• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1# Copyright 2023 Huawei Technologies Co., Ltd
2#
3# Licensed under the Apache License, Version 2.0 (the "License");
4# you may not use this file except in compliance with the License.
5# You may obtain a copy of the License at
6#
7# http://www.apache.org/licenses/LICENSE-2.0
8#
9# Unless required by applicable law or agreed to in writing, software
10# distributed under the License is distributed on an "AS IS" BASIS,
11# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12# See the License for the specific language governing permissions and
13# limitations under the License.
14# ============================================================================
15
16import numpy as np
17import mindspore.nn as nn
18from mindspore import Tensor
19from mindspore.communication.management import init, get_rank
20from mindspore.ops.operations import comm_ops
21
22np.random.seed(1)
23init()
24this_rank = get_rank()
25dest_rank = 0
26
27
28class CollectiveGatherNet(nn.Cell):
29    def __init__(self):
30        super().__init__()
31        self.collective_gather = comm_ops.CollectiveGather(dest_rank=dest_rank)
32
33    def construct(self, x):
34        out = self.collective_gather(x)
35        return out
36
37
38def generate_input(dtype):
39    if this_rank == 0:
40        return Tensor(np.array([[1, 1, 1]]).astype(dtype))
41    if this_rank == 1:
42        return Tensor(np.array([[2, 2, 2]]).astype(dtype))
43    if this_rank == 2:
44        return Tensor(np.array([[3, 3, 3]]).astype(dtype))
45    if this_rank == 3:
46        return Tensor(np.array([[4, 4, 4]]).astype(dtype))
47    return None
48
49
50def test_hccl_gather_4p_float32():
51    """
52    Feature: test 'CollectiveGather' communication operator.
53    Description: test 'CollectiveGather' communication operator.
54    Expectation: expect correct result.
55    """
56    ms_input = generate_input(np.float32)
57    net = CollectiveGatherNet()
58    output = net(ms_input)
59
60    if this_rank == dest_rank:
61        res = np.array([[1, 1, 1], [2, 2, 2], [3, 3, 3], [4, 4, 4]]).astype(np.float32)
62        assert (output.numpy() == res).all()
63
64
65def test_hccl_gather_4p_float16():
66    """
67    Feature: test 'CollectiveGather' communication operator.
68    Description: test 'CollectiveGather' communication operator.
69    Expectation: expect correct result.
70    """
71    ms_input = generate_input(np.float16)
72    net = CollectiveGatherNet()
73    output = net(ms_input)
74
75    if this_rank == dest_rank:
76        res = np.array([[1, 1, 1], [2, 2, 2], [3, 3, 3], [4, 4, 4]]).astype(np.float16)
77        assert (output.numpy() == res).all()
78