profiler_helper.h

/* Copyright (c) 2020 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. */

#pragma once

#include <algorithm>
#include <iomanip>
#include <limits>
#include <list>
#include <map>
#include <memory>
#include <mutex>  // NOLINT
#include <random>
#include <set>
#include <stack>
#include <string>
#include <unordered_map>
#include <utility>
#include <vector>
#ifdef PADDLE_WITH_CUDA
#include <cuda.h>
#endif  // PADDLE_WITH_CUDA
#ifdef PADDLE_WITH_HIP
#include <hip/hip_runtime.h>
#endif
#ifdef PADDLE_WITH_MLU
#include "paddle/fluid/platform/device/mlu/enforce.h"
#include "paddle/fluid/platform/device/mlu/mlu_info.h"
#endif

#include "paddle/fluid/memory/memory.h"
#include "paddle/fluid/platform/device/gpu/gpu_info.h"

namespace paddle {
namespace platform {

static int64_t profiler_lister_id = 0;
static bool should_send_profile_state = false;
std::mutex profiler_mu;

static TracerOption g_tracer_option = TracerOption::kDefault;
// The profiler state, the initial value is ProfilerState::kDisabled
static ProfilerState g_state = ProfilerState::kDisabled;
// To hook RecordEvent's events, use it to nvtx timeline
static bool g_enable_nvprof_hook = false;
// The thread local event list only can be accessed by the specific thread
// The thread index of each thread
static thread_local uint64_t g_thread_id;
// The g_next_thread_id is a global counter for threads, by the g_thread_id and
// g_next_thread_id, we can know how many threads have created EventList.
static uint32_t g_next_thread_id = 0;
// The global mutex
static std::mutex g_all_event_lists_mutex;
// The total event lists of all threads
static std::list<std::shared_ptr<EventList<Event>>> g_all_event_lists;
// The thread local event list only can be accessed by the specific thread
static thread_local std::shared_ptr<EventList<Event>> g_event_list;

static std::list<std::shared_ptr<EventList<MemEvent>>> g_all_mem_event_lists;
static thread_local std::shared_ptr<EventList<MemEvent>> g_mem_event_list;
static std::mutex g_all_mem_event_lists_mutex;
static thread_local int32_t g_mem_thread_id;
static uint32_t g_mem_next_thread_id = 0;

static int FindNthReversePos(const std::string &s, const char ch, const int N) {
  int found_pos = -1;
  auto pos = s.rfind('/', s.length() - 1);
  int pos_number = 1;
  while (pos != std::string::npos && pos_number < N) {
    pos = s.rfind(ch, pos - 1);
    pos_number++;
  }
  if (pos != std::string::npos) found_pos = pos;
  return found_pos;
}

inline uint64_t GetTimeInNsec() {
  using clock = std::conditional<std::chrono::high_resolution_clock::is_steady,
                                 std::chrono::high_resolution_clock,
                                 std::chrono::steady_clock>::type;
  return std::chrono::duration_cast<std::chrono::nanoseconds>(
             clock::now().time_since_epoch())
      .count();
}

inline EventList<Event> &GetEventList() {
  if (!g_event_list) {
    std::lock_guard<std::mutex> guard(g_all_event_lists_mutex);
    g_event_list = std::make_shared<EventList<Event>>();
    g_thread_id = g_next_thread_id++;
    g_all_event_lists.emplace_front(g_event_list);
    RecoreCurThreadId(g_thread_id);
  }
  return *g_event_list;
}

inline EventList<MemEvent> &GetMemEventList() {
  if (!g_mem_event_list) {
    g_mem_event_list = std::make_shared<EventList<MemEvent>>();
    std::lock_guard<std::mutex> guard(g_all_mem_event_lists_mutex);
    g_mem_thread_id = g_mem_next_thread_id++;
    g_all_mem_event_lists.emplace_front(g_mem_event_list);
  }
  return *g_mem_event_list;
}

std::vector<std::vector<MemEvent>> GetMemEvents() {
  std::lock_guard<std::mutex> guard(g_all_mem_event_lists_mutex);
  std::vector<std::vector<MemEvent>> result;
  for (auto &it : g_all_mem_event_lists) {
    result.emplace_back((*it).Reduce());
  }
  return result;
}

