Get information about OpenCL devices

platforms = cv.Utils.getPlatfomsInfo()

Output

• platforms struct-array of information about OpenCL platforms:
  • name Platform name string.
  • vendor Platform vendor string.
  • version OpenCL version string supported by the implementation.
  • device struct-array of information about OpenCL device:
    • name Device name string.
    • extensions Space separated list of extension names.
    • version OpenCL version string supported by the device.
    • vendorName Vendor name string.
    • vendorID Vendor ID string.
    • OpenCL_C_Version The highest OpenCL C version supported by the compiler for this device.
    • OpenCLVersion OpenCL version string supported by the device.
    • deviceVersionMajor OpenCL major version supported by the device.
    • deviceVersionMinor OpenCL minor version supported by the device.
    • driverVersion OpenCL software driver version string.
    • type The OpenCL device type (CPU, GPU, etc.).
    • addressBits The default compute device address space size, 32 or 64 bits.
    • available true if the device is available and false if the device is not available.
    • compilerAvailable Is false if the implementation does not have a compiler available to compile the program source. Is true if the compiler is available.
    • linkerAvailable Is false if the implementation does not have a linker available. Is true if the linker is available.
    • doubleFPConfig Describes double precision floating-point capability of the device.
    • singleFPConfig Describes single precision floating-point capability of the device:
      • Denorm denorms are supported.
      • InfNaN INF and NaNs are supported.
      • RoundToNearest round to nearest even rounding mode supported.
      • RoundToZero round to zero rounding mode supported.
      • RoundToInf round to positive and negative infinity rounding modes supported.
      • FMA IEEE754-2008 fused multiply-add is supported.
      • SoftFloat Basic floating-point operations (such as addition, subtraction, multiplication) are implemented in software.
      • halfFPConfig Describes the optional half precision floating-point capability of the device.
    • endianLittle Is true if the OpenCL device is a little endian device and false otherwise.
    • errorCorrectionSupport Is true if the device implements error correction for all accesses to compute device memory (global and constant). Is false if the device does not implement such error correction.
    • executionCapabilities Describes the execution capabilities of the device:
      • Kernel The OpenCL device can execute OpenCL kernels.
      • NativeKernel The OpenCL device can execute native kernels.
    • globalMemCacheSize Size of global memory cache in bytes.
    • globalMemCacheType Type of global memory cache supported, one of: 'NoCache', 'ReadOnlyCache', 'ReadWriteCache'.
    • globalMemCacheLineSize Size of global memory cache line in bytes.
    • globalMemSize Size of global device memory in bytes.
    • localMemSize Size of local memory arena in bytes.
    • localMemType Type of local memory supported (local or global).
    • hostUnifiedMemory Is true if the device and the host have a unified memory subsystem and is false otherwise.
    • imageSupport Is CL_TRUE if images are supported by the OpenCL device and CL_FALSE otherwise.
    • imageFromBufferSupport Is true if "cl_khr_image2d_from_buffer" extension is supported.
    • imagePitchAlignment The row pitch alignment size in pixels for 2D images created from a buffer.
    • imageBaseAddressAlignment
    • image2DMaxWidth Max width of 2D image or 1D image not created from a buffer object in pixels
    • image2DMaxHeight Max height of 2D image in pixels.
    • image3DMaxWidth Max width of 3D image in pixels.
    • image3DMaxHeight Max height of 3D image in pixels.
    • image3DMaxDepth Max depth of 3D image in pixels.
    • imageMaxBufferSize Max number of pixels for a 1D image created from a buffer object.
    • imageMaxArraySize Max number of images in a 1D or 2D image array.
    • maxClockFrequency Maximum configured clock frequency of the device in MHz.
    • maxComputeUnits The number of parallel compute units on the OpenCL device. A work-group executes on a single compute unit.
    • maxConstantArgs Max number of arguments declared with the __constant qualifier in a kernel.
    • maxConstantBufferSize Max size in bytes of a constant buffer allocation.
    • maxMemAllocSize Max size of memory object allocation in bytes.
    • maxParameterSize Max size in bytes of all arguments that can be passed to a kernel.
    • maxReadImageArgs Max number of image objects arguments of a kernel declared with the read_only qualifier.
    • maxWriteImageArgs Max number of image objects arguments of a kernel declared with the write_only qualifier.
    • maxSamplers Maximum number of samplers that can be used in a kernel.
    • maxWorkGroupSize Maximum number of work-items in a work-group executing a kernel on a single compute unit, using the data parallel execution model.
    • maxWorkItemDims Maximum dimensions that specify the global and local work-item IDs used by the data parallel execution model.
    • maxWorkItemSizes Maximum number of work-items that can be specified in each dimension of the work-group.
    • memBaseAddrAlign The minimum value is the size (in bits) of the largest OpenCL built-in data type supported by the device.
    • nativeVectorWidthChar, nativeVectorWidthShort, nativeVectorWidthInt, nativeVectorWidthLong, nativeVectorWidthFloat, nativeVectorWidthDouble, nativeVectorWidthHalf The native ISA vector width. The vector width is defined as the number of scalar elements that can be stored in the vector.
    • preferredVectorWidthChar, preferredVectorWidthShort, preferredVectorWidthInt, preferredVectorWidthLong, preferredVectorWidthFloat, preferredVectorWidthDouble, preferredVectorWidthHalf Preferred native vector width size for built-in scalar types that can be put into vectors. The vector width is defined as the number of scalar elements that can be stored in the vector.
    • printfBufferSize Maximum size in bytes of the internal buffer that holds the output of printf calls from a kernel.
    • profilingTimerResolution Resolution of timer, i.e. the number of nanoseconds elapsed before the timer is incremented.

See OpenCL docs.

If OpenCV is compiled without OpenCL support, the function returns an empty struct.