mpi_ops.py

# Copyright 2018 Uber Technologies, Inc. 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 absolute_import
from __future__ import division
from __future__ import print_function

# Load all the necessary PyTorch C types.
import torch

from horovod.torch import mpi_lib_impl
from horovod.torch import mpi_lib
from horovod.torch import rank, size


# Schema: handle -> input, output
# We keep input in order to make sure it does not get garbage collected
# before the operation is finished.
_handle_map = {}


# Null pointer.
_NULL = mpi_lib._ffi.NULL


def _check_function(function_factory, tensor):
    function = function_factory(tensor)
    if not hasattr(mpi_lib, function):
        raise ValueError('Tensor type %s is not supported.' % tensor.type())
    if not tensor.is_contiguous():
        raise ValueError('Tensor is required to be contiguous.')
    return function


def _allreduce_function_factory(tensor):
    return 'horovod_torch_allreduce_async_' + tensor.type().replace('.', '_')


def _allreduce_async(tensor, output, average, name):
    function = _check_function(_allreduce_function_factory, tensor)
    handle = getattr(mpi_lib, function)(tensor, output, average,
                                        name.encode() if name is not None else _NULL)
    _handle_map[handle] = (tensor, output)
    return handle


def allreduce_async(tensor, average=True, name=None):
    """
    A function that performs asynchronous averaging or summation of the input tensor
    over all the Horovod processes. The input tensor is not modified.

    The reduction operation is keyed by the name. If name is not provided, an incremented
    auto-generated name is used. The tensor type and shape must be the same on all
    Horovod processes for a given name. The reduction will not start until all processes
    are ready to send and receive the tensor.

    Arguments:
        tensor: A tensor to average and sum.
        average: A flag indicating whether to compute average or summation,
                 defaults to average.
        name: A name of the reduction operation.

    Returns:
        A handle to the allreduce operation that can be used with `poll()` or
        `synchronize()`.
    """
    output = tensor.new(tensor.shape)
    return _allreduce_async(tensor, output, average, name)


class HorovodAllreduce(torch.autograd.Function):
    """An autograd function that performs allreduce on a tensor."""

    @staticmethod
    def forward(ctx, tensor, average, name):
        ctx.average = average
        handle = allreduce_async(tensor, average, name)
        return synchronize(handle)

    @staticmethod
    def backward(ctx, grad_output):
        return allreduce(grad_output, ctx.average), None, None


def allreduce(tensor, average=True, name=None):
    """
    A function that performs averaging or summation of the input tensor over all the
    Horovod processes. The input tensor is not modified.

    The reduction operation is keyed by the name. If name is not provided, an incremented
    auto-generated name is used. The tensor type and shape must be the same on all
    Horovod processes for a given name. The reduction will not start until all processes
    are ready to send and receive the tensor.

    This acts as a thin wrapper around an autograd function.  If your input
    tensor requires gradients, then callings this function will allow gradients
    to be computed and backpropagated.

    Arguments:
        tensor: A tensor to average and sum.
        average: A flag indicating whether to compute average or summation,
                 defaults to average.
        name: A name of the reduction operation.

    Returns:
        A tensor of the same shape and type as `tensor`, averaged or summed across all
        processes.
    """
    return HorovodAllreduce.apply(tensor, average, name)


def allreduce_async_(tensor, average=True, name=None):
    """
    A function that performs asynchronous in-place averaging or summation of the input
    tensor over all the Horovod processes.

    The reduction operation is keyed by the name. If name is not provided, an incremented
    auto-generated name is used. The tensor type and shape must be the same on all
    Horovod processes for a given name. The reduction will not start until all processes
    are ready to send and receive the tensor.

    Arguments:
        tensor: A tensor to average and sum.
        average: A flag indicating whether to compute average or summation,
                 defaults to average.
        name: A name of the reduction operation.

    Returns:
        A handle to the allreduce operation that can be used with `poll()` or
        `synchronize()`.
    """
    return _allreduce_async(tensor, tensor, average, name)


def allreduce_(tensor, average=True, name=None):
    """
    A function that performs in-place averaging or summation of the input tensor over
    all the Horovod processes.

