I came across an updated whitepaper for Tegra3. tegra3 is now marketed as a "4+1" core system. This whitepaper goes into detail exactly how Tegra3 works. its a very good read and highly recommended to download to become more educated on how the Tegra3 in your prime really operates. it also tells how its quad core uses less power than dual cores, no matter the die size. plus tegra3 can perform same tasks as dual cores but use less power doing it. it details all its benefits. it also details the challenges it has also and how it was addressed. this chip has several patented techniques used in it. This chip is a first and revolutionary in mobile computing world. check it out. I've uploaded the file so it can be shared. after reading it, I'm even more impressed with Nvidia now and appreciate the maximum power n great battery life of Tegra3 even more.
check it out..I've enclosed it within a zip file to meet uploading requirements since original PDF was too large.
Also Tegra3 expected to be a big presence at MWC @ Barcelona. Several tablets and phones using this chipset. Here is link to more details. http://www.androidcentral.com/nvidia-dubs-their-tegra-3-architecture-4-plus-1-hints-big-mwc-presence
edit: added another good read. whitepaper on Benefits of Multi-Core CPU's in Mobile Device's
Thank you, I'm interested in reading this myself
Thanx for the info.
Thanks Demandarin, I am always interested in how things work, it's what got me into tech in the first place. I do like a good tech read, the same stuff that I find fascinating would put most folks into a coma!
Thanks for going to the trouble to zip & upload.
As I read this white paper, I'm pretty convinced that ASUS and Nvidia (probably mostly Nvidia) still have some work to do in optimizing this thing. I think it spends too much time with too many cores working at too high a frequency, and that's impacting battery life. And that's both in standby and actual use.
Not complaining, mind you, just making a note. While my Prime on .14 gets decent battery life, it doesn't get great battery life. I see too many instances of 15%/hour battery use when I'm only reading an ebook--if just the companion core were being utilized, as I think it should, I'd expect this number to be MUCH lower. And this is with the screen on the lowest brightness. Honestly, I hope to be able use the Prime for ebook reading for longer than 6.5 hours...
I'm sure they'll get things optimized at some point. I hear rumors that .15 might bring much improved battery life, and so I'm looking forward to seeing it.
wynand32 said:
As I read this white paper, I'm pretty convinced that ASUS and Nvidia (probably mostly Nvidia) still have some work to do in optimizing this thing. I think it spends too much time with too many cores working at too high a frequency, and that's impacting battery life. And that's both in standby and actual use.
Not complaining, mind you, just making a note. While my Prime on .14 gets decent battery life, it doesn't get great battery life. I see too many instances of 15%/hour battery use when I'm only reading an ebook--if just the companion core were being utilized, as I think it should, I'd expect this number to be MUCH lower. And this is with the screen on the lowest brightness. Honestly, I hope to be able use the Prime for ebook reading for longer than 6.5 hours...
I'm sure they'll get things optimized at some point. I hear rumors that .15 might bring much improved battery life, and so I'm looking forward to seeing it.
Click to expand...
Click to collapse
I get what you saying. It is weird that although it was confirmed that all primes have the same hardware, performance and battery life varies between devices. I get longer than 6.5hrs. On 1.6Ghz overclock I'm running..lmao and that's with surfing the web and playing games n videos. You are correct, no reason why just e-reading should drain 15% battery per he. On lowest brightness setting. I normally get about 10% battery drain with heavy usage. I would think something else is at play on your device. Maybe a rogue app or widget. I always get get great battery life. I deleted some apps that seemed to make battery drain faster. Plus I cut back my notification timings. Did all I could to maximize battery performance while maintaining good speeds. I usually either run stock balanced mode or cut on 1.6Ghz overclock.
I'm always down for updates improving battery life. Hopefully .15 will bring this for you. Can't see how your tablet is draining so fast only reading books. SMH. Have you installed Better Battery stats to see if any rogue apps or processes are constantly running or cutting off n on? That's what I used to narrow down poorly coded apps/Widgets not optimized yet.for the type of use you doing, only the companion core or a real low frequency main core should be running. I really think something else might be going on in your tablet. There are plenty experiencing great battery life. With only reading e-books, you should easily be able to pull 9-10hrs.+ out of tablet in balanced mode. Even longer in powersavings mode.
What mode do you usually run your tablet in?
Also I get great standby n deep sleep. It only barely sips battery in those states. I know around .11 and before .14 updates, the speeds were changed and always seemed to run high all the time. Multiple cores seemed to run at same time when not needed. Now the latest update .14 brought back our original speeds. Multiple cores don't run now when not needed. I use System Tuner to confirm what speeds processors are running at. Others confirmed also speeds going back to normal and getting the great battery life back. So I don't see my cores running excessively high anymore or running more at once if not needed.
One trick to try for your case is go into developers options and cut on force 2d hardware gpu rendering. Run it for a little to see if it helps make battery last longer. It might help since gpu being forced to do what you need for reading. Gpu is more efficient and consumes less power than CPU at that kind of stuff. Plus it'll load up and turn your pages faster n stuff n shouldn't be hardly any lag.
