#include <celero/Celero.h>

#include "common.h"

constexpr unsigned samples = 12, iterations = 10, cores = 4;
constexpr std::size_t vecSize = 1'000;
constexpr int minValue = -250, maxValue = +250;

BASELINE(ExecTwoLevelsHard, Sequential, samples, iterations) {
	auto f = alsk::edsl::implement<alsk::exec::Sequential>(bench::exprTwoS);
	f.skeleton.task.task.task<0>().size = vecSize;
	f(minValue, maxValue);
}

BENCHMARK(ExecTwoLevelsHard, FirstLevelEqui, samples, iterations) {
	auto f = alsk::edsl::implement<alsk::exec::FirstLevelEqui>(bench::exprTwoS);
	f.executor.cores = cores;
	f.skeleton.task.task.task<0>().size = vecSize;
	f(minValue, maxValue);
}

BENCHMARK(ExecTwoLevelsHard, FirstLevelGreedy, samples, iterations) {
	auto f = alsk::edsl::implement<alsk::exec::FirstLevelGreedy>(bench::exprTwoS);
	f.executor.cores = cores;
	f.skeleton.task.task.task<0>().size = vecSize;
	f(minValue, maxValue);
}

BENCHMARK(ExecTwoLevelsHard, FirstLevelNoOpti, samples, iterations) {
	auto f = alsk::edsl::implement<alsk::exec::FirstLevelNoOpti>(bench::exprTwoS);
	f.executor.cores = cores;
	f.skeleton.task.task.task<0>().size = vecSize;
	f(minValue, maxValue);
}

BENCHMARK(ExecTwoLevelsHard, DynamicPool, samples, iterations) {
	auto f = alsk::edsl::implement<alsk::exec::DynamicPool>(bench::exprTwoS);
	f.executor.cores = cores;
	f.skeleton.task.task.task<0>().size = vecSize;
	f(minValue, maxValue);
}

BENCHMARK(ExecTwoLevelsHard, StaticPool, samples, iterations) {
	auto f = alsk::edsl::implement<alsk::exec::StaticPool>(bench::exprTwoS);
	f.executor.cores = cores;
	f.skeleton.task.task.task<0>().size = vecSize;
	f(minValue, maxValue);
}

BENCHMARK(ExecTwoLevelsHard, StaticPoolId, samples, iterations) {
	auto f = alsk::edsl::implement<alsk::exec::StaticPoolId>(bench::exprTwoS);
	f.executor.cores = cores;
	f.skeleton.task.task.task<0>().size = vecSize;
	f(minValue, maxValue);
}

BENCHMARK(ExecTwoLevelsHard, StaticThread, samples, iterations) {
	auto f = alsk::edsl::implement<alsk::exec::StaticThread>(bench::exprTwoS);
	f.executor.cores = cores;
	f.skeleton.task.task.task<0>().size = vecSize;
	f(minValue, maxValue);
}