    The reduction operation is keyed by the name. If name is not provided, an incremented
    auto-generated name is used. The tensor type and shape must be the same on all
    Horovod processes for a given name. The reduction will not start until all processes
    are ready to send and receive the tensor.

    Arguments:
        tensor: A tensor to average and sum.
        average: A flag indicating whether to compute average or summation,
                 defaults to average.
        name: A name of the reduction operation.

    Returns:
        A tensor of the same shape and type as `tensor`, averaged or summed across all
        processes.
    """
    handle = allreduce_async_(tensor, average, name)
    return synchronize(handle)


def _allgather_function_factory(tensor):
    return 'horovod_torch_allgather_async_' + tensor.type().replace('.', '_')


def _allgather_async(tensor, output, name):
    function = _check_function(_allgather_function_factory, tensor)
    handle = getattr(mpi_lib, function)(
        tensor, output, name.encode() if name is not None else _NULL)
    _handle_map[handle] = (tensor, output)
    return handle


def allgather_async(tensor, name=None):
    """
    A function that asynchronously concatenates the input tensor with the same input
    tensor on all other Horovod processes. The input tensor is not modified.

    The concatenation is done on the first dimension, so the input tensors on the
    different processes must have the same rank and shape, except for the first
    dimension, which is allowed to be different.

    Arguments:
        tensor: A tensor to allgather.
        name: A name of the allgather operation.

    Returns:
        A handle to the allgather operation that can be used with `poll()` or
        `synchronize()`.
    """
    output = tensor.new()
    return _allgather_async(tensor, output, name)


class HorovodAllgather(torch.autograd.Function):
    """An autograd function that performs allgather on a tensor."""

    @staticmethod
    def forward(ctx, tensor, name):
        ctx.dim = tensor.shape[0]
        handle = allgather_async(tensor, name)
        return synchronize(handle)

    @staticmethod
    def backward(ctx, grad_output):
        grad_reduced = allreduce(grad_output, average=False)

        dim_t = torch.IntTensor([ctx.dim])
        dim = allgather(dim_t).view(size())

        r = rank()
        offset = torch.sum(dim.narrow(0, 0, r)).data[0] if r != 0 else 0
        return grad_reduced.narrow(0, offset, ctx.dim), None


def allgather(tensor, name=None):
    """
    A function that concatenates the input tensor with the same input tensor on
    all other Horovod processes. The input tensor is not modified.

    The concatenation is done on the first dimension, so the input tensors on the
    different processes must have the same rank and shape, except for the first
    dimension, which is allowed to be different.

    This acts as a thin wrapper around an autograd function.  If your input
    tensor requires gradients, then callings this function will allow gradients
    to be computed and backpropagated.

    Arguments:
        tensor: A tensor to allgather.
        name: A name of the allgather operation.

    Returns:
        A tensor of the same type as `tensor`, concatenated on dimension zero
        across all processes. The shape is identical to the input shape, except for
        the first dimension, which may be greater and is the sum of all first
        dimensions of the tensors in different Horovod processes.
    """
    return HorovodAllgather.apply(tensor, name)


def _broadcast_function_factory(tensor):
    return 'horovod_torch_broadcast_async_' + tensor.type().replace('.', '_')


def _broadcast_async(tensor, output, root_rank, name):
    function = _check_function(_broadcast_function_factory, tensor)
    handle = getattr(mpi_lib, function)(
        tensor, output, root_rank, name.encode() if name is not None else _NULL)
    _handle_map[handle] = (tensor, output)
    return handle


def broadcast_async(tensor, root_rank, name=None):
    """
    A function that asynchronously broadcasts the input tensor on root rank to the same
    input tensor on all other Horovod processes. The input tensor is not modified.

    The broadcast operation is keyed by the name. If name is not provided, an incremented
    auto-generated name is used. The tensor type and shape must be the same on all
    Horovod processes for a given name. The broadcast will not start until all processes
    are ready to send and receive the tensor.

    Arguments:
        tensor: A tensor to broadcast.
        root_rank: The rank to broadcast the value from.
        name: A name of the broadcast operation.