Well, it's interesting. I'm running a number of apps to track battery usage (it's become something of a hobby, trying to figure out what's going on ), and I've not managed to identify anything acting wonky. I run on Balanced mode pretty much all the time, but interestingly I don't notice much different when I switch to either Battery Saver or Performance.
I can run at 20%/hour or more when I'm really hitting things, such as heavy browsing or gaming. And, yes, sometimes I drop down to 10%/hour when I'm ebook reading, although that usually doesn't last long. Overall, there's enough inconsistency that I'm pretty much convinced that it's due to the Tegra 3 not making efficient use of its cores, and again I'm also convinced Nvidia and ASUS will fix it eventually.
Regarding standby, I get around .5%/hour use. While that seems good at first glance, it still seems high when you consider that only the companion core should be running, at something less than 500MHz. Configured roughly the same way, my TF101 has managed about the same .5%/hour use (sometimes better). So, while the Prime is doing OKAY, it doesn't seem to be doing what it's capable of. And, I remember on one of the builds, I was running closer to .25%/hour, so I know it's possible.
Oh, incidentally: there's also a bug in .14. Sometimes, it'll drop 2% instead of 1%. So, apps will be reporting good battery use, and then WHAM, it'll drop 2% instead of 1% and the numbers go back up (naturally).
Havent read it yet but does it explain why the tablet slows down a lot during downloading? I assumed with 4 cores it would be able to handle it no problem. Maybe its an OS issue.
Nice Thank you sir!
aznmode said:
Havent read it yet but does it explain why the tablet slows down a lot during downloading? I assumed with 4 cores it would be able to handle it no problem. Maybe its an OS issue.
Click to expand...
Click to collapse
Some downloads go really fast n others seem really slow. Very dependent on the servers where the downloads are coming from. If anything I think that's more of a software or server issue.
Wow. When they get the software aspect down (Nvidia) this chip will ROCK for battery life. Can't wait to see Wayne, 28nm ftw! Assuming 28nm transition goes smoothly, which seems like a thing of the past for every company except Intel. From 65nm down to 28nm has been a rough, expensive ride for more companies than one.
Sent from my Transformer Prime TF201 using Tapatalk
Now TI is showing off their OMAP5 vs. the Tegra3.
http://mobilesyrup.com/2012/02/23/t...core-omap-5-soc-wipes-the-floor-with-tegra-3/
Lmfao, if it comes out early 2013 that's an entire generational leap! I would hope it wipes the floor with tegra 3. If it comes out next month then that sucks for nvidia, next year who cares.
Sent from my Transformer Prime TF201 using Tapatalk
clockwork58 said:
Now TI is showing off their OMAP5 vs. the Tegra3.
http://mobilesyrup.com/2012/02/23/t...core-omap-5-soc-wipes-the-floor-with-tegra-3/
Click to expand...
Click to collapse
Wow, they comparing a chip that won't be released till 2013(article confirms it) to a chip that's already available now. Lmfao I'd say a big fail on their part. Test it against tegra4 or even an overclocked tegra3 n see how the results change
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
Nice try. TEGRA3 is best Android has to offer today and for at least a few months more. June is probably when other chips for android will finally drop.
benefit14snake said:
Lmfao, if it comes out early 2013 that's an entire generational leap! I would hope it wipes the floor with tegra 3. If it comes out next month then that sucks for nvidia, next year who cares.
Sent from my Transformer Prime TF201 using Tapatalk
Click to expand...
Click to collapse
That's what I'm saying..lmfao. when that TI chip finally drops Tegra4/Wayne will be out already. Then it'll wipe the floor with that TI chip. I think even an overclocked prime can hang with it.
Maybe it's my age. But when I think of TI, I think of calculators.
Most of this reiterates what I already know, I've paid close attention to nvidia in the months before t3s release. The thing that bothers me is that there is space for larger caches on the cpu next to the 5th core. If you look at intels cpu layouts they use the space on the chip very frugally.
I hope they will include a companion core to wayne. With 8 cores the 9th companion core will form a nice square to properly fill up the chip. I'm not sure if we'll get batman though, if tablets only get robin I don't know how steep of an increase that will be. Apparently its supposed to have a ridiculous amount of gpu cores/shaders.
As long as it gets better battery I'll be pleased. I don't lag with t3, but the battery life is poor compared to my tf101.
aznmode said:
Havent read it yet but does it explain why the tablet slows down a lot during downloading? I assumed with 4 cores it would be able to handle it no problem. Maybe its an OS issue.
Click to expand...
Click to collapse
It's an I/O issue.
Regardless how fast the processors are, if the memory (um, hard disk equivalent memory, not RAM) is slow, then the processor will be waiting on the input/output writing reading.
