Parallel Computing: PLINQ
Visual Studio 2010 has new API for LINQ (Language Integrated Query). This helps us to implement the power of Parallel Computing in declarative manner.
LINQ without Parallel Extension
It will take 3 sec for 28 thousand elements.
static void Main(string[] args)
{
Stopwatch sw = Stopwatch.StartNew();
DoIt();
Console.WriteLine("Elapsed = " + sw.ElapsedMilliseconds.ToString());
Console.ReadLine();
}
private static void DoIt()
{
IEnumerable<int> arrInt = Enumerable.Range(1, 4000000);
var q =
from n in arrInt
where IsPrime(n)
select n.ToString();
List<string> list = q.ToList();
Console.WriteLine("Elements : " + list.Count.ToString());
}
private static bool IsPrime(int p)
{
//Find the prime number
int upperBound = (int)Math.Sqrt(p);
for (int i = 2; i < upperBound; i++)
{
if (p % i == 0) return false;
}
return true;
}
LINQ with Parallel Extension
With a very simple change in query this will take around 1.5 sec for the same number of elements.
var q =
from n in arrInt.AsParallel()
where IsPrime(n)
select n.ToString();
Greenroom
What’s the magic? When we use .AsParallel() it does cast the list of integers into ParallelEnumerable class and calls a different Where method which implements the new “Task” API.
With Lambda expression this would look more clear.
Before
var q = arrant
.Where(n => IsPrime(n))
.Select(n => n.ToString());
Class: Enumerable
Method:
public static IEnumerable<TSource> Where<TSource>(this IEnumerable<TSource> source, Func<TSource, bool> predicate);
After (while using AsParallel())
Code:
var q = arrInt.AsParallel()
.Where(n => IsPrime(n))
.Select(n => n.ToString());
Class : ParallelEnumerable
Method :
public static ParallelQuery<TSource> Where<TSource>(this ParallelQuery<TSource> source, Func<TSource, bool> predicate);
The above time will vary based on CPU power and availability.
Namoskar!!!
Comments
- Anonymous
November 19, 2009
This is awesome! Thanks for posting. Just one error in your IsPrime calculation The "for (int i = 2; i < upperBound; i++)" needs to change to " for (int i = 2; i <= upperBound; i++)"