>The one thing I see missing is if you run it on a PIE and send it to
>another device over wifi then you eliminate any chance of thunder
>?crashing a whole system and I mean everything in your home! ?Of course
>I live on the beach and we do have lot of storms! ?I just remember when
>I had a bolt land 200ft from the home and it still nocked everything
>and started from scratch again. ?since that time I do not have outdoor
Good point! Did not think about that yet. That is a good reason for having a wireless (e.g. wifi) link, right?
>Wifi is going to eat your battery. The CPU does mostly nothing and will
>eat <1/10 of Wifi.
>Also, if you want to save energy, you should use one of the
>Onewire host adapters, e.g. the DS2483. Bitbanging wakes up the CPU on
>every bit sent and this will eat a lot of energy. (But still *nothing*
>compared to Wifi.)
Thanks for that info! That is quite dramatic. This actually makes me think even more I should try to reduce the wifi links as my intention was. And I understand your arguments regarding wifi now.
The point about bitbanging and DS2483 is important too.
>With only one core, there is only one process running. All the other
>processes are waiting in the scheduler queue. What effects energy
>consumption of the CPU in a non-number-crunching application is the
>number of interrupts to service. That is because each interrupt means
>waking up the CPU from sleep. All other effects are neglible.
>You don't have to count processes, you have to count interrupts. Wifi
>produces a lot. Timer produces an awful lot when the CPU has to bitbang
Of course interrupts... :) Thanks for reminding me to that - quite some time since I learned about this. Is a good point!
So do I get this right that wifi also takes a lot of power because of transmitted power? Since you mention bitbanging to throw more ints than wifi, but wifi taking more power, correct?