Our New Machine, Pandora

[Alex.]

And let's face it, as talented as the homebrewers are, they simply don't have the time and resources to make a game to justify that power. It takes years for a team of professional game developers who do it for a living to make a commercial game. How can a one- or two-man team make anything even 10% as good in their limited spare time? What we will see are Asteroids clones featuring asteroids made up of 50,000 polygons and the same basic gameplay that you've seen on the GP32 asteroids games. No offence to the homebrewers on this scene but have you ever seen a GP2X homebrew game that compares to an average Megadrive game in terms of quality and quantity?

That's not the way to go forward, Billy. The GP2X already has some neat homebrew games that surpass asteroid clones. A more powerful machine with better 2D acceleration can only improve things and open doors for more complex and demanding software.

You should also keep in mind that these handhelds are appealing most to enthusiasts and homebrew coders than to those who expect cookie-cutter games that a team of 50 worked on for 4 years. A lot of these people look forward to it, it will be a lot of fun.

 

[z32fairlady]

So you guys are going to make a brand new gaming handheld (which is built by community input)? Nice to hear that, and looking forward to seeing this jewel comes out.

But looks like many things are not decided yet, it is still possible to roll out by the end of this year?

 

[Magic Sam]

Hi everyone,

I'm really excited about this new handheld, it will be a nice replacement for my broken and unrepairable GP2X, and a far better deal than the GP2X F-200
As I'm also interested in the OpenMoko project (because my Sagem my201x sucks) I'll have to start saving my money soon
I agree with those who say that we don't need such a powerful device for homebrew gaming and emulators, as it will drain batteries and our pockets faster We're lucky to have gifted coders in this community, and in my opinion they could code great games for microwave ovens if they had to Hardware is not that important, don't you think ? But I have to admit that I would feel very great if I had a very powerful and expensive handheld in my hands

Now I'm only waiting for news from Craig, ED & Co.

Bye

Magic Sam

 

[cappuchok]

If the OMAP3430 could become a contender, I would be a very very happpy bunny indeed. Not going to gget my hopes sky high yet, but well, it'll be very interesting to see what happens....

I agree wholeheartedly. Craig himself said they intend the device to be cutting-edge on release. If we're talking about a truly cutting edge device (never mind the price), the OMAP3430 is where it's at. If we're talking "almost cutting edge" and have a specific price point to meet, the i.MX31 is a good candidate (or even better, the SiRF Titan though it requires a few addon chips for functions built into the i.MX31) but I'd rather see a device with a "full" MBX core instead of the "Lite" one. Unfortunately most SoCs utilizing "full" MBX cores seem to use the SH-4 CPU.

That being said, I'd happily pay £200 for a Linux gaming/emulation/media device using the OMAP3430.

As for homebrewers not being able to take advantage of the immense power of cutting-edge chips, just take a look at some of the recent Windows- or Linux-based 64K intros ("paradise" from 2004 is highly recommended!) at www.pouet.net, and reconsider.

A list of viable (ARM-based) SoCs mentioned so far (I hope all of these are under consideration), in order of my humble preference:
1) TI OMAP3430
2) SiRF Titan (with suitable add-on chips for BT, WiFi, TV/VGA-out and stuff)
3) i.MX31 (with a suitable TV/VGA-out chip)
4) MMSP2+ (making sure RGB/VGA is usable).

 

[Dubya']

MMSP2+ is around $12 (including TV out)
IMX.31 is around $30 + TV out chip which has to be bought separate and added to the PCB, I guess the max cost of both would not be above $40.

Although battery life is a very important factor and I can't find actual power consumption figures for any of the SoC's mentioned thus far as being under consideration, the MMSP2+ sounds like the lowest cost option and, because price is another very important factor, I'd be perfectly satisfied with a machine based upon it. With all of the other features of the new machine implemented from GP2X user input combined with the capabilities of even the MMSP2+ which blows away the current GP2X SoC, this machine sounds like a definite "must buy" to me if the price ends up being anywhere close to that of the current GP2X.

 

[naples39]

I think the people saying that the OMAP has more power than necessary are correct in some ways, but can you ever have too much power? The more powerful the machine, the longer the life of the machine will be, and the better the software will be sooner or later.

At the risk of stating the obvious, there is no magic answer for what is the correct level of processing power for this machine. This powerful OMAP SoC is probably in the area of diminishing marginal returns for what devs can do with the processing power, so it really all comes down to the prices craigix can get on these things. If it's slightly more expensive, I say hell yeah get all that power in there. If it's really gonna boost up the price into another bracket, then no, the OMAP isn't worth it.

I just personally hope this new device is a significant boost in power so that we can see big steps in SNES, Amiga, MAME emulation...etc

 

[fatblueduck]

I hope that battery life, wireless, and a good screen take priority over everything else.

If gamepark had included those upgrades with the f200, I would not buy another device. The gp2x would still be cool taday, it just lacks those few things.

 

[cappuchok]

MMSP2+ is around $12 (including TV out)
IMX.31 is around $30 + TV out chip which has to be bought separate and added to the PCB, I guess the max cost of both would not be above $40.

