How to: Measure PLINQ Query Performance
This example shows how to use the Stopwatch class to measure the time it takes for a PLINQ query to execute.
Example
This example uses an empty foreach loop (For Each in Visual Basic) to measure the time it takes for the query to execute. In real-world code, the loop typically contains additional processing steps that add to the total query execution time. Notice that the stopwatch is not started until just before the loop, because that's when the query execution begins. If you require more fine-grained measurement, you can use the ElapsedTicks property instead of ElapsedMilliseconds.
using System;
using System.Diagnostics;
using System.Linq;
class ExampleMeasure
{
static void Main()
{
var source = Enumerable.Range(0, 3000000);
var queryToMeasure =
from num in source.AsParallel()
where num % 3 == 0
select Math.Sqrt(num);
Console.WriteLine("Measuring...");
// The query does not run until it is enumerated.
// Therefore, start the timer here.
var sw = Stopwatch.StartNew();
// For pure query cost, enumerate and do nothing else.
foreach (var n in queryToMeasure) { }
sw.Stop();
long elapsed = sw.ElapsedMilliseconds; // or sw.ElapsedTicks
Console.WriteLine("Total query time: {0} ms", elapsed);
Console.WriteLine("Press any key to exit.");
Console.ReadKey();
}
}
Module ExampleMeasure
Sub Main()
Dim source = Enumerable.Range(0, 3000000)
' Define parallel and non-parallel queries.
Dim queryToMeasure = From num In source.AsParallel()
Where num Mod 3 = 0
Select Math.Sqrt(num)
Console.WriteLine("Measuring...")
' The query does not run until it is enumerated.
' Therefore, start the timer here.
Dim sw = Stopwatch.StartNew()
' For pure query cost, enumerate and do nothing else.
For Each n As Double In queryToMeasure
Next
sw.Stop()
Dim elapsed As Long = sw.ElapsedMilliseconds ' or sw.ElapsedTicks
Console.WriteLine($"Total query time: {elapsed} ms.")
Console.WriteLine("Press any key to exit.")
Console.ReadKey()
End Sub
End Module
The total execution time is a useful metric when you are experimenting with query implementations, but it doesn't always tell the whole story. To get a deeper and richer view of the interaction of the query threads with one another and with other running processes, use the Concurrency Visualizer.
