What would be the best way to test a device for FPS ability? I want to pretty much run a 1 second stress test on my loading scene and see what the device can handle. I’m looking to use this stress test number to limit things like particles for example. Anyone have any good ideas on how to do a quick test?
I was thinking about loading in a bunch of text fields and moving them around while changing the text of them maybe?
Is this something you would be running in a live app or is this just and experiment.
In either case, if you’re going to benchmark something, benchmark something that represents the heavy lifting your app will do.
If you want to improve the performance of a live app, I’d suggest detecting FPS and throttling back features if the FPS dips for too long in the app. i.e. I wouldn’t necessarily benchmark the app when it launches.
Yeah the issue with that though is once the gameplay starts it’s pretty intense, lots of moving objects and images being created and destroyed etc etc. There isn’t a good time to really capture an FPS and depending on what is going on when we capture it, it may not be during a less busy time and the FPS would appear higher than under heavy load. With the stress test in the loading I was seeing it as a way to control the test and get a pretty solid idea of FPS without having to poll the user during their gameplay and then average it out.
Is this something you would be running in a live app or is this just and experiment.
In either case, if you’re going to benchmark something, benchmark something that represents the heavy lifting your app will do.
If you want to improve the performance of a live app, I’d suggest detecting FPS and throttling back features if the FPS dips for too long in the app. i.e. I wouldn’t necessarily benchmark the app when it launches.
Yeah the issue with that though is once the gameplay starts it’s pretty intense, lots of moving objects and images being created and destroyed etc etc. There isn’t a good time to really capture an FPS and depending on what is going on when we capture it, it may not be during a less busy time and the FPS would appear higher than under heavy load. With the stress test in the loading I was seeing it as a way to control the test and get a pretty solid idea of FPS without having to poll the user during their gameplay and then average it out.