Although battery life is a very important factor and I can't find actual power consumption figures for any of the SoC's mentioned thus far as being under consideration, the MMSP2+ sounds like the lowest cost option and, because price is another very important factor, I'd be perfectly satisfied with a machine based upon it. With all of the other features of the new machine implemented from GP2X user input combined with the capabilities of even the MMSP2+ which blows away the current GP2X SoC, this machine sounds like a definite "must buy" to me if the price ends up being anywhere close to that of the current GP2X.

However, a device using MMSP2+ would be a far cry from being "cutting edge", which is Craigix's stated intention with this device. Though if the choice narrowed down to those two chips (MMSP2+ and i.MX31), and no RGB/VGA-capable TV-out chip can be found to complement the i.MX31 while the MMSP2+ can do it out of the box, I would also choose the MMSP2+ (unfortunately, the MMSP2+ info states YCbCr as the highest TV-out standard). The SiRF Titan also uses the MBX Lite just like i.MX31 (it is not specified which 3D core is used by the MMSP2+), and requires several add-on chips for stuff that's already integrated in the competitors (Bluetooth, WiFi and stuff like that).

If OMAP3430 chips can be acquired at a reasonable price and in a reasonable amount, though, that's the only chip mentioned so far that would make the device well and truly cutting edge (similarly revolutionary as the Amiga was when it was first released, and it took the competition years to catch up), and it would possibly even enable N64 emulation, but I'll settle for fullspeed PSX, Amiga and SNES with sound and no frameskip.

 

[jimid2]

Although battery life is a very important factor and I can't find actual power consumption figures for any of the SoC's mentioned thus far as being under consideration, the MMSP2+ sounds like the lowest cost option and, because price is another very important factor, I'd be perfectly satisfied with a machine based upon it....{snip!}

I agree, battery life is a huge factor! in the top three, for sure! All this talk of cutting edge, high-power processors hasn't addressed battery life to any real degree that I can see... Look at the PSP - running at 200MHz and without using the UMD (running off the memory card), battery life is still not all that impressive even with an expensive, high capacity, custom-made LIon battery pack. Jack it up to 300MHz and you're lucky to get three hours (without using the UMD drive at all!)... The cost of running a high-speed, power-hungry SoC is more than just the initial outlay for the chipset; it's the cost of more powerful LIon batteries to get acceptable battery life as well. And lets not forget that screen quality is largely corelative to power consumption as well (bigger, brighter screens with faster response time require more power). Where are the trade-offs made? Now, if the OMAP3430 uses LESS power than the MMSP2+, then I'd think the per chip cost premium would be well worth it, but it's a real juggling act trying to balance all these issues...

 

[craigix]

The original plan was to make a low cost machine with a good spec (hires screen, lots of ram, mid range but low cost soc etc) however the view of all the threads seems to point that most people want something quite high spec and don't mind paying more for it.

We are taking all this into account, and I hope we can get the balance right (probably the imx31 at the moment).

Everything is now ready to roll once the final decision is made.

 

[TheMinder]

QUOTE
however the view of all the threads seems to point that most people want something quite high spec and don't mind paying more for it.


I actually got the vibe from the relevant threads that people wouldn't be happy going much higher than around the £125 the GP2X costs, but I could have misunderstood.

 

[natxo]

Yes, everybody wants high specifications and low price but the most important issue is an homebrew-based videogame are DEVELOPERS. And developers want a machine easy to port and develop software and here there're important some things like float point hardware support and GPU processor. I think Nokia N800 it's a good example. TV-out or keyboard integrated are secondaries.

In conclusion, i.MX31 would be genial

 

[sephiroth111]

that's extremely good news. i don't mind the cost too much, provided i have fullspeed SNES, NES, and GB/C/A, Amiga, and PSX would be a plus also. If it has a "chatboard" so to speak, that'd be nice also. i'd like just a simple text editor (similar to notepad or kate) to make notes or lists. if it has wifi and bluetooth, a pictochat-esq program would be nice too, (or AIM for that matter. using libpurple from pidgin)

 

[Orkie]

(or AIM for that matter. using libpurple from pidgin)

May as well just use Pidgin .

 

[fusion_power]

Although battery life is a very important factor and I can't find actual power consumption figures for any of the SoC's mentioned thus far as being under consideration, the MMSP2+ sounds like the lowest cost option and, because price is another very important factor, I'd be perfectly satisfied with a machine based upon it....{snip!}

I agree, battery life is a huge factor! in the top three, for sure! All this talk of cutting edge, high-power processors hasn't addressed battery life to any real degree that I can see... Look at the PSP - running at 200MHz and without using the UMD (running off the memory card), battery life is still not all that impressive even with an expensive, high capacity, custom-made LIon battery pack. Jack it up to 300MHz and you're lucky to get three hours (without using the UMD drive at all!)... The cost of running a high-speed, power-hungry SoC is more than just the initial outlay for the chipset; it's the cost of more powerful LIon batteries to get acceptable battery life as well. And lets not forget that screen quality is largely corelative to power consumption as well (bigger, brighter screens with faster response time require more power). Where are the trade-offs made? Now, if the OMAP3430 uses LESS power than the MMSP2+, then I'd think the per chip cost premium would be well worth it, but it's a real juggling act trying to balance all these issues...

