Could someone give me a *crash-course* in the handheld-3D industry?

[asicnewbie]

From lurking, I've learned the following about the handheld-3D gaming market. I'm about to sound like a 5-year student old giving his/her first book report, so bear with me...

1) Cellphone/handsets (with "gaming functions")
2) Dedicated handheld gaming devices (PSP, Nintendo DS, GBA)
3) 'bigger' things (PDAs, smartphones? tablets?)

1) Gaming on Cellphones : "Biggest potential" (because of wordwide cellphone use)
But problems with ergonomics - screen-size, orientation, input-keys are POORLY suited for freeform gaming

2) dedicated gaming devices
I didn't think much about these, since I'm not interested..

3) PDAs, tablets, smartphones
Ok, I'm lost. What differentiates a PDA from a smartphone? Is there a core featureset or industry-standard convention to classify these things, or is it just up to product-marketing to position (multi-purpose) devices for distinct segments?

---
My questions are aimed at the SUPPLY-SIDE of the industry. Looking through the posts here, Imgtec's name seems to come up a lot -- do they have the most marketshare (in terms of design-wins in shipping 3D-capable mobile chipsets?) I know Imgtec doesn't sell chips, but they license IP to semicondcutor suppliers (TI, Intel, NEC, Reneases, etc.)

In looking at marketing literature for ATI and NVidia 's handheld (not PC/laptop) graphics, they often mention support for SVGA and higher resolutions. I'm confused here -- cellphone LCD-displays are still stuck at QVGA (320x240) -- do ATI/NVidia actually expect cellphone-products to operate at VGA+ resolution?

And finally, besides TI OMAP, who are the other major players in the handset/3D market?

 

[Xmas]

3) PDAs, tablets, smartphones
Ok, I'm lost. What differentiates a PDA from a smartphone? Is there a core featureset or industry-standard convention to classify these things, or is it just up to product-marketing to position (multi-purpose) devices for distinct segments?

Well, a smartphone is a phone
There is Microsoft's Pocket PC standard, but that's just one PDA platform. I don't think there is a globally accepted strict definition of what's a PDA and what isn't, just a certain set of common expectations.

My questions are aimed at the SUPPLY-SIDE of the industry. Looking through the posts here, Imgtec's name seems to come up a lot -- do they have the most marketshare (in terms of design-wins in shipping 3D-capable mobile chipsets?) I know Imgtec doesn't sell chips, but they license IP to semicondcutor suppliers (TI, Intel, NEC, Reneases, etc.)

Looking at the design wins I know I'd say most certainly yes, but I'm not without bias.

In looking at marketing literature for ATI and NVidia 's handheld (not PC/laptop) graphics, they often mention support for SVGA and higher resolutions. I'm confused here -- cellphone LCD-displays are still stuck at QVGA (320x240) -- do ATI/NVidia actually expect cellphone-products to operate at VGA+ resolution?

There are some cellphones with higher than QVGA resolution. Nokia N90 has 352x416, and Samsung announced a 2" VGA panel. I think it's inevitable that resolution will increase further.
But these graphics solutions are not limited to cellphones. The MBX Lite based Intel 2700G drives a WVGA (800x480) panel in the PepperPad, and a Dell Axim PDA using the same chip can display up to SXGA to an external monitor if you attach a VGA output cable. Car information systems use higher resolutions as well.

As long as the framebuffer doesn't have to fit into a small chunk of embedded memory, there's not much reason to limit the output resolution to QVGA anyway. The display controller is not the most complex part, and the 3D core should be able to render to higher resolutions for render-to-texture anyway. In contrast, allowing higher resolutions opens up more market segments to sell to.

 

[Lazy8s]

Telephony is the application which drives most computer sales, so the growing cellphone market is the largest target at nearly a billion units per year. 3D has reached just a small percentage of those so far at the high-end in smart phones and feature phones, but even basic phones will move on to the next generation of system-on-chips, many of which have integrated 3D, within a few years.

The other sectors of the portable 3D market consist of dedicated devices like game consoles, steady at about fifty million units per year, and also less converged devices like PDAs, whose share has mostly disappeared as almost all portable devices become phones of some sort.

The 3D chips mostly are integrated solutions on a SoC with the CPU, DSPs, and sometimes a baseband network processor, so the major companies are large semiconductor producers with their application processors. Most of them, with a few exceptions like Toshiba, don't develop their graphics in-house and go out to license the IP from a specialist like PowerVR, ATi, or Falanx.

