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

Lally Singh lally.singh at gmail.com
Wed Jun 11 17:32:21 CEST 2008

On Wed, Jun 11, 2008 at 9:04 AM, Ken Young <rtm at cfa.harvard.edu> wrote:
> On Wed, Jun 11, 2008 at 07:57:13AM +0800, Carsten Haitzler wrote:
>> the day the design mockups for the ui i see stop having alpha
>> transparency is the day i make this unimportant. until that day,
>> your "i don't care about this" is the kind of opinion that i also am
>> not interested in, because i am being shown ui designs hat REQUIRE
>> it in the long run between windows, and in the short term is being
>> faked with software within windows. i am just trying to make
>> something possible that is being requested, and has been for a long
>> time.  not just say "i don't care".
> The problem isn't that transparency effects, and other CPU/GPU intensitve
> UI enhancements, are unimportant.   On a handheld device they *are*
> important.    They make the device worse.    It is important to resist
> the push to add eye-candy to a handheld device, because every CPU/GPU
> cycle spent animating an icon, or making a window translucent, eats
> some of the energy stored in your battery, and reduces the amount of
> useful work which can be done between recharges.

I disagree with such categorical statements.  There is a trade-off
between usability and performance (e.g. user performance and device
performance).  The optimal value is in between, dependent on both user
and system capabilities.  The iPhone is success *because* of its heavy
bias for user performance over system performance.  The hardware isn't
novel, but the UI is, and it makes all the difference.

Example: Shadows on windows on Mac OS X --- the shadows indicate,
better than any titlebar hilight ever will, what window has focus.
Using the brain's innate understanding of depth provides user-side
hardware acceleration for this activity.

Example: Desktop switcher animation --- when switching virtual
desktops, having the windows slide off to the appropriate side is
critical for building a spatial model in the user's mind.

Geeks will probably want a different set of trade-offs between
usability & performance, but those are best done as customizations on
an expert platform.  One that we hope that OM will become.

H. Lally Singh
Ph.D. Candidate, Computer Science
Virginia Tech

More information about the community mailing list