Post- GTA02

Andy Green andy at openmoko.com
Fri Mar 21 11:58:18 CET 2008


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Somebody in the thread at some point said:

>>> If you are looking for something that will run on the CPU most of the

>> It would be optimal if the whole GSM side could run autonomously and the
>> CPU in fact unpowered / suspended during a call in the normal case (no
>> listening sockets on Wifi for example).
> 
> Yep, i see. But we have to drive the LCM anyway during call, no?

I don't know why we do that when it is just illuminating your earlobes.
 Maybe for users with cool earrings it is a feature otherwise we burn
maybe 200mA x 3.7V --> 750mW during the call for the CPU and LCM to do
nothing useful at all, that has to be making a noticable impact on talk
time from our 1150mA/h (or 1200 or 1250 I am no longer sure) / 3.7V -->
4255mW/h battery.

If I randomly guess we get 2hr talk time (truly a random estimate) then
we eat ~2W/h and suspending the CPU and LCM during calls would get us a
~30% increase in talk time according to these random numbers.

> Ok, clock for sth like echo cancellation will be higher than the minimum 12MHz

I think the 6400 (and probably any replacement from another vendor) has
a better arch for clock reduction with completely async PLLs for CPU and
other functions, plus they have individual dividers IIRC now.

Mark is definitely right about the acoustic echo cancellation being
tricky to get right, I used a Zarlink chip on a previous project and
this has all kinds of wonderful patented nonlinear stuff going on -- but
it was like magic in what it did.  But I guess it would be okay to have
a less awesome implementation in this case.

- -Andy
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org

iD8DBQFH45TKOjLpvpq7dMoRAq19AJ0Q4aWchOr3NU9BJL0t/lWYID9sywCglizz
gfaIdClojUnm5YPT/vk+pKk=
=5v2P
-----END PGP SIGNATURE-----




More information about the openmoko-kernel mailing list