用于目录遍历的多线程与异步

Question

I am writing a code to calculate the size of the folder, for now I have this code

private long CalculateSize(string root)
    {
        long size = 0;
        Stack<string> dirs = new Stack<string>(20);

        if (!System.IO.Directory.Exists(root))
        {
            throw new ArgumentException();
        }
        dirs.Push(root);

        while (dirs.Count > 0)
        {
            string currentDir = dirs.Pop();
            string[] subDirs;
            try
            {
                subDirs = Directory.GetDirectories(currentDir);
            }
            catch (UnauthorizedAccessException)
            {
                continue;
            }
            catch (DirectoryNotFoundException e)
            {
                continue;
            }

            string[] files = null;
            try
            {
                files = Directory.GetFiles(currentDir);
            }

            catch (UnauthorizedAccessException)
            {
                continue;
            }

            catch (DirectoryNotFoundException e)
            {
                continue;
            }
            foreach (string file in files)
            {
                try
                {
                    FileInfo fi = new System.IO.FileInfo(file);
                    size += fi.Length;
                }
                catch (FileNotFoundException e)
                {
                    Console.WriteLine(e.Message);
                    continue;
                }
            }
            foreach (string str in subDirs)
            {
                dirs.Push(str);
            }
        }
        return size;
    }

It works fine, but it is so slow if I provide some big starting directory like "C:" or "C:\Windows". I had an idea to do something like:

string[] subDirs;
subDirs = Directory.GetDirectories(currentDir);

And then start a new thread or parallel process or smth like that for each directory in subDirs, but I have no idea what should I use, and how to implement it. Also, do I have to do it for each level of subdirectories? I mean for example I have "C:" it has 3 subdirectories like "Windows", "ProgramFile", and "ProgramFIlex86", so I am starting a multithreading processes for each, and each of these directories have 50 more directories, do I have to start a new process for each?