Waits for all kernels in all streams on a CUDA device to complete.

Usage

cuda_synchronize(device = NULL)

Arguments

device

device for which to synchronize. It uses the current device given by cuda_current_device() if no device is specified.