FORUMS

OnePlus Addresses Reservation List Cheating

Jake Cooper figured out how to bump his place up on OnePlus’ reservation list and … more

OnePlus 2 Benchmark Scores Show Its Muscle

This video by FoneArena contains all the popular benchmarks being tested on the OnePlus 2, … more

Sony’s Emergence in The Middle: Is The Price Right?

Sony’s Electronics Division is not in its best days, and its smartphone … more

HTC Delivering Ads Straight to Sense Home

HTC’s 2015 has been a year marked by a desperate search for revenue.The HTC One M9 … more

[Kernel Test] Doing a blind study for UV. Kernels sent. Read for instructions.

854 posts
Thanks Meter: 720
 
By persiansown, Retired Recognized Developer on 28th April 2010, 08:22 PM
Post Reply Subscribe to Thread Email Thread
Ever since I first released an undervolted kernel (can you believe that it actually used to be a new idea? Seems so normal now), its been a mad dash to see how much further down the voltage can be pushed. In light of recent events, there has been controversy regarding whether or not dropping below ~925mV actually begins to negatively impact battery life.

Without getting into the technical details behind why I believe that going that low actually does affect battery life, I'm conducting a blind study. I'm going to produce 3 kernels at three different voltages. I'm then going to send these kernels randomly to people that sign up on the attached spreadsheet in order to collect data for 7 days.

I know that this won't be 100% accurate, but I feel that if we get a large enough sample size, over 7 days, the general trend should emerge.

What I ask for you to do if you want to participate is sign up on the attached spreadsheet and try not to change your rom/usage patterns within these 7 days. Just go about your day and record your values at the end.

You will not know what the voltage of the kernel you are receiving is, so there will have to be a certain degree of trust. I promise I will do my best to not blow up your phone

This truly has to be a community effort. The more people we have, the better. Once the study has been concluded, I will reveal which kernel had the best battery life for the most amount of people.

Please spread the word about this, the more it is on blogs and the like, the greater the sample size and the more accurate the results.

Link: http://bit.ly/9YjM5X


UPDATE:
The kernels have all be emailed to people that gave me an email address. Tues-Sat, please run the kernel that is labeled CONTROL. Do not alter your day to day usage from what you always do. At the end of each day please record how long you were off charger and your percent battery left. Use your best judgement for giving statistics. The more detailed, the better.
Sunday-Thurs we will run the TEST kernel. Again do the same thing and record your usage daily.

It is VITAL that you record the statistics at the end of the day every day.

As an added bonus, I included the memory modification by coolbho3k because I know you all want the latest and greatest
Last edited by persiansown; 3rd May 2010 at 11:27 PM.
 
 
28th April 2010, 08:33 PM |#2  
Member
Flag Little Rock
Thanks Meter: 0
 
More
Wink
Will it matter what battery we have? Or is stock battery required?
28th April 2010, 08:37 PM |#3  
OP Retired Recognized Developer
Thanks Meter: 720
 
More
Quote:
Originally Posted by Four Fourty Four

Will it matter what battery we have? Or is stock battery required?

doesn't matter as long as you post your previous results so we have something to base it against.
28th April 2010, 08:39 PM |#4  
Daswolven's Avatar
Senior Member
Flag Richmond Hill
Thanks Meter: 139
 
More
This is a fantastic idea. Might actually help me recover from my kernel flashing compulsion. Prolly not though....
28th April 2010, 08:41 PM |#5  
Senior Member
Flag Beltsville
Thanks Meter: 472
 
Donate to Me
More
Persiansown, what do you mean by "record your values at the end"? Are we to take a log of battery usage or just writing down on the sheet what the usage was like each day?

thanks
28th April 2010, 08:44 PM |#6  
OP Retired Recognized Developer
Thanks Meter: 720
 
More
Quote:
Originally Posted by jblazea50

Persiansown, what do you mean by "record your values at the end"? Are we to take a log of battery usage or just writing down on the sheet what the usage was like each day?

thanks

Its going to be subjective but I'm trying to figure out the best way to record them right now. I don't need any feedback values until I actually send out the kernels though
28th April 2010, 08:51 PM |#7  
Senior Member
Flag Beltsville
Thanks Meter: 472
 
Donate to Me
More
Quote:
Originally Posted by persiansown

Its going to be subjective but I'm trying to figure out the best way to record them right now. I don't need any feedback values until I actually send out the kernels though

yeah, i understand that it's not needed until after we start using the kernel; just wanted to know what you were actually looking for, and if it wasn't subjective

thanks and looking forward to this
28th April 2010, 09:51 PM |#8  
Senior Member
Flag London
Thanks Meter: 2
 
More
im really interested in this. would it also be possible to give feedback on reception? because as of late my signal has been worse with all the uv kernels...
28th April 2010, 10:00 PM |#9  
pjcforpres's Avatar
Senior Member
Flag CR
Thanks Meter: 28
 
More
How are we going to set the baseline? The best way to eliminate the potential for usage differences to skew the results is for us to all run the exact same kernel for X amount of time to set a baseline, then do the blind test of the 3 different kernels, and compare the difference between before and after.
28th April 2010, 10:04 PM |#10  
OP Retired Recognized Developer
Thanks Meter: 720
 
More
Quote:
Originally Posted by pjcforpres

How are we going to set the baseline? The best way to eliminate the potential for usage differences to skew the results is for us to all run the exact same kernel for X amount of time to set a baseline, then do the blind test of the 3 different kernels, and compare the difference between before and after.

Yeah I've been trying to think of a way that would take less time, but I think that way would be best. Anyone have a better suggestion?
28th April 2010, 10:14 PM |#11  
pjcforpres's Avatar
Senior Member
Flag CR
Thanks Meter: 28
 
More
Quote:
Originally Posted by persiansown

Yeah I've been trying to think of a way that would take less time, but I think that way would be best. Anyone have a better suggestion?

We could also look at time spent "running" in spare parts... but that would be much more complicated and less accurate (I would hate to do the statistical break down on that)... but it would be much quicker and would provide some nice info as well... ie 5 hours spent running and 7 hours sleeping used 80% battery... where as 1 hour running and 11 hours sleeping used 50%... but we would also have to factor in so many other aspects and track those to get a truly accurate picture, ie WiFi or 3G or 2G... what is your screen brightness... and so forth.

Read More
Post Reply Subscribe to Thread
Previous Thread Next Thread
Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes