Utilities
General-purpose functions for managing arrays and devices.
af.alloc_device() - Allocate memory on device |
af.alloc_host() - Allocate memory on host. |
af.alloc_pinned() - Allocate pinned memory on device. |
af.cublas_set_math_mode() - Sets the cuBLAS math mode for the internal handle |
af.delete_image_memory() - Delete memory created by saveImageMem |
af.det() - Find the determinant of a matrix. |
af.device_gc() - Call the garbage collection routine |
af.device_info() - Gets the information about device and platform as strings. |
af.device_mem_info() - Gets information about the memory manager |
af.free_device() - Free memory allocated on device internally by ArrayFire. |
af.free_host() - Free memory allocated on host internally by ArrayFire. |
af.free_pinned() - Free pinned memory allocated by ArrayFire’s memory manager. |
af.get_backend() - Gets the backend enum for the active backend. |
af.get_dbl_support() - Gets if the device supports double floating point |
af.get_device() - Get the current device ID. |
af.get_device_count() - Gets the number of compute devices on the system. |
af.get_half_support() - Gets if the device supports half floating point |
af.get_kernel_cache_directory() - Returns directory where ArrayFire JIT kernels are being stored |
af.get_mem_step_size() - Get the minimum memory chunk size. |
af.get_native_id() - Get the native device id of the CUDA device with id in ArrayFire context |
af.get_stream() - Returns the current cuda stream |
af.info() - Display ArrayFire and device info. |
af.info_string() - Returns a string with information of current device and backend |
af.init() - Initializes ArrayFire |
af.is_image_io_available() - Checks if Image IO is available |
af.is_lapack_available() - Check if lapack runtimes are available |
af.load_image() - Load an image from disk to an array |
af.load_image_memory() - Load an image from memory which is stored as a FreeImage stream |
af.load_image_native() - Load an image as is original type. |
af.print_mem_info() - Prints buffer details from the ArrayFire Device Manager |
af.read_array() - Load an array from a file. |
af.save_array() - Save an array to a binary file |
af.save_image() - Save an array to disk as an image |
af.save_image_memory() - Save an array to memory as an image using FreeImage stream |
af.save_image_native() - Save an image as is original type. |
af.set_backend() - Set the current backend |
af.set_device() - Change current device to specified device. |
af.set_fft_plan_cache_size() - Sets fft plan cache size |
af.set_kernel_cache_directory() - Sets the directory for JIT kernel caching |
af.set_mem_step_size() - Get the minimum memory chunk size. |
af.set_native_id() - Set the CUDA device with given native id as the active device for ArrayFire |