The Next Gp*x Revision


sam fisher

Well-Known Member
Joined
Apr 11, 2004
Messages
9,452
Location
Bristol, UK
Website
blog.peter-r.co.uk
I know there was a topic about this long ago, started by Epicenter, supposedly doing research for Craigix, and i'm pretty sure it turned into a GP2x vs XGP flame war as Epicenter ranted about the superior XGP SoC.

Well, what would be your MINIMUM specifications and price range?

As I stated in another topic.

As it would be in the distant future, not too distant hopefully, I'd require it to have AT LEAST two 400+mhz ARM cores. Not one good one, and one crippled with a tiny cache and no MMU etc. Two fully fledged ones. It would also have to have some decent 2D accelerator, and what the hell by that time most things will be 3D, I suppose it should have 3D acceleration hardware by then.

Not a wide screen screen, a 4:3 screen at either, still, 320*240 or at, VGA, 640*480.

Price: £300-400. I mean, if i knew for sure it was decent for the price, i'd happily pay a larger price.

Edit: AND A D-PAD...Not a crappy stick.
 
mayby you should have stuck to the other topics.
but screw £300-£400.
what the hell, we dont have money to chuck around like that.
£150 maximum, considering we then need to buy a SD card and other accesories after that.
 
I kind of doubt it's going to happen the way I'd hope, since Craig seemed so .. resistant to logic, shall we say. Strange fixations on software rendering, high-clocked CPUs for 'big numbers for marketing value, like big engines', and no 2D or 3D acceleration of any sort. Logical, right? :rolleyes: GPH follows similar patterns of thinking, insane in their own unique ways as well.

So if there is a new handheld out of GPH it probably won't make nearly as much sense as the design I pitched. But a guy can dream. I don't know of any SoCs based on 2 discrete ARM processors unless they are weird and obnoxious implementations like Magiceyes' in the GP2X's SoC, MMSP2/MP2520F. Some offer an ARM processor/DSP combo, like Texas Instruments' high end OMAP2 and newer series. Honestly, just one fast processor is probably the best, and most affordable, solution.

The call for a 2D accelerator is best satisfied by a fast 3D accelerator that can perform all the same functions but also brings a broader range of possible applications to the realistic scope.

My original suggestion was Samsung's S3C2460. Sadly the SoC doesn't really exist at this point in time due to lack of public demand. The design is cemented but there is no mass production. Freescale's i.MX31 (ARM11 at ~533 MHz, onboard ARM11-based FPU, MBX-RS based GPU supporting 2D and 3D acceleration and full support for the OpenGL-ES API being the major features) or a high-end TI OMAP2 chipset (e.g. ARM926EJ-S processor at ~300 MHz, ~200 MHz DSP, MBX-RS 2D/3D accelerator) would be good choices. A good balance between price and performance. If Linux is used as the OS rather than a custom firmware (most likely), 64MB+ of RAM, ideally 96MB, would be crucial. So would the use of MobileDDR memory for power consumption and speed reasons (since it would likely double as a framebuffer and scratchspace for intensive graphics operations.)
 
mayby you should have stuck to the other topics.
but screw £300-£400.
what the hell, we dont have money to chuck around like that.
£150 maximum, considering we then need to buy a SD card and other accesories after that.
Inflation, I want something that really actaully has potential, GP2x was outdated by release date.

I kind of doubt it's going to happen the way I'd hope, since Craig seemed so .. resistant to logic, shall we say. Strange fixations on software rendering, high-clocked CPUs for 'big numbers for marketing value, like big engines', and no 2D or 3D acceleration of any sort. Logical, right? :rolleyes: GPH follows similar patterns of thinking, insane in their own unique ways as well.

So if there is a new handheld out of GPH it probably won't make nearly as much sense as the design I pitched. But a guy can dream. I don't know of any SoCs based on 2 discrete ARM processors unless they are weird and obnoxious implementations like Magiceyes' in the GP2X's SoC, MMSP2/MP2520F. Some offer an ARM processor/DSP combo, like Texas Instruments' high end OMAP2 and newer series. Honestly, just one fast processor is probably the best, and most affordable, solution.

The call for a 2D accelerator is best satisfied by a fast 3D accelerator that can perform all the same functions but also brings a broader range of possible applications to the realistic scope.

My original suggestion was Samsung's S3C2460. Sadly the SoC doesn't really exist at this point in time due to lack of public demand. The design is cemented but there is no mass production. Freescale's i.MX31 (ARM11 at ~533 MHz, onboard ARM11-based FPU, MBX-RS based GPU supporting 2D and 3D acceleration and full support for the OpenGL-ES API being the major features) or a high-end TI OMAP2 chipset (e.g. ARM926EJ-S processor at ~300 MHz, ~200 MHz DSP, MBX-RS 2D/3D accelerator) would be good choices. A good balance between price and performance. If Linux is used as the OS rather than a custom firmware (most likely), 64MB+ of RAM, ideally 96MB, would be crucial. So would the use of MobileDDR memory for power consumption and speed reasons (since it would likely double as a framebuffer and scratchspace for intensive graphics operations.)
I've got to say I agree with you there, but I would want something with a fairly fast CPU as well, so if I have a choice it eould be the freescale SoC. Didn't they design the GP32 themselves? Can't they do that again?
 
