Check our online training feedbacks Here

10 November 2010

Value types, reference types, boxing and Unboxing

Value types are types which hold both data and the memory on the same location. While a reference type has a pointer which points to the memory location.

Below is a simple integer data type with name ‘i’ whose value is assigned to an other integer data type with name ‘j’. Both these memory values are allocated on the stack.

When we assign the ‘int’ value to the other ‘int’ value it creates a complete different copy. In other word if you change either of them the other does not change. These kinds of data types are called as ‘Value types’.
When we create an object and when we assign one object to the other object, they both point to the same memory location as show in the below code snippet. So when we assign ‘obj’ to ‘obj1’ they both point to the same memory location.

In other words if we change one of them the other object is also affected this is termed as ‘Reference types’.

So which data types are ref type and value type?
In .NET depending on data types the variable is either assigned on the stack or on the heap. ‘String’ and ‘Objects’ are reference types and any other .NET primitive data types are assigned on the stack. Below figure explains the same in a more detail manner.

Boxing and Unboxing
WOW, you have given so much knowledge, so what’s the use of it in actual programming. One of the biggest implications is to understand the performance hit which is incurred due to data moving from stack to heap and vice versa. 

Consider the below code snippet. When we move a value type to reference type the data is moved from the stack to the heap. When we move reference type to a value type the data is moved from the heap to the stack.

This movement of data from the heap to stack and vice-versa creates a performance hit.

When the data moves from value types to reference types its termed as ‘Boxing’ and the vice versa is termed as ‘UnBoxing’.
If you compile the above code and see the same in ILDASM you can see in the IL code how ‘boxing’ and ‘unboxing’ looks, below figure demonstrates the same.

Performance implication of Boxing and unboxing
In order to see how the performance is impacted we ran the below two functions 10,000 times. One function has boxing and the other function is simple. We used a stop watch object to monitor the time taken. 

The boxing function was executed in 3542 MS while without boxing the code was executed in 2477 MS. In other words try to avoid boxing and unboxing. In project you always need boxing and unboxing , use it when it’s absolutely necessary.

With the same article the sample code is attached which demonstrates this performance implication.
Currently I have not put a source code for unboxing but the same hold true for the same. You can write the same and experiment it by using stopwatch class.

No comments :

Post a Comment

Your comments, Feedbacks and Suggestions are very much valuable to me :)