mindspore-Ecosystem
/
mindspore
21
0
Fork 27
Code Issues Pull Requests 6 Projects Releases Wiki Activity

!28712 update alltoall op api 

Merge pull request !28712 from zhoufeng/code_docs_alltoall
This commit is contained in:
i-robot 2022-01-10 06:18:07 +00:00 committed by Gitee
parent 0a52e1723c 342667ef5c
commit 8b29b805ce
No known key found for this signature in database
GPG Key ID: 173E9B9CA92EEF8F
1 changed files with 20 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
mindspore/python/mindspore/ops/operations/comm_ops.py
View File

@ -48,7 +48,7 @@ class ReduceOp:
For more, refer to example. This needs to run in an environment with multiple graphics cards.
The user needs to preset
communication environment variables before running the following example, please check the details on the
official websit of `MindSpore \
official website of `MindSpore \
<https://www.mindspore.cn/docs/api/zh-CN/master/api_python/mindspore.ops.html#communication-operators>`_.
Supported Platforms:
@ -103,7 +103,7 @@ class AllReduce(PrimitiveWithInfer):
The operation of AllReduce does not support "prod" currently.
The tensors must have the same shape and format in all processes of the collection. The user needs to preset
communication environment variables before running the following example, please check the details on the
official websit of `MindSpore \
official website of `MindSpore \
<https://www.mindspore.cn/docs/api/zh-CN/master/api_python/mindspore.ops.html#communication-operators>`_.
Args:
@ -181,7 +181,7 @@ class AllGather(PrimitiveWithInfer):
Note:
The tensors must have the same shape and format in all processes of the collection. The user needs to preset
communication environment variables before running the following example, please check the details on the
official websit of `MindSpore \
official website of `MindSpore \
<https://www.mindspore.cn/docs/api/zh-CN/master/api_python/mindspore.ops.html#communication-operators>`_.
Args:
@ -384,7 +384,7 @@ class ReduceScatter(PrimitiveWithInfer):
The back propagation of the op is not supported yet. Stay tuned for more.
The tensors must have the same shape and format in all processes of the collection. The user needs to preset
communication environment variables before running the following example, please check the details on the
official websit of `MindSpore \
official website of `MindSpore \
<https://www.mindspore.cn/docs/api/zh-CN/master/api_python/mindspore.ops.html#communication-operators>`_.
Args:
@ -517,7 +517,7 @@ class Broadcast(PrimitiveWithInfer):
Note:
The tensors must have the same shape and format in all processes of the collection. The user needs to preset
communication environment variables before running the following example, please check the details on the
official websit of `MindSpore \
official website of `MindSpore \
<https://www.mindspore.cn/docs/api/zh-CN/master/api_python/mindspore.ops.html#communication-operators>`_.
Args:
@ -651,9 +651,13 @@ class NeighborExchange(Primitive):
Note:
The user needs to preset
communication environment variables before running the following example, please check the details on the
official websit of `MindSpore \
official website of `MindSpore \
<https://www.mindspore.cn/docs/api/zh-CN/master/api_python/mindspore.ops.html#communication-operators>`_.
This operator requires a full-mesh network topology, each device has the same vlan id, and the ip & mask are
in the same subnet, please check the details on the official website of `MindSpore \
<https://www.mindspore.cn/docs/programming_guide/zh-CN/master/distributed_training_ops.html#id2>`_.
Args:
send_rank_ids (list(int)): Ranks which the data is sent to.
recv_rank_ids (list(int)): Ranks which the data is received from.
@ -723,9 +727,13 @@ class AlltoAll(PrimitiveWithInfer):
Note:
The tensors must have the same shape and format in all processes of the collection. The user needs to preset
communication environment variables before running the following example, please check the details on the
official websit of `MindSpore \
official website of `MindSpore \
<https://www.mindspore.cn/docs/api/zh-CN/master/api_python/mindspore.ops.html#communication-operators>`_.
This operator requires a full-mesh network topology, each device has the same vlan id, and the ip & mask are
in the same subnet, please check the details on the official website of `MindSpore \
<https://www.mindspore.cn/docs/programming_guide/zh-CN/master/distributed_training_ops.html#id2>`_.
Args:
split_count (int): On each process, divide blocks into split_count number.
split_dim (int): On each process, split blocks along the split_dim.
@ -809,9 +817,13 @@ class NeighborExchangeV2(Primitive):
Note:
The user needs to preset
communication environment variables before running the following example, please check the details on the
official websit of `MindSpore \
official website of `MindSpore \
<https://www.mindspore.cn/docs/api/zh-CN/master/api_python/mindspore.ops.html#communication-operators>`_.
This operator requires a full-mesh network topology, each device has the same vlan id, and the ip & mask are
in the same subnet, please check the details on the official website of `MindSpore \
<https://www.mindspore.cn/docs/programming_guide/zh-CN/master/distributed_training_ops.html#id2>`_.
Args:
send_rank_ids (list(int)): Ranks which the data is sent to. 8 rank_ids represents 8 directions, if one
direction is not send to , set it -1.