From: Daniel Stenberg <daniel_at_haxx.se>
Date: Fri, 13 Dec 2002 08:48:07 +0100 (MET)

On Fri, 13 Dec 2002, George Michaelson wrote:

> Extrapolating from this, I infer that the in-memory cost of games and demos
> has to be marginal as well, ie if one imagined building a version of the
> code without them, the increased memory would give <5% extended battery
> life if the memory saved was measured as 1Mb (unlikely?). for a 10hr
> battery life, thats 30min extra play. For me, thats on the cusp of being
> worth it. But since I expect the SH3 codesize is dense, and so the saved
> memory will be smaller, the net increase in battery life begins to look
> marginal (this is only theory: until I make my build env with a gcc-sh3, I
> can't say what the cost is)

It is an SH1, SH7034. And I'd guess that the size of all games together is
more like in the area 30-50 KB or so. It would be easy to check.

> What other things tend to improving battery life? The backlight costs and
> LCD update also look to be marginal savings at this time.

Not much more can be done, AFAIK. We can verify that we're using the best
possible ATA code that keeps the power consuption at its lowest. We can work
on using the CPU SLEEP mode to reduce power even more, but it requires some
neat hacks.

Björn and I were talking the other day about adding a mode where we shut down
all external circuits and go to SLEEP. That would bring us down to almost
40mA power consumption and thus we could get a "wake up" function that could
survive ~40 hours on 1600mAh batteries.

It could for example be used as an alarm clock, or to make it survive a night
in the car until you start it again next morning and charges your Archos
again. Etc.

Of course we have no code for this at this point. We were only dreaming of
new and better worlds! ;-)

-- 
Daniel "Bagder" Stenberg -- http://rockbox.haxx.se/

Received on 2002-12-13