchs();
};
// Gives information about the given GPU
class CV_EXPORTS DeviceInfo
{
public:
// Creates DeviceInfo object for the current GPU
DeviceInfo() : device_id_(getDevice()) { query(); }
// Creates DeviceInfo object for the given GPU
DeviceInfo(int device_id) : device_id_(device_id) { query(); }
std::string name() const { return name_; }
// Return compute capability versions
int majorVersion() const { return majorVersion_; }
int minorVersion() const { return minorVersion_; }
int multiProcessorCount() const { return multi_processor_count_; }
size_t freeMemory() const;
size_t totalMemory() const;
// Checks whether device supports the given feature
bool supports(FeatureSet feature_set) const;
// Checks whether the GPU module can be run on the given device
bool isCompatible() const;
int deviceID() const { return device_id_; }
private:
void query();
void queryMemory(size_t& free_memory, size_t& total_memory) const;
int device_id_;
std::string name_;
int multi_processor_count_;
int majorVersion_;
int minorVersion_;
};
CV_EXPORTS void printCudaDeviceInfo(int device);
CV_EXPORTS void printShortCudaDeviceInfo(int device);
//////////////////////////////// CudaMem ////////////////////////////////
// CudaMem is limited cv::Mat with page locked memory allocation.
// Page locked memory is only needed for async and faster coping to GPU.