Power Consumption

Here's a video showing my ammeter in series with the Pandora battery. I wanted to double check the fuel gauge calibration because with lite settings, 15 hours is being presented. I thought that was crazy, but it's true. With an average current of 278mA and a battery testing at 4200mAh, that really is 15 hours (under specific settings talked about below of course).

Testing was done at 500MHz and the screen brightness set just about as low as it will go before shutting off. Increasing the screen brightness to 100% upped the average current to 398mA which is only 10.5 hours of continuous use.

In the second half, I upped the clock to 750MHz using the little menu option but I'm not certain if it actually did anything. Some stuff seems to work better for other people than me but this little program is one that seems to bug out on me.

This video also shows the power consumption with wifi running. You can see the power go way up when web pages are loaded or when new cover art is loaded in Mplayer. If you load pages and then spend a long time reading them before loading the next one, perhaps 10 hours is still possible. At the max I saw of approx. 600mA, that would be 7 hours and you'd have to be hammering wireless hard to maintain that as you can see. Notaz did a fantastic job of getting the power management for wifi working and it will really help battery life for normal use. I guess it's like any other device you would buy now, but for the developers it's such a thrill to see it working. :)

Category:

Gaming

Tags:

• pandora
• openpandora
• wifi
• gaming
• handheld

[JayFoxRox]

Any tests with some SGX demo running? I suspect that will be used in so many games in the near future that a test of that would be more useful than just browsing the internet - who does that on the go all the time anyway? You'd have extra power draw from the usb gadget which allows you to connect whatever service you use.

[MWeston2]

@JayFoxRox Quake 2 - Lowest LCD brightness - 410mA peak, Maximum LCD brightness - 530mA peak

Quake 3 - Lowest LCD brightness - 370mA peak, Maximum LCD brightness - 500mA peak

[ArchNME]

FAKE!!!!!!!! Why don't you show the whole 15 hours???

[Sammi84]

@MWeston2 Q3 really draws less electricity?! Interesting...

[oddyoh]

Cool, Our Lady Peace...great album. Oh yeah, Pandora too!

[Jhim2000]

Floating point math is a big deal to me. I will use fixed point on the A8, because I have to. I'd rather not though.

NEON seems like it has uses - complex geometry for one. But the overheads involved make it redundant in terms of pushing pixels. I *really* hope to see some libs that make use of it's advantages. A 32 bit data bus? That is no more effective than fixed point done on the CPU itself.

Be good to off-load some non-critical calcs though.

More interested in the DSP to be honest.

[JayFoxRox]

@Jhim2000 The NEON should be part of the Cortex-A8 package - it shouldn't matter too much wether its used or not (actually it could speed up calculations and leave the CPU longer in idle/power-saving state). The DSP and the SGX on the other hand were added by TI and draw additional current. The DSP won't be used much if the compiler / doc situation stays like this. The DSP can also be used to unstress the CPU (media). Games often require SGX for rendering AND CPU for precalc, logic, physics etc.

[JayFoxRox]

@Jhim2000 I'm not "offended". Just wanted to clarify it because others might not understand my post neither ;)

I'm just one large fail when it comes to communication.

[Jhim2000]

@JayFoxRox

Sry Jay. Didn't realise you would have been offended :)

JhimBhoy out

[JayFoxRox]

@Jhim2000 I'm not interested in how the SGX demos run - I have seen them before on beagleboard and other devices. I'm interested in how much power that will draw.

I should have been more specific from the beginning