    Returns:
        A handle to the broadcast operation that can be used with `poll()` or
        `synchronize()`.
    """
    output = tensor.new(tensor.shape)
    return _broadcast_async(tensor, output, root_rank, name)


class HorovodBroadcast(torch.autograd.Function):
    """An autograd function that broadcasts a tensor."""

    @staticmethod
    def forward(ctx, tensor, root_rank, name):
        ctx.root_rank = root_rank
        handle = broadcast_async(tensor, root_rank, name)
        return synchronize(handle)

    @staticmethod
    def backward(ctx, grad_output):
        grad_reduced = allreduce(grad_output, average=False)
        if rank() != ctx.root_rank:
            grad_reduced *= 0
        return grad_reduced, None, None


def broadcast(tensor, root_rank, name=None):
    """
    A function that broadcasts the input tensor on root rank to the same input tensor
    on all other Horovod processes. The input tensor is not modified.

    The broadcast operation is keyed by the name. If name is not provided, an incremented
    auto-generated name is used. The tensor type and shape must be the same on all
    Horovod processes for a given name. The broadcast will not start until all processes
    are ready to send and receive the tensor.

    This acts as a thin wrapper around an autograd function.  If your input
    tensor requires gradients, then callings this function will allow gradients
    to be computed and backpropagated.

    Arguments:
        tensor: A tensor to broadcast.
        root_rank: The rank to broadcast the value from.
        name: A name of the broadcast operation.

    Returns:
        A tensor of the same shape and type as `tensor`, with the value broadcasted
        from root rank.
    """
    return HorovodBroadcast.apply(tensor, root_rank, name)


def broadcast_async_(tensor, root_rank, name=None):
    """
    A function that asynchronously broadcasts the input tensor on root rank to the same
    input tensor on all other Horovod processes. The operation is performed in-place.

    The broadcast operation is keyed by the name. If name is not provided, an incremented
    auto-generated name is used. The tensor type and shape must be the same on all
    Horovod processes for a given name. The broadcast will not start until all processes
    are ready to send and receive the tensor.

    Arguments:
        tensor: A tensor to broadcast.
        root_rank: The rank to broadcast the value from.
        name: A name of the broadcast operation.

    Returns:
        A handle to the broadcast operation that can be used with `poll()` or
        `synchronize()`.
    """
    return _broadcast_async(tensor, tensor, root_rank, name)


def broadcast_(tensor, root_rank, name=None):
    """
    A function that broadcasts the input tensor on root rank to the same input tensor
    on all other Horovod processes. The operation is performed in-place.

    The broadcast operation is keyed by the name. If name is not provided, an incremented
    auto-generated name is used. The tensor type and shape must be the same on all
    Horovod processes for a given name. The broadcast will not start until all processes
    are ready to send and receive the tensor.

    Arguments:
        tensor: A tensor to broadcast.
        root_rank: The rank to broadcast the value from.
        name: A name of the broadcast operation.

    Returns:
        A tensor of the same shape and type as `tensor`, with the value broadcasted
        from root rank.
    """
    handle = broadcast_async_(tensor, root_rank, name)
    return synchronize(handle)


def poll(handle):
    """
    Polls an allreduce, allgather or broadcast handle to determine whether underlying
    asynchronous operation has completed. After `poll()` returns `True`, `synchronize()`
    will return without blocking.

    Arguments:
        handle: A handle returned by an allreduce, allgather or broadcast asynchronous
                operation.

    Returns:
        A flag indicating whether the operation has completed.
    """
    return mpi_lib.horovod_torch_poll(handle) != 0


def synchronize(handle):
    """
    Synchronizes an asynchronous allreduce, allgather or broadcast operation until
    it's completed. Returns the result of the operation.

    Arguments:
        handle: A handle returned by an allreduce, allgather or broadcast asynchronous
                operation.

    Returns:
        An output tensor of the operation.
    """
    if handle not in _handle_map:
        return
    mpi_lib.horovod_torch_wait_and_clear(handle)
    _, output = _handle_map.pop(handle)
    return output