dygraph_hybrid_dp.py 2.2 KB
Newer Older
R
Roc 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67
# 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 print_function

import numpy as np
import argparse
import os
import sys
import signal
import time
import socket
from contextlib import closing
from six import string_types
import math
import paddle
import paddle.fluid as fluid
import paddle.fluid.profiler as profiler
import paddle.fluid.unique_name as nameGen
from paddle.fluid import core
import paddle.distributed.fleet as fleet
from paddle.fluid.incubate.fleet.base import role_maker
import unittest
from multiprocessing import Process
import paddle.fluid.layers as layers
from functools import reduce
from test_collective_multi_nodes import TestCollectiveAPIRunnerBase, runtime_main


class TestDygrapgHybridDP(TestCollectiveAPIRunnerBase):

    def __init__(self):
        pass

    def check_pass(self, *args, **kwargs):
        from common import init_parallel_env
        import paddle
        from paddle.distributed import fleet
        hcg = init_parallel_env("DP16-MP1-PP1-SH1-O1", 2)
        import numpy as np
        dp_group = hcg.get_data_parallel_group()
        np.random.seed(1024)
        data = np.random.random((10 * dp_group.nranks, 100)).reshape(
            (dp_group.nranks, -1, 100))
        data_part = paddle.to_tensor(data[dp_group.rank])
        paddle.distributed.collective.all_reduce(data_part)
        data_reduced = data_part
        data_sumed = np.sum(data, axis=0)
        assert np.allclose(data_sumed,
                           data_reduced.numpy(),
                           rtol=1e-8,
                           atol=1e-8)


if __name__ == "__main__":
    runtime_main(TestDygrapgHybridDP, "dp")