[libc++][PSTL] Implement std::merge

Reviewed By: ldionne, #libc

Spies: pcwang-thead, libcxx-commits

Differential Revision: https://reviews.llvm.org/D151375
This commit is contained in:
Nikolas Klauser 2023-05-30 09:27:10 -07:00
parent 48e5f704c5
commit bf63b15bd4
14 changed files with 330 additions and 124 deletions

View File

@ -77,6 +77,7 @@ set(files
__algorithm/pstl_backends/cpu_backends/fill.h
__algorithm/pstl_backends/cpu_backends/find_if.h
__algorithm/pstl_backends/cpu_backends/for_each.h
__algorithm/pstl_backends/cpu_backends/merge.h
__algorithm/pstl_backends/cpu_backends/serial.h
__algorithm/pstl_backends/cpu_backends/thread.h
__algorithm/pstl_backends/cpu_backends/transform.h
@ -85,6 +86,7 @@ set(files
__algorithm/pstl_find.h
__algorithm/pstl_for_each.h
__algorithm/pstl_frontend_dispatch.h
__algorithm/pstl_merge.h
__algorithm/pstl_transform.h
__algorithm/push_heap.h
__algorithm/ranges_adjacent_find.h

View File

@ -72,6 +72,15 @@ implemented, all the algorithms will eventually forward to the basis algorithms
template <class _ExecutionPolicy, class _Iterator, class _SizeT, class _Tp>
void __pstl_fill_n(_Backend, _Iterator __first, _SizeT __n, const _Tp& __value);
template <class _ExecutionPolicy, class _terator1, class _Iterator2, class _OutIterator, class _Comp>
_OutIterator __pstl_merge(_Backend,
_Iterator1 __first1,
_Iterator1 __last1,
_Iterator2 __first2,
_Iterator2 __last2,
_OutIterator __result,
_Comp __comp);
// TODO: Complete this list
*/

View File

@ -20,6 +20,20 @@
// Cancel the execution of other jobs - they aren't needed anymore
void __cancel_execution();
template <class _RandomAccessIterator1,
class _RandomAccessIterator2,
class _RandomAccessIterator3,
class _Compare,
class _LeafMerge>
void __parallel_merge(
_RandomAccessIterator1 __first1,
_RandomAccessIterator1 __last1,
_RandomAccessIterator2 __first2,
_RandomAccessIterator2 __last2,
_RandomAccessIterator3 __outit,
_Compare __comp,
_LeafMerge __leaf_merge);
TODO: Document the parallel backend
*/
@ -27,6 +41,7 @@
#include <__algorithm/pstl_backends/cpu_backends/fill.h>
#include <__algorithm/pstl_backends/cpu_backends/find_if.h>
#include <__algorithm/pstl_backends/cpu_backends/for_each.h>
#include <__algorithm/pstl_backends/cpu_backends/merge.h>
#include <__algorithm/pstl_backends/cpu_backends/transform.h>
#endif // _LIBCPP___ALGORITHM_PSTL_BACKENDS_CPU_BACKEND_H

View File

@ -0,0 +1,79 @@
//===----------------------------------------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
#ifndef _LIBCPP___ALGORITHM_PSTL_BACKENDS_CPU_BACKENDS_MERGE_H
#define _LIBCPP___ALGORITHM_PSTL_BACKENDS_CPU_BACKENDS_MERGE_H
#include <__algorithm/merge.h>
#include <__algorithm/pstl_backends/cpu_backends/backend.h>
#include <__config>
#include <__iterator/iterator_traits.h>
#include <__type_traits/is_execution_policy.h>
#include <__utility/move.h>
#include <__utility/terminate_on_exception.h>
#if !defined(_LIBCPP_HAS_NO_PRAGMA_SYSTEM_HEADER)
# pragma GCC system_header
#endif
#if !defined(_LIBCPP_HAS_NO_INCOMPLETE_PSTL) && _LIBCPP_STD_VER >= 17
_LIBCPP_BEGIN_NAMESPACE_STD
template <class _ExecutionPolicy,
class _ForwardIterator1,
class _ForwardIterator2,
class _ForwardOutIterator,
class _Comp>
_LIBCPP_HIDE_FROM_ABI _ForwardOutIterator __pstl_merge(
__cpu_backend_tag,
_ForwardIterator1 __first1,
_ForwardIterator1 __last1,
_ForwardIterator2 __first2,
_ForwardIterator2 __last2,
_ForwardOutIterator __result,
_Comp __comp) {
if constexpr (__is_parallel_execution_policy_v<_ExecutionPolicy> &&
__has_random_access_iterator_category<_ForwardIterator1>::value &&
__has_random_access_iterator_category<_ForwardIterator2>::value &&
__has_random_access_iterator_category<_ForwardOutIterator>::value) {
return std::__terminate_on_exception([&] {
__par_backend::__parallel_merge(
__first1,
__last1,
__first2,
__last2,
__result,
__comp,
[](_ForwardIterator1 __g_first1,
_ForwardIterator1 __g_last1,
_ForwardIterator2 __g_first2,
_ForwardIterator2 __g_last2,
_ForwardOutIterator __g_result,
_Comp __g_comp) {
return std::__pstl_merge<__remove_parallel_policy_t<_ExecutionPolicy>>(
__cpu_backend_tag{},
std::move(__g_first1),
std::move(__g_last1),
std::move(__g_first2),
std::move(__g_last2),
std::move(__g_result),
std::move(__g_comp));
});
return __result + (__last1 - __first1) + (__last2 - __first2);
});
} else {
return std::merge(__first1, __last1, __first2, __last2, __result, __comp);
}
}
_LIBCPP_END_NAMESPACE_STD
#endif // !defined(_LIBCPP_HAS_NO_INCOMPLETE_PSTL) && _LIBCPP_STD_VER >= 17
#endif // _LIBCPP___ALGORITHM_PSTL_BACKENDS_CPU_BACKENDS_MERGE_H