void SynchronizeAllDevice() {
#ifdef PADDLE_WITH_CUDA
  int count = GetGPUDeviceCount();
  for (int i = 0; i < count; i++) {
    SetDeviceId(i);
    PADDLE_ENFORCE_GPU_SUCCESS(cudaDeviceSynchronize());
  }
#endif
#ifdef PADDLE_WITH_HIP
  int count = GetGPUDeviceCount();
  for (int i = 0; i < count; i++) {
    SetDeviceId(i);
    PADDLE_ENFORCE_GPU_SUCCESS(hipDeviceSynchronize());
  }
#endif
#ifdef PADDLE_WITH_MLU
  int count = GetMLUDeviceCount();
  for (int i = 0; i < count; i++) {
    SetMLUDeviceId(i);
    PADDLE_ENFORCE_MLU_SUCCESS(cnrtSyncDevice());
  }
#endif
}

static double ToMegaBytes(size_t bytes) {
  return static_cast<double>(bytes) / (1 << 20);
}

// Print results
void PrintMemProfiler(
    const std::map<Place, std::unordered_map<std::string, MemoryProfierReport>>
        &annotation_report,
    const size_t name_width, const size_t data_width) {
  // Output header information
  std::cout << "\n------------------------->"
            << "    Memory Profiling Report     "
            << "<-------------------------\n\n";

#if defined(PADDLE_WITH_CUDA) || defined(PADDLE_WITH_HIP)
  int num_gpus = GetGPUDeviceCount();
  std::cout.setf(std::ios::left);
  if (num_gpus > 0) {
    std::cout << "GPU Memory Usage (MB):\n";
    for (int dev_id = 0; dev_id < num_gpus; ++dev_id) {
      int64_t allocated =
          memory::DeviceMemoryStatCurrentValue("Allocated", dev_id);
      int64_t reserved =
          memory::DeviceMemoryStatCurrentValue("Reserved", dev_id);
      size_t available = 0, total = 0, actual_available = 0, actual_total = 0;
      RecordedGpuMemGetInfo(&available, &total, &actual_available,
                            &actual_total, dev_id);

      std::ostringstream system_gpu_memory;
      system_gpu_memory << "System GPU Memory (gpu:" << dev_id << ")";
      std::cout << "  " << std::setw(30) << system_gpu_memory.str()
                << "Total: " << std::setw(12) << ToMegaBytes(total)
                << "Allocated: " << std::setw(12)
                << ToMegaBytes(total - available) << "Free: " << std::setw(12)
                << ToMegaBytes(available) << "\n";
      std::ostringstream software_memory_pool;
      software_memory_pool << "Software Memory Pool (gpu:" << dev_id << ")";
      std::cout << "  " << std::setw(30) << software_memory_pool.str()
                << "Total: " << std::setw(12) << ToMegaBytes(reserved)
                << "Allocated: " << std::setw(12)
                << ToMegaBytes(reserved - allocated)
                << "Free: " << std::setw(12) << ToMegaBytes(allocated) << "\n";
    }
    std::cout << "\n";
  }
#endif

  // Output events table
  std::cout.setf(std::ios::left);
  std::cout << std::setw(name_width) << "Event" << std::setw(data_width)
            << "Alloc Calls" << std::setw(data_width) << "Size(MB)"
            << std::setw(data_width) << "Free Calls" << std::setw(data_width)
            << "Size(MB)" << std::endl;

  for (auto &tmp : annotation_report) {
    for (auto &e : tmp.second) {
      auto event_name = string::Sprintf("%s:%s", tmp.first, e.first);
      std::cout << std::setw(name_width) << event_name;
      std::cout << std::setw(data_width) << e.second.alloc_times;
      std::cout << std::setw(data_width)
                << e.second.alloc_size / (1024.0 * 1024.0);
      std::cout << std::setw(data_width) << e.second.free_times;
      std::cout << std::setw(data_width)
                << e.second.free_size / (1024.0 * 1024.0) << std::endl;
    }
  }
  std::cout << std::endl;
}

