Pandora "console Edition"

[Exophase]

A low power PC? For game programmers like me, consoles are a great target because they have one big advantage: uniform hardware. That drastically reduces the time you need to write a game. Good luck with that on the PC platform where you have a plethora of graphic cards, sound cards, processor speeds, operating systems, etc, etc, and you have to write tons of hacks to get around those problems. PC platform is open but not uniform.

PC's aren't like that to game programmers at all these days - you program for standard uniform APIs or at worst OS specific ones. The most cutting edge games do end up using different engines for different video cards, but this is only necessary to get the most out of performance. Using a lowest common denominator will give you better performance than backing down to the level of a cheap custom console.

 

[Beoran]

Hehe, can I get a beer too? ^_~

Seriously guys, I really didn't want to rile up anyone. Again, I'm sorry if anyone felt that way. My ideas and questions are honest, if a bit naive, perhaps? >_<

 

[Beoran]

Exophase, yes I know that. That's how I'm doing it now. You can use standard API's like SDL, OpenGL, OpenAL, etc, etc. Still, that approach has may problems, because these standard API's don't cover everything you need for a game. There are very few uniform, standardized API's for things like filesystem, filesystem layout, GUI subsystem, how the main() function should be called (no thanks to you, WinMain()), etc. And then there are all those subtle diferrences between the different compilers and build systems. And of course, every time I have to recompile my game or every operating system I want to support. Porting is still much of a pain. Both Java and Flash prove one thing: Binary compatibility is great.

As for the hardware, what is the lowest common denominator for the PC platform? PC's up to 1 years old, 3 years old, 5 years old, 10 years old? We shouldn't be blined by the clock speeds, a contemporary 300 Euro PC with an intel graphics chip probably has worse performance that the Pandora. Nevermind all those people who still use a 5 year old PC!

Pandora uses an ARM CPU that has high performance even at low clock speeds and has a very nice GFX chip built in. The hardware looks like it could be cheap if you cut out the LCD screen. That's why I think there could be a niche for a "Pandora Console Version" around 100 to 150 Euro or a "Pandora docking station" around 50 to 100 Euro. Perhaps the latter is the most viable idea.

 

[TaG]

If you are happy with the SoC chip and graphics but don't want the rest of Pandora, maybe beagleboard would be suitable.
They are likely to have the same base OS, it has DVI rather than VGA/HDMI and is expected to cost $149, currently about £80 or €100.

Once it is out it should be possible to create a 'recipe' so that people can make their own 'console', ie add this hardware to the board / USB add these drivers and software that you will need. Then games could potentially be written that would work on both. If tweaks are needed there would still only be two versions, most likely for screen size or controls at a guess.

I want a Pandora for the portability / pocket size. I think it will be my most used gadget/toy/whatever. I will also keep an eye on beagleboard because if it ends up being able to do something / interface with other hardware that the Pandora can't nd does something really cool it could be easily worth £80.

Pandora docking station isn't likely to be able to link to too much non-USB hardware, eg vga/hdmi, hence why people talk of USB/wifi/BT devices - there are other threads on these topics so it might be worth looking at those.

 

[Beoran]

Thanks for that link!

Hmmm... the beagleboard show that it's possible to make a moterboard for 100 Euro that could be a used in a Pandora-compatible console setup. It's still a bit expensive to be completely viable for a console, though, as the casing, connectors and power supply would probably cost more than 50 euro. I guess it would be neccessary to design an even more cost-cutting motherboard.

 

[danboid]

Pandora 'console version' would be actually be a downgrade from Pandora normal seeing as it loses one of its key selling points- its high portability- and hence would be doomed to absolute failure.

I can't think of a single console or computer that was released as a portable first and then as a non-upgraded console later. There's a good reason for that because non-portability is a de-volution. Megadrive (Genesis anybody?) begat Nomad, iMac begat iBook, PS(2) begat PSP etc etc.

If the team had neglected to put TV out and USB 2.0 host on the original Pandora but these were added to the console, things might've been different but obviously thats not the case. Plug pand into your TV, AC and USB pads/USB DVD or whatever and you have your console right there.

Woe betide the next being who suggests we remove the keyboard- mods could you please just delete all threads that run along the lines of "I have this great idea for the Pandora hardware- totally cripple it. Remove the keyboard, make it non-portable, stop it doing anything useful" or whatever

 

[Beoran]

Danboid,

Point taken! I hadn't looked at it from a consumer point of view. When I think of now, it there are only a few reasons why people still use consoles over portables: screen size, and being able to connect it TV, and general performance. Why else would they pay more for a console than a handheld? But the Pandora has a TV out, so basically, it can already function like a console... :huh:

The only argument I have against what you have to say is price, that is, a "console version" could arguably be a lot cheaper. The VTech VSmile Console costs about 45$, while the VSmile portable is around 75$, and they both have the same perfomance (they even use the same cartridges). Admitted, this is the low end of the market, and performance is SNES-like.