View File

@ -30,6 +30,22 @@ _LIBCPP_HIDE_FROM_ABI void __parallel_for(_RandomAccessIterator __first, _Random
_LIBCPP_HIDE_FROM_ABI inline void __cancel_execution() {}
template <class _RandomAccessIterator1,
class _RandomAccessIterator2,
class _RandomAccessIterator3,
class _Compare,
class _LeafMerge>
_LIBCPP_HIDE_FROM_ABI void __parallel_merge(
_RandomAccessIterator1 __first1,
_RandomAccessIterator1 __last1,
_RandomAccessIterator2 __first2,
_RandomAccessIterator2 __last2,
_RandomAccessIterator3 __outit,
_Compare __comp,
_LeafMerge __leaf_merge) {
__leaf_merge(__first1, __last1, __first2, __last2, __outit, __comp);
}
// TODO: Complete this list
} // namespace __serial_cpu_backend

View File

@ -33,6 +33,22 @@ _LIBCPP_HIDE_FROM_ABI void __parallel_for(_RandomAccessIterator __first, _Random
_LIBCPP_HIDE_FROM_ABI inline void __cancel_execution() {}
template <class _RandomAccessIterator1,
class _RandomAccessIterator2,
class _RandomAccessIterator3,
class _Compare,
class _LeafMerge>
_LIBCPP_HIDE_FROM_ABI void __parallel_merge(
_RandomAccessIterator1 __first1,
_RandomAccessIterator1 __last1,
_RandomAccessIterator2 __first2,
_RandomAccessIterator2 __last2,
_RandomAccessIterator3 __outit,
_Compare __comp,
_LeafMerge __leaf_merge) {
__leaf_merge(__first1, __last1, __first2, __last2, __outit, __comp);
}
} // namespace __thread_cpu_backend
} // namespace __par_backend

View File

@ -0,0 +1,56 @@
//===----------------------------------------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
#ifndef _LIBCPP___ALGORITHM_PSTL_MERGE_H
#define _LIBCPP___ALGORITHM_PSTL_MERGE_H
#include <__algorithm/pstl_backend.h>
#include <__config>
#include <__functional/operations.h>
#include <__type_traits/is_execution_policy.h>
#include <__type_traits/remove_cvref.h>
#if !defined(_LIBCPP_HAS_NO_PRAGMA_SYSTEM_HEADER)
# pragma GCC system_header
#endif
#if !defined(_LIBCPP_HAS_NO_INCOMPLETE_PSTL) && _LIBCPP_STD_VER >= 17
_LIBCPP_BEGIN_NAMESPACE_STD
template <class _ExecutionPolicy,
class _ForwardIterator1,
class _ForwardIterator2,
class _ForwardOutIterator,
class _Comp = std::less<>,
class _RawPolicy = __remove_cvref_t<_ExecutionPolicy>,
enable_if_t<is_execution_policy_v<_RawPolicy>, int> = 0>
_LIBCPP_HIDE_FROM_ABI _ForwardOutIterator
merge(_ExecutionPolicy&&,
_ForwardIterator1 __first1,
_ForwardIterator1 __last1,
_ForwardIterator2 __first2,
_ForwardIterator2 __last2,
_ForwardOutIterator __result,
_Comp __comp = {}) {
using _Backend = typename __select_backend<_RawPolicy>::type;
return std::__pstl_merge<_RawPolicy>(
_Backend{},
std::move(__first1),
std::move(__last1),
std::move(__first2),
std::move(__last2),
std::move(__result),
std::move(__comp));
}
_LIBCPP_END_NAMESPACE_STD
#endif // !defined(_LIBCPP_HAS_NO_INCOMPLETE_PSTL) && _LIBCPP_STD_VER >= 17
#endif // _LIBCPP___ALGORITHM_PSTL_MERGE_H