People have suggested here in other threads that the memory used in the Prime isn't all that good, bc there are low scores on the Bonnie++ and other tests.
clockwork58 said:
Maybe it's my age. But when I think of TI, I think of calculators.
Click to expand...
Click to collapse
Same here..lol. they were the first on the mobile chip/computing scene. I believe. I remember those graphing calculators they had in school. Some people hascked those to run simple programs since they were like simple mini computers. The newer more expensive ones came out in color. Who knows how advanced their calculators are now. I use to think those were so cool. Even had a port on it to hook it up to stuff.
Talking about this now got me remembering when electronic organizers first came out. I had several. Man, I thought I had the coolest devices n those were even more like mini computers. Always dreamed of hacking stuff with them. Those were the days. I even had one of the more expensive Rolodex ones. It was like a box clam shell. The whole top piece was the screen n whole bottom half was the keyboard. Then you had the remote control watches. Changing the channels in class n teacher wondering how OTA happening..lol there was organizer watches also along with the watches that played videogames on them.
Let me stop before I go on n on about history of tech n how its evolved so much..lol
Thanks for the flashback, I'm still young, but old enough to remember all the old stuff. Damn, its been 14yrs. Since I graduated in 98. Now I'm feeling old lol still seems just like yesterday thinkn back to it all.
ickkii said:
Most of this reiterates what I already know, I've paid close attention to nvidia in the months before t3s release. The thing that bothers me is that there is space for larger caches on the cpu next to the 5th core. If you look at intels cpu layouts they use the space on the chip very frugally.
I hope they will include a companion core to wayne. With 8 cores the 9th companion core will form a nice square to properly fill up the chip. I'm not sure if we'll get batman though, if tablets only get robin I don't know how steep of an increase that will be. Apparently its supposed to have a ridiculous amount of gpu cores/shaders.
As long as it gets better battery I'll be pleased. I don't lag with t3, but the battery life is poor compared to my tf101.
Click to expand...
Click to collapse
Are you talking about the legendary STARKS chipset? Once Nvidia releases that beast, Its a Whole New worldddddd...lol. cpu's n gpu's out da a$$.
I actually get great battery life. My prime lasts longer than my Ipad. Even on 1.6Ghz overclock I get good battery life compared to the speed increase.
Related
Hello android fellas..i am an sgs 2 owner and a log time android user..for the most of android devices the most screen time i could ever get is 5 hours (screen on, not idle..idle means ntn..with uv i can get week++)..
So the question i rise here is, will ice cream sandwitch help our battery effeciency with hw acceleration?i noticed that my phone can do as much as 10 hours+ of video playback, but can only surf the web for around 4-4,5 ...which is amazingly ridiculous..if i let the phone idle , with screen turned on, at all times(there are actually apps to do that) i can get almost 11 hours of screen on time..but if i use my phone to as much as navigate through menus, i am lucky if i get 30% of that time.
That led me to the conclusion that cpu usage for 2d ui elements drawing is at best idiotic, but other than that, its that its not my device(s) problem..its the os that sucks it up..
So, do you believe, that ICS will fix that?is there hope on the horizon?4 hours screen on time/day , just wont cut it for a tought work day for me..
( i would like to close that section by saying that i am a 2 year android user and i know what i need to shut down, when to have on and what so that my batt is at its max effeciency..)
Opinions please? Will our GPUs make us look less embarassing?
Unlikely... partially, because the majority of actions needed by a web browser is unable to utilize the GPU (it's really only usable for compositing and rendering, which occurs surprisingly seldom outside of games: really only while the document loads... of course that's bound to change at some point, but for now, there really isn't much to gain other than more fluid animations), but mostly because it's a bit of an unfair comparison: Video playback today is more or less handled by special-purpose silicon, not what you'd usually call "The GPU".
Hans Schmucker said:
Unlikely... partially, because the majority of actions needed by a web browser is unable to utilize the GPU (it's really only usable for compositing and rendering, which occurs surprisingly seldom outside of games: really only while the document loads... of course that's bound to change at some point, but for now, there really isn't much to gain other than more fluid animations), but mostly because it's a bit of an unfair comparison: Video playback today is more or less handled by special-purpose silicon, not what you'd usually call "The GPU".
Click to expand...
Click to collapse
no, i am aware that java still needs to be rendered in cpu... but i was not only bein referred to in browsing...menu browsing as well..ui navigation should be humongousl benetited am i right?
Accelerated is usually faster and smoother, especially multiple layers benefit greatly from additional acceleration (note however, that Android already uses that in most cases)... it just doesn't really translate to power saving.
Java doesn't really have anything to do with that: it may put a bit of additional load on the CPU, but that's not what's eating your battery, especially since virtually all computation-intensive parts of Android are written in C++. The real problem is that while the GPU is faster in these "burst" situations when there's suddenly a whole lot of layers to render, it's hard to predict those. So the CPU most always run at an acceptable rate... which in the end often means that the accelerated version uses more power since the system isn't able to clock down the CPU nor the GPU.
