mirror of
https://github.com/PaddlePaddle/FastDeploy.git
synced 2026-04-23 00:17:25 +08:00
gongweibao
8906e09e0f
* [Feature] Add batch-invariant RMSNorm kernel and TP embedding Custom AR path - Add Triton-based rms_norm_batch_invariant kernel for M-invariant RMSNorm - Add linear/linear_v2 tracking wrappers in batch_invariant_mode - Route TP VocabParallelEmbedding through Custom AR instead of NCCL - Increase FD_CUSTOM_AR_MAX_SIZE_MB default from 8 to 64 - Add unit tests for RMSNorm and TP embedding invariance * [Fix] Fix test tolerances for bfloat16 RMSNorm and custom AR buffer size - Relax bfloat16 atol from 1e-3 to 1e-2 for D=3584 in RMSNorm numerical correctness test (0.0078125 diff is expected at bfloat16 precision) - Update test_communication expected buffer size from 8MB to 64MB to match FD_CUSTOM_AR_MAX_SIZE_MB default change in envs.py Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Add RMSNorm layer batch_invariant_mode unit test for coverage Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Add pragma no cover for Triton kernel and multi-GPU embedding path Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> --------- Co-authored-by: gongweibao <gognweibao@baidu.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
108 lines
4.0 KiB
Python
108 lines
4.0 KiB
Python
# Copyright (c) 2025 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.
|
|
|
|
import unittest
|
|
from unittest.mock import MagicMock, patch
|
|
|
|
import paddle
|
|
|
|
from fastdeploy.distributed import communication
|
|
|
|
|
|
class TestCommunicationBasic(unittest.TestCase):
|
|
def setUp(self):
|
|
communication._TP_AR = None
|
|
|
|
def test_capture_custom_allreduce_no_tp_ar(self):
|
|
with communication.capture_custom_allreduce():
|
|
pass
|
|
|
|
def test_capture_custom_allreduce_with_tp_ar(self):
|
|
mock_tp_ar = MagicMock()
|
|
mock_context = MagicMock()
|
|
mock_tp_ar.capture.return_value = mock_context
|
|
communication._TP_AR = mock_tp_ar
|
|
with communication.capture_custom_allreduce():
|
|
pass
|
|
mock_tp_ar.capture.assert_called_once()
|
|
|
|
@patch("paddle.distributed.fleet.get_hybrid_communicate_group")
|
|
@patch("fastdeploy.distributed.custom_all_reduce.CustomAllreduce")
|
|
def test_use_custom_allreduce(self, mock_custom_ar, mock_get_hcg):
|
|
mock_hcg = MagicMock()
|
|
mock_get_hcg.return_value = mock_hcg
|
|
|
|
# fake group with required attributes used by CustomAllreduce
|
|
fake_group = MagicMock()
|
|
fake_group.rank = 0
|
|
fake_group.world_size = 2
|
|
mock_hcg.get_model_parallel_group.return_value = fake_group
|
|
|
|
communication.use_custom_allreduce()
|
|
|
|
self.assertIsNotNone(communication._TP_AR)
|
|
mock_custom_ar.assert_called_once_with(fake_group, 64 * 1024 * 1024)
|
|
|
|
def test_custom_ar_clear_ipc_handles(self):
|
|
mock_tp_ar = MagicMock()
|
|
communication._TP_AR = mock_tp_ar
|
|
communication.custom_ar_clear_ipc_handles()
|
|
mock_tp_ar.clear_ipc_handles.assert_called_once()
|
|
|
|
@patch("fastdeploy.distributed.communication.dist.all_reduce")
|
|
@patch("paddle.distributed.fleet.get_hybrid_communicate_group")
|
|
def test_tensor_model_parallel_all_reduce(self, mock_get_hcg, mock_all_reduce):
|
|
# ensure group exists
|
|
mock_hcg = MagicMock()
|
|
mock_get_hcg.return_value = mock_hcg
|
|
fake_group = MagicMock()
|
|
fake_group.world_size = 2
|
|
mock_hcg.get_model_parallel_group.return_value = fake_group
|
|
|
|
# make all_reduce callable
|
|
def fake_all_reduce(x, group=None):
|
|
return x
|
|
|
|
mock_all_reduce.side_effect = fake_all_reduce
|
|
|
|
x = paddle.to_tensor([1.0])
|
|
# call should not raise, ensure all_reduce was invoked
|
|
_ = communication.tensor_model_parallel_all_reduce(x)
|
|
mock_all_reduce.assert_called()
|
|
|
|
@patch("fastdeploy.distributed.communication.stream.all_reduce")
|
|
@patch("paddle.distributed.fleet.get_hybrid_communicate_group")
|
|
def test_tensor_model_parallel_all_reduce_custom(self, mock_get_hcg, mock_stream_ar):
|
|
# ensure group exists
|
|
mock_hcg = MagicMock()
|
|
mock_get_hcg.return_value = mock_hcg
|
|
fake_group = MagicMock()
|
|
fake_group.world_size = 2
|
|
mock_hcg.get_model_parallel_group.return_value = fake_group
|
|
|
|
# stream.all_reduce may not return value in source; ensure callable
|
|
def fake_stream_all_reduce(x, **kwargs):
|
|
return None
|
|
|
|
mock_stream_ar.side_effect = fake_stream_all_reduce
|
|
|
|
x = paddle.to_tensor([2.0])
|
|
# the function does not return input_ in source; ensure call succeeds and stream.all_reduce used
|
|
_ = communication.tensor_model_parallel_all_reduce_custom(x)
|
|
mock_stream_ar.assert_called_once()
|
|
|
|
|
|
if __name__ == "__main__":
|
|
unittest.main()
|