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

[Carsten Haitzler (The Rasterman)]

On Wed, 11 Jun 2008 09:48:10 +0200 "Dotan Cohen" <dotancohen at gmail.com> babbled:

> 2008/6/11 The Rasterman Carsten Haitzler <raster at openmoko.org>:
> > if the vocal group here are to be accounted for going to these screens
> > would be utterly bad and we should accept nothing less than vga so wvga
> > (800x480) is the only way up into the future.
> >
> 
> So you argue to put a lesser screen on the device now, so that you can
> have a lesser upgrade in the future? That is not the type of company
> that I want to deal with. Give us what is possible now, now.

i didn't say that. you have to remember not everything is possible. we make
compromises all the time. but you are saving that the only options we have for
you to be happy is vga or higher. no matter what. even if we make a smaller
phone? some people want big phones, some want small. we might have multiple
products and you choose the one that is best for you. some may go for the
smaller device that weight less, uses less pocket space and has a low-res small
screen, some may go for one that is the size of an n800 with a massive high-res
screen. these phones are not necessarily a linear progression from 1st to 2nd
to 3rd version. some may be, some may not be.

> > ouch. poor cpu.
> >
> 
> Other than price, what are the constraints about using a more powerful
> CPU? And what is the price hit? If we are talking about having four
> times the screen resolution for an additional 30% of the price of the
> device, then I say that is worth it. The target market for this phone
> will buy it almost regardless of price. That is not to say that price
> is not important (it is) but it should be more flexible than the
> hardware.

right now - constraints vary from soc's we can actually buy in volume to being
open (eg nvidia have an soc... do we want that along with the closed graphics
system?), to just development time. a new soc means a whole lot of kernel work
and driver work normally - unless we stick to the kernel provided by the
vendor, and that means you have a kernel behind many versions. we want to bring
things to market as soon as we can. i'd love to see us improve our soc, but
that just takes time and development. that's no promises one way or another or
indications on whats going on - it's just where things are right now. :)

-- 
Carsten Haitzler (The Rasterman) <raster at openmoko.org>