Stop Using the Pipeline in PowerShell

Last week I blogged about Fred Bainbridge and Ryan Ephgrave’s MMSMOA session where they showed us that we shouldn’t be using the pipeline anymore.  Before beginning the session, I was as confused as you might be about why.  The guys did a great job explaining it and I’m going to show you why in the same manner they showed me.

Let’s start with a really basic array of numbers:

Now, you want to do something with these numbers, right?  Let’s say we’re going to write them out to a file:

Easy enough, and if I throw a Measure-Command around it (example at the bottom of this post), this takes 4.44s to execute.

Now, let’s try this again, but without using the pipe to pass the input-object:

Measuring this command, I’m completing in just 1.434!  That’s slightly more than 3x faster. This is consistent, every time, within a few milliseconds, and it doesn’t matter if I put one test on top or the other.  And it goes up exponentially – it’s not linear movement.

If I double the number to 400,000 and rerun the tests, the pipeline code runs in 9.0s while not using it runs in 2.8s.  If I double that again to 800,000 we’re looking at 18.69s for the pipeline method, while the non-pipeline method runs in just 5.81s.  You can see here that the larger our set gets, the more of a performance hit we take using the pipeline.

Over multiple runs, the numbers change by milliseconds but there is always a measurable difference.  Here is just one example of a run on my laptop:

Array CountUsing PipelineWithout Pipeline

Thanks to the awesome presenters at MMSMAO for this awesome new knowledge!

Completed test script:



Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.