// parse memory events
void ParseMemEvents(const std::vector<std::vector<MemEvent>> &events) {
  if (g_state == ProfilerState::kDisabled) return;
  // place, annotation, alloc times,  alloc size
  std::map<Place, std::unordered_map<std::string, MemoryProfierReport>>
      annotation_report;

  for (auto &tmp : events) {
    for (auto &e : tmp) {
      if (e.type() == EventType::kPushRange) {
        annotation_report[e.place()][e.annotation()].alloc_times += 1;
        annotation_report[e.place()][e.annotation()].alloc_size += e.bytes();
      } else if (e.type() == EventType::kPopRange) {
        annotation_report[e.place()][e.annotation()].free_times += 1;
        annotation_report[e.place()][e.annotation()].free_size += e.bytes();
      }
    }
  }
  PrintMemProfiler(annotation_report, 55, 18);
}

void DealWithShowName() {
  std::unordered_map<std::string, std::vector<std::string>> profiler_name_info;
  for (auto it = g_all_event_lists.begin(); it != g_all_event_lists.end();
       ++it) {
    for (auto &block : (*it)->event_blocks) {
      for (auto &r : block) {
        auto event_name = r.name();
        auto origin_event_name = event_name;
        size_t start = origin_event_name.find('%', 0);
        size_t end = origin_event_name.find('%', start + 1);
        size_t start_replace = start;
        size_t end_replace = end;
        std::string prefix_str = origin_event_name.substr(0, start);
        while (start != std::string::npos && end != std::string::npos &&
               start_replace != std::string::npos &&
               end_replace != std::string::npos) {
          auto search_str = origin_event_name.substr(start, end - start + 1);
          std::string replace_str = "";
          int replace_index = 0;

          auto it = profiler_name_info.find(prefix_str);
          if (it == profiler_name_info.end()) {
            std::vector<std::string> op_name_vector{search_str};
            profiler_name_info[prefix_str] = op_name_vector;
          } else {
            auto op_name_vector = it->second;
            auto iter =
                find(op_name_vector.begin(), op_name_vector.end(), search_str);
            if (iter == op_name_vector.end()) {
              replace_index = profiler_name_info[prefix_str].size();
              profiler_name_info[prefix_str].push_back(search_str);
            } else {
              replace_index = iter - op_name_vector.begin();
            }
          }
          replace_str = std::to_string(replace_index);
          event_name.replace(start_replace, end_replace - start_replace + 1,
                             replace_str);
          start = start + 1;
          start = origin_event_name.find('%', start);
          end = origin_event_name.find('%', start + 1);
          start_replace = event_name.find('%', 0);
          end_replace = event_name.find('%', start_replace + 1);
          prefix_str = origin_event_name.substr(0, start);
        }
        r.set_name(event_name);
      }
    }
  }
}

std::function<bool(const EventItem &, const EventItem &)> SetSortedFunc(
    EventSortingKey sorted_by, std::string *domain) {
  std::string sorted_domain;
  std::function<bool(const EventItem &, const EventItem &)> sorted_func;
  switch (sorted_by) {
    case EventSortingKey::kCalls:
      sorted_domain = "number of calls";
      sorted_func = [](const EventItem &a, const EventItem &b) {
        return a.calls > b.calls;
      };
      break;
    case EventSortingKey::kTotal:
      sorted_domain = "total time";
      sorted_func = [](const EventItem &a, const EventItem &b) {
        return a.total_time > b.total_time;
      };
      break;
    case EventSortingKey::kMin:
      sorted_domain = "minimum time";
      sorted_func = [](const EventItem &a, const EventItem &b) {
        return a.min_time > b.min_time;
      };
      break;
    case EventSortingKey::kMax:
      sorted_domain = "maximum time";
      sorted_func = [](const EventItem &a, const EventItem &b) {
        return a.max_time > b.max_time;
      };
      break;
    case EventSortingKey::kAve:
      sorted_domain = "average time";
      sorted_func = [](const EventItem &a, const EventItem &b) {
        return a.ave_time > b.ave_time;
      };
      break;
    case EventSortingKey::kGPUTime:
      sorted_domain = "average time";
      sorted_func = [](const EventItem &a, const EventItem &b) {
        return a.gpu_time > b.gpu_time;
      };
      break;
    case EventSortingKey::kCPUTime:
      sorted_domain = "average time";
      sorted_func = [](const EventItem &a, const EventItem &b) {
        return a.cpu_time > b.cpu_time;
      };
      break;
    default:
      sorted_domain = "event first end time";
  }
  *domain = sorted_domain;
  return sorted_func;
}

void SetEvent(bool merge_thread, const Event &analyze_event,
              size_t *max_name_width, std::list<Event> *pushed_events,
              std::vector<EventItem> *event_items,
              std::unordered_map<std::string, int> *event_idx,
              const std::set<std::string> &main_thread_event_name) {
  if (analyze_event.type() == EventType::kPushRange) {
    pushed_events->push_back(analyze_event);
  } else if (analyze_event.type() == EventType::kPopRange) {
    std::list<Event>::reverse_iterator rit = pushed_events->rbegin();
    while (rit != pushed_events->rend() &&
           rit->name() != analyze_event.name()) {
      ++rit;
    }

