Hi,
I need to port a game (already finished for offline and browser) to the iPhone. It should support also the older iPhones. I am new on iOS developing and building a small development devices-park. Already got an iPad2 and a iPod-touch4 and of course my developer account and unity ios advanced. Now i think for realworld-testing i should also have something with fewer ram and the small display. So i wanted to buy either a 3g or a 3gs. My guess is, it might be wizer to choose the lamer one with less ram and slower gpu to go sure my game also runs there. But i read the 3G is not supported by iOS 4.3 any more.
edit: The game is a LERPZ-like platformgame completely with 3D-Meshes and 2 boned characters.
What would you recommend? Grabbing a G3 or a 3GS?
Thanks in advance 
Gregor
If it really needs to support the older models go for the 3G. The 3GS is miles ahead of the original iPhone and the 3G so you can’t really test a 3G oriented app on it.
That being said, pre 3GS models are starting to be discarded and you should really think seriously about supporting them at all.
For the cost of one of those phones, you should be able to pick up a first gen iPod Touch and a third gen. That way you have both generations covered (make sure it is a proper 3rd gen (32 or 64GB) as they are the first of the armv7 devices along with the 3GS).
@azzogat: i totally agree thinking about discarding 3G or even 3GS-devices. But its often the decision of the client. Of course they want as much target devices as possible. On the other side they want top gfx with lots of nice meshes and textures.
@ moonjump: The iPods might be an alternative. I read a LOT in the forum to learn and learned that the iPhone is running lots of background tasks, minimizing available RAM and making it more likely to crash than the iPods. I also learned that lots of installed, started and stopped apps also chews up ram and making it more likely to crash the own app at startup. Dont know if it’s better to choose the more ‘stable’ device (iPod) or the wider spread device (iPhone).
Is there a common sense about which generation to choose as the bottom end of supported devices? I guess it always depends on which game to develop (2D, 3D, …). So let’s say an entirely 3D-Game with about 10.000 faces and 30 drawcalls without physics (except charactercontroller, collision-platforms and triggerspheres) - maybe a 3G is out of the race then?
10k triangles and 30 draw calls would run decently on a 3G (or the original iphone since they are identical hardware wise). It all boils down to how you optimize everything.
do not buy a 3G unless a client requires you to get one, its a dead platform.
iOS already by now does not even support that generation anymore, 3GS+ is the only devices supported as of now and the future by the current iOS SDKs, GameCenter and various other things
I think 3GS will be supported on iOS 5 (the last iOS update for that generation). All 2G devices and below will not be supported in 3 month’s time when iOS 5 come out so they will all be stuck at iOS 4.
Apple calls the 3g a dead device, doesn’t make them any more and will soon be phasing out the 3gs after xmas. So do you need to support it? also it uses fixed function, so none of your 4g/ipad/3gs shaders will work on it.
If I was doing a standard 2D game, I’d support the 3g, however it doesn’t have much muscle for anything half decent in 3D.
Yeah, sure, if you want it at 30fps and having no animation at all, plus no lighting. More realistic to go half that quote.
thanks, this helped a lot
That’s just about right. Although I’ve managed to squeeze in 3 animated meshes, a good couple of non-kinematic rigidbodies (a max of 15) and some script animated chunks. No lighting, ofc.
But that was at 12k verts (not 10k tris) with everything on 2 texture atlases and very little update dependent scripts + everything was pooled. = 15 draw calls + another 10-15 for the inverted normals for the outlines.
AppControler.mm was set to 1.0 only, switched to nstimer, enabled the draw link, reduced physics solvers and fixed time step. Basically every optimization possible. A bit too much work for a really simple title imho.