Designing The Gp2x's Successor - Hardware Selection Poll

Select one:

  • Option 1: ARM926EJ-S 266 MHz Processor, Floating Point/Vector Coprocessor, 2D/3D Accelerator

    Votes: 0 0.0%
  • Option 2: ARM920T 400 MHz Processor, No FPU/Vector Unit or Graphics Accelerators

    Votes: 0 0.0%
  • Option 3: These are both inadequate (post alternative)

    Votes: 0 0.0%

  • Total voters
    0

The problem is, ARM11s are awesome in all sorts of ways. But, they're in such recent SoCs that for a small company like GPH or GamePark to produce a machine with one of these, especially in relatively small quantities, would probably make each unit cost several hundred dollars. Not too realistic.

As for Crusoes and Efficeons and VIA C3s .. integrating one into a handheld would require a lot of engineering, designing a full platform with independent motherboard design, video chipset, north/south bridges, and interface hardware of all sorts for the LCD, storage cards like SD, audio controller, and many other things. As it stands GP/GPH design very, VERY little hardware outside of the system-on-a-chip and I doubt they have the manpower or budget to design a full computer-in-a-handheld without an SoC doing it for them. Not to mention that the Crusoe/Transmeta/C3 route entails a processor that uses far, FAR more power than any ARM processor at similar clockrate, so battery life would suffer quite dramatically. Since those processors DO use heatsinks too (Just not a fan) the unit would probably get quite toasty in your hands and need to be bigger to accomodate a decent cooling apparatus.
 
Probably a lot more than the ARM11.. :) I don't really see a reason for hardware much more powerful than an ARM9 around 300 Mhz for gaming applications, though.

Let's take for example, the originally-planned specs for the XGP. The 2D/3D accelerator in the originally-selected MagicEyes VRender-3D chipset (believed to be the same one used in Samsung's S3C2460) ... it is able to render 1.0-1.5 Million polygons per second, and has comparable 2D capability. This places the graphical capabilities of a handheld using it several notches above the N64's graphical capabilities (most games did not utilize greater than 30% of the N64 GPU's capability for price-related reasons.) Its 200 MHz ARM9 processor, running integer-only or fixed-point code, is roughly on par with the Dreamcast SH-4 processor's integer capability. The comparatively immense quantity of RAM available (very fast RAM at that, MobileDDR) and support for very large games given maximum SD card size make it very clear that homebrew games have the potential to soar far and beyond anything available on the GP2X or GP32-- treated not like just a homebrew platform for emulation, but a console in its own right, such a machine would be a truly incredible gaming platform. The mistake that is maken all too often is relegating the GP2X's capabilities to be that of the systems it emulates-- which isn't fair in any sense given the enormous overhead emulation carries.

This is not to even mention the possibilities with the more powerful S3C2460. A 66 MHz faster processor, bringing it up to 266 MHz. A full floating point processor that is as powerful as the main one, and can perform vector processing-- not only will floating-point heavy code run at full speed, but it will run FASTER than an integer/fixed-point only program. A 16-bit DSP at 100 MHz will provide even more processing capability to developers who how know to effectively utilize it-- and due to the superior internal architecture compared to the GP2X's MMSP2's layout, it should be much easer to utilize than the ARM940T, due to lack of the memory contention, cache and bus accessibility issues that plague the 940T. Other improvements are that SD Direct Memory Access (DMA) should be fully functional (it is broken in the MMSP2, causing the CPU to stall whenever the SD card is read from, or written to) allowing realtime loading/saving of data without slowing the rest of the system; as well as that the hardware JPEG decoding support would allow large 3D textures to be decompressed into memory far faster than is possible with purely software decoding.

When you look at CPU performance rivalling high-end consoles and PDAs, 3D capabilities exceeding the N64 and probably matching those of the PSP, 2D capabilities to an absurd degree, and an incredible about of RAM and game storage space available, it's hard to see why anything beefier and more expensive is necessary-- a system with either of these SoCs, epsecially the S3C2460, would be an incredible platform for any developer or user, and possibly the greatest gift to the open-source gaming community since Linux... :)
 
because sony or microsoft will bring out faster handhelds.. 300 mhz isn't enough for me :)
 
Clock speed really doesn't mean anything on its own - a 300MHz ARM9 should be enough for you (assuming the system has dedicated hardware for graphics/sound), the DS has a 67MHz ARM9 and a 33MHz ARM7, and that copes fine.
 
