整数或引用字典关键-性能？

Question

I am interested to know the difference in Dictionary read and write performance between using an int value type key and an integer wrapped into a reference type ( IntId ).

Instantiating with int :

new Dictionary<int, SomeValue>();

Instantiating with IntId :

new Dictionary<IntId, SomeValue>();

Averaged performance results yielded the following on writing and reading 200,000 items:

• IntId is 2X faster at writing
• int is 3X faster at reading

What is the discrepancy between these two results?

The IntId class:

public class IntId
{
    public int Value { get; private set; }

    public IntId(int value)
    {
        Value = value;
    }

    public override bool Equals(object other)
    {
        var entity = other as IntId;

        if (entity == null)
        {
            return false;
        }
        else
        {
            return (this.Value.Equals(entity.Value));
        }
    }

    public override int GetHashCode()
    {
        return Value.GetHashCode();
    }
}

Answer 1

In all likelihood the discrepancy between these results will be strongly related to the method used to obtain them.

Your IntId is a small reference object. Creating one requires a memory allocation.

An integer is a value object, but may require boxing to use in your collection.

The values assigned to the keys affect the efficiency of the hashing function.

There are many factors to consider. I would not accept the results on face value until after a careful scrutiny of how they were obtained.