Agreed to, like I posted before. Battery Life is important (an WORKING battery warning LCD to by the way :lol: )
I prefer a looooong battery life of the new device which means MUCH longer than GP2X battery life. GBA or Nintendo DS-times I want to see.
Would be cool, if there were infos about the power consumption of the chipsets. Now, everyone speeks about this OMAP3430....do this thing use an ARM processor? (for GP2X compatibility of course)

The original plan was to make a low cost machine with a good spec (hires screen, lots of ram, mid range but low cost soc etc) however the view of all the threads seems to point that most people want something quite high spec and don't mind paying more for it.

but not all people. I don't want to buy more than - lets say- 250€ for such an Device. If the price is higher you reach the region of modern powerful PDA's where you can do nearly the same stuff on it. To prevent this choice because of equal price regions, the new Handheld should be more in an Handheld price segment instead of an PDA price segment.

We are taking all this into account, and I hope we can get the balance right (probably the imx31 at the moment).

To make "KeyboardCowboy" Happy... :lol:

Everything is now ready to roll once the final decision is made.

LETS PUSH THE <START> BUTTON!

 

[cappuchok]

The original plan was to make a low cost machine with a good spec (hires screen, lots of ram, mid range but low cost soc etc) however the view of all the threads seems to point that most people want something quite high spec and don't mind paying more for it.

We are taking all this into account, and I hope we can get the balance right (probably the imx31 at the moment).

While I wouldn't mind paying a fair bit extra for the OMAP3430, I can definitely support the choice of the i.MX31 (with a good TV-out solution complementing it). All of the discussed SoCs are, in my mind, good choices, however the i.MX31 seems to strike the best balance between cost-per-unit, performance and amount of integrated functions.

In summary, the Titan has better performance (higher clock speed and dual core) than the i.MX31, dual fully concurrent SD interfaces and integrated GPS at the cost of less integrated functions, thereby adding the need for separate WiFi and Bluetooth chipsets, raising the cost. So that one pretty much goes out the window, although it seems to be a fascinating chip but maybe not quite right for this particular device.

The MMSP2+ is apparently rock bottom cheap (would get my vote if final price of the unit was the most important issue) and has everything integrated but is worse-performing and less cutting-edge than the other choices. It also lacks USB 2.0 Host (or USB-on-the-Go which also provides USB 2.0 Host) which is absolutely needed for any practical use of USB mass storage devices or other high-throughput USB accessories (DVB cards for example).

The OMAP3430 is the only real cutting-edge choice (Cortex A8 core, SGX accellerator and everything but the kitchen sink ) and would get my vote if the final price of the device wasn't an issue, but from what we know (or think we know) it comes at a high per-unit cost and must be ordered in high quantities.

So that leaves the i.MX31 plus a decent TV/VGA-out chip (which if chosen wisely may give more available options for connections than the pre-integrated alternatives). As none of the other "cheaper-than-OMAP" chips have the "full" MBX instead of MBX Lite anyway, the i.MX31 seems to be the best choice of these.

I would feel more comfortable, though, if the OMAP3430 had at least been properly looked at and a compelling reason given (the actual per-unit cost or minimum order volume required, for example, for comparison with the alternatives) for not choosing it.

 

[jadon]

I think the people saying that the OMAP has more power than necessary are correct in some ways, but can you ever have too much power? The more powerful the machine, the longer the life of the machine will be, and the better the software will be sooner or later.

A good SoC implementation will allow the clocks to scale and regions to be powered-off when not in use to avoid burning a lot of power when you don't need it. Setting the right operating point can be a system design challenge, but can certainly be done for each of the various emulators if they are the only primary application running.

 

[cappuchok]

EDIT: Should have edited, not quoted my post...

 

[atomicthumbs]

Maybe once everything is done we should look into getting Opera Mini on the device as the web browser. Opera did port it to the Wii and DS, remember?

 

[Squidge]

I don't see why we should restrict clock speed to encourage battery life when we can simply set the clock speed in software. We can do the same as the PSP - have a 200Mhz speed by default which can be altered by a running application or the menu/launcher itself (thus solving the problem of people writing applications and just running them at max speed with lots of delays rather than just turning the speed down).

At the moment, I'd lean towards the i.MX31 for it's open nature, mature status, and wealth of existing knowledge (Freescale have a Linux BSP, lots of people have played with the silicon and know how it works, and people know how to reliably nail it to pcb's). Oh, and lots of those people speak English rather than Engrish

I've no knowledge of the OMAP, but it sems TI wants to keep everything underwraps, shrinking the number of people who have the used the processor to big companies who are unlikely to communicate. This could make it more difficult to develop for, and increase the time taken to get a reliable platform available for development.