A console version of the Pandora would have to be dirt cheap (I guess about 100$) to be viable. However, I finally found out the the price of the CPU, and I realised that that would be a tough nut to crack. With the CPU of the Pandora costing 50$ even at high volume, it's simply too expensive yet to go for the low end console market... So, my idea simply comes too soon. I guess I'd have to wait a year or two for Moore's law to work, and bring the chip's price down. Or I'd have to use a cheaper CPU which would make my idea completely Pandora-unrelated so I won't discuss that idea here further.

And I can understand that you'd get a bit miffed at people suggesting all sorts of crazy stuff here. But the reason I'm suggesting my wild ideas here is to get feedback on it, and see if my ideas would be viable or not. You all have given me a lot to think about, and a lot of interesting feedback, which helped me see things in a more realistic way than I could in my initial wild enthusiasm. Thanks for that, people. ^_^

And finally, I do hope the Pandora becomes a successful. If it does, I hope to port my games to it.

 

[danboid]

np beoran!

Just another classic case of gotta post/post before you think/read any docs- we've all done it

I look forward to seeing what you will (help) create for our power pocket pal the andora1:

 

[CyruzDraxs]

My only complaint with using the Pandora as a console is the wimpy S-Video. On my HDTV that would look like a blurry mess. That's where my suggestion for a dock came from; if you could put HDMI, DVI or at least Component video on it you could have a higher resolution output on TV-Out and make everything look even better. I know the Pandora is very capable of such a thing--only noticeable slowdown might be in PSX emulation, but not by much if what I've heard about it's progress so far is correct.

 

[Beoran]

Well, the problem wit HDMI output is that the CPU doesn't support it. It does support S-Video "out of the box", so I guess TV out was added to the Pandora because they essentially "got if for free". HDMI would probably require the addition ofyet another chip on the board, which is not realistic at this stage.

 

[gaterooze]

Dude... if you were going to build a console, why would your first port of call be a handheld SoC? Pandora was specifically designed for portability and low power use above all else, two factors that are irrelevant to a console.

You could find far more powerful hardware at lower cost that has higher drain and a larger form factor, and go from there.

And I still don't see what it would do that a small linux box (cobbled together for $100 or so) hooked up to a TV wouldn't, and better.

 

[trooper]

Ok!, I was wrong.

The beer`s on me.

Trooper

 

[MarkoeZ]

i'm usually restricted to serious reply's for being a noob here. But since i followed this thread and a beer is offered:

Cheers!!!!

 

[CyruzDraxs]

Well, the problem wit HDMI output is that the CPU doesn't support it

Uh...yes it does. The Beagleboard had HDMI before they decided to switch it to DVI-D, and it's the exact same CPU.

Dude... if you were going to build a console, why would your first port of call be a handheld SoC? Pandora was specifically designed for portability and low power use above all else, two factors that are irrelevant to a console.

You could find far more powerful hardware at lower cost that has higher drain and a larger form factor, and go from there.

And I still don't see what it would do that a small linux box (cobbled together for $100 or so) hooked up to a TV wouldn't, and better.

Because then it wouldn't run Pandora's software and games. I don't want it to be it's own separate console, I want it to be a way to play Pandora's games on a higher res TV.

 

[Squidge]

Well, the problem wit HDMI output is that the CPU doesn't support it

Uh...yes it does. The Beagleboard had HDMI before they decided to switch it to DVI-D, and it's the exact same CPU.


Uh, no it doesn't.

HDMI and DVI are provided by an external IC that converts the output of the CPUs frame buffer. The CPU itself doesn't support HDMI.

 

[Beoran]

I quote http://focus.ti.com/docs/prod/folders/prin...tml#qualitydata
QUOTE

Display Subsystem
* Parallel Digital Output
o Up to 24-Bit RGB
o HD Maximum Resolution
o Supports Up to 2 LCD Panels
o Support for Remote Frame Buffer Interface (RFBI) LCD Panels
* 2 10-Bit Digital-to-Analog Converters (DACs) Supporting:
o Composite NTSC/PAL Video
o Luma/Chroma Separate Video (S-Video)
* Rotation 90-, 180-, and 270-degrees
* Resize Images From 1/4x to 8x
* Color Space Converter
* 8-bit Alpha Blending



So the omap3530 chip-on-a-chip supports composite out, S-Video and LCD panels. No HDMI.

GunPei2X,

Point taken! The only counterargument to that is compatibility, but, the omap3530 doesn't have any more powerful versions yet. I guess the Super-H4 would better place to start for a console (proven to work well in Dreamcasts.) On the other hand, for a cheap handheld, I want whatever the VG Pocket series is using, they sell full color 2' LD screens that can run arcade classics starting from 10$. Add a USB controller and some nand memory, and we're set for a very cheap programmable retro-machine. ( http://www.pdp.com/prod_by_brand.php?brand=vgpocket )

 

[Exophase]

Exophase, yes I know that. That's how I'm doing it now. You can use standard API's like SDL, OpenGL, OpenAL, etc, etc. Still, that approach has may problems, because these standard API's don't cover everything you need for a game.

I disagree...

