parallel.py 9.0 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14
# Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except jin 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.
import os
15
import six
Y
Yan Xu 已提交
16
import numpy as np
17
from collections import OrderedDict
18
from .. import core
19
from . import layers
C
chengduo 已提交
20
from . import parallel_helper
21 22
from .. import framework
from ..layers import collective
J
Jiabin Yang 已提交
23
from . import to_variable, no_grad
24 25 26 27 28 29

__all__ = ["prepare_context"]

ParallelStrategy = core.ParallelStrategy


C
chengduo 已提交
30 31 32 33 34 35 36 37 38
def prepare_context(strategy=None):
    if strategy is None:
        strategy = ParallelStrategy()
        strategy.nranks = Env().nranks
        strategy.local_rank = Env().local_rank
        strategy.trainer_endpoints = Env().trainer_endpoints
        strategy.current_endpoint = Env().current_endpoint
    if strategy.nranks < 2:
        return
39
    assert framework.in_dygraph_mode() is True, \
C
chengduo 已提交
40
        "dygraph.parallel.prepare_context should be used with dygrahp mode."
41
    place = framework._current_expected_place()
C
chengduo 已提交
42 43
    assert place is not None, \
        "dygraph.parallel.prepare_context should be used in fluid.dygraph.guard(place) guard."
44
    if isinstance(place, core.CUDAPlace):
C
chengduo 已提交
45 46
        parallel_helper._set_parallel_ctx(
            core.NCCLParallelContext(strategy, place))
47 48 49
    else:
        # TODO(Yancey1989): add Gloo Parallel Context to support CPU parallel computation
        assert ("Only support CUDAPlace for now.")
C
chengduo 已提交
50 51
    parallel_helper._init_parallel_ctx()
    return strategy
52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77


class Env(object):
    def __init__(self):
        self._nranks = int(os.getenv("PADDLE_TRAINERS_NUM", "1"))
        self._local_rank = int(os.getenv("PADDLE_TRAINER_ID", "0"))
        self._dev_id = int(os.getenv("FLAGS_selected_gpus", "0"))
        self._trainer_endpoints = os.getenv("PADDLE_TRAINER_ENDPOINTS",
                                            "").split(",")
        self._current_endpoint = os.getenv("PADDLE_CURRENT_ENDPOINT", "")

    @property
    def nranks(self):
        return self._nranks

    @property
    def local_rank(self):
        return self._local_rank

    @property
    def dev_id(self):
        return self._dev_id

    @property
    def current_endpoint(self):
        return self._current_endpoint
78 79 80 81 82 83 84

    @property
    def trainer_endpoints(self):
        return self._trainer_endpoints


class DataParallel(layers.Layer):
C
chengduo 已提交
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139
    """
    Runs the module with data parallelism.

    Currently, DataParallel only supports to run the dynamic graph
    with multi-process. The usage is:
    `python -m paddle.distributed.launch --gpus 2 dynamic_graph_test.py`.
    And the content of `dynamic_graph_test.py` is the code of examples.

    Examples:
        .. code-block:: python

           import numpy as np
           import paddle.fluid as fluid
           import paddle.fluid.dygraph as dygraph
           from paddle.fluid.optimizer import AdamOptimizer
           from paddle.fluid.dygraph.nn import FC
           from paddle.fluid.dygraph.base import to_variable

           place = fluid.CUDAPlace(0)
           with fluid.dygraph.guard(place=place):

               # prepare the data parallel context
               strategy=dygraph.parallel.prepare_context()

               fc_layer = FC("FC", 10, act="softmax")
               adam = fluid.optimizer.AdamOptimizer()

               # make the module become the data parallelism module
               fc_layer = dygraph.parallel.DataParallel(fc_layer, strategy)

               x_data = np.random.random(size=[10, 1]).astype(np.float32)
               data = to_variable(x_data)

               hidden = fc_layer(data)
               avg_loss = fluid.layers.mean(hidden)

               # scale the loss according to the number of trainers.
               avg_loss = fc_layer.scale_loss(avg_loss)

               avg_loss.backward()

               # collect the gradients of trainers.
               fc_layer.apply_collective_grads()

               adam.minimize(avg_loss)
               fc_layer.clear_gradients()

    Args:
        layers(Layer): The module that should be executed by data parallel.
        strategy(ParallelStrategy): The strategy of data parallelism.

    Returns:
        Layer: The data paralleled module.
    """

Y
Yan Xu 已提交
140
    def __init__(self, layers, strategy):
141 142
        super(DataParallel,
              self).__init__(layers.full_name() + "_data_parallel")
C
chengduo 已提交
143

144
        self._layers = layers
Y
Yan Xu 已提交
145
        self._strategy = strategy
146 147

    def forward(self, *inputs, **kwargs):
Y
Yan Xu 已提交
148 149 150
        return self._layers(*inputs, **kwargs)

    def scale_loss(self, loss):
C
chengduo 已提交
151 152 153 154 155 156 157 158 159 160 161 162
        """
        Scale the loss. In data parallel mode, the loss should be scale with
        the number of trainers. If not in data parallel mode, return the loss
        directly.

