QVGA V/s VGA for GTA03 (was something about yummy CPU-GPU combos!)

[Dr. H. Nikolaus Schaller]

Am 06.06.2008 um 09:45 schrieb Carsten Haitzler (The Rasterman):

> On Fri, 6 Jun 2008 09:09:55 +0200 "Dr. H. Nikolaus Schaller" <hns at computer.org 
> >
> babbled:
>
>> IMHO, reducing any Openmoko device to QVGA will be the end of the
>> Openmoko project. And I have business reasons to wish that OM
>> survives...
>
> why would it be the end? the majority of phones in the non-japan  
> world are
> qvga. it'd be no worse than most things out there.

Yes, the majority is. And has been brought out last year. So, is this  
the scale for the
2009 market?

There was some shift with high end PDA displays. In approx. 2006 most  
QVGA
models were dropped and only VGA remained. E.g. Acer n30 -> n310.

Another point from my discussions at LinuxTag: people want to have a  
successor for the Sharp Zaurus.
The C750 introduced VGA in 2005 (or so). So, we don't get anyone who  
wanted to have a Zaurus successor!

>> There are plenty of QVGA designs out there and I even have an offer
>> and a sample of a Linux smartphone at half the price of the OM. It
>> works, has 90g, Quadband, and an integrated stylus. But QVGA. This
>> makes the difference.

This is my core argument: I can buy a "GTA03 / QVGA" today. I don't  
have to wait for OM to develop such
a reduced device. Therefore I want OM to stay with VGA because it is  
leading and unique...

>> I have shown a QVGA Acer n30 and the VGA Neo on FOSDEM and LinuxTag  
>> to
>> the public. Everybody considered the VGA as better although the Acer
>> is approx. 30% faster (because it has less rendering to do).
>
> for us qvga would be give or take 3 times faster. thats a massie  
> difference to
> 30%.


The Acer n30 has a 266 MHz Samsung S3C2410, QVGA
The Neo1973 has a S3C2410AL-26 with 266 MHz, VGA

Rendering a quarter of pixels appears to be not all the tasks the  
processor has to do.

I have not done a FPS comparison between both.

>> So, where would be the uniqueness of OM devices? The average mass
>> market customer doesn't care about full openness.
>
> the problem is - if you have a nice screen but the engine to power  
> it is
> underpowered, you will suffer from complaints of it just being slow  
> then
> instead.

Yes, slowness is the most critical complaint. But how should I  
convince anyone
to buy an OM if I also have a QVGA device which feels fast (based on  
OMAP 730
and some Qt 2.x)?

>> Please consider basic rules of marketing. And, consider purchasing a
>> Device Feature Roadmap from Strategy Analytics (they are really good)
>> and ask what the percentage prediction for VGA vs. QVGA is for 2009.
>> They are good in predictions because they look into the roadmaps of
>> the component manufacturers.
>>
>> http://www.strategyanalytics.com/default.aspx?mod=NavigationHeader&a0=42
>
> don't need them. just go to tokyo and look what's there on sale now.  
> what's
> there will be what the rest of the world gets to doing 2 years later.

Not necessarily. The iPhone wasn't there 3 years ago. And i-mode never  
became
popular. But I agree that you can see some technology there already  
built into
devices.

> 834x480 at 3.2" is now the top of the line display. but again - you  
> need/want the
> graphics back end to fill that many pixels smoothly. right now we  
> are pushing
> that at best. :)

How is the iPhone doing it? It has Half-VGA resolution and "feels"  
very smooth.

> but even so - in the ultra-high-tech of japan's mobile phone telco's  
> qvga is
> still VERY COMMON. if we want to play the "my specs are better than  
> your specs"
> game right now, we will lose. we do not have the sourcing power of the

Unfortuntely, I have to play that game if I want to sell OM devices...

> competing vendors. we often can't even get the better components at  
> all, let
> alone for a good price. we get what we can get. sure - we have a vga  
> screen,
> but really, how good is that if the rest of the device can't  
> smoothly handle
> the screen? it isn't just dpi. if all you measure a device on is dpi  
> and pixel

There is one strategic aspect to consider: If you keep the same  
display model and vendor,
it saves engineering time and you can drive down cost by getting
more and more volume from a single vendor. This may outweight any  
better price
of a different vendor.