There are very few uniform, standardized API's for things like filesystem, filesystem layout,

Why does a game need to poke around the filesystem? You should not need more than to access files relative to the current directory you're in, and the filesystem used to implement this is irrelevant. Pretty much any programming language you'd use abstracts this. At worst there are different separator conventions, except if you use gcc everywhere. Or if you need directory enumeration, except if you use gcc everywhere... pick the right compiler.

GUI subsystem

Use Qt, or write your own GUI that's suitable for the game. Were you planning for games on this console to heavily use standard GUIs? That's not the norm for most consoles made...

, how the main() function should be called (no thanks to you, WinMain()),

Abstracted by SDL.

etc. And then there are all those subtle diferrences between the different compilers and build systems. And of course, every time I have to recompile my game or every operating system I want to support. Porting is still much of a pain. Both Java and Flash prove one thing: Binary compatibility is great.

Differences between compilers and build systems is irrelevant to differences between platforms - use gcc.

I think the thing I really don't get about all this is that you're talking about software/OS/tool differences, but isn't it just as well to say "I'm just going to be developing this for Windows or Linux, and computers greater than 5 years old", rather than limit it to a niche platform that will certainly have less users than any limitation you can pick for PCs? And I don't really understand the hang up over having to recompile a program for another platform, after you've probably recompiled it thousands of times while developing it. I think this is more people inventing advantages to support technologies after the fact.

Personally, I don't do games, I do emulators - but where portability was a concern early on it has been easy to port to different operating systems, devices, etc. If you were to do games for this platform only you'd probably piss off a lot of people anyway so I'm sure most people developing for it would just develop for PC as well anyway. There's no real point to any of this, really.

As for the hardware, what is the lowest common denominator for the PC platform? PC's up to 1 years old, 3 years old, 5 years old, 10 years old? We shouldn't be blined by the clock speeds, a contemporary 300 Euro PC with an intel graphics chip probably has worse performance that the Pandora. Nevermind all those people who still use a 5 year old PC!

Pandora uses an ARM CPU that has high performance even at low clock speeds and has a very nice GFX chip built in. The hardware looks like it could be cheap if you cut out the LCD screen. That's why I think there could be a niche for a "Pandora Console Version" around 100 to 150 Euro or a "Pandora docking station" around 50 to 100 Euro. Perhaps the latter is the most viable idea.

I think you have a very inflated opinion of the OMAP3530. Sure, it's great for handheld technology, but desktop technology surpassed it well over 5 years ago. As far as CPU is concerned a low end machine you would buy today would cream Pandora. The belief that clock for clock ARM is so much more incredible than x86 Pentium and greater is a total myth, and I'd love to know where it originated (although ARM is catching up, especially with Cortex-A9... but that's not out yet)

Because then it wouldn't run Pandora's software and games. I don't want it to be it's own separate console, I want it to be a way to play Pandora's games on a higher res TV.

Aside from upscaling (which has arguable value), how do you intend to play Pandora software at a higher resolution than it was programmed for...

And with that in mind, why would the TV out it offers be so insufficient?

 

[Beoran]

Exophase,

Of course, I program my games as portably as possible, but you'll have to admit that doing so is more difficult than programing focused for one system.

As for the common lowest denominator, I admit that I may have overestimated the OMAP. However, that doesn't really change the problem of deciding what the common lowest denominator should be. I only have access to a few PC's so , all I can do is try my games on those, and hope that they will run well on other systems of comparable age.

I do agree with you on the TV out thing. Although, theoretically, the graphics driver on the OMAP can do HDMI resolution, it would probably be very slow at anything seriously above the LCD resolution. So component or S-Video out should be enough I guess.

 

[CyruzDraxs]

Hmm...I guess I was mistaken then. I didn't know that the HDMI/DVI on the Beagleboard was handled seperately. Still though, if the Beagleboard can do all that for $150 there shouldn't be any reason we can't do that with a Pandora console.

 

[Exophase]

Exophase,

Of course, I program my games as portably as possible, but you'll have to admit that doing so is more difficult than programing focused for one system.

If your target from the start is "PC" then I generally disagree, or at least I would say that it's only slightly more difficult at most. That is, if you keep it in mind from the beginning.

As for the common lowest denominator, I admit that I may have overestimated the OMAP. However, that doesn't really change the problem of deciding what the common lowest denominator should be. I only have access to a few PC's so , all I can do is try my games on those, and hope that they will run well on other systems of comparable age.

You don't really have to decide anything. It either works fast enough on someone's computer or it doesn't. If you have no problem writing software for a niche platform and excluding everyone else then why should you feel bad about excluding people with ancient hardware, people who would most likely benefit in many other ways if they were to upgrade?

I do agree with you on the TV out thing. Although, theoretically, the graphics driver on the OMAP can do HDMI resolution, it would probably be very slow at anything seriously above the LCD resolution. So component or S-Video out should be enough I guess.

Right, and that means that aside from meeting lower price requirements and perhaps providing a standard external controller Pandora is already the game console you're interested in. I've actually seen people mention several times now that they want an open console to develop on, and while I'm not the least bit interested myself every time I tell them to look to the Pandora for it.