o "jJ@sddlZddlmZddlmZddlmZddlmZm Z gdZ edd dd d d,d d Z eddddd d,ddZ ddZ ddZddZd,ddZd,ddZd,ddZd,ddZd d!Zedd"dd#d ed$d%Zd,d&d'Zd,d(d)Zd,d*d+ZdS)-N)core)signature_safe_contextmanager) deprecated)StreamEvent)rrcurrent_stream synchronize device_count empty_cachemax_memory_allocatedmax_memory_reservedmemory_allocatedmemory_reserved stream_guardget_device_propertiesget_device_nameget_device_capabilityz2.5.0zpaddle.device.current_streamz>current_stream in paddle.device.cuda will be removed in future)ZsinceZ update_tolevelreasoncCsDd}|durt|tr|}nt|tjr|}ntdt|S)a Return the current CUDA stream by the device. Args: device(paddle.CUDAPlace()|int, optional): The device or the ID of the device which want to get stream from. If device is None, the device is the current device. Default: None. Returns: CUDAStream: the stream to the device. Examples: .. code-block:: python >>> # doctest: +REQUIRES(env:GPU) >>> import paddle >>> paddle.device.set_device('gpu') >>> s1 = paddle.device.cuda.current_stream() >>> s2 = paddle.device.cuda.current_stream(0) >>> s3 = paddle.device.cuda.current_stream(paddle.CUDAPlace(0)) N+device type must be int or paddle.CUDAPlace) isinstanceintr CUDAPlace get_device_id ValueErrorZ_get_current_streamdevice device_idr \/var/www/html/Deteccion_Ine/venv/lib/python3.10/site-packages/paddle/device/cuda/__init__.pyr(s     rzpaddle.device.synchronizez;synchronize in paddle.device.cuda will be removed in futurecCsl|durt|tr |}n%t|tjr|}ntdtj}t r/t|tjr/|}nd}t |S)a Wait for the compute on the given CUDA device to finish. Args: device(paddle.CUDAPlace()|int, optional): The device or the ID of the device. If device is None, the device is the current device. Default: None. Examples: .. code-block:: python >>> # doctest: +REQUIRES(env:GPU) >>> import paddle >>> paddle.device.cuda.synchronize() >>> paddle.device.cuda.synchronize(0) >>> paddle.device.cuda.synchronize(paddle.CUDAPlace(0)) Nrr) rrrrrrpaddleZ frameworkZ_current_expected_placeis_compiled_with_cudaZ_device_synchronize)rrZplacer r r!r Us       r cCsttdr t}|Sd}|S)z Return the number of GPUs available. Returns: int: the number of GPUs available. Examples: .. code-block:: python >>> import paddle >>> paddle.device.cuda.device_count() get_cuda_device_countr)hasattrrr$)Znum_gpusr r r!r sr cCstr tdSdS)a Releases idle cached memory held by the allocator so that those can be used in other GPU application and visible in `nvidia-smi`. In most cases you don't need to use this function, Paddle does not release the memory back to the OS when you remove Tensors on the GPU, Because it keeps gpu memory in a pool so that next allocations can be done much faster. Examples: .. code-block:: python >>> # doctest: +REQUIRES(env:GPU) >>> import paddle >>> paddle.device.set_device('gpu') >>> tensor = paddle.randn([512, 512, 512], "float") >>> del tensor >>> paddle.device.cuda.empty_cache() N)rr#Zcuda_empty_cacher r r r!r s r cCs|durtSt|tr|}n.t|tjr|}n#t|tr6|dr.t|dd}ntd ||td |||dksJJd|d|t ksZJd |d t |S) a Return the id of the given cuda device. It is just a utility that will not be exposed to users. Args: device(paddle.CUDAPlace or int or str): The device, the id of the device or the string name of device like 'gpu:x'. Default: None. Return: int: The id of the given device. If device is None, return the id of current device. Ngpu:zThe current string {} is not expected. Because {} only support string which is like 'gpu:x'. Please input appropriate string again!zThe device type {} is not expected. Because {} only support int, str or paddle.CUDAPlace. Please input appropriate device again!rz4The device id must be not less than 0, but got id = .zThe device id z exceeds gpu card number ) rZget_cuda_current_device_idrrrrstr startswithrformatr )rop_namerr r r!extract_cuda_device_ids0         r-cC4d}tstd|dt||d}td|S)a Return the peak size of gpu memory that is allocated to tensor of the given device. Note: The size of GPU memory allocated to