Question Object Access Time vs. Function Processing Time

[Dylvan]

Hello all!

I'm new to VB.NET and have been toying around with creating functions, subs, classes, etc. One test program I've written tests the differences in processing times via two methods: variable access vs. function invocation.

What I mean by this is I have one module and one function. In the module I've declared a String object that is set by calling my function 'ReturnString()'. I also have two DateTime objects for recording elapsed seconds.

In one instance, I call 'Console.Writeline(MyString)' and report the time it took for that line to process. Then I called 'Console.WriteLine(ReturnString())' and reported the time it took to process that line.

My impression is that the first 'Console.WriteLine(MyString)' invocation would take up a much shorter time frame to execute than the 'Console.WriteLine(ReturnString())' call, however when running my program I found that to be false.



So, my question is such: How is it that TWO function invocations take less time than only one?

Module Module1


    Sub Main()
        Dim TestInteger As Integer = 100
        Dim TestString As String = ReturnString(TestInteger)
        Dim StartTime As DateTime = New DateTime()
        Dim EndTime As DateTime = New DateTime()


        StartTime = DateTime.Now
        Console.WriteLine(TestString)
        EndTime = DateTime.Now


        Console.WriteLine("Storing in variable: {0}", EndTime.Subtract(StartTime).TotalSeconds)


        StartTime = DateTime.Now
        Console.WriteLine(ReturnString(TestInteger))
        EndTime = DateTime.Now


        Console.WriteLine("Calling function on each instance: {0}", EndTime.Subtract(StartTime).TotalSeconds)


        Console.Read()
    End Sub


    Function ReturnString(ByVal Int As Integer)
        Dim MyString As String = Int.ToString


        MyString = MyString & " Test"


        Return MyString
    End Function


End Module

 

[JohnH]

A few general issues that could affect your tests:
interacting with the user interface (writing to console - are you trying to measure how fast that is? that itself could change based on machine work load)
doing the test the same time the application is starting up
many CPU's today changes clock frequency dynamically depending on work load, this happens very fast and could result in slower processing earlier in processing than later.

A specific issue that affects your tests:
neither operation is not something that takes any significant time

todo:
let computer finish application startup before test (console.readkey)
try assigning the result to a local variable instead of writing to console
do test enough times that it could matter, for example loop 10000000 times, now you may see test1 about 0 seconds and test2 about one second.