    // to find the father name event name
    if (rit != pushed_events->rend()) {
      double event_time = 0;
      double gpu_time = 0.0f;
#if defined(PADDLE_WITH_CUDA) || defined(PADDLE_WITH_HIP)
      gpu_time = rit->CudaElapsedMs(analyze_event);
#endif
      double cpu_time = rit->CpuElapsedMs(analyze_event);
      if (g_state == ProfilerState::kCUDA) {
        event_time = gpu_time;
      } else if (g_state == ProfilerState::kCPU) {
        event_time = cpu_time;
      } else {
        event_time = gpu_time + cpu_time;
      }

      std::string event_name;
      if (merge_thread) {
        event_name = rit->name();
      } else {
        if (!main_thread_event_name.empty()) {
          auto origin_name = rit->name();
          int index = 1;
          int split_pos = 0;
          while ((split_pos = FindNthReversePos(origin_name, '/', index)) !=
                 -1) {
            auto prefix_str = origin_name.substr(0, split_pos);
            if (main_thread_event_name.count(prefix_str)) {
              break;
            }
            index++;
          }
          if (split_pos == -1 && !main_thread_event_name.count(rit->name())) {
            event_name = "thread" + std::to_string(rit->thread_id()) +
                         "::" + rit->name();
          } else {
            if (!main_thread_event_name.count(rit->name())) {
              event_name =
                  origin_name.substr(0, split_pos + 1) + "thread" +
                  std::to_string(rit->thread_id()) + "::" +
                  origin_name.substr(split_pos + 1, origin_name.length() - 1);
            } else {
              event_name = rit->name();
            }
          }
        } else {
          event_name =
              "thread" + std::to_string(rit->thread_id()) + "::" + rit->name();
        }
      }
      auto print_name_size = event_name.size();
      int found_pos = 0;
      if (rit->role() == EventRole::kInnerOp &&
          g_tracer_option != TracerOption::kDefault &&
          (found_pos = FindNthReversePos(event_name, '/', 2)) != -1) {
        print_name_size = event_name.size() - (found_pos + 1);
      } else if ((found_pos = FindNthReversePos(event_name, '/', 1)) != -1 &&
                 (rit->role() != EventRole::kInnerOp ||
                  g_tracer_option == TracerOption::kDefault)) {
        print_name_size = event_name.size() - (found_pos + 1);
      }
      *max_name_width = std::max(*max_name_width, print_name_size);

      if (event_idx->find(event_name) == event_idx->end()) {
        event_idx->insert({event_name, event_items->size()});
        EventItem event_item = {event_name, 1,          event_time, event_time,
                                event_time, event_time, cpu_time,   gpu_time,
                                0.,         rit->role()};
        event_items->push_back(event_item);
      } else {
        int index = event_idx->at(event_name);
        event_items->at(index).calls += 1;
        // total time
        event_items->at(index).total_time += event_time;
        // min time
        event_items->at(index).min_time =
            std::min(event_time, event_items->at(index).min_time);
        // max time
        event_items->at(index).max_time =
            std::max(event_time, event_items->at(index).max_time);
        event_items->at(index).gpu_time += gpu_time;
        event_items->at(index).cpu_time += cpu_time;
      }

      // remove the push marker from the list
      pushed_events->erase((++rit).base());
    } else {
      LOG(WARNING) << "Cannot find the push marker of event \'"
                   << analyze_event.name()
                   << "\', which will be ignored in profiling report.";
    }
  }
}

void UpdateGpuMemcpy(const EventItem &item, EventItem *memcpy_async,
                     EventItem *memcpy_sync) {
  if (item.name.find("GpuMemcpyAsync") != std::string::npos) {
    memcpy_async->calls += item.calls;
    memcpy_async->total_time += item.total_time;
    memcpy_async->ratio += item.ratio;
  } else if (item.name.find("GpuMemcpySync") != std::string::npos) {
    memcpy_sync->calls += item.calls;
    memcpy_sync->total_time += item.total_time;
    memcpy_sync->ratio += item.ratio;
  }
}