tensor is 256-byte aligned in Paddle, which may larger than the memory size that tensor actually need. For instance, a float32 tensor with shape [1] in GPU will take up 256 bytes memory, even though storing a float32 data requires only 4 bytes. Args: device(paddle.CUDAPlace or int or str, optional): The device, the id of the device or the string name of device like 'gpu:x'. If device is None, the device is the current device. Default: None. Return: int: The peak size of gpu memory that is allocated to tensor of the given device, in bytes. Examples: .. code-block:: python >>> # doctest: +REQUIRES(env:GPU) >>> import paddle >>> paddle.device.set_device('gpu') >>> max_memory_allocated_size = paddle.device.cuda.max_memory_allocated(paddle.CUDAPlace(0)) >>> max_memory_allocated_size = paddle.device.cuda.max_memory_allocated(0) >>> max_memory_allocated_size = paddle.device.cuda.max_memory_allocated("gpu:0") z'paddle.device.cuda.max_memory_allocatedThe API l is not supported in CPU-only PaddlePaddle. Please reinstall PaddlePaddle with GPU support to call this API.r, Allocatedrr#rr-Zdevice_memory_stat_peak_valuernamerr r r!r    r cCr.)a Return the peak size of GPU memory that is held by the allocator of the given device. Args: device(paddle.CUDAPlace or int or str, optional): The device, the id of the device or the string name of device like 'gpu:x'. If device is None, the device is the current device. Default: None. Return: int: The peak size of GPU memory that is held by the allocator of the given device, in bytes. Examples: .. code-block:: python >>> # doctest: +REQUIRES(env:GPU) >>> import paddle >>> paddle.device.set_device('gpu') >>> max_memory_reserved_size = paddle.device.cuda.max_memory_reserved(paddle.CUDAPlace(0)) >>> max_memory_reserved_size = paddle.device.cuda.max_memory_reserved(0) >>> max_memory_reserved_size = paddle.device.cuda.max_memory_reserved("gpu:0") z&paddle.device.cuda.max_memory_reservedr/r0r1Reservedr3r4r r r!r    r cCr.)a Return the current size of gpu memory that is allocated to tensor of the given device. Note: The size of GPU memory allocated to tensor is 256-byte aligned in Paddle, which may be larger than the memory size that tensor actually need. For instance, a float32 tensor with shape [1] in GPU will take up 256 bytes memory, even though storing a float32 data requires only 4 bytes. Args: device(paddle.CUDAPlace or int or str, optional): The device, the id of the device or the string name of device like 'gpu:x'. If device is None, the device is the current device. Default: None. Return: int: The current size of gpu memory that is allocated to tensor of the given device, in bytes. Examples: .. code-block:: python >>> # doctest: +REQUIRES(env:GPU) >>> import paddle >>> paddle.device.set_device('gpu') >>> memory_allocated_size = paddle.device.cuda.memory_allocated(paddle.CUDAPlace(0)) >>> memory_allocated_size = paddle.device.cuda.memory_allocated(0) >>> memory_allocated_size = paddle.device.cuda.memory_allocated("gpu:0") z#paddle.device.cuda.memory_allocatedr/r0r1r2rr#rr-Z device_memory_stat_current_valuer4r r r!rr6rcCr.)ap Return the current size of GPU memory that is held by the allocator of the given device. Args: device(paddle.CUDAPlace or int or str, optional): The device, the id of the device or the string name of device like 'gpu:x'. If device is None, the device is the current device. Default: None. Return: int: The current size of GPU memory that is held by the allocator of the given device, in bytes. Examples: .. code-block:: python >>> # doctest: +REQUIRES(env:GPU) >>> import paddle >>> paddle.device.set_device('gpu') >>> memory_reserved_size = paddle.device.cuda.memory_reserved(paddle.CUDAPlace(0)) >>> memory_reserved_size = paddle.device.cuda.memory_reserved(0) >>> memory_reserved_size = paddle.device.cuda.memory_reserved("gpu:0") z"paddle.device.cuda.memory_reservedr/r0r1r7r9r4r r r!rCr8rcCs<t|tjjjs tdt}t|t|kr|St |S)z