Tag: wp7

One of the last things in my slides for Gamefest was a simple little object I had written to help us detect when our code was allocating memory we weren’t expecting. Due to the size limitations of slides I had attempted to to shorten the object to the minimalist version, and in doing so lost a few things. Here is the object I used in its entirety:

Not very much different than what’s in the slides, but a few subtle differences. The most important being whether or not ‘true’ or “false” is passed in during the two calls to GetTotalMemory. What does this code actually do though?

I wrote it basically to allow myself to “wrap” a section of code and measure any allocations from it. At the constructor of the helper object I measure the total memory in use and store how much allocation (in bytes) I expect the code block to take. Then at the end I measure the memory once more and if more memory was allocated than I was expecting I break into the debugger.

A few things to note. First, when you pass in ‘true’ to GetTotalMemory a full collection will occur, so if you are doing this often you will dramatically slow down your application.

Second, you’ll notice that the slides pass in ‘true’ in both spots which is probably not the behavior you want (you’ll notice in the code above that ‘false’ is passed at the end in the second constructor). If you pass in true in both spots, you will only catch allocations that were not garbage because any garbage would have been collected due to the ‘true’ being passed in!

Now, there are times when you do want to know this, so having the option is certainly fine, but the majority of time you want to know about allocations that create garbage, in which case you need to pass in false to the second call.

Another point I brought up during the talk that isn’t obvious is that this code completely ignores other threads. Which means if you have another thread allocating memory in between the two calls, you will get “false positives” that can prove difficult to track down. When I used this method I kept the code I wrapped in this object small to help eliminate them.

To see an example of using this helper in code see below:

I guess I’m impatient, because there are very few things that annoy me more than loading screens. In a world of instant gratification, waiting for any significant amount of time is unacceptable. Hell, I’ve only just started typing this and I’m already annoyed that I’m not finished yet. Loading just bugs me, and I see it creep a lot on Windows Phone 7 games. Perhaps I should ramble on about that for a while.

The first goal of the game should be to show me something on screen as soon as possible. While you can “somewhat” use the splashscreen.jpg trick to show something quick, in XNA applications, it doesn’t stay on screen long enough and looks awkward (in my opinion).  I personally would rather show my splash screen on the first render.  However, that leads me to another common mistake I see people making, and that’s loading a bunch of content (or all!) in the LoadContent method.  This is certainly a bad idea if you want to show something on screen quickly. This is essentially the order you will see things happen at startup:

1. Game’s Constructor is called
2. Initialize is called
3. LoadContent is called
4. Update is called
5. Draw is called
6. Goto 4.

While there are certainly nuances here, that’s the basic order, and notice that Draw isn’t called until way at the end.  I’ve seen people decide that the best way to fix this would be to draw something in LoadContent, and then they implement that, see that it works and call it a success. If only it were so easy. If your application is starting up once more after being tombstoned, essentially the same sequence of events happens, however there is a caveat this time.  While the “startup sequence” is happening, the phone will be displaying the “Resuming…” screen (my how I dislike that screen), and will not show anything from your application until the “startup sequence” has finished. In this case, the sequence doesn’t finish until *after* LoadContent has finished. So if you showed something on screen early in your LoadContent method, and then loaded a bunch of content, it would work in the initial startup scenario, and potentially fail in the restore from tombstone scenarios.

The easy way to get around these issues is to do all of your content loading (aside from the first thing you want to see on screen, such as a splash screen) on the second update call.  This allows you to show something on screen quickly in both the initial startup and coming back from tombstoning cases, and if you don’t have much content, should be sufficient enough. Your splash screen image will stay static on screen (and your application is essentially non-interactive) while the content is loading. If it takes more than a few seconds to get to this state though (ie, you have too much content), you have more work to do.

The most common thing I see happening here is the game will load only the content needed for the main UI portions, and will force you behind a loading screen after you choose the level or what have you. Again, this works perfectly fine, but we can do so much better! Why can’t you load everything you need while I’m playing around with the UI? For ease of explanation, let’s say that you are making a 2D game, and you have a large list of strings that are all the sprite textures you will need to load. What if you had a class such as this:

So long as you called BackgroundLoad.Go(this); sometime after you started up, your game would dutifully start loading all of your assets behind the scenes in a background thread.  This isn’t a foolproof solution though. For one, you have very little control over background threads on Windows Phone 7. You cannot change priority, and you have no control over how much they will execute (although on average for every 4ms your main thread is executing your background thread will run for 2ms).  In this example, you are also using the graphics device on another thread, which will take out a lock to avoid multiple threads manipulating it at the same time. In the larger scheme of things though, it’s still probably faster in this mechanism than without.

You will notice two potentially useless pieces of code here as well. First, a property to dictate when the loading has been complete, and a second in which the actual asset load is happening within an action rather than directly in code. The property is self explanatory if you think about it, you still need to know when all of your content has been loaded so you can allow the user to start the game! This just gives you easy access to it.  The action delegate is slightly less obvious though.

What should you do when the user has gone through the UI and is ready to play the game but the content isn’t done loading? As was mentioned earlier, you don’t have any control over the background threads execution time, and because it is running half the speed as the main thread, as the code is written now, your loading time could potentially be *increased*! This is most definitely not what we wanted. What you really want is a mechanism where you would load things in the background if the user was still messing around with the UI, but will switch to loading in the foreground if the user was waiting on loading, and that is what the action delegate is for. If you replaced the code in the thread method with the following, you will get this behavior:

Now your code is loading in the background while the user is messing around with the UI and will switch to the foreground once they’re done. In the degenerate case, this is still slightly slower than simply forcing them to wait at a loading screen, but given the degenerate case is almost certain to never happen, a mechanism such as this will (for most intents and purposes) always be faster. It certainly took me long enough to exlain all this, I should have had a loading screen for this post…

As a final caveat, this was an extremely simple example of multithreading. If you do anything more complex and don’t have knowledge of multithreading (or what the ‘lock’ keyword does), you should probably read up a bit on that!