Skip to content

Optimize is_permutation for vector<bool> #6148

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
AlexGuteniev wants to merge 13 commits into
base: main
Choose a base branch
from
+167 −1
Open

Optimize is_permutation for vector<bool> #6148

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
13 commits
Select commit Hold shift + click to select a range
408cd46
Test
AlexGuteniev Mar 7, 2026
90e156c
Benchmark
AlexGuteniev Mar 7, 2026
bf9ff31
Mutation!
AlexGuteniev Mar 7, 2026
549a9d3
Generalize
AlexGuteniev Mar 7, 2026
5a0ba81
benchmark plain dynamic array
AlexGuteniev Mar 8, 2026
443e021
Restore mismatch
AlexGuteniev Mar 8, 2026
08b2657
format
AlexGuteniev Mar 8, 2026
4ca22be
Admit we rely on random access iterators, extract
AlexGuteniev Mar 8, 2026
21888e2
Expand coverage
AlexGuteniev Mar 8, 2026
9aad0b0
Merge branch 'main' into order-mutant-vector-bool
AlexGuteniev Apr 29, 2026
b767508
copypasta error
AlexGuteniev Apr 29, 2026
3f5385f
Merge branch 'microsoft:main' into order-mutant-vector-bool
AlexGuteniev May 30, 2026
d01fcda
Ok let's check the unwrapped iterators
AlexGuteniev May 30, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions benchmarks/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -147,4 +147,5 @@ add_benchmark(vector_bool_copy_n src/vector_bool_copy_n.cpp)
add_benchmark(vector_bool_count src/vector_bool_count.cpp)
add_benchmark(vector_bool_meow_of src/vector_bool_meow_of.cpp)
add_benchmark(vector_bool_move src/vector_bool_move.cpp)
add_benchmark(vector_bool_permute src/vector_bool_permute.cpp)
add_benchmark(vector_bool_transform src/vector_bool_transform.cpp)
83 changes: 83 additions & 0 deletions benchmarks/src/vector_bool_permute.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,83 @@
// Copyright (c) Microsoft Corporation.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception

#include <benchmark/benchmark.h>
//
#include <algorithm>
#include <cstddef>
#include <random>
#include <vector>

#include "utility.hpp"

using namespace std;

enum class equality { eq, neq };
enum class args { three, four };


template <equality Eq, args Args>
void perm_arr_check(benchmark::State& state) {
const auto size = static_cast<size_t>(state.range(0));
vector<bool> v1 = random_vector<bool>(size);
vector<bool> v2;
if constexpr (Eq == equality::eq) {
v2 = v1;
} else {
v2 = random_vector<bool>(size, 1u);
}

auto a1 = make_unique<bool[]>(size);
auto a2 = make_unique<bool[]>(size);
copy(v1.begin(), v1.end(), a1.get());
copy(v2.begin(), v2.end(), a2.get());

for (auto _ : state) {
benchmark::DoNotOptimize(a1.get());
benchmark::DoNotOptimize(a2.get());
bool r;
if constexpr (Args == args::three) {
r = is_permutation(a1.get(), a1.get() + size, a2.get());
} else {
r = is_permutation(a1.get(), a1.get() + size, a2.get(), a2.get() + size);
}
benchmark::DoNotOptimize(r);
}
}


template <equality Eq, args Args>
void perm_vbool_check(benchmark::State& state) {
const auto size = static_cast<size_t>(state.range(0));
vector<bool> v1 = random_vector<bool>(size);
vector<bool> v2;
if constexpr (Eq == equality::eq) {
v2 = v1;
} else {
v2 = random_vector<bool>(size, 1u);
}

for (auto _ : state) {
benchmark::DoNotOptimize(v1);
benchmark::DoNotOptimize(v2);
bool r;
if constexpr (Args == args::three) {
r = is_permutation(v1.begin(), v1.end(), v2.begin());
} else {
r = is_permutation(v1.begin(), v1.end(), v2.begin(), v2.end());
}
benchmark::DoNotOptimize(r);
}
}

BENCHMARK(perm_arr_check<equality::eq, args::three>)->RangeMultiplier(64)->Range(64, 64 << 10);
BENCHMARK(perm_arr_check<equality::eq, args::four>)->RangeMultiplier(64)->Range(64, 64 << 10);
BENCHMARK(perm_arr_check<equality::neq, args::three>)->RangeMultiplier(64)->Range(64, 64 << 10);
BENCHMARK(perm_arr_check<equality::neq, args::four>)->RangeMultiplier(64)->Range(64, 64 << 10);