Set the current stream. Parameters: stream(paddle.device.cuda.Stream): The selected stream. Returns: CUDAStream: The previous stream. /stream type should be paddle.device.cuda.Stream) rr"rcudar TypeErrorridr_set_current_stream)stream cur_streamr r r!r>cs   r>zpaddle.device.stream_guardz>> # doctest: +REQUIRES(env:GPU) >>> import paddle >>> paddle.device.set_device('gpu') >>> s = paddle.device.cuda.Stream() >>> data1 = paddle.ones(shape=[20]) >>> data2 = paddle.ones(shape=[20]) >>> with paddle.device.cuda.stream_guard(s): ... data3 = data1 + data2 Nr:) rr"rr;rr<rr=r>)r?r@Z pre_streamr r r!rxs  rcCststd|durBt|tr|}n0t|tjr|}n%t|tr:|dr2t|dd}ntd|dtd|dd }t |S) a Return the properties of given device. Args: device(paddle.CUDAPlace or int or str, optional): The device, the id of the device or the string name of device like 'gpu:x' which to get the properties of the device from. If device is None, the device is the current device. Default: None. Returns: _gpuDeviceProperties: The properties of the device which include ASCII string identifying device, major compute capability, minor compute capability, global memory available and the number of multiprocessors on the device. Examples: .. code-block:: python >>> # doctest: +REQUIRES(env:GPU) >>> import paddle >>> paddle.device.set_device('gpu') >>> paddle.device.cuda.get_device_properties() >>> # _gpuDeviceProperties(name='A100-SXM4-40GB', major=8, minor=0, total_memory=40536MB, multi_processor_count=108) >>> paddle.device.cuda.get_device_properties(0) >>> # _gpuDeviceProperties(name='A100-SXM4-40GB', major=8, minor=0, total_memory=40536MB, multi_processor_count=108) >>> paddle.device.cuda.get_device_properties('gpu:0') >>> # _gpuDeviceProperties(name='A100-SXM4-40GB', major=8, minor=0, total_memory=40536MB, multi_processor_count=108) >>> paddle.device.cuda.get_device_properties(paddle.CUDAPlace(0)) >>> # _gpuDeviceProperties(name='A100-SXM4-40GB', major=8, minor=0, total_memory=40536MB, multi_processor_count=108) zThe API paddle.device.cuda.get_device_properties is not supported in CPU-only PaddlePaddle. Please reinstall PaddlePaddle with GPU support to call this API.Nr&r'zThe current string z is not expected. Because paddle.device.cuda.get_device_properties only support string which is like 'gpu:x'. Please input appropriate string again!zThe device type z is not expected. Because paddle.device.cuda.get_device_properties only support int, str or paddle.CUDAPlace. Please input appropriate device again!r) rr#rrrrrr)r*rrr r r!rs(%        rcCs t|jS)a Return the name of the device which is got from CUDA function `cudaDeviceProp `_. Parameters: device(paddle.CUDAPlace|int, optional): The device or the ID of the device. If device is None (default), the device is the current device. Returns: str: The name of the device. Examples: .. code-block:: python >>> # doctest: +REQUIRES(env:GPU) >>> import paddle >>> paddle.device.set_device('gpu') >>> paddle.device.cuda.get_device_name() >>> paddle.device.cuda.get_device_name(0) >>> paddle.device.cuda.get_device_name(paddle.CUDAPlace(0)) )rr5)rr r r!rs rcCst|}|j|jfS)a Return the major and minor revision numbers defining the device's compute capability which are got from CUDA function `cudaDeviceProp `_. Parameters: device(paddle.CUDAPlace|int, optional): The device or the ID of the device. If device is None (default), the device is the current device. Returns: tuple(int,int): the major and minor revision numbers defining the device's compute capability. Examples: .. code-block:: python >>> # doctest: +REQUIRES(env:GPU) >>> import paddle >>> paddle.device.set_device('gpu') >>> paddle.device.cuda.get_device_capability() >>> paddle.device.cuda.get_device_capability(0) >>> paddle.device.cuda.get_device_capability(paddle.CUDAPlace(0)) )rmajorminor)rpropr r r!r s r)N)r"Z paddle.baserZpaddle.base.wrapped_decoratorrZ paddle.utilsrstreamsrr__all__rr r r r-r r rrr>rrrrr r r r!sN    ' % + $ $   ' F