void ComputeOverhead(const std::vector<EventItem> &main_event_items,
                     const std::multimap<std::string, EventItem> &sub_child_map,
                     OverHead *overhead) {
  EventItem memcpy_async = {
      "GpuMemcpyAsync", 0, 0., 0., 0., 0., 0., 0., 0.0f, EventRole::kOrdinary};
  EventItem memcpy_sync = {"GpuMemcpySync",     0, 0., 0., 0., 0., 0., 0., 0.0f,
                           EventRole::kOrdinary};
  // GpuMemcpy may be in main_event_items
  for (auto &item : main_event_items) {
    if (item.role != EventRole::kSpecial) {
      overhead->accumulated_time += item.total_time;
    }
    UpdateGpuMemcpy(item, &memcpy_async, &memcpy_sync);
  }

  for (auto it = sub_child_map.begin(); it != sub_child_map.end(); it++) {
    if (it->first == "ParallelExecutor::Run") {
      overhead->accumulated_time += it->second.total_time;
    }
    if (it->second.name.find("compute") != std::string::npos &&
        it->second.name.find("compute/") == std::string::npos) {
      overhead->compute_time += it->second.total_time;
    }
    UpdateGpuMemcpy(it->second, &memcpy_async, &memcpy_sync);
  }
  overhead->framework_time =
      overhead->accumulated_time - overhead->compute_time;
  overhead->memcpy_item.calls = memcpy_async.calls + memcpy_sync.calls;
  overhead->memcpy_item.total_time =
      memcpy_async.total_time + memcpy_sync.total_time;
  overhead->memcpy_item.ratio = memcpy_async.ratio + memcpy_sync.ratio;
  overhead->sub_memcpy_items = {memcpy_async, memcpy_sync};
}

std::string FindOrdinaryParent(
    const std::multimap<std::string, EventItem> &sub_child_map,
    std::string name) {
  bool find_name = false;
  std::string parent = name;
  EventRole role;
  for (auto it = sub_child_map.begin(); it != sub_child_map.end(); it++) {
    if (it->second.name == name) {
      role = it->second.role;
      parent = it->first;
      find_name = true;
      break;
    }
  }
  if (find_name && role == EventRole::kOrdinary) {
    return name;
  } else if (find_name && role != EventRole::kOrdinary) {
    return FindOrdinaryParent(sub_child_map, parent);
  } else {
    return parent;
  }
}

// When TracerOption is KDefault, OpDetail will be recorded but only default
// profile result will be printed.
// GpuMemcpy should be printed in kDefault setting, however it offten occurs
// during 'compute' or 'prepare data' process, so the elements of sub_child_map
// need to be changed before being inserted into child_map. for instance:
// it->first: OpType/compute => OpType
// it->second.name: OpType/compute/GpuMemcpyAsync => OpType/GpuMemcpyAsync.
void GetChildMap(const std::multimap<std::string, EventItem> &sub_child_map,
                 std::multimap<std::string, EventItem> *child_map) {
  if (platform::GetTracerOption() != TracerOption::kDefault) {
    for (auto it = sub_child_map.begin(); it != sub_child_map.end(); it++) {
      child_map->insert(
          std::pair<std::string, EventItem>(it->first, it->second));
    }
  } else {
    for (auto it = sub_child_map.begin(); it != sub_child_map.end(); it++) {
      if (it->second.name.find("GpuMemcpy") != std::string::npos) {
        std::string parent_name = FindOrdinaryParent(sub_child_map, it->first);
        auto item = it->second;
        auto right_pos = item.name.rfind("/");
        if (right_pos != std::string::npos) {
          std::string child_name = item.name.substr(
              right_pos + 1, item.name.length() - right_pos - 1);
          item.name = parent_name + "/" + child_name;
        }
        child_map->insert(std::pair<std::string, EventItem>(parent_name, item));
      } else if (it->second.role == EventRole::kOrdinary) {
        child_map->insert(
            std::pair<std::string, EventItem>(it->first, it->second));
      }
    }
  }
}

