reduce_scatter

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.

Function

Functions as the operation API of the ReduceScatter operator to evenly divide the input data of all ranks in a communicator into rank size parts, perform reduction (sum, prod, max, and min) on 1/rank size part of data of each rank, and distributes the result to the output buffer of each rank based on the number.

Prototype

def reduce_scatter(tensor, reduction, rank_size, group="hccl_world_group", fusion=0, fusion_id=-1)

Parameters

Parameter	Input/Output	Description
tensor	Input	TensorFlow tensor type. Atlas 350 Accelerator Card: The supported data types are int8, int16, int32, int64, uint64, float16, float32, float64, and bfp16. Data types int64, uint64, and float64 supports only intra-node communication. Atlas A3 training product/Atlas A3 inference product: The supported data types are int8, int16, int32, int64, float16, float32, and bfp16. Atlas A2 training product/Atlas A2 inference product: The supported data types are int8, int16, int32, int64, float16, float32, and bfp16. Note that the performance will deteriorate for the int64 data type. Atlas training product: The supported data types are int8, int32, int64, float16, and float32. Atlas 300I Duo Inference Card: The supported data types are int8, int16, int32, float16, and float32. Note that the size of the first dimension of a tensor must be an integer multiple of the rank size.
reduction	Input	Reduction operation, string type. Atlas 350 Accelerator Card: The supported operation types are sum, max, and min. Atlas A3 training product/Atlas A3 inference product: The supported operation types are sum, max, min, and prod. In the current version, the prod operation does not support the int16 or bfp16 data type. Atlas A2 training product/Atlas A2 inference product: The supported operation types are sum, max, min, and prod. In the current version, the prod operation does not support the int16 or bfp16 data type. Atlas 300I Duo Inference Card: The supported operation types are sum, max, min, and prod. In the current version, the max, min, and prod operations do not support the int16 data type.
rank_size	Input	Number of devices in a group, int type. Maximum value: 32768.
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.
fusion	Input	reducescatter operator fusion flag, int type. The value can be one of the following: 0: The ReduceScatter operator is not fused with other ReduceScatter operators during network compilation. 2: ReduceScatter operators with the same fusion_id are fused during network compilation.
fusion_id	Input	reducescatter operator fusion ID, int type. If fusion is set to 2, ReduceScatter operators with the same fusion_id are fused during network compilation.

Returns

The result tensor after the reducescatter operation is performed on the input 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 input tensor size must be less than or equal to 8 GB.
For the reducescatter operator fusion, only the reduction type sum is supported.

Example

from npu_bridge.hccl import hccl_ops
tensor = tf.random_uniform((2, 3), minval=1, maxval=10, dtype=tf.float32)
rank_size = 2
result = hccl_ops.reduce_scatter(tensor, "sum", rank_size)

Parent topic: npu_bridge.hccl.hccl_ops