Hans Schmucker said:
Accelerated is usually faster and smoother, especially multiple layers benefit greatly from additional acceleration (note however, that Android already uses that in most cases)... it just doesn't really translate to power saving.
Java doesn't really have anything to do with that: it may put a bit of additional load on the CPU, but that's not what's eating your battery, especially since virtually all computation-intensive parts of Android are written in C++. The real problem is that while the GPU is faster in these "burst" situations when there's suddenly a whole lot of layers to render, it's hard to predict those. So the CPU most always run at an acceptable rate... which in the end often means that the accelerated version uses more power since the system isn't able to clock down the CPU nor the GPU.
Click to expand...
Click to collapse
Damn..The load that cpu took from being less effecient to draw 2d, was my best explaination as to why our batteries suck so much..
It is definately not the phones..As i said earlier if i let the phone screen on, just idling, at, say auto brighness, it can go for 10-11 hours..but if during that time i, as much as navigate at menus, or settings , or whatever ui...BAM..4-5,5 hours maximum..
It HAS to be something wrong with the os..heck i can do 10 hours of .avi playback on a single charge , but i cannot go over 5 for just panning around the app drawer..(and my screen is amoled so it greatly benefits from black background of the drawer)...
If you could shed some light to that, for me i'd love ya even more!
Btw on my second post i meant to type javascript, not java my bad!
Well I downloaded cpu master (free) to just mess around with my photon, and come to find out, the governor for the cpu is set to performance, but gave me the option to change it to powersave, so I switch it and I'll report back to you guys and see if I've found the holy Grail to even better battery life
Sent from my MoPho using XDA App
That's awesome
Sent from my MB855 using xda premium
Wait until I or another dev gets onDemand governor enabled...
I have SetCPU, and honestly, I haven't noticed a huge difference in battery life with powersave unless I seriously scale the CPU back to like 300 mhz. Then it'll last a while, if I don't do ANYTHING with it. For example, when I'm sleeping, otherwise, it just makes the phone laggy and doesn't seem to help enough to make it worth while. Just my opinion from screwing with it.
xTMFxOffshore said:
I have SetCPU, and honestly, I haven't noticed a huge difference in battery life with powersave unless I seriously scale the CPU back to like 300 mhz. Then it'll last a while, if I don't do ANYTHING with it. For example, when I'm sleeping, otherwise, it just makes the phone laggy and doesn't seem to help enough to make it worth while. Just my opinion from screwing with it.
Click to expand...
Click to collapse
The general consensus regarding over/underclocking when I had my HTC Hero was that overclocking would save battery because you could get what you wanted done faster. If you scale the CPU back massively while it is set to sleep, however, you will save a lot of battery.
xTMFxOffshore said:
I have SetCPU, and honestly, I haven't noticed a huge difference in battery life with powersave unless I seriously scale the CPU back to like 300 mhz. Then it'll last a while, if I don't do ANYTHING with it. For example, when I'm sleeping, otherwise, it just makes the phone laggy and doesn't seem to help enough to make it worth while. Just my opinion from screwing with it.
Click to expand...
Click to collapse
well, i don't know about your phone or what else you have setup, but my phone didn't lag and it seemed to help, i currently have my evo 3d active so my photon just sits there, so i can give a good feedback of idle time, but i can tell you after switching the governor, it went down 1% in 2 hours, now i say that's an improvement, so when i get back home i'll really give you guys some feedback
P.S. with any phone i had that had a fully custom kernal, i always used conservative governor
Well, perhaps it's just the apps I have running then. As I said, when it's set to sleep, it works pretty well with the powersave mode, otherwise, doesn't seem to make any real difference. Guess it's different for everyone cause of the **** they're running on their phone.
mrinehart93 said:
The general consensus regarding over/underclocking when I had my HTC Hero was that overclocking would save battery because you could get what you wanted done faster.
Click to expand...
Click to collapse
You know, I've never bought into this argument. To me, it is like saying that if I drive 100 mph I will get there faster, so I use less gas....which we all know is not how it works.
Maybe the physics are different for processors then they are for engines, but I think there is probably a happy medium somewhere. And I have a feeling that the manufactures really take this into consideration when they develop the kernels and ROMs. However, I might be wrong.
This is by no means an effort to discourage your awesome work. Everyone gets different results, but stock always seems to have the best battery life for me once all the bloat is gone. However, custom kernels do perform better. That is the trade off, in my opinion.
my2cents said:
You know, I've never bought into this argument. To me, it is like saying that if I drive 100 mph I will get there faster, so I use less gas....which we all know is not how it works.
Maybe the physics are different for processors then they are for engines, but I think there is probably a happy medium somewhere. And I have a feeling that the manufactures really take this into consideration when they develop the kernels and ROMs. However, I might be wrong.
This is by no means an effort to discourage your awesome work. Everyone gets different results, but stock always seems to have the best battery life for me once all the bloat is gone. However, custom kernels do perform better. That is the trade off, in my opinion.
