tests/utils/test_torch_profile.py

# Copyright 2026 Bytedance Ltd. and/or its affiliates
#
# 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 torch

from verl.utils.profiler.config import ProfilerConfig, TorchProfilerToolConfig
from verl.utils.profiler.torch_profile import Profiler, get_torch_profiler


class TestTorchProfile(unittest.TestCase):
    def setUp(self):
        # Reset Profiler class state
        Profiler._define_count = 0

    @patch("torch.profiler.profile")
    def test_get_torch_profiler(self, mock_profile):
        # Test wrapper function
        get_torch_profiler(contents=["cpu", "cuda", "stack"], save_path="/tmp/test", rank=0)
        mock_profile.assert_called_once()
        _, kwargs = mock_profile.call_args

        # Verify activities
        activities = kwargs["activities"]
        self.assertIn(torch.profiler.ProfilerActivity.CPU, activities)
        self.assertIn(torch.profiler.ProfilerActivity.CUDA, activities)

        # Verify options
        self.assertTrue(kwargs["with_stack"])
        self.assertFalse(kwargs["record_shapes"])
        self.assertFalse(kwargs["profile_memory"])

    @patch("verl.utils.profiler.torch_profile.get_torch_profiler")
    def test_profiler_lifecycle(self, mock_get_profiler):
        # Mock the underlying torch profiler object
        mock_prof_instance = MagicMock()
        mock_get_profiler.return_value = mock_prof_instance

        # Initialize
        tool_config = TorchProfilerToolConfig(contents=["cpu"], discrete=False)
        config = ProfilerConfig(save_path="/tmp/test", enable=True, tool_config=tool_config)
        profiler = Profiler(rank=0, config=config, tool_config=tool_config)

        # Test Start
        profiler.start()
        mock_get_profiler.assert_called_once()
        mock_prof_instance.start.assert_called_once()

        # Test Step
        profiler.step()
        mock_prof_instance.step.assert_called_once()

        # Test Stop
        profiler.stop()
        mock_prof_instance.stop.assert_called_once()

    @patch("verl.utils.profiler.torch_profile.get_torch_profiler")
    def test_discrete_mode(self, mock_get_profiler):
        # Mock for discrete mode
        mock_prof_instance = MagicMock()
        mock_get_profiler.return_value = mock_prof_instance

        tool_config = TorchProfilerToolConfig(contents=["cpu"], discrete=True)
        config = ProfilerConfig(save_path="/tmp/test", enable=True, tool_config=tool_config)
        profiler = Profiler(rank=0, config=config, tool_config=tool_config)

        # In discrete mode, start/stop shouldn't trigger global profiler immediately
        profiler.start()
        mock_get_profiler.assert_not_called()

        profiler.stop()
        mock_prof_instance.stop.assert_not_called()


if __name__ == "__main__":
    unittest.main()