void PrintOverHead(const OverHead &overhead, const size_t data_width) {
  float compute_ratio = overhead.compute_time / overhead.accumulated_time;
  float framework_ratio = 1 - compute_ratio;
  std::cout << "-------------------------"
            << "     Overhead Summary      "
            << "-------------------------\n\n";
  if (overhead.print_explanation) {
    std::cout
        << "The Overhead Summary divides the cost of each event into framework "
           "overhead or computation time."
        << "\nThe `Accumulated time of events` is higher than the `Elapsed "
           "time of events`."
        << "\nBecause the OP is executed asynchronously. For example,"
        << "\nEvent                   Timeline"
        << "\nParallelExecutor::Run   "
           "---------------------------------------------------------"
        << "\n  thread1::OP1                 -----------------------------"
        << "\n  thread2::OP2                      "
           "---------------------------------------------"
        << "\nOP1.time + OP2.time > ParallelExecutor::Run.time\n\n";
    std::cout << "Elapsed time of events: " << overhead.elapsed_time
              << std::endl;
    std::cout << "Accumulated time of events: " << overhead.accumulated_time
              << std::endl;
  } else {
    std::cout << "Total time: " << overhead.elapsed_time << std::endl;
  }
  std::cout.setf(std::ios::left);
  std::cout << std::setw(25) << "  Computation time"
            << "Total: " << std::setw(data_width) << overhead.compute_time
            << "Ratio: " << compute_ratio * 100 << "%" << std::endl;
  std::cout << std::setw(25) << "  Framework overhead"
            << "Total: " << std::setw(data_width) << overhead.framework_time
            << "Ratio: " << framework_ratio * 100 << "%" << std::endl;

  std::cout << "\n-------------------------"
            << "     GpuMemCpy Summary     "
            << "-------------------------\n\n";
  std::cout << std::setw(25) << "GpuMemcpy"
            << "Calls: " << std::setw(data_width) << overhead.memcpy_item.calls
            << "Total: " << std::setw(data_width)
            << overhead.memcpy_item.total_time
            << "Ratio: " << overhead.memcpy_item.ratio * 100 << "%"
            << std::endl;
  for (size_t i = 0; i < overhead.sub_memcpy_items.size(); ++i) {
    EventItem item = overhead.sub_memcpy_items[i];
    if (item.calls != 0) {
      std::cout << std::setw(25) << "  " + item.name
                << "Calls: " << std::setw(data_width) << item.calls
                << "Total: " << std::setw(data_width) << item.total_time
                << "Ratio: " << item.ratio * 100 << "%" << std::endl;
    }
  }
}

// Print results
void PrintProfiler(
    const std::vector<std::vector<EventItem>> &events_table,
    const std::multimap<std::string, EventItem> &child_map,
    std::function<bool(const EventItem &, const EventItem &)> sorted_func,
    EventSortingKey sorted_by, const OverHead &overhead,
    const std::string &sorted_domain, const size_t name_width,
    const size_t data_width, bool merge_thread, int print_depth) {
  if (print_depth == 0) {
    // Output header information
    std::cout << "\n------------------------->"
              << "     Profiling Report     "
              << "<-------------------------\n\n";
    std::string place;
    if (g_state == ProfilerState::kCPU) {
      place = "CPU";
    } else if (g_state == ProfilerState::kCUDA) {
      place = "CUDA";
    } else if (g_state == ProfilerState::kAll) {
      place = "All";
    } else {
      PADDLE_THROW(platform::errors::InvalidArgument(
          "Except profiler state must to be one of ['CPU', 'GPU' 'ALL'], but "
          "received Invalid profiler state."));
    }

    if (merge_thread) {
      std::cout << "Note! This Report merge all thread info into one."
                << std::endl;
    }
    std::cout << "Place: " << place << std::endl;
    std::cout << "Time unit: ms" << std::endl;
    std::cout << "Sorted by " << sorted_domain
              << " in descending order in the same thread\n\n";

    if (overhead.print_overhead) {
      PrintOverHead(overhead, data_width);
    }
    std::cout << "\n-------------------------"
              << "       Event Summary       "
              << "-------------------------\n\n";
    // Output events table
    std::cout.setf(std::ios::left);
    std::cout << std::setw(name_width) << "Event" << std::setw(data_width)
              << "Calls" << std::setw(data_width) << "Total";
    if (g_state == ProfilerState::kAll) {
      std::cout << std::setw(data_width * 2) << "CPU Time (Ratio)"
                << std::setw(data_width * 2) << "GPU Time (Ratio)";
    }
    std::cout << std::setw(data_width) << "Min." << std::setw(data_width)
              << "Max." << std::setw(data_width) << "Ave."
              << std::setw(data_width) << "Ratio." << std::endl;
  }