BENCHMARK(perm_vbool_check<equality::eq, args::three>)->RangeMultiplier(64)->Range(64, 64 << 10);
BENCHMARK(perm_vbool_check<equality::eq, args::four>)->RangeMultiplier(64)->Range(64, 64 << 10);
BENCHMARK(perm_vbool_check<equality::neq, args::three>)->RangeMultiplier(64)->Range(64, 64 << 10);
BENCHMARK(perm_vbool_check<equality::neq, args::four>)->RangeMultiplier(64)->Range(64, 64 << 10);

BENCHMARK_MAIN();
39 changes: 38 additions & 1 deletion stl/inc/algorithm
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1270,6 +1270,21 @@ namespace ranges {
} // namespace ranges
#endif // _HAS_CXX20

template <class _FwdIt1, class _FwdIt2>
_NODISCARD _CONSTEXPR20 bool _Is_permutation_of_bool(_FwdIt1 _First1, _FwdIt1 _Last1, _FwdIt2 _First2, _FwdIt2 _Last2) {
Comment on lines +1273 to +1274

@AlexGuteniev AlexGuteniev May 30, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think this one is premature. I don't attempt to handle std::ranges::is_permutation yet, so there's no distinct sentinel type. When we handle it, we'll see how it should be done.

if constexpr (!_Is_vb_iterator<_FwdIt1> && !_Is_vb_iterator<_FwdIt2>) {
auto _Pair = _STD mismatch(_First1, _Last1, _First2);
Comment on lines +1275 to +1276

@AlexGuteniev AlexGuteniev May 30, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No.

This is intentional optimization decision. It is explained in the PR description.

mismatch is good for non-vb iterators (either vectorized, or better than count), but for just one vb iterator it flips as count is SWAR with popcount and mismatch is individual bit matching.

_First1 = _STD move(_Pair.first);
_First2 = _STD move(_Pair.second);

if (_First1 == _Last1) {
return true;
}
}

return _STD count(_First1, _Last1, true) == _STD count(_First2, _Last2, true);
}

_EXPORT_STD template <class _FwdIt1, class _FwdIt2, class _Pr>
_NODISCARD _CONSTEXPR20 bool is_permutation(_FwdIt1 _First1, _FwdIt1 _Last1, _FwdIt2 _First2, _Pr _Pred) {
// test if [_First1, _Last1) == permuted [_First2, ...)
Expand All @@ -1278,6 +1293,18 @@ _NODISCARD _CONSTEXPR20 bool is_permutation(_FwdIt1 _First1, _FwdIt1 _Last1, _Fw
const auto _ULast1 = _STD _Get_unwrapped(_Last1);
auto _UFirst2 = _STD _Get_unwrapped_n(_First2, _STD _Idl_distance<_FwdIt1>(_UFirst1, _ULast1));

if constexpr (is_same_v<_Iter_value_t<decltype(_UFirst1)>, bool>
&& is_same_v<_Iter_value_t<decltype(_UFirst2)>, bool> //
&& _Is_ranges_random_iter_v<decltype(_UFirst1)> //
&& _Is_ranges_random_iter_v<decltype(_UFirst2)>
Comment on lines +1296 to +1299

@AlexGuteniev AlexGuteniev May 30, 2026

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The 4-arg version checks the wrapped iterator instead. The check it pre-existing.
Maybe I need to change it to check the unwrapped iterators too though?

@AlexGuteniev AlexGuteniev May 30, 2026

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Checking the unwrapped iterators.

&& _Is_any_of_v<_Pr,
#if _HAS_CXX20
_RANGES equal_to,
#endif
equal_to<>, equal_to<bool>>) {
return _STD _Is_permutation_of_bool(_UFirst1, _ULast1, _UFirst2, _UFirst2 + (_ULast1 - _UFirst1));
}

for (;; ++_UFirst1, (void) ++_UFirst2) { // trim matching prefix
if (_UFirst1 == _ULast1) { // everything matched
return true;
Expand Down Expand Up @@ -1310,11 +1337,21 @@ _NODISCARD _CONSTEXPR20 bool is_permutation(
auto _ULast1 = _STD _Get_unwrapped(_Last1);
auto _UFirst2 = _STD _Get_unwrapped(_First2);
auto _ULast2 = _STD _Get_unwrapped(_Last2);
if constexpr (_Is_ranges_random_iter_v<_FwdIt1> && _Is_ranges_random_iter_v<_FwdIt2>) {
if constexpr (_Is_ranges_random_iter_v<decltype(_UFirst1)> && _Is_ranges_random_iter_v<decltype(_UFirst2)>) {
if (_ULast1 - _UFirst1 != _ULast2 - _UFirst2) {
return false;
}

