Notices

Take a look at a picture of a N900 and the question should answer itself.

Regarding Pandora's "gaming optimizations" vs N900: Afaik, both the N900 and Pandora are based on TI's full-featured OMAP3430 / 3530 System On a Chip, respectively. That means that they share the same cores, capabilities, coprocessors, interfaces etc. There was no way for MWeston to for e.g. 'drop in a more powerful graphics card' onto the board.

At the moment, it looks like the PowerVR SGX openGL ES texture surfaces will not be updateable fast enough to use for scaled 50-60fps screen-updates from emulators.

That's pretty aggravating, since the concept of maemo5 is to map application windows to openGL textures. Unless some new info comes along, I think we are looking at no high fps emulators (or *any* games, or high framerate video...) within scaled GL-ES windows.

I'm ignorant of what is available and planned for fremantle in the hardware-scaling department. The SoC also has some video hardware which should handle such things easily. Video players, games, emus should at least have 1990s tech. for drawing to screen. That means providing the basics. Scaled, Vsync'd Blitting, preferably with support for double/triple buffers.
--

Back on topic, Exophase's (gpsp) and Marat's (vgba) should run fine, but only Marat has contributed emus to the tablets so far; we'll see...

Thanks ArnimS. I'm not a developer, but could at least a software scaling, via pixel doubling or something be done without major problems?

Just attach a bluetooth based gamecontroller, then plug the TV-out into your telly :-)

Depends on what you consider major problems. I can think of more than two people who got bogged-down trying to work around similar issues, and a lot of things that they might have otherwise achieved.

Doubling is gone afaik. Of course you can software scale, but every bit of waste adds up. A 2x scaler means you must copy 4x the data. I'm not too worried about getting some decent/simple screen updating, but I am concerned that the apps that need to go that route will end up necessarily breaking the windowmanager concept - bastard OS stepchildren swept out in front of the doorstep.

I'm wandering a bit here but i hope this bit is interesting: one of the advantages of the maemo5 compositing task-switch/fullscreen metaphor is it's even more agnostic to historic linux programs. Managing those apps as fullscreen units means less putzing with layout changes for the smaller screen. In other words the new taskswitcher helps visually integrate those non-native (e.g. SDL) apps with the rest. But what's the point of havin an overview of your *running* programs when the game window is either black, or boggin down your scrolly-overview with 15fps updates?

These are the things I wonder about.

[edit] while i'm goin..

How about allowing for badly ported things or apps requiring higher resolution to expand the desktop to higher (and/or higher/wider) than 800x480. User would have to pan around desktop with stylus+key combo, or possibly by tilting unit around. Hey as long as we are doing Gl...

https://gforge.ti.com/gf/project/dmai/ This might be an alternative - daVinci multimedia application interface.

Not exactly. I am currently using accelerometer to control the actual GBA gamepad. I can also use it to control the emulated GBA tilt sensor, but not right now.

Good luck for development, I'll donate to you when this version is out and my n900 in my hands

I do not take donations. You can "donate" by buying any of the commercial versions (WIndows or Symbian).

Hm, yes ok

How hard would it be to use a Wii controller as a pad for those times you have a desk to prop your N900 on?

I see there is already a project cwiid which does this for desktop Linux.

An initial port of Cwiid to Maemo was done more than a year ago. See: "Port: Cwiid on OS2008 - Wiimotes on maemo, now you can play those games properly!"