This might be a very stupid question, but I have the following lines of coding that convert RAW images to BitmapImages:
public async void CreateImageThumbnails(string imagePath, int imgId)
{
await Task.Run(() => controlCollection.Where(x => x.ImageId == imgId).FirstOrDefault().ImageSource = ThumbnailCreator.CreateThumbnail(imagePath));
}
which calls this method CreateThumbnail()
public static BitmapImage CreateThumbnail(string imagePath)
{
var bitmap = new BitmapImage();
using (var stream = new FileStream(imagePath, FileMode.Open, FileAccess.Read))
{
bitmap.BeginInit();
bitmap.DecodePixelWidth = 283;
bitmap.CacheOption = BitmapCacheOption.OnLoad;
bitmap.StreamSource = stream;
bitmap.EndInit();
}
bitmap.Freeze();
GC.WaitForPendingFinalizers();
GC.Collect();
return bitmap;
}
When using async Void
instead of async Task
in my CreateImageThumbnails
method, my application processes the images(29 of them) about 11 seconds faster than async Task
. Why would this be?
The memory usage is much more using void
, but the operation is completed much quicker. I have little knowledge of threading, this is why I am asking this question. Can someone please explain why this is happening?
Also I have done some research on on when and when not to use async void
, but I could not find an answer to my question. (I might just not have searched very well).
Thank you.
When you call an async void
method, or call an async Task
method without awaiting it (if the called method contains an await
, so it doesn't block), your code will continue right away, without waiting for the method to actually complete. This means that several invocations of the method can be executing in parallel, but you won't know when they actually complete, which you usually need to know.
You can take advantage of executing in parallel like this, while also being able to wait for all the invocations to complete by storing the Task
s in a collection and then using await Task.WhenAll(tasks);
.
Also keep in mind that if you want to execute code in parallel, you have to make sure it's safe to do it. This is commonly called "thread-safety".