Embarrassingly parallel problem solved with different approaches

sequential-vs-plinq-vs-mapreduce.cs

void Main()
{
	Stopwatch sw;
	var ints = Enumerable.Range(0, 25000);
	
	Func<int, double> func = (x) => {
		double result = 0;
		for(var i = 0; i < x; i++) {
			result += Math.Cos(i);
		}
		return result;
	};
	
	// sequential
	sw = Stopwatch.StartNew();
	var result1 = ints.Select(x => func(x)).Sum();
	sw.Stop();
	
	result1.Dump();
	sw.Elapsed.Dump();
	
	// PLINQ
	sw = Stopwatch.StartNew();
	var result2 = ints.AsParallel().Select(x => func(x)).Sum();
	sw.Stop();
	
	result2.Dump();
	sw.Elapsed.Dump();
	
	// map-reduce
	double result3 = 0;
	var obj = new Object();
	sw = Stopwatch.StartNew();
	
	var task = Parallel.ForEach(ints,
	() => new List<double>(),
	(data, loopControl, localList) => 
	{
		localList.Add(func(data));
		return localList;
	},
	(localList) => 
	{
		lock(obj)
		{
			result3 += localList.Sum();
		}
	});
	sw.Stop();

	result3.Dump();
	sw.Elapsed.Dump();
}