mindspore.communication.create_group

mindspore.communication.create_group(group, rank_ids, options=None)[source]

Create a user collective communication group.

Note

This method isn't supported in GPU and CPU versions of MindSpore. The size of rank_ids should be larger than 1, rank_ids should not have duplicate data. This method should be used after init(). Only support global single communication group in PyNative mode if you do not start with mpirun.

Parameters

group (str) – The name of the communication group to be created.
rank_ids (list) – A list of device IDs.
options (GroupOptions, optional) –
Additional communication group configuration parameters. The backend will automatically select supported parameters and apply them during group initialization. i.e. for the HCCL backend, hccl_config can be specified so that group initialization configurations can be applied. Default is None.

GroupOptions is defined as a class that can be instantiated as a python object.
```
GroupOptions {
    hccl_config(dict)
}
```

Raises

TypeError – If group is not a string or rank_ids is not a list.
ValueError – If rank_ids size is not larger than 1, or rank_ids has duplicate data, or backend is invalid.
RuntimeError – If HCCL is not available or MindSpore is GPU/CPU version.

Supported Platforms:: Ascend GPU CPU

Examples

Note

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

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

>>> import mindspore as ms
>>> from mindspore import set_context, ops
>>> from mindspore._c_expression import GroupOptions
>>> from mindspore.communication import init, create_group, get_rank
>>> set_context(mode=ms.GRAPH_MODE)
>>> ms.set_device(device_target="Ascend")
>>> init()
>>> group = "0-7"
>>> rank_ids = [0,7]
>>> options = GroupOptions()
>>> options.hccl_config = {"hccl_buffer_size": 400}
>>> if get_rank() in rank_ids:
...     create_group(group, rank_ids, options)
...     allreduce = ops.AllReduce(group)