include/llvm/Support/Parallel.h

   1 //===- llvm/Support/Parallel.h - Parallel algorithms ----------------------===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8
   9 #ifndef LLVM_SUPPORT_PARALLEL_H
  10 #define LLVM_SUPPORT_PARALLEL_H
  11
  12 #include "llvm/ADT/STLExtras.h"
  13 #include "llvm/Config/llvm-config.h"
  14 #include "llvm/Support/MathExtras.h"
  15
  16 #include <algorithm>
  17 #include <condition_variable>
  18 #include <functional>
  19 #include <mutex>
  20
  21 #if defined(_MSC_VER) && LLVM_ENABLE_THREADS
  22 #pragma warning(push)
  23 #pragma warning(disable : 4530)
  24 #include <concrt.h>
  25 #include <ppl.h>
  26 #pragma warning(pop)
  27 #endif
  28
  29 namespace llvm {
  30
  31 namespace parallel {
  32 struct sequential_execution_policy {};
  33 struct parallel_execution_policy {};
  34
  35 template <typename T>
  36 struct is_execution_policy
  37     : public std::integral_constant<
  38           bool, llvm::is_one_of<T, sequential_execution_policy,
  39                                 parallel_execution_policy>::value> {};
  40
  41 constexpr sequential_execution_policy seq{};
  42 constexpr parallel_execution_policy par{};
  43
  44 namespace detail {
  45
  46 #if LLVM_ENABLE_THREADS
  47
  48 class Latch {
  49   uint32_t Count;
  50   mutable std::mutex Mutex;
  51   mutable std::condition_variable Cond;
  52
  53 public:
  54   explicit Latch(uint32_t Count = 0) : Count(Count) {}
  55   ~Latch() { sync(); }
  56
  57   void inc() {
  58     std::lock_guard<std::mutex> lock(Mutex);
  59     ++Count;
  60   }
  61
  62   void dec() {
  63     std::lock_guard<std::mutex> lock(Mutex);
  64     if (--Count == 0)
  65       Cond.notify_all();
  66   }
  67
  68   void sync() const {
  69     std::unique_lock<std::mutex> lock(Mutex);
  70     Cond.wait(lock, [&] { return Count == 0; });
  71   }
  72 };
  73
  74 class TaskGroup {
  75   Latch L;
  76
  77 public:
  78   void spawn(std::function<void()> f);
  79
  80   void sync() const { L.sync(); }
  81 };
  82
  83 #if defined(_MSC_VER)
  84 template <class RandomAccessIterator, class Comparator>
  85 void parallel_sort(RandomAccessIterator Start, RandomAccessIterator End,
  86                    const Comparator &Comp) {
  87   concurrency::parallel_sort(Start, End, Comp);
  88 }
  89 template <class IterTy, class FuncTy>
  90 void parallel_for_each(IterTy Begin, IterTy End, FuncTy Fn) {
  91   concurrency::parallel_for_each(Begin, End, Fn);
  92 }
  93
  94 template <class IndexTy, class FuncTy>
  95 void parallel_for_each_n(IndexTy Begin, IndexTy End, FuncTy Fn) {
  96   concurrency::parallel_for(Begin, End, Fn);
  97 }
  98
  99 #else
 100 const ptrdiff_t MinParallelSize = 1024;
 101
 102 /// Inclusive median.
 103 template <class RandomAccessIterator, class Comparator>
 104 RandomAccessIterator medianOf3(RandomAccessIterator Start,
 105                                RandomAccessIterator End,
 106                                const Comparator &Comp) {
 107   RandomAccessIterator Mid = Start + (std::distance(Start, End) / 2);
 108   return Comp(*Start, *(End - 1))
 109              ? (Comp(*Mid, *(End - 1)) ? (Comp(*Start, *Mid) ? Mid : Start)
 110                                        : End - 1)
 111              : (Comp(*Mid, *Start) ? (Comp(*(End - 1), *Mid) ? Mid : End - 1)
 112                                    : Start);
 113 }
 114
 115 template <class RandomAccessIterator, class Comparator>
 116 void parallel_quick_sort(RandomAccessIterator Start, RandomAccessIterator End,
 117                          const Comparator &Comp, TaskGroup &TG, size_t Depth) {
 118   // Do a sequential sort for small inputs.
 119   if (std::distance(Start, End) < detail::MinParallelSize || Depth == 0) {
 120     llvm::sort(Start, End, Comp);
 121     return;
 122   }
 123
 124   // Partition.
 125   auto Pivot = medianOf3(Start, End, Comp);
 126   // Move Pivot to End.
 127   std::swap(*(End - 1), *Pivot);
 128   Pivot = std::partition(Start, End - 1, [&Comp, End](decltype(*Start) V) {
 129     return Comp(V, *(End - 1));
 130   });
 131   // Move Pivot to middle of partition.
 132   std::swap(*Pivot, *(End - 1));
 133
 134   // Recurse.
 135   TG.spawn([=, &Comp, &TG] {
 136     parallel_quick_sort(Start, Pivot, Comp, TG, Depth - 1);
 137   });
 138   parallel_quick_sort(Pivot + 1, End, Comp, TG, Depth - 1);
 139 }
 140
 141 template <class RandomAccessIterator, class Comparator>
 142 void parallel_sort(RandomAccessIterator Start, RandomAccessIterator End,
 143                    const Comparator &Comp) {
 144   TaskGroup TG;
 145   parallel_quick_sort(Start, End, Comp, TG,
 146                       llvm::Log2_64(std::distance(Start, End)) + 1);
 147 }
 148
 149 template <class IterTy, class FuncTy>
 150 void parallel_for_each(IterTy Begin, IterTy End, FuncTy Fn) {
 151   // TaskGroup has a relatively high overhead, so we want to reduce
 152   // the number of spawn() calls. We'll create up to 1024 tasks here.
 153   // (Note that 1024 is an arbitrary number. This code probably needs
 154   // improving to take the number of available cores into account.)
 155   ptrdiff_t TaskSize = std::distance(Begin, End) / 1024;
 156   if (TaskSize == 0)
 157     TaskSize = 1;
 158
 159   TaskGroup TG;
 160   while (TaskSize < std::distance(Begin, End)) {
 161     TG.spawn([=, &Fn] { std::for_each(Begin, Begin + TaskSize, Fn); });
 162     Begin += TaskSize;
 163   }
 164   std::for_each(Begin, End, Fn);
 165 }
 166
 167 template <class IndexTy, class FuncTy>
 168 void parallel_for_each_n(IndexTy Begin, IndexTy End, FuncTy Fn) {
 169   ptrdiff_t TaskSize = (End - Begin) / 1024;
 170   if (TaskSize == 0)
 171     TaskSize = 1;
 172
 173   TaskGroup TG;
 174   IndexTy I = Begin;
 175   for (; I + TaskSize < End; I += TaskSize) {
 176     TG.spawn([=, &Fn] {
 177       for (IndexTy J = I, E = I + TaskSize; J != E; ++J)
 178         Fn(J);
 179     });
 180   }
 181   for (IndexTy J = I; J < End; ++J)
 182     Fn(J);
 183 }
 184
 185 #endif
 186
 187 #endif
 188
 189 template <typename Iter>
 190 using DefComparator =
 191     std::less<typename std::iterator_traits<Iter>::value_type>;
 192
 193 } // namespace detail
 194
 195 // sequential algorithm implementations.
 196 template <class Policy, class RandomAccessIterator,
 197           class Comparator = detail::DefComparator<RandomAccessIterator>>
 198 void sort(Policy policy, RandomAccessIterator Start, RandomAccessIterator End,
 199           const Comparator &Comp = Comparator()) {
 200   static_assert(is_execution_policy<Policy>::value,
 201                 "Invalid execution policy!");
 202   llvm::sort(Start, End, Comp);
 203 }
 204
 205 template <class Policy, class IterTy, class FuncTy>
 206 void for_each(Policy policy, IterTy Begin, IterTy End, FuncTy Fn) {
 207   static_assert(is_execution_policy<Policy>::value,
 208                 "Invalid execution policy!");
 209   std::for_each(Begin, End, Fn);
 210 }
 211
 212 template <class Policy, class IndexTy, class FuncTy>
 213 void for_each_n(Policy policy, IndexTy Begin, IndexTy End, FuncTy Fn) {
 214   static_assert(is_execution_policy<Policy>::value,
 215                 "Invalid execution policy!");
 216   for (IndexTy I = Begin; I != End; ++I)
 217     Fn(I);
 218 }
 219
 220 // Parallel algorithm implementations, only available when LLVM_ENABLE_THREADS
 221 // is true.
 222 #if LLVM_ENABLE_THREADS
 223 template <class RandomAccessIterator,
 224           class Comparator = detail::DefComparator<RandomAccessIterator>>
 225 void sort(parallel_execution_policy policy, RandomAccessIterator Start,
 226           RandomAccessIterator End, const Comparator &Comp = Comparator()) {
 227   detail::parallel_sort(Start, End, Comp);
 228 }
 229
 230 template <class IterTy, class FuncTy>
 231 void for_each(parallel_execution_policy policy, IterTy Begin, IterTy End,
 232               FuncTy Fn) {
 233   detail::parallel_for_each(Begin, End, Fn);
 234 }
 235
 236 template <class IndexTy, class FuncTy>
 237 void for_each_n(parallel_execution_policy policy, IndexTy Begin, IndexTy End,
 238                 FuncTy Fn) {
 239   detail::parallel_for_each_n(Begin, End, Fn);
 240 }
 241 #endif
 242
 243 } // namespace parallel
 244 } // namespace llvm
 245
 246 #endif // LLVM_SUPPORT_PARALLEL_H