Click to expand...
Click to collapse
Lol honestly I never bought into it either. I was just posting was the other devs at the time said. Even using an OC kernel, I never overclocked my phone.
mrinehart93 said:
Lol honestly I never bought into it either. I was just posting was the other devs at the time said. Even using an OC kernel, I never overclocked my phone.
Click to expand...
Click to collapse
Interesting...I hadn't thought about using an OC kernel and underclocking it back to stock. I wonder what that does, if anything, to performance.
The other concern that I have is that overclocking typically means more heat, which means more battery use... Just figured I would throw that out there too.
well, the results are in, now granted i already had the photon of the charge for more then 24 hours, so at 1d 15hr 57m and 10s i'm at 48%, but from the time i started the cpu test, 1:30pm, it was at 68% so in 8 horus there was only a 20% drop while idle, i say that's a good score , you guys tell me otherwise
A2CKilla said:
well, the results are in, now granted i already had the photon of the charge for more then 24 hours, so at 1d 15hr 57m and 10s i'm at 48%, but from the time i started the cpu test, 1:30pm, it was at 68% so in 8 horus there was only a 20% drop while idle, i say that's a good score , you guys tell me otherwise
Click to expand...
Click to collapse
and just an update, in another 4 hours, it's only gone down 2%!!!!!! come on guys, these numbers can't lie, but remember this is idle feedback, i'll re-activate my photon at the end of the week (missing the beast!) to give more detail feedback on heavy usage and most importantly 4g!!
Development. Development. Development.
Sent from my MB855 using xda premium
Not sure if this will help but let me explain the car analogy. All motors have an effeciency range at x amount of throttle. So same cars will get better gas mileage at a higher speed vs a lower one. Its getting up to that speed where most energy is used.
So let me move this over to electronics. If you run a faster clock speed while on, your apps will open faster so that ia less on time for the screen and other processes that have to run. So using more watts for less time does not always equal more than using less watts for more time. We just need to fill in those blanks and obviously overclocking will not benefit the nook or internet reader as it eould someone who opens a lot of apps for short periods. Same is true for a gamer .
Hope that makes sense and this is all IMHO of course.
Sent from my MB855 using XDA App
scoobdude said:
Not sure if this will help but let me explain the car analogy. All motors have an effeciency range at x amount of throttle. So same cars will get better gas mileage at a higher speed vs a lower one. Its getting up to that speed where most energy is used.
So let me move this over to electronics. If you run a faster clock speed while on, your apps will open faster so that ia less on time for the screen and other processes that have to run. So using more watts for less time does not always equal more than using less watts for more time. We just need to fill in those blanks and obviously overclocking will not benefit the nook or internet reader as it eould someone who opens a lot of apps for short periods. Same is true for a gamer .
Hope that makes sense and this is all IMHO of course.
Sent from my MB855 using XDA App
Click to expand...
Click to collapse
Let me start this off by saying that I am an agricultural engineer by trade. With that said, I would agree that engines have an efficiency range. However, I would not agree that it is at x throttle. Rather, it is a x load. And most cars are designed to be at optimal load at about 60 mph (wind resist, weight, etc. play a role in this). You're right, it does require more fuel to get to that speed because the load is higher until that speed is reached. Furthermore, higher speeds (greater than 60 mph) do NOT translate into higher fuel mileages because the load increases to maintain the higher speed (because there is more wind resistance, among other things).
I know a bit about electricity too, but I don't fully understand the physics behind circuit boards. However, I think the analogy still holds. I can get to 60 mph as fast as I want, but the faster I do it the more energy is required. Therefore, even though it is done faster it still requires more energy, which also creates more heat, both of which use the battery. So, I continue to contend that there is a happy medium that most be found and I think electronic engineers aren't to far off.
Here is a little more reading about car efficiency, if you are interested: http://www.mpgforspeed.com/
I believe your confusing overclocking and overvolting. We are putting x volts into the processor so the more cycles we can get in x volts the better. If we have to overvolt to overclock then we see big battery hits.
Sent from my MB855 using Tapatalk
2cents, that link is interesting but real world examples have proven otherwise to me. Our saab will do better on mpg at 70 to 75 (30 on cruise control) vs 55 to 65(27 to 28 on cruise). Now the wrx is another story as well. With the old 3.9 final drive i would pull in more air at the airflow meter at 65 than i would at 70 with stoich as the target a/f ratio, and because that motor was doing under 2500 rpms the turbo was out of the equation further taking out efficiency with it. Now with the 4.44 and a better 1-2 gear ratio car gets better at the lower engine speeds and accelerates even better and that was before i retuned it.
Another misconception is bigger motors use more gas, one of the recent corvettes get 30 on the highway proving that there is more to this as well.
I think i have taken this off topic enoigh for now. But in general i think my formula still needs to be applied to see the results as a valid number to compare overclocking and underclocking to running stock.
