Nikolas Klauser
925f4f27b7
Instead of updating the status array every time `operator()` is called, we can instead update it once every time the entire array has been read. That allows the compiler to vectorize the updating code, improving performance. Apple M4: ``` Benchmark old new Difference % Difference ---------------------------------------- ------- ------- ------------ -------------- std::mt19937::operator() 1.39 0.65 -0.74 -53.26% std::mt19937::operator()_(into_array) 172.92 80.69 -92.23 -53.34% std::mt19937_64::operator() 1.46 0.99 -0.47 -31.99% std::mt19937_64::operator()_(into_array) 187.31 108.10 -79.21 -42.29% ``` Fixes #197221
63 lines
1.9 KiB
C++
63 lines
1.9 KiB
C++
//===----------------------------------------------------------------------===//
|
|
//
|
|
// 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
|
|
|
|
#include <algorithm>
|
|
#include <array>
|
|
#include <cstddef>
|
|
#include <cstdint>
|
|
#include <functional>
|
|
#include <random>
|
|
|
|
#include "benchmark/benchmark.h"
|
|
|
|
constexpr std::size_t MAX_BUFFER_LEN = 256;
|
|
constexpr std::size_t MAX_SEED_LEN = 16;
|
|
|
|
static void BM_SeedSeq_Generate(benchmark::State& state) {
|
|
std::array<std::uint32_t, MAX_BUFFER_LEN> buffer;
|
|
std::array<std::uint32_t, MAX_SEED_LEN> seeds;
|
|
{
|
|
std::random_device rd;
|
|
std::generate(std::begin(seeds), std::begin(seeds) + state.range(0), [&]() { return rd(); });
|
|
}
|
|
std::seed_seq seed(std::begin(seeds), std::begin(seeds) + state.range(0));
|
|
for (auto _ : state) {
|
|
seed.generate(std::begin(buffer), std::begin(buffer) + state.range(1));
|
|
}
|
|
}
|
|
BENCHMARK(BM_SeedSeq_Generate)->Ranges({{1, MAX_SEED_LEN}, {1, MAX_BUFFER_LEN}});
|
|
|
|
template <class Engine>
|
|
static void BM_engine(benchmark::State& state) {
|
|
Engine engine;
|
|
|
|
for (auto _ : state) {
|
|
benchmark::DoNotOptimize(engine());
|
|
}
|
|
}
|
|
BENCHMARK(BM_engine<std::mt19937_64>)->Name("std::mt19937_64::operator()");
|
|
BENCHMARK(BM_engine<std::mt19937>)->Name("std::mt19937::operator()");
|
|
|
|
template <class Engine>
|
|
static void BM_engine_array(benchmark::State& state) {
|
|
Engine engine;
|
|
|
|
typename Engine::result_type buffer[128];
|
|
|
|
for (auto _ : state) {
|
|
std::generate(std::begin(buffer), std::end(buffer), std::ref(engine));
|
|
benchmark::DoNotOptimize(buffer);
|
|
}
|
|
}
|
|
BENCHMARK(BM_engine_array<std::mt19937_64>)->Name("std::mt19937_64::operator() (into array)");
|
|
BENCHMARK(BM_engine_array<std::mt19937>)->Name("std::mt19937::operator() (into array)");
|
|
|
|
BENCHMARK_MAIN();
|