paddlepaddle--paddle/python/paddle/distributed/communication/reduce_scatter.py

# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

from __future__ import annotations

from typing import TYPE_CHECKING

import paddle
from paddle.distributed.communication import stream
from paddle.distributed.communication.reduce import ReduceOp
from paddle.distributed.communication.stream.reduce_scatter import (
    _reduce_scatter_base as _reduce_scatter_base_stream,
)

if TYPE_CHECKING:
    from paddle import Tensor
    from paddle.base.core import task
    from paddle.distributed.communication.group import Group
    from paddle.distributed.communication.reduce import _ReduceOp


def reduce_scatter(
    tensor: Tensor,
    tensor_list: list[Tensor],
    op: _ReduceOp = ReduceOp.SUM,
    group: Group | None = None,
    sync_op: bool = True,
) -> task:
    """
    Reduces, then scatters a list of tensors to all processes in a group

    Args:
        tensor (Tensor): The output tensor on each rank. The result will overwrite this tenor after communication. Support
            float16, float32, float64, int32, int64, int8, uint8 or bool as the input data type.
        tensor_list (List[Tensor]]): List of tensors to reduce and scatter. Every element in the list must be a Tensor whose data type
            should be float16, float32, float64, int32, int64, int8, uint8, bool or bfloat16.
        op (ReduceOp.SUM|ReduceOp.MAX|ReduceOp.MIN|ReduceOp.PROD|ReduceOp.AVG, optional): The reduction used. If none is given, use ReduceOp.SUM as default.
        group (Group, optional): Communicate in which group. If none is given, use the global group as default.
        sync_op (bool, optional): Indicate whether the communication is sync or not. If none is given, use true as default.

    Returns:
        Return a task object.

    Warning:
        This API only supports the dygraph mode.


    Examples:
        .. code-block:: pycon

            >>> # doctest: +REQUIRES(env: DISTRIBUTED)
            >>> import paddle
            >>> import paddle.distributed as dist

            >>> dist.init_parallel_env()
            >>> if dist.get_rank() == 0:
            ...     data1 = paddle.to_tensor([0, 1])
            ...     data2 = paddle.to_tensor([2, 3])
            >>> else:
            ...     data1 = paddle.to_tensor([4, 5])
            ...     data2 = paddle.to_tensor([6, 7])
            >>> dist.reduce_scatter(data1, [data1, data2])
            >>> print(data1)
            >>> # [4, 6] (2 GPUs, out for rank 0)
            >>> # [8, 10] (2 GPUs, out for rank 1)

    """
    if op not in [
        ReduceOp.AVG,
        ReduceOp.MAX,
        ReduceOp.MIN,
        ReduceOp.PROD,
        ReduceOp.SUM,
    ]:
        raise RuntimeError(
            "Invalid ``op`` function. Expected ``op`` to be of type ``ReduceOp.SUM``, ``ReduceOp.Max``, ``ReduceOp.MIN``, ``ReduceOp.PROD`` or ``ReduceOp.AVG``."
        )
    # AVG is only supported when nccl >= 2.10
    if op == ReduceOp.AVG and paddle.base.core.nccl_version() < 21000:
        group = (
            paddle.distributed.collective._get_global_group()
            if group is None
            else group
        )
        tensor.scale_(1.0 / group.nranks)
        return stream.reduce_scatter(
            tensor,
            tensor_list,
            op=ReduceOp.SUM,
            group=group,
            sync_op=sync_op,
            use_calc_stream=False,
        )
    return stream.reduce_scatter(
        tensor,
        tensor_list,
        op=op,
        group=group,
        sync_op=sync_op,
        use_calc_stream=False,
    )


def _reduce_scatter_base(
    output: Tensor,
    input: Tensor,
    op: _ReduceOp = ReduceOp.SUM,
    group: Group | None = None,
    sync_op: bool = True,
) -> task | None:
    """
    Reduces, then scatters a flattened tensor to all processes in a group.

    Args:
        output (Tensor): Output tensor. Its data type should be float16, float32, float64, int32, int64, int8, uint8, bool or bfloat16.
        input (Tensor): Input tensor that is of size output tensor size times world size. Its data type
            should be float16, float32, float64, int32, int64, int8, uint8, bool or bfloat16.
        op (ReduceOp.SUM|ReduceOp.MAX|ReduceOp.MIN|ReduceOp.PROD): Optional. The operation used. Default: ReduceOp.SUM.
        group (ProcessGroup, optional): The process group to work on. If None,
            the default process group will be used.
        sync_op (bool, optional): Whether this op is a sync op. The default value is True.

    Returns:
        Async task handle, if sync_op is set to False.
        None, if sync_op or if not part of the group.

    Examples:
        .. code-block:: pycon

            >>> # doctest: +REQUIRES(env: DISTRIBUTED)
            >>> import paddle
            >>> import paddle.distributed as dist

            >>> dist.init_parallel_env()
            >>> rank = dist.get_rank()
            >>> data = paddle.arange(4) + rank
            >>> # [0, 1, 2, 3] (2 GPUs, for rank 0)
            >>> # [1, 2, 3, 4] (2 GPUs, for rank 1)
            >>> output = paddle.empty(shape=[2], dtype=data.dtype)
            >>> dist.collective._reduce_scatter_base(output, data)
            >>> print(output)
            >>> # [1, 3] (2 GPUs, out for rank 0)
            >>> # [5, 7] (2 GPUs, out for rank 1)

    """
    if op not in [ReduceOp.MAX, ReduceOp.MIN, ReduceOp.PROD, ReduceOp.SUM]:
        raise RuntimeError(
            "Invalid ``op`` function. Expected ``op`` to be of type ``ReduceOp.SUM``, ``ReduceOp.Max``, ``ReduceOp.MIN`` or ``ReduceOp.PROD``."
        )
    return _reduce_scatter_base_stream(
        output,
        input,
        op=op,
        group=group,
        sync_op=sync_op,
        use_calc_stream=False,
    )