Sent from my MB855 using XDA App
i'm surprised no one even thought of this topic, regardless if things are getting done "faster" you are overclocking the cpu, making it go faster then what it's suppose to, which makes it use more power i.e. more battery, every phone that i had when i overclocked it, the battery wouldn't last too long, even with my 3500mah OG evo, if i overclocked it, i couldn't get a whole day, so underclocking will have the same effect in a sense since the processor now has to work harder to do what it does at it's stock clocked speed, well, hope anything i said here makes any logical sense, but on another note, the photon has gone almost 3 days unplugged!!!
again it depends on how the overclock is achieved. In most cases overclock is achieved by dumping more electricity into the cpu this will impact battery life, however alot of chips these days can be overclocked at the same volts essentially upping the "MPG" or clocks per volt.
A2CKilla said:
i'm surprised no one even thought of this topic, regardless if things are getting done "faster" you are overclocking the cpu, making it go faster then what it's suppose to, which makes it use more power i.e. more battery, every phone that i had when i overclocked it, the battery wouldn't last too long, even with my 3500mah OG evo, if i overclocked it, i couldn't get a whole day, so underclocking will have the same effect in a sense since the processor now has to work harder to do what it does at it's stock clocked speed, well, hope anything i said here makes any logical sense, but on another note, the photon has gone almost 3 days unplugged!!!
Click to expand...
Click to collapse
I agree with this for the most part. When it comes to power the end result is watts, which is essentially equal to volts x amps. It is not perfect because of a power factor, but it is close. Therefore, if you lower the voltage, the amperage goes up because the same watts are required to run the processor. The inverse is also true. These processors have voltage ranges that they will safely run in, but in the end, they require the same energy (in watts) to function at a given load. Change one a little bit (the voltage for example) and the other (amperage) compensates. Change it a lot and it likely wont work. Again, this is how it works for your typical home appliances. For a circuit board, it might be a slightly different story, but I imagine the science does not change.
---------- Post added at 03:57 PM ---------- Previous post was at 03:47 PM ----------
scoobdude said:
2cents, that link is interesting but real world examples have proven otherwise to me. Our saab will do better on mpg at 70 to 75 (30 on cruise control) vs 55 to 65(27 to 28 on cruise). Now the wrx is another story as well. With the old 3.9 final drive i would pull in more air at the airflow meter at 65 than i would at 70 with stoich as the target a/f ratio, and because that motor was doing under 2500 rpms the turbo was out of the equation further taking out efficiency with it. Now with the 4.44 and a better 1-2 gear ratio car gets better at the lower engine speeds and accelerates even better and that was before i retuned it.
Another misconception is bigger motors use more gas, one of the recent corvettes get 30 on the highway proving that there is more to this as well.
I think i have taken this off topic enoigh for now. But in general i think my formula still needs to be applied to see the results as a valid number to compare overclocking and underclocking to running stock.
Sent from my MB855 using XDA App
Click to expand...
Click to collapse
You may be right. Perhaps Saab designed their fuel efficiency at 70 mph. It's possible because many speed limits are now at or near that, but in general, optimal fuel consumption is going to be at or near 60 mph. Obviously gear ratios and such play a huge role in fuel economy. It is like using a custom rom, typically your mods will make it perform better, but the best fuel economy or battery life will come with a stock like setup.
Sure a corvette can have HP and economy. There is a power to weight ratio and lots of aerodynamics involved, which again is designed at a specific speed. But there is no way that a dragster will have a good fuel efficiency. To my point, there is a balance...
I agree, we are off topic a little bit, but the conversation is interesting, nonetheless.
By the way, can a mod move this to general, please?
Ok, I had a prime that rebooted, swapped it and now have one with BT/wifi issues.
I always thought the reboots were like a bad PC overclock, aka happend to some but not others.
Well now I see by looking at CPU spy, ASUS has not only 1600mhz and 1500mhz marked unused but 1400mhz is now unused as well.
Also they added 370mhz and 204mhz and 102mhz.
So because *some* locked up we all have lower top end performance and more power drain by having 475 vs 102 be the lowest clock.
Dunno about you but this was not the fix I was hoping for.
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
Wait what? ASUS lowered the processing speed of Performance mode back to 1.3Ghz!?
Yep! In performance mode I can get 1300, balance seems to cap out at 1200 max even with glowball going.
So we "fixed" the crashing by Under-clocking? I'm not sure the 90% will be happy.
You would have thought they would have at least let us have 1400 in performance mode.
Does anyone have a app that will show all 5 tegra 3 cores in real time?
Guys by the end of the month they'll release the bootloader unlock tool and then you will ne able to build your own kernels and use whatever clock you want.
I think its more important that the device is stable for the normal user that doesnt know about clocks and voltage. CPUs are different, some can handle extreme values while others chicken out at the slightest overclock. The standard settings have to work on all devices so yeah...
Funny thing: at the same time users report a smoother experience because of the new gpu driver from nvidia I dont see the need to OC the fastest tablet out there anyway. But thats just me.
