在Parallel.ForEach中传递列表

Question

This is the first time I try to use the "Parallel.ForEach" loop to see if I can improve performance using many cores in the system. I have tried to look at MSDN and other examples but I can't find a way to pass in 3 lists as below.

How can this loop be exchanged with a Parallel.ForEach loop? I think I am looking for the basics how to pass in all those 3 lists so I can work with them in the Parallel.ForEach loop?

Thank you!

        List<double> nums = new List<double> { 0.0005, 0.00035, 0.00205 };
        List<double> list1 = new List<double>();
        List<double> list2 = new List<double>();
        List<double> list3 = new List<double>(); Random random = new Random(); double calc1 = 0;
        List<double> resultLIST = new List<double>();
        for (int i = 0; i < 4000000; i++)
        {
            list1.Add(nums[random.Next(0, 7)]);
            list2.Add(nums[random.Next(0, 7)]);
            list3.Add(nums[random.Next(0, 7)]);
        }

        //How can the below loop be replaced with a: Parallel.ForEach loop?
        for (int i = 0; i < list1.Count; i++)
        {
            calc1 = list1[i] * list2[i] * list3[i];
            resultLIST.Add(calc1);
        }

        //Now sort the list 
        resultLIST.Sort();

        //Here I will write "resultLIST" to a .txt file

Answer 1

Personally, I would use LINQ with AsParallel :

var result = Enumerable.Range(0, list1.Count)
    .AsParallel()
    .Select(index => list1[index] + list2[index] + list3[index])
    .OrderBy(v => v)
    .ToList();

Note that list1 , list2 and list3 are not expected to change during this calculation.

Answer 2

You use the Parallel.Foreach when the operations on each list can run independently in parallel. In your case you cannot. Also remember using Parallel.Foreach or AsParallel for trivial operations like * or + can actual degrade performance because there is more overhead.

In your case the best approach will be to split all three lists with 1000000 elements each. Add them into an array of array and call Parallel.Foreach on them. That way you will have 4 threads doing 1000000 operations each and if your machine has 4 cores they can run independently on each core.