When using the partitioner object in a parallel construct "Parallel.ForEach", why do we have to keep the size of the partitioned chunks large and why do we have to keep the number of locks small?

static double ParallelPartitionerPi()
        {
            double sum = 0.0;
            double step = 1.0 / (double)num_steps;
            object monitor = new object();
            Parallel.ForEach(Partitioner.Create(0, num_steps), () => 0.0,
            (range, state, local) =>
            {
                for (int i = range.Item1; i < range.Item2; i++)
                {
                    double x = (i + 0.5) * step;
                    local += 4.0 / (1.0 + x * x);
                }
                return local;
            }, local => { lock (monitor) sum += local; });
            return step * sum;
        }