        Args:
            loss(Layer): The loss of the current Model.

        Returns:
            Layer: the scaled loss.
        """
        if not self._is_data_parallel_mode():
Y
Yan Xu 已提交
163
            return loss
C
chengduo 已提交
164

Y
Yan Xu 已提交
165 166 167 168 169 170
        loss_scale = to_variable(
            np.array([self._strategy.nranks]).astype("float32"))
        loss_scale.stop_gradient = True
        loss = loss / loss_scale
        return loss

171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186
    def _coalesce_tensors(self, var_groups):
        from ..layers import nn
        coalesced_grads_and_grad_vars = []
        for group_id, grad_vars in var_groups.items():
            flattened_vars = []
            g_var_shapes = []
            for g_var in grad_vars:
                g_var_shapes.append(g_var.shape)
                flattened_vars.append(
                    nn.reshape(
                        x=g_var, shape=[np.prod(g_var.shape)], inplace=True))
            coalesced_grad = nn.concat(flattened_vars)
            coalesced_grads_and_grad_vars.append(
                [coalesced_grad, grad_vars, g_var_shapes])
        return coalesced_grads_and_grad_vars

187 188 189 190 191 192 193 194 195
    def _reshape_inplace(self, x, shape):
        x_shape = self._helper.create_variable_for_type_inference(dtype=x.dtype)
        self._helper.append_op(
            type="reshape2",
            inputs={'X': x},
            attrs={'shape': shape},
            outputs={'Out': x,
                     'XShape': x_shape})

196 197 198 199
    def _split_tensors(self, coalesced_grads_and_grad_vars):
        from ..layers import nn
        for coalesced_grad, origin_grad_vars, grad_shapes in coalesced_grads_and_grad_vars:
            grad_var_len = [np.prod(g_shape) for g_shape in grad_shapes]
200 201 202 203 204 205 206
            self._helper.main_program.current_block().append_op(
                type='split',
                inputs={'X': coalesced_grad},
                outputs={'Out': origin_grad_vars},
                attrs={'sections': grad_var_len,
                       'axis': 0})
            for g_var, g_shape in zip(origin_grad_vars, grad_shapes):
207 208
                self._reshape_inplace(x=g_var, shape=g_shape)
                assert g_var.shape == g_shape
209

J
Jiabin Yang 已提交
210
    @no_grad
Y
Yan Xu 已提交
211
    def apply_collective_grads(self):
C
chengduo 已提交
212 213 214 215
        """
        AllReduce the Parameters' gradient.
        """
        if not self._is_data_parallel_mode():
Y
Yan Xu 已提交
216 217
            return

218 219
        grad_var_set = set()
        grad_vars = []
Y
Yan Xu 已提交
220
        for param in self._layers.parameters():
C
chengduo 已提交
221
            # NOTE(zcd): The grad_ivar maybe no generated.
Y
Yan Xu 已提交
222 223 224 225 226 227
            if param.trainable and param._ivar._grad_ivar():
                g_var = framework.Variable(
                    block=self._helper.main_program.current_block(),
                    name=param._ivar._grad_name(),
                    stop_gradient=True,
                    ivar=param._ivar._grad_ivar())
228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257
                grad_vars.append(g_var)
                assert g_var not in grad_var_set
                grad_var_set.add(g_var)

        # FIXME(zcd): the type of the var should be LoDTensor, i.e
        # the gradients should be dense, otherwise, the following
        # logic should be updated.
        # 128 MB as a group
        mega_bytes = 128 * 1024 * 1024
        group_idx = 0
        memory_counter = 0
        grad_var_groups = OrderedDict()
        dtype = grad_vars[0].dtype
        for g_var in grad_vars:
            # Note: the dtype of the same group should be the same.
            bytes = np.prod(g_var.shape) * core.size_of_dtype(g_var.dtype)
            if memory_counter < mega_bytes and dtype == g_var.dtype:
                memory_counter += bytes
            else:
                memory_counter = bytes
                group_idx += 1
            grad_var_groups.setdefault(group_idx, []).append(g_var)

        coalesced_grads_and_vars = self._coalesce_tensors(grad_var_groups)

        for coalesced_grad, g_vars, g_shapes in coalesced_grads_and_vars:
            collective._allreduce(
                coalesced_grad, coalesced_grad, sync_mode=False)

        self._split_tensors(coalesced_grads_and_vars)
C
chengduo 已提交
258 259 260

    def _is_data_parallel_mode(self):
        return self._strategy.nranks > 1