I can almost buy the 'lets slow everyone down to stop the crashing issue'. But why increase the minimum from 102 to 475? It just uses the battery quicker.
I would like to have the advertised clock speed without voiding my ability to RMA if something breaks. At the VERY least they could add it to the performance setting.
Also while you are looking at the 1.4 lockout, they also locked out 3 low mhz options, and that will have a huge effect on battery life.
This feels more like a hack than a fix.
It would be cool just to have a over clocked kernel just to say that it can be even faster but not really need to utilize it hahaha
I'm not complaining right now my Prime is very stable at the moment & i have been throwing alot at it. But we shall see.
Ok, so from what I can see, rather than have some cores on and other off, and 1 core only when the other 4 are not used etc. It now as all 4 cores on, at 1200 or 475!!!!
All 4 cores on all the time is no doubt going to give us a LOT less battery time!
Sounds like there are some issues with the Tegra3 and ICS in how they handle the "magic" of turning on and off the cores.
From what I have found to test with my prime just cycles between 475X4 and 1200X4 in balanced mode. Anyone else show something different?
Power saving mode seems to be 600mhz X 4 or 475X4
performance is 1300 X 4 and 475X4
fenturi said:
Power saving mode seems to be 600mhz X 4 or 475X4
performance is 1300 X 4 and 475X4
Click to expand...
Click to collapse
For what its worth, I am seeing exactly the same as you fenturi
For me having if I don't see a performance issue by being 100MHz below the spec for now, I'll take it.
It's like my computer.. I have an i-7600k and GTX 580.. I play lot of stuff on max and a good deal of things I do don't peg the GPU or CPU to the max so other than just having the overheard, it's nothing to fret over for me.
For those that just care about benching and raw numbers, yeah it's an issue.
It's more so on a mobile device.. I just want a solid smooth experience with good battery life. We'll have to see how much battery life this adjustment does shave off, but if I can still get a full days usage out of it without having to run to a charger, I'm satisfied. Something like it now runs 9.5 hours instead of 10 hours would be fine by me.
Hopefully, they can mature the drivers and such to eek more performance without having to get the CPU to max out. I'd rather have that then to just have it run at 1400MHz. I'll take optimization over using raw power any day.
I'm a little loss here with the original poster?
You say you would just like the advertised speed, but 1.3 is the advertised speed of the Prime. With Honeycomb, that was all you got. It wasn't until ICS was released that we saw the 1.5 and 1.6 settings, no?
Anyhow, I think what Asus is doing here, is helping those people with tablets with mediocre cores that either cannot handle the lower frequencies while in deep sleep (sleep of death syndrome), and those who have reboots for no other apparent reason, other than there chips cannot handle a bit more juice. So to save face, they lowered back the max clock to the advertised speed, which was 1.3 on Tegra3, and up'ed the deep sleep speed to help alleviate any sleep of death syndrome's.
Like someone posted above, by the end of the month we should hopefully have the bootloader unlocked, the kernal source is already available, so custom kernals should follow, allowing those of us with chips that can go lower and higher those frequencies again.
Remember, this is nVidia, which has been known to release chips on the edge for the past several years! Asus does not make the Tegra3, they just oem it. Ultimately nVidia writes the core drivers, and Asus can modify them if need be, or add there "skin" to them. They have been doing this for years with graphic drivers.
Just like the framebuffer bug that is still in the new release firmware. This is a nVidia issue. Its there graphic driver, and they need to keep tweaking it until they get it right.
jyan_osu said:
For those that just care about benching and raw numbers, yeah it's an issue.
We'll have to see how much battery life this adjustment does shave off, but if I can still get a full days usage out of it without having to run to a charger, I'm satisfied. Something like it now runs 9.5 hours instead of 10 hours would be fine by me.
Click to expand...
Click to collapse
Um... OK, so, if you take the minimum power it can use... Multiply it by 4... And operate four times across all the cores...
This will not be a mere thirty minutes' difference. They apparently did this on 11.1 as well. Can anyone comment on 11? I had already noticed worse battery life but just hadn't thought about it. This is not ok! My device is draining so much faster while doing nothing...
Forget the upper end, OC'ing isn't even needed at this point. Everyone should be livid about the UC'ing, they're murdering our battery just to have it on standby!
1.4Hgz only cuts on when absolutely needed. It doesn't just cut on for hell of it just because you put it in Performance mode. It still stays lower clocked till you do something to need a higher clock. Plus the 1.4Ghz will only be on 1 core. Any other time you will get all cores at 1.3Ghz. Which better or more powerful than just [email protected] 1.4Ghz. People are to quick to jump to conclusions. They get CPU spy for the first time and see something then quick to jumpbto conclusions with nothing tobreally back it up. If you just messing around on home screen, of course the 1.4ghz speed isn't going to cut on. PLAY shadowgun or something n watch that speed easily cut on. Plus when 1.4ghz does cut on, its only briefly. Then other cores take over. Unless the game is taxing it enough to need it on longer.
