C# Tip: Initialize lists size to improve performance
Lists have an inner capacity. Every time you add more items than the current Capacity, you add performance overhead. How to prevent it?
Table of Contents
Just a second! π«·
If you are here, it means that you are a software developer. So, you know that storage, networking, and domain management have a cost .
If you want to support this blog, please ensure that you have disabled the adblocker for this site. I configured Google AdSense to show as few ADS as possible - I don't want to bother you with lots of ads, but I still need to add some to pay for the resources for my site.
Thank you for your understanding.
- Davide
Some collections, like List<T>
, have a predefined initial size.
Every time you add a new item to the collection, there are two scenarios:
- the collection has free space, allocated but not yet populated, so adding an item is immediate;
- the collection is already full: internally, .NET resizes the collection, so that the next time you add a new item, we fall back to option #1.
Clearly, the second approach has an impact on the overall performance. Can we prove it?
Here’s a benchmark that you can run using BenchmarkDotNet:
[Params(2, 100, 1000, 10000, 100_000)]
public int Size;
[Benchmark]
public void SizeDefined()
{
int itemsCount = Size;
List<int> set = new List<int>(itemsCount);
foreach (var i in Enumerable.Range(0, itemsCount))
{
set.Add(i);
}
}
[Benchmark]
public void SizeNotDefined()
{
int itemsCount = Size;
List<int> set = new List<int>();
foreach (var i in Enumerable.Range(0, itemsCount))
{
set.Add(i);
}
}
Those two methods are almost identical: the only difference is that in one method we specify the initial size of the list: new List<int>(itemsCount)
.
Have a look at the result of the benchmark run with .NET 7:
Method | Size | Mean | Error | StdDev | Median | Gen0 | Gen1 | Gen2 | Allocated |
---|---|---|---|---|---|---|---|---|---|
SizeDefined | 2 | 49.50 ns | 1.039 ns | 1.678 ns | 49.14 ns | 0.0248 | - | - | 104 B |
SizeNotDefined | 2 | 63.66 ns | 3.016 ns | 8.507 ns | 61.99 ns | 0.0268 | - | - | 112 B |
SizeDefined | 100 | 798.44 ns | 15.259 ns | 32.847 ns | 790.23 ns | 0.1183 | - | - | 496 B |
SizeNotDefined | 100 | 1,057.29 ns | 42.100 ns | 121.469 ns | 1,056.42 ns | 0.2918 | - | - | 1224 B |
SizeDefined | 1000 | 9,180.34 ns | 496.521 ns | 1,400.446 ns | 8,965.82 ns | 0.9766 | - | - | 4096 B |
SizeNotDefined | 1000 | 9,720.66 ns | 406.184 ns | 1,184.857 ns | 9,401.37 ns | 2.0142 | - | - | 8464 B |
SizeDefined | 10000 | 104,645.87 ns | 7,636.303 ns | 22,395.954 ns | 99,032.68 ns | 9.5215 | 1.0986 | - | 40096 B |
SizeNotDefined | 10000 | 95,192.82 ns | 4,341.040 ns | 12,524.893 ns | 92,824.50 ns | 31.2500 | - | - | 131440 B |
SizeDefined | 100000 | 1,416,074.69 ns | 55,800.034 ns | 162,771.317 ns | 1,402,166.02 ns | 123.0469 | 123.0469 | 123.0469 | 400300 B |
SizeNotDefined | 100000 | 1,705,672.83 ns | 67,032.839 ns | 186,860.763 ns | 1,621,602.73 ns | 285.1563 | 285.1563 | 285.1563 | 1049485 B |
Notice that, in general, they execute in a similar amount of time; for instance when running the same method with 100000 items, we have the same magnitude of time execution: 1,416,074.69 ns vs 1,705,672.83 ns.
The huge difference is with the allocated space: 400,300 B vs 1,049,485 B. Almost 2.5 times better!
Ok, it works. Next question: How can we check a List capacity?
We’ve just learned that capacity impacts the performance of a List.
How can you try it live? Easy: have a look at the Capacity
property!
List<int> myList = new List<int>();
foreach (var element in Enumerable.Range(0,50))
{
myList.Add(element);
Console.WriteLine($"Items count: {myList.Count} - List capacity: {myList.Capacity}");
}
If you run this method, you’ll see this output:
Items count: 1 - List capacity: 4
Items count: 2 - List capacity: 4
Items count: 3 - List capacity: 4
Items count: 4 - List capacity: 4
Items count: 5 - List capacity: 8
Items count: 6 - List capacity: 8
Items count: 7 - List capacity: 8
Items count: 8 - List capacity: 8
Items count: 9 - List capacity: 16
Items count: 10 - List capacity: 16
Items count: 11 - List capacity: 16
Items count: 12 - List capacity: 16
Items count: 13 - List capacity: 16
Items count: 14 - List capacity: 16
Items count: 15 - List capacity: 16
Items count: 16 - List capacity: 16
Items count: 17 - List capacity: 32
Items count: 18 - List capacity: 32
Items count: 19 - List capacity: 32
Items count: 20 - List capacity: 32
Items count: 21 - List capacity: 32
Items count: 22 - List capacity: 32
Items count: 23 - List capacity: 32
Items count: 24 - List capacity: 32
Items count: 25 - List capacity: 32
Items count: 26 - List capacity: 32
Items count: 27 - List capacity: 32
Items count: 28 - List capacity: 32
Items count: 29 - List capacity: 32
Items count: 30 - List capacity: 32
Items count: 31 - List capacity: 32
Items count: 32 - List capacity: 32
Items count: 33 - List capacity: 64
Items count: 34 - List capacity: 64
Items count: 35 - List capacity: 64
Items count: 36 - List capacity: 64
Items count: 37 - List capacity: 64
Items count: 38 - List capacity: 64
Items count: 39 - List capacity: 64
Items count: 40 - List capacity: 64
Items count: 41 - List capacity: 64
Items count: 42 - List capacity: 64
Items count: 43 - List capacity: 64
Items count: 44 - List capacity: 64
Items count: 45 - List capacity: 64
Items count: 46 - List capacity: 64
Items count: 47 - List capacity: 64
Items count: 48 - List capacity: 64
Items count: 49 - List capacity: 64
Items count: 50 - List capacity: 64
So, as you can see, List capacity is doubled every time the current capacity is not enough.
Further readings
To populate the lists in our Benchmarks we used Enumerable.Range
. Do you know how it works? Have a look at this C# tip:
π C# Tip: LINQ’s Enumerable.Range to generate a sequence of consecutive numbers
This article first appeared on Code4IT π§
Wrapping up
In this article, we’ve learned that just a minimal change can impact our application performance.
We simply used a different constructor, but the difference is astounding. Clearly, this trick works only if already know the final length of the list (or, at least, an estimation). The more precise, the better!
I hope you enjoyed this article! Let’s keep in touch on Twitter or on LinkedIn, if you want! π€π€
Happy coding!
π§
ABOUT THE AUTHOR
Davide Bellone is a software developer with more than 10 years of professional experience with Microsoft platforms and frameworks.
He loves learning new things and sharing these learnings with others: thatβs why he writes on this blog and is involved as speaker at tech conferences.
He's a Microsoft MVP π and content creator on LinkedInπ.