receive

Applicability

Product

Supported (√/x)

Atlas 350 Accelerator Card

Atlas A3 training product/Atlas A3 inference product

Atlas A2 training product/Atlas A2 inference product

Atlas 200I/500 A2 inference product

Atlas inference product

Atlas training product

For the Atlas inference products, only the Atlas 300I Duo inference card is supported.

Description

Receives data from a rank within a collective communication group.

Prototype

1
def receive(shape, data_type, sr_tag, src_rank, group="hccl_world_group")

Parameters

Parameter

Input/Output

Description

shape

Input

Shape of the received tensor.

data_type

Input

Data type of the received data.

For the Atlas 350 Accelerator Card, the supported data types are int8, uint8, int16, uint16, int32, uint32, int64, uint64, float16, float32, float64, and bfp16.

Atlas A3 training product/Atlas A3 inference product: The supported data types are int8, uint8, int16, uint16, int32, uint32, int64, uint64, float16, float32, float64, and bfp16.

Atlas A2 training product/Atlas A2 inference product: The supported data types are int8, uint8, int16, uint16, int32, uint32, int64, uint64, float16, float32, float64, and bfp16.

For the Atlas training product, the supported data types are int8, uint8, int16, uint16, int32, uint32, int64, uint64, float16, float32, and float64.

Atlas 300I Duo Inference Card: The supported data types are int8, uint8, int16, uint16, int32, uint32, int64, uint64, float16, float32, and float64.

sr_tag

Input

Message tag. The send/recv pairs with the same sr_tag can receive and send data, int type.

src_rank

Input

Source rank of the received data. This rank indicates the rank ID in the group, int type.

group

Input

A string containing a maximum of 128 bytes, including the end character.

Group name, which can be a user-defined value or hccl_world_group.

Returns

The result tensor.

Restrictions

  • The caller rank must be within the range defined by the group argument passed to this API call. Otherwise, the API call fails.
  • The send and receive APIs must be used in pairs. That is, after the send API is called, the next API can be called only after the paired receive API receives data.

Example

1
2
3
4
5
from npu_bridge.hccl import hccl_ops
tensor = tf.random_uniform((1, 3), minval=1, maxval=10, dtype=tf.float32)
sr_tag = 0
src_rank = 0
tensor = hccl_ops.receive(tensor.shape, tensor.dtype, sr_tag, src_rank)