Bump. Can anyone give insight into this?
BTW, should I be using GPU mode, because my UI has lots of components on the same screen that are sized depending on the device width and height.
I need this for precise positioning of my view's components.
Would using the GPU mode possibly help in this situation?
Your app will run much slower in debug mode, but it sounds like you might also be dealing with a bulky view, which must go through it's entire creation and validation stages before transitioning in. This can get quite expensive if you are using a lot of static MXML content.
For this, I am using a home-baked solution for deferred instantiation, which creates the bulk of the view once the transition has completed. Part of the view is created right away using static MXML, which shows a basic outline of what the view will look like (basic chrome, empty stylized containers, etc.). The rest of the view is created after the transition. If any of my views take longer then about half a second to transition in, I will optimize them in this manner until the start up time is bellow this.
I think for most Android devices, CPU mode will yield better results. Mine is just set to auto, and I am satisfied with the performance. I've never used the tap event, so I can't speak on which one is faster. I just use a standard click event.