Try with free() applied to the DeviceAllocation object (in this case a_gpu ) import pycuda.driver as cuda a ...
確定! 回上一頁