Improve Your Code To Reduce Churn

04-15 Improve Your Code To Reduce Churn

We can make a small adjustment in
our code to prevent excessive churn.

Let's take a look at a comparison view.

Rather than concatenate one cell value
at a time to build each row, let's use

a StringBuilder instance, and construct
each row using a single string.

Note that the StringBuilder's
instantiated outside the loop,

and thus its memory is allocated once.

And then we simply use
it as a buffer for

each iteration of the loop where we
first clear it and then we append

a single string of ints to represent
the row for that loop iteration.

Now see the instructor notes for
more details into this code segment.

All right, now it's time to verify.

You want to go ahead and
load the improved branch of code,

which is called memory_churn_optimized,
into both trace view and memory monitor

to confirm we've reduced the amount of
GC's occurring in the short time window.

You may also use allocation
tracker to verify.

If you use allocation tracker,

or if you got something unexpected
in trace view, or memory monitor,

share a screenshot of your
output in the discussion forums.

We're interested in
seeing what you've got.

Now for us, even with these changes,
the Perf pirate still pauses.

But this time for less time.

Now at this point, this also might mean
that this function is probably a good

candidate to be backgrounded.