  if (events_table.size() <= 0) return;

  for (size_t i = 0; i < events_table.size(); ++i) {
    for (size_t j = 0; j < events_table[i].size(); ++j) {
      auto event_item = events_table[i][j];
      std::vector<std::vector<EventItem>> child_table;
      std::vector<EventItem> table;
      for (auto it = child_map.begin(); it != child_map.end(); it++) {
        if (it->first == event_item.name) {
          table.push_back(it->second);
        }
      }

      if (sorted_by != EventSortingKey::kDefault) {
        std::sort(table.begin(), table.end(), sorted_func);
      }
      if (!table.empty()) child_table.push_back(table);

      auto name_len = event_item.name.length();
      int remove_len = 0;
      int Nth = 1;
      int found_pos = 0;
      if (event_item.role == EventRole::kInnerOp) Nth = 2;
      found_pos = FindNthReversePos(event_item.name, '/', Nth);
      if (found_pos != -1) remove_len = found_pos + 1;

      std::string print_name = event_item.name.substr(remove_len, name_len);
      std::string delimiter;
      for (int i = 0; i < print_depth; i++) {
        delimiter = "  " + delimiter;
      }
      print_name = delimiter + print_name;

      std::cout << std::setw(name_width) << print_name << std::setw(data_width)
                << event_item.calls << std::setw(data_width)
                << event_item.total_time;
      if (g_state == ProfilerState::kAll) {
        std::cout << std::setw(data_width * 2)
                  << string::Sprintf(
                         "%f (%f)", event_item.cpu_time,
                         (event_item.cpu_time / event_item.total_time))
                  << std::setw(data_width * 2)
                  << string::Sprintf(
                         "%f (%f)", event_item.gpu_time,
                         (event_item.gpu_time / event_item.total_time));
      }
      std::cout << std::setw(data_width) << event_item.min_time
                << std::setw(data_width) << event_item.max_time
                << std::setw(data_width) << event_item.ave_time;
      if (event_item.name.find("ext_reorder") != std::string::npos ||
          event_item.name.find("int_reorder") != std::string::npos) {
        std::cout << event_item.ratio << '*';
      } else {
        std::cout << std::setw(data_width) << event_item.ratio;
      }
      std::cout << std::endl;

      PrintProfiler(child_table, child_map, sorted_func, sorted_by, overhead,
                    sorted_domain, name_width, data_width, merge_thread,
                    print_depth + 1);
    }
  }
}

void AnalyzeEvent(
    const std::vector<std::vector<Event>> *analyze_events,
    std::vector<std::vector<EventItem>> *events_table,
    std::multimap<std::string, EventItem> *child_map,
    std::function<bool(const EventItem &, const EventItem &)> sorted_func,
    EventSortingKey sorted_by, size_t *max_name_width, OverHead *overhead,
    bool merge_thread) {
  // In oreder to deal with special event in main thread
  std::set<std::string> main_thread_event_name;
  for (size_t i = 0; i < (*analyze_events).size(); i++) {
    for (size_t j = 0; j < (*analyze_events)[i].size(); j++) {
      Event event = (*analyze_events)[i][j];
      if (event.role() == EventRole::kSpecial) {
        main_thread_event_name.insert(event.name());
      }
    }
  }
  for (size_t i = 0; i < (*analyze_events).size(); i++) {
    double total = 0.;  // the total time in one thread
    std::list<Event> pushed_events;
    std::vector<EventItem> event_items;
    std::vector<EventItem> main_event_items;
    std::unordered_map<std::string, int> event_idx;
    std::multimap<std::string, EventItem> sub_child_map;

    for (size_t j = 0; j < (*analyze_events)[i].size(); j++) {
      Event analyze_event = (*analyze_events)[i][j];
      if (!(analyze_event.role() == EventRole::kSpecial && !merge_thread)) {
        SetEvent(merge_thread, analyze_event, max_name_width, &pushed_events,
                 &event_items, &event_idx, main_thread_event_name);
      }
    }

