#include <celero/Celero.h>
#include <alsk/alsk.h>

#include "common.h"

using namespace bench;

constexpr unsigned samples = 30, iterations = 10, cores = 4;

constexpr unsigned n          = 64;
constexpr std::size_t vecSize = 1'000;

constexpr auto eFarm = n*eTaskV<vecSize>;

BASELINE(Farm, Handwritten, samples, iterations) {
	for(unsigned i = 0; i < n; ++i) taskV<vecSize>();
}

BENCHMARK(Farm, Skeleton, samples, iterations) {
	auto farm = alsk::edsl::implement<alsk::exec::Sequential>(eFarm);
	farm();
}

BASELINE(FarmPar, Handwritter, samples, iterations) {
#pragma omp parallel for num_threads(cores)
	for(unsigned i = 0; i < n; ++i) taskV<vecSize>();
}

BENCHMARK(FarmPar, Parallel, samples, iterations) {
	auto farm = alsk::edsl::implement<alsk::exec::StaticThread>(eFarm);
	farm.executor.cores = cores;

	farm();
}