[Tarwin]

Hi guys,

as a huge fan of everything Freespace, and one of the maniacs that payed a lot of money for a product yet in the making, i am asking how to combine these two.
The product i am talking about is a new handheld console, the Pandora aka OpenPandora. Its a very powerful handheld, powered by a ARM based TI Cortex CPU, OpenGL2.0 ES compliant 3D hardware, running Linux. More infos here:

http://www.openpandora.org/index_en.php

As this is 100% community driven, a lot of homebrew software is starting to show up (and it doesn't even get sold yet), including ported open source games. Some, like "Battle for Wesnoth", were redicusly simple to get to run on the ARM hardware. Some, like Sauerbraten/Cube need more work, but can be ported (as proven by the iPhone port).

FS2_open would be perfect for this little gem. My questions now, is anyone aware off an Linux/ARM port, played around with the concept or at least agrees with me, that this would be a great idea?

- Tarwin 

[Wobble73]

This has already been discussed I believe, please see this thread!

Also..........




 

[Tarwin]

This has already been discussed I believe, please see this thread!

Also..........




 

Good to be here. =)

Sorry, i stressed the forum search, but found nothing.

The question persists, has anyone ever tried porting FSo on ARM?

[taylor]

The question persists, has anyone ever tried porting FSo on ARM?

I don't think that getting it running on ARM would be that big of a deal, though properly optimizing it might be a bit more involved.

The biggest problem would likely be making it compatible with OpenGL ES.  A good deal of code would have to be added/rewritten for that to work and even then it would mostly just be quick hacks with a negative performance impact.

[chief1983]

Does the Pandora setup put the ARM processor into Little-endian mode?  That alone would be a huge hurdle if not I'd imagine.  Also what are the sizes of common variable types like int on ARM vs x86?  I found an interesting article on porting x86 to ARM and other RISC hardware types here, might be worth reading for anyone interested in seeing if this is possible.  Also keep in mind, many mods and enhancements have started to omit the low-res interface art which would cause the engine to break since it would be looking for that, at a resolution of 800x480.  Still, retail would be nice to have running on it.  I just imagine there's a lot of issues with the code that would have to be fixed.  All someone has to do is start working on that though.

Yeah I missed that ES part too.  It could be a lot of work to get it within that spec.

[taylor]

FSO already supports both big and little-endian systems, so that isn't an issue.  Alignment isn't a big issue either since I already worked to remove that as a barrier while working on the PPC port.  Also, type sizes on ARM-8 are the same as standard x86.  And SDL is available which makes it a lot easier to port over.  Vorbis, Theora and OpenAL should also be available (not 100% sure if they are now or not, but I'm pretty sure it's being worked on if not).  So, getting it ported over processor wise shouldn't be a big deal.

[Wobble73]

Thanks Linux guru!
 I imagine the graphic capabilities would be low though? I mean we are talking retail level, maybe a little better, yes?

I kinda interested in getting one of these if possible in the future!

[taylor]

I imagine the graphic capabilities would be low though? I mean we are talking retail level, maybe a little better, yes?

I don't really want to guess at the performance without ever having used a PowerVR chipset.   But it is OpenGL 2.0/DirectX 10 level hardware, so it's definitely not some crap card to deal with.  It should be SM4.0 compatible or better as far as I know, meaning that it is probably better than what many people use now with FSO.  CPU and memory would likely be the bigger issue.  FSO is pretty CPU limited, though with a proper graphics upgrade for OpenGL ES support it should be possible to greatly reduce that.  And with memory being so limited (by FSO standards anyway) it will certainly cause problems with some mods/features unless custom content is created (lower-res maps, etc.).

Assuming the work was put into it all though, it might be possible to get something up and going which could actually be graphically superior to current FSO.

[Tarwin]

Well, sounds like I am in the right forum. =)

Some technical data (OMAP3530):
http://focus.ti.com/download/tidc/pdfs/asia/2008/cn_T3_3_Bob_Lee_Ingenient.pdf

Shader Effects
http://www.youtube.com/watch?v=24TXpqa9jG0

More Eyecandy from the cortex platform:
http://www.youtube.com/watch?v=evJ0Vr8LjYk

Interesting Thread in the pandora forum
http://www.gp32x.com/board/lofiversion/index.php/t45129.html

The first working prototype has just been assembled! =)
http://www.openpandora.org/blog.php

The pandora has 256MB of RAM (specs were upgraded 2 month ago). The hardware is OpenGL2.0 capable, only the driver is missing. Maybe TI will work in that direction, we can only hope.

[Tarwin]

I thought about it a little more. Porting to OG2.0 ES would enable ports to other mobile devices, such as a iPhone or an Nokia N97.
These should well be powerful enough to run FS2_open (lets think about input another time).

Other mobile devices that should have no problem running FS2_open can be expected to show up in the near future. Most N-Series Nokia devices should already be able to display the basic FS2 engine. Porting it would surly bring attention from other coders who could use the engine for something FS like on mobiles. The possibilities!

[Tarwin]

There is a better Datasheet:
http://focus.ti.com/docs/prod/folders/print/omap3530.html

[Drack]

Sorry to dig this topic from the grave.

FreeSpace 2 is one of the best games I ever played.
I'm not a coder with loads of experiences. I have some knowledge into C, C++, java but I never used OpenGL.
Since I'll get a Pandora when the first batch will be shipped, I would like to try to port FreeSpace 2 on it.

As I said, I'm not experienced, do you think it would still be possible for me to give it a try and would it be a good "introduction" to game programming?
I know it's a big project, I don't hope to be able to finish it over a short amount of time.
You know, if I could make a ship without textures spin at 2fps on the screen, I would already be happy and proud of myself 

(Excuse my english, I try my best to be understandable)

[chief1983]

It would be a terrible introduction, as the engine isn't exactly well organized.  It's also never been compiled for ARM yet, to my knowledge.  However it does work on PowerPC so it might run on ARM with a little tinkering.

[Drack]

Thanks, then I'll keep doing tutorials to learn OpenGL handling.
Maybe I'll try later, if it's not already done by some talented coders 

[chief1983]

Yeah I don't want to be offputting or anything, I just don't want you to look at FS and get turned off to game programming altogether or anything.  It's been rough for a lot of more experienced coders to get into.