Last edited by a moderator:
You've got GPH and GamePark mixed up-- GPH consists of some people who split from GPH to make a media player. Only 1 person, who was an intern at GP during GP32's development, was actually involved in the GP2X's internal design.

At any rate, the GP32 was based on a System-on-a-Chip as well. It was a Samsung part that encompassed all the system's core hardware, including the CPU. This is much cheaper in the end in terms of engineering costs rather than trying to make a lot of distinct components play happily together. With what robust SoCs are out there these days, for decent prices in large enough volume, it doesn't make much sense for a tiny company to custom-engineer a whole game system and it's less likely to turn out as well.

ARM's got a rock-solid platform designed these days that major SoC makers adhere to, that dictate how an ARM processor, RAM, FPU, 2D/3D accelerator, and numerous controllers should work in harmony across an industry-standard bus. With such strict and tried-and-true standards, there's far less chance of incident. The MMSP2/MP2520F in the GP2X likely derives many of its deficiencies and bugs from its deviation or complete abandonment of ARM reference standards in favor of strange, proprietary (and in the end, non-beneficial) implementations of the same basic technology.
 
Who in their right mind is going to pay £300-400 for a hand held game machine (even with a D-pad :)). Even Sony don't charge that that much!

You'd need some pretty amazing games (and plenty of them) for that much money, only a company like Sony or Nintendo might have the marketing power to get away with that.

I'd buy a proper hand held computer or a lap top.

Part of the appeal of the GP2X is that it is cheap.
 
That sort of price isn't necessary. I'm sure a system with the i.MX31 and a 320x240 touchscreen could fit under a $300 price point if the production run was large enough.
 
Who in their right mind is going to pay £300-400 for a hand held game machine (even with a D-pad :)). Even Sony don't charge that that much!

You'd need some pretty amazing games (and plenty of them) for that much money, only a company like Sony or Nintendo might have the marketing power to get away with that.

I'd buy a proper hand held computer or a lap top.

Part of the appeal of the GP2X is that it is cheap.
Thats the point, it would be fairly future proof, and have so much potential, it could draw in developers. I believe that most software currently availible has reached a bottleneck on the GP2x.
 
Last edited by a moderator:
I'm probably living proof that most developers don't want to code for platforms that won't have a decent user base who will enjoy their work. You just wouldn't move enough units at that price. And honestly, even for a high-end machine, there's no reason for it to get that steep.
 
At that price I would forget console and get this;
http://www.dynamism.com/sa1/main.shtml

With the great addition that you can use it standing up, just like a portable console. On the negative side GBA would look quite bad scaled to that screen. :) I'm thinking portable PSX, plus a few older pc titles like morrowind, or NWN...

Now what about a qualcomm Snapdragon. Okay its going to need Li-ion battery, but then, so is a 400mhz dual core proc. A 1GHz processor. Single core means its easier to develop for too...

http://www.cdmatech.com/products/snapdragon.jsp

Personally though I'm not too fussed on a next gen gp2x, and I'd like to see the HW utilised to the max before then. I mean come on, we've just had a 3d engine added to the GP2x repertoire, and that is still only in its infancy! It isn't much of a challenge to develop a wonderfully made system such as the psp. I begin to think PC's with there gazillion mhz have spoilt programmers. If you're doing it for a hobby don't you want a challenge, don't you want to support the underdog. I can only see that as the reason why someone would try to build and opengl engine, or try to make a psx emu for the GP2x... It's all to do with defeating adversity, and so one can respect rixed, zodttd and others that have produced crazy stuff. Yeah DJwillis for the scumm engine on the GP2x.

Anyway I suppose I'm done with that rant, well in conclusion, it wouldn't matter if GPH/GP produced a better console than Sony (apart from being unlikely), Sony would easily come back around an trump them. Not to mention that they'll always have more commercial games. The only real reason I suppose for sticking with the GP2x is philosophy. It also happens to be one of the reasons (I suspect) that open source works.
 
I'll be frank. If as a programmer I have to 'defeat adversity' as an excuse for the hardware being fucked up right from the factory, nay, the drawing board, my dollars are better spent somewhere else. I imagine this is not an uncommon sentiment.
 
The thing is, it's always going to be a hobbyist machine with a small user base. I can't see anyone forking out huge amounts for a homebrew system, no matter how good the SNES emulation is.
 
Again though, why would anyone buy one when they can get that on a, by then, much cheaper PSP? I get the feeling the type of system you're after is more likely to come from one of the 2 big companies. The smaller ones like GPH are better off sticking with cheap and cheerful devices.
 
I'm much more interested in what can be done with original software. That said, a design with an ARM-based processor should achieve much better battery life than the PSP.
 
Back
Top