Orkie posted on Oct 7 2006 at 10:40 AM said:
Clock speed really doesn't mean anything on its own - a 300MHz ARM9 should be enough for you (assuming the system has dedicated hardware for graphics/sound), the DS has a 67MHz ARM9 and a 33MHz ARM7, and that copes fine.
Bingo. It has a main CPU over 3 times slower than the GP2X's, but its 2D and 3D capabilities vastly exceed the GP2X's. :)
 
Last edited by a moderator:
Epicenter posted on Oct 7 2006 at 06:11 PM said:
Orkie posted on Oct 7 2006 at 10:40 AM said:
Clock speed really doesn't mean anything on its own - a 300MHz ARM9 should be enough for you (assuming the system has dedicated hardware for graphics/sound), the DS has a 67MHz ARM9 and a 33MHz ARM7, and that copes fine.
Bingo. It has a main CPU over 3 times slower than the GP2X's, but its 2D and 3D capabilities vastly exceed the GP2X's. :)


That is why the Genesis emu is so much faster with sound on the DS.. Oh wait it is unplayably slow on the DS ;)
 
Last edited by a moderator:
Um. :p Most of the work involved in emulating the MegaDrive/Genesis is emulation of the M68000 and the Z80 processors. The system's VDP is very simple and the YM2612, DAC and PSG aren't that stressful to emulate half-decently, either. The DS' superior 2D/3D capabilities accelerate rendering of the frames but can't help with the raw emulation. But run an original 2D (or 3D!) game on both systems that is extremely graphically intense but does not overwork the CPU with math and you will see a dramatic improvement on the DS over the GP2X.
 
DaveC posted on Oct 7 2006 at 07:16 PM said:
That is why the Genesis emu is so much faster with sound on the DS.. Oh wait it is unplayably slow on the DS ;)
Let's twist that around:
That is why the SNES emu is so much faster with sound on the GP2X.. Oh wait it is about the same with a fraction of the clock speed and is improving all the time ;)

I never said anything about emulation (and despite what you seem to think, the DS is actually quite good for emulation of Nintendo's other consoles in particular), I was merely pointing out that two different systems clocked at 300MHz will perform differently.
 
Last edited by a moderator:
Orkie posted on Oct 7 2006 at 01:23 PM said:
Let's twist that around:
That is why the SNES emu is so much faster with sound on the GP2X.. Oh wait it is about the same with a fraction of the clock speed and is improving all the time ;)
Indeed, most of the difficulty in emulating the SNES is intensity of handling all of its graphical functions, not emulating its CPU or other hardware.
 
Last edited by a moderator:
The big problem with the DS for emulation is the poor screen resolution. It is not really adequate for anything above speccy or ColecoVision.
 
Haha, this is great, everyone's riding their own horse on their own road...

- Alex
 
stupid simple AND simple best

best sw AND best hw

mass production ;)


sony killed the nintendo star,
microsoft will kill the sony star,
GP will kill the microsoft star
 
__g_blade posted on Oct 9 2006 at 03:54 AM said:
stupid simple AND simple best

best sw AND best hw

mass production ;)

sony killed the nintendo star,
microsoft will kill the sony star,
GP will kill the microsoft star
This post is pretty baffling. "best sw and best hw"? As for 'simple is best', well it depends. For some things raw CPU horsepower is more important. Like MD, Sega CD emulation, some higher-end arcade games , more CPU horsepower is the most important bit. But for most applications, 266-300 MHz-ish is plenty and graphics acceleration will do more good (high-end 2D homebrew/commercial titles-- the GP2X chokes on very heavy 2D work with large sprites .. PSX, SNES emulation, Amiga emulation, any sort of polygonal 3D game [quake, quake 2 especially].)

"best sw and best hw", no idea what you mean.

As for the Nintendo, Sony, MS, GP bit .. "Sony killed the Nintendo star"? I assume you mean PSP vs. DS. The DS is outselling the PSP, especially in Japan, and Nintendo makes a profit on the DS. Sony loses money making each PSP unit. MS, well, we don't know that they are producing anything. At least not yet. I'll agree with the last bit for enjoyment's sake. A well-built high end handheld from GP would blow anything else on the market out of the water for me.
 
Last edited by a moderator:
think bigger

playstation killed (s)nes
xbox(360) killing playstation(2/3)
gp(3x with turbo-hardware) would kill m$

for example... games like www.planeshift.it.. need cpu power + 3d
 
Back
Top