Parallel.Foreach and Parallel Execution in C#

Question

I am having a task of scanning all the folders name start with "xyz" parallely. I meant if one folder getting scan same time other one should also getting scan. I don't want one by one scanning.

For that I used Parallel Foreach.

Question is? Is it correct or not? and How to know is it running parallely(to put any message some where)?

Parallel.ForEach(path, currentPath =>
   {
    var output = programReader.GetData(currentPath, durReader.dirPattern);

    foreach (var item in output)
      {
        foreach (var project in item.Name)
           Console.WriteLine(item.serverName + " " + item.serverNumber + " " + fileName);
        }
     }

EDIT:

Is Parallel.Foreach only works on multicore systems or it could work on single core system also to perform show parallelism

As this is I/O bound, running it in parallel may actually be slower. Make sure you time the results. — Jonathan Allen
– Jonathan Allen, Commented Feb 14, 2011 at 8:38

TomTom · Accepted Answer · 2011-02-14 08:37:30Z

3

Foirst - if you ask a question, make it a question. !! is a good indication it is not a question.

Second, your approach makes little sense. Parallel will go VERY parallel. Good. Bad: you still have only one disc. Result: tasks waiting. It makes no sense to paralellize IO operations over the degree supported by the hardware.

answered Feb 14, 2011 at 8:37

TomTom

62.4k11 gold badges95 silver badges152 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

PawanS Over a year ago

Oh!! thanx for pointing. I made changes. I am in doubt here is Parallel will divide the task on different proccessors. How could I know for each path folders are scanning parallely . I m totally confused.

TomTom Over a year ago

Again, number of processors is irrelevant. Your disc is not fast enough to handle multiple requests at the same time. Even ONE processsor, even 10 years old, would be too fast. Disc IO can not be efficiently paralellized unless you have a large SAN or SSD's available.

PawanS Over a year ago

Oh... so you want to say, this Parallel will not work in ONE core Systems. Then How should I do Parallel scanning of Directories?

TomTom Over a year ago

No, imean it is pretty irrelevant how much performance you try to get out of the processor. Your limiting factor is not the processor in this particular example, it is the DISC. What you do here is building a formula one car, jsut to find out you need to transport a 10 ton machine. Does not fit. Directory info analysis is limited by DISC speed primarily.

Andras Zoltan · Accepted Answer · 2011-02-14 09:05:39Z

1

The Parallel extensions split the load per core - although I'm not sure if they take into account hyperthreaded cores.

But, to add another answer to steer you in the right direction:

Don't try and do this in parallel. The disk can only serve one request at a time, so you're likely to just slow everything down as the disk queue is just likely to get bigger.

The only scenario where you might be able to get increased performance is if the location you're scanning is actually a SAN where the storage is distributed and highly replicated.

answered Feb 14, 2011 at 9:05

Andras Zoltan

42.4k13 gold badges109 silver badges160 bronze badges

Comments

Jonathan Allen · Accepted Answer · 2011-02-14 08:36:52Z

0

You can print the Thread.ManagedThreadId value to see which threads are being used.

answered Feb 14, 2011 at 8:36

Jonathan Allen

70.6k77 gold badges276 silver badges459 bronze badges

1 Comment

PawanS Over a year ago

Oh yes!! that is one way. Let me try it.

Thanigainathan · Accepted Answer · 2011-02-14 09:02:49Z

0

Parallel.ForEach or Parallel.Execute uses the cores of the processor. So if your processor is having more thn one core they will be used equally to run this thread. Here you are m

answered Feb 14, 2011 at 9:02

Thanigainathan

1,54714 silver badges25 bronze badges

Collectives™ on Stack Overflow

Parallel.Foreach and Parallel Execution in C#

4 Answers 4

4 Comments

Comments

1 Comment

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

4 Answers 4

4 Comments

Comments

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Related