View File

@ -2869,89 +2869,6 @@ _RandomAccessIterator __pattern_remove_if(
});
}
//------------------------------------------------------------------------
// merge
//------------------------------------------------------------------------
template <class _ForwardIterator1, class _ForwardIterator2, class _OutputIterator, class _Compare>
_OutputIterator __brick_merge(
_ForwardIterator1 __first1,
_ForwardIterator1 __last1,
_ForwardIterator2 __first2,
_ForwardIterator2 __last2,
_OutputIterator __d_first,
_Compare __comp,
/* __is_vector = */ std::false_type) noexcept {
return std::merge(__first1, __last1, __first2, __last2, __d_first, __comp);
}
template <class _RandomAccessIterator1, class _RandomAccessIterator2, class _RandomAccessIterator3, class _Compare>
_RandomAccessIterator3 __brick_merge(
_RandomAccessIterator1 __first1,
_RandomAccessIterator1 __last1,
_RandomAccessIterator2 __first2,
_RandomAccessIterator2 __last2,
_RandomAccessIterator3 __d_first,
_Compare __comp,
/* __is_vector = */ std::true_type) noexcept {
// TODO: vectorize
return std::merge(__first1, __last1, __first2, __last2, __d_first, __comp);
}
template <class _Tag,
class _ExecutionPolicy,
class _ForwardIterator1,
class _ForwardIterator2,
class _OutputIterator,
class _Compare>
_OutputIterator __pattern_merge(
_Tag,
_ExecutionPolicy&&,
_ForwardIterator1 __first1,
_ForwardIterator1 __last1,
_ForwardIterator2 __first2,
_ForwardIterator2 __last2,
_OutputIterator __d_first,
_Compare __comp) noexcept {
return __internal::__brick_merge(
__first1, __last1, __first2, __last2, __d_first, __comp, typename _Tag::__is_vector{});
}
template <class _IsVector,
class _ExecutionPolicy,
class _RandomAccessIterator1,
class _RandomAccessIterator2,
class _RandomAccessIterator3,
class _Compare>
_RandomAccessIterator3 __pattern_merge(
__parallel_tag<_IsVector> __tag,
_ExecutionPolicy&& __exec,
_RandomAccessIterator1 __first1,
_RandomAccessIterator1 __last1,
_RandomAccessIterator2 __first2,
_RandomAccessIterator2 __last2,
_RandomAccessIterator3 __d_first,
_Compare __comp) {
using __backend_tag = typename decltype(__tag)::__backend_tag;
__par_backend::__parallel_merge(
__backend_tag{},
std::forward<_ExecutionPolicy>(__exec),
__first1,
__last1,
__first2,
__last2,
__d_first,
__comp,
[](_RandomAccessIterator1 __f1,
_RandomAccessIterator1 __l1,
_RandomAccessIterator2 __f2,
_RandomAccessIterator2 __l2,
_RandomAccessIterator3 __f3,
_Compare __comp) { return __internal::__brick_merge(__f1, __l1, __f2, __l2, __f3, __comp, _IsVector{}); });
return __d_first + (__last1 - __first1) + (__last2 - __first2);
}
//------------------------------------------------------------------------
// inplace_merge
//------------------------------------------------------------------------

View File

@ -763,37 +763,6 @@ is_sorted(_ExecutionPolicy&& __exec, _ForwardIterator __first, _ForwardIterator
}
// [alg.merge]
template <class _ExecutionPolicy,
class _ForwardIterator1,
class _ForwardIterator2,
class _ForwardIterator,
class _Compare>
__pstl::__internal::__enable_if_execution_policy<_ExecutionPolicy, _ForwardIterator>
merge(_ExecutionPolicy&& __exec,
_ForwardIterator1 __first1,
_ForwardIterator1 __last1,
_ForwardIterator2 __first2,
_ForwardIterator2 __last2,
_ForwardIterator __d_first,
_Compare __comp) {
auto __dispatch_tag = __pstl::__internal::__select_backend(__exec, __first1, __first2, __d_first);
return __pstl::__internal::__pattern_merge(
__dispatch_tag, std::forward<_ExecutionPolicy>(__exec), __first1, __last1, __first2, __last2, __d_first, __comp);
}
template <class _ExecutionPolicy, class _ForwardIterator1, class _ForwardIterator2, class _ForwardIterator>
__pstl::__internal::__enable_if_execution_policy<_ExecutionPolicy, _ForwardIterator>
merge(_ExecutionPolicy&& __exec,
_ForwardIterator1 __first1,
_ForwardIterator1 __last1,
_ForwardIterator2 __first2,
_ForwardIterator2 __last2,
_ForwardIterator __d_first) {
return std::merge(
std::forward<_ExecutionPolicy>(__exec), __first1, __last1, __first2, __last2, __d_first, std::less<>());
}
template <class _ExecutionPolicy, class _BidirectionalIterator, class _Compare>
__pstl::__internal::__enable_if_execution_policy<_ExecutionPolicy, void>
inplace_merge(_ExecutionPolicy&& __exec,

View File

@ -98,16 +98,6 @@ __parallel_stable_sort(__pstl::__internal::__serial_backend_tag, _ExecutionPolic
__leaf_sort(__first, __last, __comp);
}
template <class _ExecutionPolicy, typename _RandomAccessIterator1, typename _RandomAccessIterator2,
typename _RandomAccessIterator3, typename _Compare, typename _LeafMerge>
_LIBCPP_HIDE_FROM_ABI void
__parallel_merge(__pstl::__internal::__serial_backend_tag, _ExecutionPolicy&&, _RandomAccessIterator1 __first1,
_RandomAccessIterator1 __last1, _RandomAccessIterator2 __first2, _RandomAccessIterator2 __last2,
_RandomAccessIterator3 __outit, _Compare __comp, _LeafMerge __leaf_merge)
{
__leaf_merge(__first1, __last1, __first2, __last2, __outit, __comp);
}
template <class _ExecutionPolicy, typename _F1, typename _F2>
_LIBCPP_HIDE_FROM_ABI void
__parallel_invoke(__pstl::__internal::__serial_backend_tag, _ExecutionPolicy&&, _F1&& __f1, _F2&& __f2)

View File

@ -1805,6 +1805,7 @@ template <class BidirectionalIterator, class Compare>
#include <__algorithm/pstl_fill.h>
#include <__algorithm/pstl_find.h>
#include <__algorithm/pstl_for_each.h>
#include <__algorithm/pstl_merge.h>
#include <__algorithm/pstl_transform.h>
#include <__algorithm/push_heap.h>
#include <__algorithm/ranges_adjacent_find.h>

View File

@ -338,6 +338,9 @@ module std [system] {
module pstl_backends_cpu_backends_for_each {
private header "__algorithm/pstl_backends/cpu_backends/for_each.h"
}
module pstl_backends_cpu_backends_merge {
private header "__algorithm/pstl_backends/cpu_backends/merge.h"
}
module pstl_backends_cpu_backends_serial {
private header "__algorithm/pstl_backends/cpu_backends/serial.h"
}

View File

@ -120,6 +120,7 @@ END-SCRIPT
#include <__algorithm/pstl_backends/cpu_backends/fill.h> // expected-error@*:* {{use of private header from outside its module: '__algorithm/pstl_backends/cpu_backends/fill.h'}}
#include <__algorithm/pstl_backends/cpu_backends/find_if.h> // expected-error@*:* {{use of private header from outside its module: '__algorithm/pstl_backends/cpu_backends/find_if.h'}}
#include <__algorithm/pstl_backends/cpu_backends/for_each.h> // expected-error@*:* {{use of private header from outside its module: '__algorithm/pstl_backends/cpu_backends/for_each.h'}}
#include <__algorithm/pstl_backends/cpu_backends/merge.h> // expected-error@*:* {{use of private header from outside its module: '__algorithm/pstl_backends/cpu_backends/merge.h'}}
#include <__algorithm/pstl_backends/cpu_backends/serial.h> // expected-error@*:* {{use of private header from outside its module: '__algorithm/pstl_backends/cpu_backends/serial.h'}}
#include <__algorithm/pstl_backends/cpu_backends/thread.h> // expected-error@*:* {{use of private header from outside its module: '__algorithm/pstl_backends/cpu_backends/thread.h'}}
#include <__algorithm/pstl_backends/cpu_backends/transform.h> // expected-error@*:* {{use of private header from outside its module: '__algorithm/pstl_backends/cpu_backends/transform.h'}}

