It only says 10.8 v ~ 2400 mAh

That makes it a 26W/hr battery.
More than I thought.

Thanks Ctitanic.

Steve.

Ctitanic

Two more test figures needed!!!

Whats the average CPU util when watching the DIVx?
Whats the bitrate of the Divx?

Regards
Steve.

Ctitanic.

When you did the C3 state testing, was the device idle? Or were you watching the film?

Regards
Steve.

Quote:

Device was idle.

...and that was without the USB stick in right?
Steve.

Movie properties (AVi)


cpu ut. with windvd


cpu ut. using vlc player

Quote:

no usb during the C3 test. I plugged a few times and unplugged to see if any reaction there, but no luck.

The CPU utilization charts were taken with the movie playing from the USB stick.

Just in case. I retested the CPU utilization playing the movie from my HDD and got the same results.

Thanks Ctitanic.

Thats good figures. Thanks for testing.
This is getting compicated but I think i'm on top of it!

Main open question is: Why is the CPU not going into C3 power-saving states during idle (note however that the C7-M has other sleep states than C3. You should see if you can test for C1 and C2 states. ) You have proved that P-states (voltage and clockrate reduction) are working already.

75% utilisation for a 1mbps DIVX on the CPU is very high but the VLC figures show 50% which is much more reasonable.

The VN800 unichrome pro GPU can not help with MPEG4 decoding unfortunately so all the processing is done by the CPU. Essentialy, watching a high bitrate Divx is almost a full-power operation. (The wifi + BT probably wont effect battery power too much if you turn them on. They are quite low-power drain devices.) and my video duration prediction is wrong when you consider high-bitrate mpeg-4 (or other advanced video codecs - AVC's)

I expected 18watts full-power drain from a UMPC (and a bit less from a VIA) and what you're getting there is something like 18w drain from an estimated 90% power-utilisation. (The other 10% being wifi, full lcd brightness and bt.) which is slightly more than predicted.
My guess is that if you turn everything else on and put the brightness up, you will get 1:15 useage - about 20Watts max power drain. Not far off my prediction.


The really interesting test would be to play an MPEG2 video and to check the CPU utilisation. I guess you'll see much much less utilisation and if the MPEG is being decoded completely by the CPU, you might even see the CPU going into other C states and starting to power-save. I don't know what causes a P-state change but if we're very very lucky, the voltage and clockrate on the CPU could drop and you'll see some big changes in lifetime.

However, having said all that, I wouldn;t like to have to convert my Divx's into MPEG2 films just to watch them on a UMPC.
One answer would be to bounce the Divx through VLC on a server, use the server to convert on-the-fly to MPEG2 and then stream that over wifi to the UMPC. A great test if you're bored!!!!! Maybe the sling-box can stream in MPEG-2?

I'll journal all this over at Carrypad later but would be interested to see if you can run an MPEG2 file through the Eo before I do that. Please say so if you want to stop testing. I don't want to take all your time becuase of my power-saving madness!!

Regards and thanks again for all the testing.

Steve.

Playing mpg the CPU stays at the same level in vcl and it went down to 55% in Windvd. Vcl, still use less CPU. I'll be back in about 4 hours vi case you need any other test.

Quote:

You said you played an MPG.
What was the coded video inside it though? The audio also needs to be determined.
A good source of MPEG2 is a VOB file from a DVD. I think WinDVD can play them. Van also mentioned in his tests that you have to 'enable' the optimisations in WinDVD. There must be a setting. Did you check that? (I don't think its going to make much diff. with a DIvx though as I mentioned before.)

However, look at this:
I went over Vans C7 testing again and something is definately wrong.
You are seeing at least 17watts power drain. (1.5 hours from a 26watt/hr battery)
In Vans tests, using his OpenSourceMark maxpowerdraw he saw only 15watts drain from a 1.5Ghz C7-M ULV on a laptop with a 12" screen. (3.2 hours from a 49watt/hr battery) There's no way an Eo should be drawing more power than a bigger laptop. I'm getting very supicious of your battery at this stage!! How do we test the battery capacity?!!

Steve.

It's not just my battery. I have checked my result with others and they are getting the same 1.5 hours figure.

Ok, played a VOB file. CPU around 55%. WindVd set to use hardware acceleration and using as much memory as it can use.

Check the link posted a few messages ago and you will see that the problem seems to me is the C3 time.