cudaProfilerUtils.cpp 2.78 KB
Newer Older
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
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
/*
 * Copyright (c) 2022-2024, NVIDIA CORPORATION.  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.
 */

#include "tensorrt_llm/common/cudaProfilerUtils.h"
#include "tensorrt_llm/common/logger.h"
#include "tensorrt_llm/common/stringUtils.h"
#include <cstdint>
#include <optional>

namespace
{

std::tuple<std::unordered_set<int32_t>, std::unordered_set<int32_t>> populateIterationIndexesImpl(
    std::string const& envVarName)
{
    auto envVarVal = std::getenv(envVarName.c_str());
    auto envVarValStr = std::string{envVarVal != nullptr ? envVarVal : ""};
    auto values = tensorrt_llm::common::str2set(envVarValStr, ',');
    std::unordered_set<int32_t> startSet;
    std::unordered_set<int32_t> endSet;
    for (std::string const& value : values)
    {
        size_t dashIdx = value.find("-");
        if (dashIdx != std::string::npos)
        {
            int32_t start = std::stoi(value.substr(0, dashIdx));
            startSet.insert(start);
            int32_t end = std::stoi(value.substr(dashIdx + 1));
            endSet.insert(end);
        }
        else
        {
            int32_t start_end = std::stoi(value);
            startSet.insert(start_end);
            endSet.insert(start_end);
        }
    }

    return std::make_pair(startSet, endSet);
}

} // namespace

namespace tensorrt_llm::common
{

std::pair<std::unordered_set<int32_t>, std::unordered_set<int32_t>> populateIterationIndexes(
    std::string const& envVarName, std::optional<std::string> const& legacyEnvVarName)
{
    auto [profileIterIdxs, stopIterIdxs] = populateIterationIndexesImpl(envVarName);

    // If empty, try to use legacy env var name
    if (legacyEnvVarName && profileIterIdxs.empty() && stopIterIdxs.empty())
    {
        std::tie(profileIterIdxs, stopIterIdxs) = populateIterationIndexesImpl(legacyEnvVarName.value());

        if (!profileIterIdxs.empty() || !stopIterIdxs.empty())
        {
            TLLM_LOG_WARNING(
                "Using deprecated environment variable %s to specify cudaProfiler start and stop iterations. "
                "Please "
                "use %s "
                "instead.",
                legacyEnvVarName.value().c_str(), envVarName.c_str());
        }
    }

    return std::make_pair(profileIterIdxs, stopIterIdxs);
}

} // namespace tensorrt_llm::common