U d3@sddlmZmZmZmZddlZddlmZddlm Z ee e ej fZeej dddZeeefeej ej fd d d Zeeeej d d dZGdddeZdS))DictListOptionalUnionN) _TensorPipeRpcBackendOptionsBase) constants)devicereturncCs*t|}|jdkr&td|jd|S)NZcudazA`set_devices` expect a list of CUDA devices, but got device type .)torchr type ValueError)r rA/tmp/pip-unpacked-wheel-ua33x9lu/torch/distributed/rpc/options.py _to_device s    r) device_mapr cCsji}i}|D]T\}}t|t|}}||krTtd|d||d||||<|||<q|S)Nz9`device_map` only supports 1-to-1 mapping, trying to map  and  to )itemsr r r)rfull_device_mapZ reverse_mapkvrrr_to_device_maps r)devicesr cCsttt|S)N)listmaprrrrr_to_device_list&src seZdZdZejejejdddddee e e e e e e e ffe ee e ee edfddZe e e e fdfdd Zee d d d ZZS) TensorPipeRpcBackendOptionsa' The backend options for :class:`~torch.distributed.rpc.TensorPipeAgent`, derived from :class:`~torch.distributed.rpc.RpcBackendOptions`. Args: num_worker_threads (int, optional): The number of threads in the thread-pool used by :class:`~torch.distributed.rpc.TensorPipeAgent` to execute requests (default: 16). rpc_timeout (float, optional): The default timeout, in seconds, for RPC requests (default: 60 seconds). If the RPC has not completed in this timeframe, an exception indicating so will be raised. Callers can override this timeout for individual RPCs in :meth:`~torch.distributed.rpc.rpc_sync` and :meth:`~torch.distributed.rpc.rpc_async` if necessary. init_method (str, optional): The URL to initialize the distributed store used for rendezvous. It takes any value accepted for the same argument of :meth:`~torch.distributed.init_process_group` (default: ``env://``). device_maps (Dict[str, Dict], optional): Device placement mappings from this worker to the callee. Key is the callee worker name and value the dictionary (``Dict`` of ``int``, ``str``, or ``torch.device``) that maps this worker's devices to the callee worker's devices. (default: ``None``) devices (List[int, str, or ``torch.device``], optional): all local CUDA devices used by RPC agent. By Default, it will be initialized to all local devices from its own ``device_maps`` and corresponding devices from its peers' ``device_maps``. When processing CUDA RPC requests, the agent will properly synchronize CUDA streams for all devices in this ``List``. N)num_worker_threads rpc_timeout init_method device_mapsr _transports _channelsc sN|dkr indd|D}|dkr*gnt|} t||||||| dS)NcSsi|]\}}|t|qSr)r).0rrrrr Zsz8TensorPipeRpcBackendOptions.__init__..)rrsuper__init__) selfr r!r"r#rr$r%Zfull_device_mapsZfull_device_list __class__rrr)Ls z$TensorPipeRpcBackendOptions.__init__)torc szt|}tj}||krh|D]F\}}|||kr ||||kr td|d|d|||q t||dS)a  Set device mapping between each RPC caller and callee pair. This function can be called multiple times to incrementally add device placement configurations. Args: worker_name (str): Callee name. device_map (Dict of int, str, or torch.device): Device placement mappings from this worker to the callee. This map must be invertible. Example:: >>> # both workers >>> def add(x, y): >>> print(x) # tensor([1., 1.], device='cuda:1') >>> return x + y, (x + y).to(2) >>> >>> # on worker 0 >>> options = TensorPipeRpcBackendOptions( >>> num_worker_threads=8, >>> device_maps={"worker1": {0: 1}} >>> # maps worker0's cuda:0 to worker1's cuda:1 >>> ) >>> options.set_device_map("worker1", {1: 2}) >>> # maps worker0's cuda:1 to worker1's cuda:2 >>> >>> rpc.init_rpc( >>> "worker0", >>> rank=0, >>> world_size=2, >>> backend=rpc.BackendType.TENSORPIPE, >>> rpc_backend_options=options >>> ) >>> >>> x = torch.ones(2) >>> rets = rpc.rpc_sync("worker1", add, args=(x.to(0), 1)) >>> # The first argument will be moved to cuda:1 on worker1. When >>> # sending the return value back, it will follow the invert of >>> # the device map, and hence will be moved back to cuda:0 and >>> # cuda:1 on worker0 >>> print(rets[0]) # tensor([2., 2.], device='cuda:0') >>> print(rets[1]) # tensor([2., 2.], device='cuda:1') z=`set_device_map` only supports 1-to-1 mapping, trying to map rrN)rr(r#rrZ_set_device_map)r*r-rrZcurr_device_mapsrrr+rrset_device_mapgs,z*TensorPipeRpcBackendOptions.set_device_maprcCst||_dS)ab Set local devices used by the TensorPipe RPC agent. When processing CUDA RPC requests, the TensorPipe RPC agent will properly synchronize CUDA streams for all devices in this ``List``. Args: devices (List of int, str, or torch.device): local devices used by the TensorPipe RPC agent. N)rr)r*rrrr set_devicess z'TensorPipeRpcBackendOptions.set_devices)__name__ __module__ __qualname____doc__ rpc_contantsZDEFAULT_NUM_WORKER_THREADSZDEFAULT_RPC_TIMEOUT_SECZDEFAULT_INIT_METHODintfloatstrrr DeviceTyperr)r.r/ __classcell__rrr+rr*s&$ 9r)typingrrrrr Ztorch._C._distributed_rpcrrr4r5r7r r8rrrrrrrrs