Underclocking? How much does it really extend battery life?

[Paul Robinson]

I've run a number of tests now using PxaClocker Lite to underclock my T3. I used Battery Graph to track the time and voltage and to estimate the total run time that could be expected. In most cases, I've run the T3 for an hour and a half to two hours and then estimated what total run time would be. I'm generally reading ebooks. I've used Docs to Go to enter the data and estimate the total time. I do not assume that the starting voltage is 4240, but rather enter the actual initial voltage.

The results have been underwhelming, to say the least.

Here is a series of estimates, with the primary activity being ebook reading:

without Clocker running:

3.2-3.7, 3.3, 3.2-3.6, 3.0

with Clocker running my ebook programs at 100 (mem and CPU), and in some cases, everthing at 100 (M & CPU)

3.8, 3.0-3.9, 4.8, 4.1- 4.5, 3.9, 3.0, 3.8, 4.1, 3.6, 3.3

The dashes represent multiple estimates from a given session.

If I take the worst estimates for no PXA Clocker, I get an average of

3.2

Now, in a couple of cases, PXA Clocker may have turned itself off (it apparently doesn't operate 2% of the time). So, If I take only the best for w/ PXA on (and drop the two lowest ones, 3. and 3.3, assuming they were anomalous), I get an average of ONLY

4.1 !!

Or, put it another way. My worst time w/o PXA was 3.0 and my best time with it was 4.8.

So, only a 60% improvement in the best case and nowhere near the doubling or 6 hour figure that I sometimes see bandied about.

I'd say 50% at MAX (which mirrors the more modest and common stated improvement), and probably more like 30% more in the typical case.

It's really too bad as it sounded so promising.

(And it's also too bad that PXA Clocker has such built in crippling--how can a user really know how much it really helps?).

I've stopped using it.

[acolythe]

I have not run any scientific tests, but i too have stopped using it. I can not detect a significant change in day to day use. I suppose it must save something but I cannot see it. I wish i had realized it before i bought the program. But there you go.

[jabberwocky]

I have the registered version of PXA Clocker, and my unscientific opinion is that it gives you somewhere between 1.5 to 1.8x the normal usage time. It would be very difficult to be more accurate, because it depends in part on what mix of programs you use and for how long each and how clocked etc.

One of the major factors (probably the biggest of all) in usage time is the brightness that you set your screen, which of course is out of PXA Clocker's control. I keep my T3 at minimum brightness unless I happen to be outside in bright light.

But I can positively state that I have gotten 6 hours of use (per Batterygraph) from my T3 between charges. This was mostly running Wordsmith for ebook reading and clocked at 100MHz. And if that means I would have only had 4 hours without PXA Clocker then I'd say that it's worthwhile buying! YMMV!!

However, I agree that it is unfortunate that the author made the Lite version crippleware, since that makes it hard to evaluate...

[Peppy]

I've been getting into PXA Clocker lately so I was curious as to Jabberwocky's findings. I have done NO scientific testing but I was surprised that he didn't get better results as my impression is that it does extend battery life.

I have varied the speed of a whole bunch of different applications anywhere from 100mhz to 400mhz. Before using PXA Clocker my T3 would be at 3.7mhz per XLauncher at the end of the work day. Now I find it to be between 3.85 and 3.9 AND I've increased my backlight to 1/4 brightness. To me, it seems worth it.

One thing I am dissapointed about though in Palm Reader. I tried underclocking that to 300mhz and I noticed a significant slow down. It just isn't running well on my T3 at that speed. Once I bumped it up to 400mhz again all was fine. I wouldn't think PalmReader would need that much speed.

Oh well,
Peppy

[jabberwocky]

Originally posted by Peppy
I've been getting into PXA Clocker lately so I was curious as to Jabberwocky's findings. I have done NO scientific testing but I was surprised that he didn't get better results as my impression is that it does extend battery life.

I assume you are talking about Paul's results? In fact, I said that I thought using PXA Clocker DID extend my battery life by a factor of about 1.5 to 1.8....

[aym]

I don't have hard numbers either, but I've noticed that primarily with PalmReader at 118MHz I got at least 5 hours one time before I got a warning. If you only have the demo version you may not be able to test this, but I also have as many peripherals as possible on the apps I underclock. Perhaps this is the difference between the results some people are seeing.

I'm currently running PalmReader with just Sync Serial, LCD, and PWM1 checked.

OH - I also keep the brightness at minimum.

Yes, PalmReader is slow as mud when underclocked that far, but I've gotten used to it. Turning pages isn't that bad, but you have to wait after clicking on any icon - including those in the system's status bar. It's worth it to me - I can read for hours without interruption, so it's annoying to have the battery message popping up because it always feels like "I just charged it!".

Aym

[ryanamur]

If you really want to extend your battery life, try "Energy Dimmer". With a screen setting of 5% I read an E-book off a SD Card this morning for 4.5 straight hours (no power at the office).

The battery of my T2 dropped from 78% to 52%.

[steve505]

How is the stability with PXA Clocker? Any hard resets or other odd behavior?

[Peppy]

Originally posted by ryanamur
If you really want to extend your battery life, try "Energy Dimmer". With a screen setting of 5% I read an E-book off a SD Card this morning for 4.5 straight hours (no power at the office).

The battery of my T2 dropped from 78% to 52%.

I'v used energy dimmer. I don't see what it does for the T3. Unless there is an update I'm not aware of it, it doesn't dim the T3's screen any lower than the built in program does.

So, what exactly are you doing with this program?

[Possum]

Underclocking never has given much of a battery improvement, in my opinion. Overclocking on the other hand tends to really suck up power. When they decide what speed to actually run a processor on hand held device the decision must be based on the best performance to power consumption.

I have a T3 and typically toggle bluetooth off when I am not using it. When I was using a Tungsten T, I had Bluetooth Toggle on it, but was getting unexpected disconnects if I was on-line for extended periods of time. If you do a lot of book reading, like I do, turning off bluetooth and ir seems to give a little bit more time.