View File

@ -0,0 +1,132 @@
//===----------------------------------------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
// UNSUPPORTED: c++03, c++11, c++14
// UNSUPPORTED: libcpp-has-no-incomplete-pstl
// template<class ExecutionPolicy, class ForwardIterator1, class ForwardIterator2,
// class ForwardIterator>
// ForwardIterator
// merge(ExecutionPolicy&& exec,
// ForwardIterator1 first1, ForwardIterator1 last1,
// ForwardIterator2 first2, ForwardIterator2 last2,
// ForwardIterator result);
//
// template<class ExecutionPolicy, class ForwardIterator1, class ForwardIterator2,
// class ForwardIterator, class Compare>
// ForwardIterator
// merge(ExecutionPolicy&& exec,
// ForwardIterator1 first1, ForwardIterator1 last1,
// ForwardIterator2 first2, ForwardIterator2 last2,
// ForwardIterator result, Compare comp);
#include <algorithm>
#include <array>
#include <cassert>
#include <iterator>
#include <numeric>
#include <vector>
#include "type_algorithms.h"
#include "test_execution_policies.h"
#include "test_iterators.h"
template <class Iter1, class Iter2>
struct Test {
template <class Policy>
void operator()(Policy&& policy) {
{ // simple test
int a[] = {1, 3, 5, 7, 9};
int b[] = {2, 4, 6, 8, 10};
std::array<int, std::size(a) + std::size(b)> out;
std::merge(
policy, Iter1(std::begin(a)), Iter1(std::end(a)), Iter2(std::begin(b)), Iter2(std::end(b)), std::begin(out));
assert((out == std::array{1, 2, 3, 4, 5, 6, 7, 8, 9, 10}));
}
{ // check that it works with the first range being empty
std::array<int, 0> a;
int b[] = {2, 4, 6, 8, 10};
std::array<int, std::size(a) + std::size(b)> out;
std::merge(
policy, Iter1(std::begin(a)), Iter1(std::end(a)), Iter2(std::begin(b)), Iter2(std::end(b)), std::begin(out));
assert((out == std::array{2, 4, 6, 8, 10}));
}
{ // check that it works with the second range being empty
int a[] = {2, 4, 6, 8, 10};
std::array<int, 0> b;
std::array<int, std::size(a) + std::size(b)> out;
std::merge(
policy, Iter1(std::begin(a)), Iter1(std::end(a)), Iter2(std::begin(b)), Iter2(std::end(b)), std::begin(out));
assert((out == std::array{2, 4, 6, 8, 10}));
}
{ // check that it works when the ranges don't have the same length
int a[] = {2, 4, 6, 8, 10};
int b[] = {3, 4};
std::array<int, std::size(a) + std::size(b)> out;
std::merge(
policy, Iter1(std::begin(a)), Iter1(std::end(a)), Iter2(std::begin(b)), Iter2(std::end(b)), std::begin(out));
assert((out == std::array{2, 3, 4, 4, 6, 8, 10}));
}
{ // check that large ranges work
std::vector<int> a(100);
std::vector<int> b(100);
{
int i = 0;
for (auto& e : a) {
e = i;
i += 2;
}
}
{
int i = 1;
for (auto& e : b) {
e = i;
i += 2;
}
}
std::vector<int> out(std::size(a) + std::size(b));
std::merge(
Iter1(a.data()), Iter1(a.data() + a.size()), Iter2(b.data()), Iter2(b.data() + b.size()), std::begin(out));
std::vector<int> expected(200);
std::iota(expected.begin(), expected.end(), 0);
assert(std::equal(out.begin(), out.end(), expected.begin()));
}
{ // check that the predicate is used
int a[] = {10, 9, 8, 7};
int b[] = {8, 4, 3};
std::array<int, std::size(a) + std::size(b)> out;
std::merge(
policy,
Iter1(std::begin(a)),
Iter1(std::end(a)),
Iter2(std::begin(b)),
Iter2(std::end(b)),
std::begin(out),
std::greater{});
assert((out == std::array{10, 9, 8, 8, 7, 4, 3}));
}
}
};
int main(int, char**) {
types::for_each(types::forward_iterator_list<int*>{}, types::apply_type_identity{[](auto v) {
using Iter = typename decltype(v)::type;
types::for_each(
types::forward_iterator_list<int*>{},
TestIteratorWithPolicies<types::partial_instantiation<Test, Iter>::template apply>{});
}});
return 0;
}