Benchmark and Battery Life Tests

  h2210/15

Toshiba

e740

VOBenchMark

400 MHz/

PXBus 199 MHz

472 MHz/

PXBus 236 MHz

528 MHz/

PXBus 264 MHz

472 

MHz/PXBus 118 MHz

CPU

       
CPU - Floating Point 12.65 15.00 16.89 15.02
CPU - Integer 26.90 31.86 35.89 31.92

Graphics

       
Bitmaps        
BitBlt 78.10 78.17 78.21 121.43
StrechBlt Grow 76.40 84.30 89.60
StrechBlt Shrink 28.50 31.90 34.70
Filled        
   Ellipse 5.01 5.94 6.33 1.98
   Rectangle 12.96 13.23 13.29 9.97
   Rounded Rectangle 4.08 4.61 4.92 1.6
GAPI (Direct Video)        
    Max FPS 58.765 66.666 73.394
    Max Sprites 178 236 282
Lines 197.60 225.50 247.40
Memory        
   Allocation 11.22 13.31 14.96 14.09
   Fill 1.96 2.33 2.62 1.13
   Move 1.24 1.47 1.70 0.46
Text 5.10 6.10 6.50 5.06

Pocket Quake Tests

Pocket PC

FPS

H2215 (at 528MHz) 18.5
H2215 (at 400MHz) 14.7
e740 (at 472MHz) 10.1
e750/e755 (at 400MHz & PPC2002) 10.3
e750/e755 (at 400MHz & WM2003) 14.9
H5555 (at 400MHz) 14.0

 

How the overclocking affects the battery life.

 

A frequently question is how the overclocking affects the battery life. The battery life depends of many parameters including what applications that you are running, what is the CPU Load, how many applications you are running at the same time, what are your backlight settings, etc, and it's very difficult to predict. So I did not tried to predict what would be the battery life in every case or situation, just I prepared a few tables representing different situations (CPU Load)  where 400 MHz has a value of 1 and the rest of the speeds have a value that you should apply to your battery life  to find out how it's going to be affected at those speeds with same conditions of CPU load, backlight, brightness, etc. I used a little program created by me in eVB to measure the time that the battery level takes to go down in 3 percentage (starting to measure the time after the level goes down in one percentage, in this way I'm sure that I'm measuring full percentages of the battery life) and I measure that time in each speed having the Pocket PC in the same conditions of CPU Load and back light.

 

CPU Normalized Load of 8 %

(Just the program used to measure the time and battery level running)

199 MHz/PXBus 99.5 MHz

354 MHz/PXBus 118 MHz

400 MHz/PXBus 199 MHz

472 MHz/PXBus 236 MHz

528 MHz/PXBus 264 MHz

0.82

0.75

1

0.67

0.60

 

As you can see from the above table in the case of XScale PXA255, having the device running in a lower speed with a PXBus smaller than 199 MHz wont help to increase the battery life if the CPU Load is minimal. You can find a good  explanation to this issue at Anton's Tomov forum (creator of Pocket Hack Master)

 

CPU Normalized Load of 18 %

(Playing in background a WMV in a loop )

199 MHz/PXBus 99.5 MHz

400 MHz/PXBus 199 MHz

472 MHz/PXBus 236 MHz

1.06

1

0.93

 

In another hand, the XScale processor works better when the CPU Load is higher as you can see from the above table where at 472 MHz the battery level dropped less than what it dropped at the same speed but with less CPU load.

 

When reading an eBook, every time you change a page the CPU Load raises to around 28 % or more and then it goes back to 0 during the time when you are reading the page. So this situation is a balance between this two cases described in the above two tables. Taking in consideration that the saving of energy in the second case is smaller than the waste of the first case, my conclusion is that the 200 MHz speed with a PXBus of 99.5 % (27x2x1) do not safe any battery while reading eBooks or the saving is minimal.

 

Reading ebook the battery life of h2215 at 400 MHz  should be around 4.5 ~ 5 hours no having any other application running in background and taking in consideration that your Pocket PC wont work with less than 10 % of battery. 

 

What is next?

 

Well, I will run some more tests with a higher CPU Load to see how the CPU load affects the battery life.

 

 

Back to PC Counselor's h221x page.

Hit Counter