Separate 3D co-processor chips are offered by the major graphics companies, nVidia, ATi, and Intel (Intel licenses their IP from PowerVR.)

A good resource for information on the devices, APIs, and platforms can be found at the San Diego Supercomputer Center Mobile Data Visualization's listing:
http://mobile.sdsc.edu/devices.html

PowerVR based solutions are listed at Wikipedia:
http://en.wikipedia.org/wiki/Powervr

Qualcomm, now licensing ATi's Imageon IP, is a major 3D SoC supplier:
http://www.qualcomm.com/technology/index.html

 

[asicnewbie]

Thanks, guys, for the quick responses!

Separate 3D co-processor chips are offered by the major graphics companies, nVidia, ATi, and Intel (Intel licenses their IP from PowerVR.)

That's interesting how handheld devices mimic laptops and PCs, with "integrated" vs. "discrete" graphics. Do you guys think this situation is a momentary blip (with market evolution and consolidation eventually eliminating the discrete graphics chips), or will two-chip solutions co-exist with single-chip solutions, long-term?

I also wonder whether NVidia and ATI will eventually enter the application-processor market.

license the IP from a specialist like PowerVR, ATi, or Falanx.

So for the handheld market, ATI both licenses its IP and sells graphics processors?

Speaking of platforms, besides Nokia's ill-fated 'N-Gage', has the cellphone/handset industry pushed forward any semblance of a stable 3D-platform?

 

[Rys]

So for the handheld market, ATI both licenses its IP and sells graphics processors?

I don't think ATI licenses Imageon out, but I could be wrong (especially now they have Bitboys).

 

[Lazy8s]

Integrated 3D, with its typically lower cost, is likely to dominate because: battery life concerns should prevent graphics solutions from escalating to the scale where only discrete chips would be competitve in performance; integrating the graphics core nearer to the rest of the cores has its own technological advantages; cellphones are closed, embedded platforms and therefore leave no market for the consumer to choose what kind of chips they would want to use to make upgrades; and the decision of whether to choose integrated 3D versus a discrete chip in the first place is done by the cellphone designer and not the consumer.

Discrete 3D may not fully disappear anytime soon, though, as long as influential companies like nVidia and ATi with attractive software/hardware technology platforms are still pushing it. That could be a possibility especially in the event that their IP didn't get licensed enough to SoC makers and they consequently were compelled to sell discrete chips directly to the cellphone designers. nVidia has already succeeded in getting GoForce 3D adopted into an impressive number of phones this way.

Producing their own SoC application processors would put nVidia and ATi in competition with the likes of Texas Instruments, Samsung, and Intel. They might be able to do it for the portable market.

ATi does offer both discrete graphics chips as well as IP for their Imageon technology, apparently. Qualcomm lists that their third generation 3D MSM7200, MSM7500, and MSM7600 chipsets will use a high performance ATi technology, midway through their Mobile 3D Gaming Brochure:
http://www.cdmatech.com/download_library/pdf/3dgaming_brochure.pdf

The N-Gage didn't sell well because it was just one, distinct model of Nokia cellphone, but they're bringing the gaming platform back as a new, high-performance 3D standard built in to their traditional line of smartphones, starting with the N93.

 

[asicnewbie]

Integrated 3D, with its typically lower cost, is likely to dominate because: battery life concerns should prevent graphics solutions from escalating to the scale where only discrete chips would be competitve in performance;...

Thanks again. I'm really a dinosaur when it comes to the subject of cellphones and mobile-gaming. I guess I'm shocked a cellphone/handset can cram in a discrete GPU, yet still retain usable battery-life. (Granted, the GPU is 'off' most of the time.)

Producing their own SoC application processors would put nVidia and ATi in competition with the likes of Texas Instruments, Samsung, and Intel. They might be able to do it for the portable market.

I wonder how that would play out? It would ATI/NVidia would need to acquire an embedded processor core (ARM?), and probably much more, to produce a competitive application-processor.

The N-Gage didn't sell well because it was just one, distinct model of Nokia cellphone,

Thanks for the correction.

 

[Lazy8s]

Presumably, nVidia and ATi won't pursue development of their own application processors next generation if they can gain some decent prospects for selling just graphics technology, either as IP or chips.

ATi is already focused on licensing graphics IP more successfully next time. Also, the possibilities that nVidia and ATi could win the significant graphics design contracts for the next generation portable video game systems from Sony and Nintendo, respectively, are reasonable speculation, as are the possibilities that those contributions would be in the form of IP to be integrated onto a SoC due to the cost-sensitive nature of the console market.