    auto table_size = event_items.size();
    std::vector<int> child_index(table_size, 0);
    for (size_t j = 0; j < table_size; ++j) {
      std::string fname = event_items[j].name;
      std::string grad_name = event_items[j].name + "_grad";
      for (size_t k = 0; k < table_size; ++k) {
        std::string cname = event_items[k].name;
        bool condition = cname.length() > fname.length() &&
                         cname.rfind(fname, 0) == 0 &&
                         cname.rfind(grad_name, 0) != 0 &&
                         (cname[fname.length()] == '/' &&
                          cname.rfind('/') == fname.length());
        if (condition) {
          sub_child_map.insert(
              std::pair<std::string, EventItem>(fname, event_items[k]));
          child_index[k] = 1;
        }
      }
    }
    for (size_t j = 0; j < table_size; ++j) {
      if (child_index[j] == 0) {
        main_event_items.push_back(event_items[j]);
        total += event_items[j].total_time;
      } else if ((child_index[j] == 1 &&
                  (event_items[j].name.find("ext_reorder") !=
                       std::string::npos ||
                   event_items[j].name.find("int_reorder") !=
                       std::string::npos)) &&
                 platform::GetTracerOption() != TracerOption::kAllOpDetail) {
        size_t first_slash_pos = event_items[j].name.find('/');
        if (first_slash_pos != std::string::npos) {
          std::string fname = event_items[j].name.substr(0, first_slash_pos);
          child_map->insert(
              std::pair<std::string, EventItem>(fname, event_items[j]));
        }
      }
    }
    // average time
    for (auto &item : main_event_items) {
      item.ave_time = item.total_time / item.calls;
      item.ratio = item.total_time / total;
      if (platform::GetTracerOption() != TracerOption::kAllOpDetail) {
        for (auto it = child_map->begin(); it != child_map->end(); ++it) {
          if ((*it).first == item.name) {
            (*it).second.ratio = (*it).second.total_time / item.total_time;
            break;  // to find only first item
          }
        }
      }
    }
    for (auto it = sub_child_map.begin(); it != sub_child_map.end(); it++) {
      it->second.ratio = it->second.total_time / total;
      it->second.ave_time = it->second.total_time / it->second.calls;
    }
    // When multi-threaded, overhead are printed only if merge_thread is true
    if ((*analyze_events).size() == 1) {
      if (!main_thread_event_name.empty()) {
        overhead->print_explanation = true;
      }
      overhead->elapsed_time = total;
      overhead->print_overhead = true;
      ComputeOverhead(main_event_items, sub_child_map, overhead);
    }
    // sort
    if (sorted_by != EventSortingKey::kDefault) {
      std::sort(main_event_items.begin(), main_event_items.end(), sorted_func);
    }

    events_table->push_back(main_event_items);
    // log warning if there are events with `push` but without `pop`
    std::list<Event>::reverse_iterator rit = pushed_events.rbegin();
    while (rit != pushed_events.rend()) {
      LOG(WARNING) << "Cannot find the pop marker of event \'" << rit->name()
                   << "\', which will be ignored in profiling report.";
      ++rit;
    }

    GetChildMap(sub_child_map, child_map);
  }
}
// Parse the event list and output the profiling report
void ParseEvents(const std::vector<std::vector<Event>> &events,
                 bool merge_thread,
                 EventSortingKey sorted_by = EventSortingKey::kDefault) {
  if (g_state == ProfilerState::kDisabled) return;
  if (merge_thread && events.size() < 2) return;

  std::string sorted_domain;
  std::function<bool(const EventItem &, const EventItem &)> sorted_func;
  sorted_func = SetSortedFunc(sorted_by, &sorted_domain);

  const std::vector<std::vector<Event>> *analyze_events;
  std::vector<std::vector<Event>> merged_events_list;
  if (merge_thread) {
    std::vector<Event> merged_events;
    for (size_t i = 0; i < events.size(); ++i) {
      for (size_t j = 0; j < events[i].size(); ++j) {
        merged_events.push_back(events[i][j]);
      }
    }
    merged_events_list.push_back(merged_events);
    analyze_events = &merged_events_list;
  } else {
    analyze_events = &events;
  }

  std::vector<std::vector<EventItem>> events_table;
  std::multimap<std::string, EventItem> child_map;
  size_t max_name_width = 0;
  OverHead overhead;
  AnalyzeEvent(analyze_events, &events_table, &child_map, sorted_func,
               sorted_by, &max_name_width, &overhead, merge_thread);

  // Print report
  PrintProfiler(events_table, child_map, sorted_func, sorted_by, overhead,
                sorted_domain, max_name_width + 8, 12, merge_thread, 0);
}

}  // namespace platform
}  // namespace paddle