Re: [jornada] Jornada 72x PM current status?

Michael Gernoth wrote:

>On Tue, May 17, 2005 at 09:59:54PM +0100, Colin Sauze wrote:
>
>
>>if/when we get 2.6 kernels working we should be able to throttle the CPU
>>speed back which should help.
>>
>>
>
>After looking at the cpufreq code in kernel 2.6, I believe
>clock-throttling is currently unsupported on the Jornada 720.
>There are no sdram timing-values for the j720 in the 2.6 cpufreq-driver,
>which automatically disables it on this machine.
>
>I just hacked throttling to 59mhz when the fake "suspend" mode
>is entered into my 2.4.30 port and added timing values for
>the samsung sdrams at least some j720s use. This works fine on
>my machine, but I can't tell if it will work on a 728...
>
>The new kernel linux-2.4.30-vrs1-pxa1-j720-2 should be on my website
>in a few minutes and you should see something like the following in
>dmesg, when pressing the power-button:
>
>Suspend!
>Display OFF
>Old CPU frequency: 206400, new CPU frequency: 59000
>
>and "resume" again:
>Suspend!
>Old CPU frequency: 59000, new CPU frequency: 206400
>Display ON
>
>

Any ideas how much battery improvement this gives?
This can be measured simply by doing something like the following shell
script:

modprobe j720_control

while [ "0" = "0" ] ; do
    grep "battery charge:" /proc/jornada/battery | awk '{print $3}'
    date
    echo
    sleep 60
done

Or If you want something you can make pretty graphs from:

modprobe j720_control

#get time in seconds since jan 1st 1970
start_time=`date +%s`

while [ "0" = "0" ] ; do
    battery_level=`grep "battery charge:" /proc/jornada/battery | awk
'{print $3}'`

    #get the difference between the start and current times
    curr_time=`date +%s`
    diff_time=`expr $curr_time - $start_time`
    echo $diff_time $battery_level >> gnuplot.data

    echo
    sleep 10
done

to plot in gnuplot do:

type gnuplot and press enter,

then type:

plot "gnuplot.data" using 1:2 with lines

And you should get a nice graph showing how the battery level went.

>
>
>>I think some advantage might also be
>>gained by turning off PCMCIA/CF cards and halting processes.
>>
>>
>
>I'm working on it, but the kernel doesn't like when it's root device
>is suddenly switched off. Need to find a solution for that first.
>
>
>
Isn't that what happens when you suspend on a desktop/laptop PC? Could
always settle for just turning off devices other than the root cf card.
Received on Wed May 18 2005 - 07:51:41 EDT