if constexpr (is_same_v<_Iter_value_t<decltype(_UFirst1)>, bool>
&& is_same_v<_Iter_value_t<decltype(_UFirst2)>, bool>
&& _Is_any_of_v<_Pr,
#if _HAS_CXX20
_RANGES equal_to,
#endif
equal_to<>, equal_to<bool>>) {
return _STD _Is_permutation_of_bool(_UFirst1, _ULast1, _UFirst2, _ULast2);
}

for (; _UFirst1 != _ULast1; ++_UFirst1, (void) ++_UFirst2) { // trim matching prefix
if (!_Pred(*_UFirst1, *_UFirst2)) {
// found first inequality, check match counts in suffix
Expand Down
45 changes: 45 additions & 0 deletions tests/std/tests/GH_000625_vector_bool_optimization/test.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1518,6 +1518,49 @@ CONSTEXPR20 bool test_copy_part_2() {
return true;
}

CONSTEXPR20 bool test_is_permutation() {
constexpr bool perm_source_raw[] = {
false, true, false, false, false, true, false, true, true, false, false, true, false, false, true, true};
const vector<bool> perm_source(begin(perm_source_raw), end(perm_source_raw));

// Actually a permutation
constexpr bool perm_raw[] = {
true, false, true, false, false, true, false, false, true, false, true, false, true, false, false, true};
const vector<bool> perm(begin(perm_raw), end(perm_raw));
assert(is_permutation(perm_source.begin(), perm_source.end(), perm.begin()));
assert(is_permutation(perm_source.begin(), perm_source.end(), perm.begin(), perm.end()));
assert(is_permutation(begin(perm_source_raw), end(perm_source_raw), perm.begin()));
assert(is_permutation(begin(perm_source_raw), end(perm_source_raw), perm.begin(), perm.end()));
assert(is_permutation(perm_source.begin(), perm_source.end(), begin(perm_raw)));
assert(is_permutation(perm_source.begin(), perm_source.end(), begin(perm_raw), end(perm_raw)));
assert(is_permutation(begin(perm_source_raw), end(perm_source_raw), begin(perm_raw)));
assert(is_permutation(begin(perm_source_raw), end(perm_source_raw), begin(perm_raw), end(perm_raw)));

// One extra true value
constexpr bool extra_true_raw[] = {
true, false, true, false, false, true, false, false, true, true, true, false, true, false, false, true};
const vector<bool> extra_true(begin(extra_true_raw), end(extra_true_raw));
assert(!is_permutation(perm_source.begin(), perm_source.end(), extra_true.begin()));
assert(!is_permutation(perm_source.begin(), perm_source.end(), extra_true.begin(), extra_true.end()));
assert(!is_permutation(begin(perm_source_raw), end(perm_source_raw), extra_true.begin()));
assert(!is_permutation(begin(perm_source_raw), end(perm_source_raw), extra_true.begin(), extra_true.end()));
assert(!is_permutation(perm_source.begin(), perm_source.end(), begin(extra_true_raw)));
assert(!is_permutation(perm_source.begin(), perm_source.end(), begin(extra_true_raw), end(extra_true_raw)));
assert(!is_permutation(begin(perm_source_raw), end(perm_source_raw), begin(extra_true_raw)));
assert(!is_permutation(begin(perm_source_raw), end(perm_source_raw), begin(extra_true_raw), end(extra_true_raw)));

// One element longer
constexpr bool longer_raw[] = {
true, false, true, false, false, true, false, false, true, false, true, false, true, false, false, true, false};
const vector<bool> longer(begin(longer_raw), end(longer_raw));
assert(!is_permutation(perm_source.begin(), perm_source.end(), longer.begin(), longer.end()));
assert(!is_permutation(begin(perm_source_raw), end(perm_source_raw), longer.begin(), longer.end()));
assert(!is_permutation(perm_source.begin(), perm_source.end(), begin(longer_raw), end(longer_raw)));
assert(!is_permutation(begin(perm_source_raw), end(perm_source_raw), begin(longer_raw), end(longer_raw)));

return true;
}

void initialize_randomness(mt19937_64& gen) {
constexpr size_t n = mt19937_64::state_size;
constexpr size_t w = mt19937_64::word_size;
Expand Down Expand Up @@ -1710,6 +1753,7 @@ static_assert(test_meow_of());
#if defined(__clang__) || defined(__EDG__) // TRANSITION, VSO-2574489
static_assert(test_copy_part_1());
static_assert(test_copy_part_2());
static_assert(test_is_permutation());
#endif // ^^^ no workaround ^^^
#endif // _HAS_CXX20

Expand All @@ -1721,6 +1765,7 @@ int main() {
test_meow_of();
test_copy_part_1();
test_copy_part_2();
test_is_permutation();

test_huge_vector_bool();

Expand Down