mindspore.ops.communication.barrier

View Source On AtomGit
mindspore.ops.communication.barrier(group=None, async_op=False, device_ids=None)[source]

Synchronizes all processes in the specified group. Once the process call this operation, it will be blocked until all processes call this operation. After all processes finish calling the operations, the blocked processes will be woken and continue their task.

Parameters

  • group (str, optional) – The communication group to work on. Default: None, which means "hccl_world_group" in Ascend.

  • async_op (bool, optional) – Whether this operator should be an async operator. Default: False.

  • device_ids (list[int], optional) – Currently It is a reserved Parameter. Default is None.

Returns

CommHandle. If async_op is set to True, CommHandle is an async work handle. If async_op is set to False, CommHandle will be None.

Raises

  • TypeError – group is not a str or async_op is not a bool.

  • RuntimeError – If backend is invalid, or distributed initialization fails.

Supported Platforms:

Ascend CPU

Examples

Note

Before running the following examples, you need to configure the communication environment variables.

For Ascend devices, it is recommended to use the msrun startup method without any third-party or configuration file dependencies. Please see the msrun startup for more details.

This example should be run with 2 devices.

>>> from mindspore.ops.communication import init_process_group
>>> from mindspore.communication.comm_func import barrier
>>> # Launch 2 processes.
>>> init_process_group()
>>> barrier()
>>> print("barrier finish!")
barrier finish!