> count, you are being silly. how it looks matters even more. dpi  
> helps there,
> but so does compositing, translucency, smooth animation etc. in fact  
> these
> probably have a much greater "buy me" effect. by far more. i'll put  
> money on
> that bet actually (this is just speaking from having done eyecandy  
> for over a
> decade - on linux, and having seen what it can do to attract  
> people). to make

Yes. But this is in some conflict with providing an open platform that  
others
can adapt to their user's needs. If it has QVGA it rules out many  
commercial
projects where eye-candy and video speed is not that important but high
information density.

> things like compositing fast, smooth and nice, you must lower  
> resolution to do
> it, or increase graphics power grunt. so given that graphicws grunt  
> is not
> changing, cpu is not, the only 2 things that can change are screen  
> resolution
> or the "eyecandy" has to remain toned down. so does vga buy you more  
> sales for
> the average joe than a sexy bit of eyecandy at qvga? i'm leaning to  
> qvga +
> eyecandy myself.
>
>> Tracking 30+ enabling technologies in the mobile devices market and
>> provides analytical views on the major issues likely to drive or
>> hinder diffusion of these technologies into the global wireless
>> devices markets. The WDS service was first to market with a quarterly
>> camera phone analysis and market share, and has extended that
>> leadership position on technologies including WLAN, digital TV, CMOS
>> and CCD cameras, Operating Systems and software, removable storage
>> media, GPS, and other wireless connectivity technologies like Zigbee,
>> NFC, and USB.
>> Device Feature Roadmaps and evolution across key global markets.
>> Research in this area includes analysis of feature phone evolution
>> with forecasts by device type; device vendor share by device type;
>> device segmentation analysis by ASP tier, with forecasts; extended
>> analysis of ultra-low handset diffusion drivers and forecasts; and  
>> in-
>> depth analysis of key device types, i.e. TV enabled handsets and MP3/
>> music enabled devices.
>>
>> http://www.strategyanalytics.com/default.aspx?mod=ReportAbstractViewer&a0=3736
>> http://www.strategyanalytics.com/default.aspx?mod=ReportAbstractViewer&a0=3899
>>
>> Nikolaus
>>
>>
>>
>> Am 05.06.2008 um 14:32 schrieb rakshat hooja:
>>
>>>
>>> quick question - would you prefer a qvga lcd (save a bit of cost)
>>> since we'e
>>> going to need to software-drive all graphics - the fewer pixels you
>>> have to
>>> fill, the better for speed. i'm really tossing up if the speed of
>>> qvga is worth
>>> the loss of resolution. i'm just not sure.
>>>
>>>
>>>
>>> I have a Sharp 903 with qvga, 2.4 inch, Nokia N95 with qvga, 2.4
>>> inch and the Neo 1973 with VGA, 2.8 inch. By far the best screen for
>>> reading is the Sharp one. On closer examination you can see pixels
>>> which you dont on the Neo but the display just feels better, crisper
>>> and better on the eyes. When you view higher res photos the Neo
>>> display seems better but not by much. The N95 is also good when you
>>> look at it on its own and one has no problems reading anything but
>>> when kept next to the Neo 1973 and Sharp 903 one can tell the
>>> display is not in the same league. (the Sharp is also visible in the
>>> sun though I dont think its trans-reflective)
>>>
>>> The point I am trying to make is that the quality of LCM being used
>>> matters as much as qvga or vga. Qvga is sufficient for almost all
>>> needs on a mobile phone size device and would be great if it
>>> provides cost and speed improvements. But it has to be a really good
>>> quality QVGA.
>>>
>>> Personally I love the resolution and form factor of the PSP Slim LCM
>>> and would love to see something similar on GTA0X.
>>>
>>> Rakshat
>>> _______________________________________________
>>> Openmoko community mailing list
>>> community at lists.openmoko.org
>>> http://lists.openmoko.org/mailman/listinfo/community
>>
>>
>
>
> -- 
> Carsten Haitzler (The Rasterman) <raster at openmoko.org>