108 lines
3.7 KiB
C++
108 lines
3.7 KiB
C++
// SPDX-License-Identifier: BSD-2-Clause
|
|
|
|
// This code is part of the sfizz library and is licensed under a BSD 2-clause
|
|
// license. You should have receive a LICENSE.md file along with the code.
|
|
// If not, contact the sfizz maintainers at https://github.com/sfztools/sfizz
|
|
|
|
#include "SIMDHelpers.h"
|
|
#include "Macros.h"
|
|
#include <benchmark/benchmark.h>
|
|
#include <random>
|
|
#include <numeric>
|
|
#include <vector>
|
|
#include <cmath>
|
|
#include <iostream>
|
|
|
|
class AddArray : public benchmark::Fixture {
|
|
public:
|
|
void SetUp(const ::benchmark::State& state) {
|
|
std::random_device rd { };
|
|
std::mt19937 gen { rd() };
|
|
std::uniform_real_distribution<float> dist { 0, 1 };
|
|
input = std::vector<float>(state.range(0));
|
|
output = std::vector<float>(state.range(0));
|
|
std::generate(output.begin(), output.end(), [&]() { return dist(gen); });
|
|
std::generate(input.begin(), input.end(), [&]() { return dist(gen); });
|
|
}
|
|
|
|
void TearDown(const ::benchmark::State& state) {
|
|
UNUSED(state);
|
|
}
|
|
|
|
std::vector<float> input;
|
|
std::vector<float> output;
|
|
};
|
|
|
|
BENCHMARK_DEFINE_F(AddArray, Value_Scalar)(benchmark::State& state) {
|
|
for (auto _ : state)
|
|
{
|
|
sfz::setSIMDOpStatus<float>(sfz::SIMDOps::add1, false);
|
|
sfz::add1<float>(1.1f, absl::MakeSpan(output));
|
|
}
|
|
}
|
|
|
|
BENCHMARK_DEFINE_F(AddArray, Value_SIMD)(benchmark::State& state) {
|
|
for (auto _ : state)
|
|
{
|
|
sfz::setSIMDOpStatus<float>(sfz::SIMDOps::add1, true);
|
|
sfz::add1<float>(1.1f, absl::MakeSpan(output));
|
|
}
|
|
}
|
|
|
|
BENCHMARK_DEFINE_F(AddArray, Value_Scalar_Unaligned)(benchmark::State& state) {
|
|
for (auto _ : state)
|
|
{
|
|
sfz::setSIMDOpStatus<float>(sfz::SIMDOps::add1, false);
|
|
sfz::add1<float>(1.1f, absl::MakeSpan(output).subspan(1));
|
|
}
|
|
}
|
|
|
|
BENCHMARK_DEFINE_F(AddArray, Value_SIMD_Unaligned)(benchmark::State& state) {
|
|
for (auto _ : state)
|
|
{
|
|
sfz::setSIMDOpStatus<float>(sfz::SIMDOps::add1, true);
|
|
sfz::add1<float>(1.1f, absl::MakeSpan(output).subspan(1));
|
|
}
|
|
}
|
|
|
|
BENCHMARK_DEFINE_F(AddArray, Scalar)(benchmark::State& state) {
|
|
for (auto _ : state)
|
|
{
|
|
sfz::setSIMDOpStatus<float>(sfz::SIMDOps::add, false);
|
|
sfz::add<float>(input, absl::MakeSpan(output));
|
|
}
|
|
}
|
|
|
|
BENCHMARK_DEFINE_F(AddArray, SIMD)(benchmark::State& state) {
|
|
for (auto _ : state)
|
|
{
|
|
sfz::setSIMDOpStatus<float>(sfz::SIMDOps::add, true);
|
|
sfz::add<float>(input, absl::MakeSpan(output));
|
|
}
|
|
}
|
|
|
|
BENCHMARK_DEFINE_F(AddArray, Scalar_Unaligned)(benchmark::State& state) {
|
|
for (auto _ : state)
|
|
{
|
|
sfz::setSIMDOpStatus<float>(sfz::SIMDOps::add, false);
|
|
sfz::add<float>(absl::MakeSpan(input).subspan(1), absl::MakeSpan(output).subspan(1));
|
|
}
|
|
}
|
|
|
|
BENCHMARK_DEFINE_F(AddArray, SIMD_Unaligned)(benchmark::State& state) {
|
|
for (auto _ : state)
|
|
{
|
|
sfz::setSIMDOpStatus<float>(sfz::SIMDOps::add, true);
|
|
sfz::add<float>(absl::MakeSpan(input).subspan(1), absl::MakeSpan(output).subspan(1));
|
|
}
|
|
}
|
|
|
|
BENCHMARK_REGISTER_F(AddArray, Value_Scalar)->RangeMultiplier(4)->Range(1 << 2, 1 << 12);
|
|
BENCHMARK_REGISTER_F(AddArray, Value_SIMD)->RangeMultiplier(4)->Range(1 << 2, 1 << 12);
|
|
BENCHMARK_REGISTER_F(AddArray, Value_Scalar_Unaligned)->RangeMultiplier(4)->Range(1 << 2, 1 << 12);
|
|
BENCHMARK_REGISTER_F(AddArray, Value_SIMD_Unaligned)->RangeMultiplier(4)->Range(1 << 2, 1 << 12);
|
|
BENCHMARK_REGISTER_F(AddArray, Scalar)->RangeMultiplier(4)->Range(1 << 2, 1 << 12);
|
|
BENCHMARK_REGISTER_F(AddArray, SIMD)->RangeMultiplier(4)->Range(1 << 2, 1 << 12);
|
|
BENCHMARK_REGISTER_F(AddArray, Scalar_Unaligned)->RangeMultiplier(4)->Range(1 << 2, 1 << 12);
|
|
BENCHMARK_REGISTER_F(AddArray, SIMD_Unaligned)->RangeMultiplier(4)->Range(1 << 2, 1 << 12);
|
|
BENCHMARK_MAIN();
|