Function cutensorReadKernelCacheFromFile 

pub unsafe extern "C" fn cutensorReadKernelCacheFromFile(
    handle: cutensorHandle_t,
    filename: *const c_char,
) -> cutensorStatus_t

Reads a kernel cache from file and adds all non-existing JIT compiled kernels to the kernel cache.

A cache is only valid for the same cuTENSOR version and CUDA version; moreover, the GPU architecture (incl. multiprocessor count) must match, otherwise cutensorStatus_t::CUTENSOR_STATUS_INVALID_VALUE will be returned.

Remark

non-blocking, no reentrant, and thread-safe.

Parameters

• handle: Opaque handle holding cuTENSOR’s library context.
• filename: Specifies the filename (including the absolute path) to the file that holds all the cache information that have previously been written by cutensorWriteKernelCacheToFile.

Return value

• cutensorStatus_t::CUTENSOR_STATUS_INVALID_VALUE: if the stored cache was created by a different cuTENSOR- or CUDA-version or if the GPU architecture (incl. multiprocessor count) doesn’t match.
• cutensorStatus_t::CUTENSOR_STATUS_IO_ERROR: if the file cannot be read.
• cutensorStatus_t::CUTENSOR_STATUS_NOT_INITIALIZED: if the handle is not initialized.
• cutensorStatus_t::CUTENSOR_STATUS_NOT_SUPPORTED: if the function is not available for the operating system, CUDA Toolkit, or compute capability of the device.
• cutensorStatus_t::CUTENSOR_STATUS_SUCCESS: The operation completed successfully, or the file pointed to by filename was empty.