This is nothing to be pressed about at all. People not understanding how these cores really work or how and when they cut on. Hasn't even been 24hrs. Since update n people already jumping to conclusions. If you so concerned about speed then use ATP tweaks to enable 1.6ghz overclock. Still works just fine n all speeds enabled.
---------- Post added at 09:38 AM ---------- Previous post was at 09:36 AM ----------
1BadHEMI said:
I'm a little loss here with the original poster?
You say you would just like the advertised speed, but 1.3 is the advertised speed of the Prime. With Honeycomb, that was all you got. It wasn't until ICS was released that we saw the 1.5 and 1.6 settings, no?
Anyhow, I think what Asus is doing here, is helping those people with tablets with mediocre cores that either cannot handle the lower frequencies while in deep sleep (sleep of death syndrome), and those who have reboots for no other apparent reason, other than there chips cannot handle a bit more juice. So to save face, they lowered back the max clock to the advertised speed, which was 1.3 on Tegra3, and up'ed the deep sleep speed to help alleviate any sleep of death syndrome's.
Like someone posted above, by the end of the month we should hopefully have the bootloader unlocked, the kernal source is already available, so custom kernals should follow, allowing those of us with chips that can go lower and higher those frequencies again.
Remember, this is nVidia, which has been known to release chips on the edge for the past several years! Asus does not make the Tegra3, they just oem it. Ultimately nVidia writes the core drivers, and Asus can modify them if need be, or add there "skin" to them. They have been doing this for years with graphic drivers.
Just like the framebuffer bug that is still in the new release firmware. This is a nVidia issue. Its there graphic driver, and they need to keep tweaking it until they get it right.
Click to expand...
Click to collapse
Yeah. Advertised speed is 1.3ghz on multiple cores n 1.4ghz on one, in performance mode.
Well said!
To be honest, I'm more concerned about the increase of idle speed and how it will affect battery life.
Just like they removed GPS from the product specs, they're going to have to ammend the "up to 18 hours battery life" part now too...
demandarin said:
1.4Hgz only cuts on when absolutely needed. It doesn't just cut on for hell of it just because you put it in Performance mode. It still stays lower clocked till you do something to need a higher clock. Plus the 1.4Ghz will only be on 1 core. Any other time you will get all cores at 1.3Ghz. Which better or more powerful than just [email protected] 1.4Ghz. People are to quick to jump to conclusions. They get CPU spy for the first time and see something then quick to jumpbto conclusions with nothing tobreally back it up. If you just messing around on home screen, of course the 1.4ghz speed isn't going to cut on. PLAY shadowgun or something n watch that speed easily cut on. Plus when 1.4ghz does cut on, its only briefly. Then other cores take over. Unless the game is taxing it enough to need it on longer.
This is nothing to be pressed about at all. People not understanding how these cores really work or how and when they cut on. Hasn't even been 24hrs. Since update n people already jumping to conclusions. If you so concerned about speed then use ATP tweaks to enable 1.6ghz overclock. Still works just fine n all speeds enabled.
Click to expand...
Click to collapse
Just cut it on
Even if it has absolutely zero impact on usability and battery life, i would still like a technical explanation from Gary as to why the frequency stepping was changed.
Keep in mind, a code change in the stepping could cause CPU spy to report incorrectly.
Probably because the devices worst about crashing would do so both at the lower and higher frequencies, so they just cut 'em both off.
Sent from my HTC Vision using xda premium
What do you mean "cut it on"?
demandarin said:
Just cut it on
Click to expand...
Click to collapse
seeknom said:
Well said!
To be honest, I'm more concerned about the increase of idle speed and how it will affect battery life.
Just like they removed GPS from the product specs, they're going to have to ammend the "up to 18 hours battery life" part now too...
Click to expand...
Click to collapse
It makes sense that there is a possibility they increased the idle speed to address sleep of death or screens not cutting immediately on. Or actually lowering the top speed to reduce random rebooted for reasons they can't figure out. I can see reading the idle speed. Lowering top speed seems unlikely. Even still people that are rooted who are that concerned with speeds can easily just use ATP tweaks or ViperControl to increase speeds. If someone concerned about increased idle speed or the lowest speed, then just use System tuner. Then set the lowest speed to what it was before. Be warned though, your device just might not wake up..lmfao. you can also experiment with the different modes n ATP tweaks, vipercontrol, or in system tuner. Just pay attention to what you doing and do stuff in increments.
---------- Post added at 09:56 AM ---------- Previous post was at 09:51 AM ----------
seeknom said:
What do you mean "cut it on"?
Click to expand...
Click to collapse
I just activated ATP tweaks turbo2 mode, which is 1.6Ghz overclock, to show other speeds enabled. That if someone really wants higher speed they can use this and then use system tuner to back off the Max speed to 1.4Ghz. This will override anything Asus might have possibly done. If they did reduce the speed there's probably a really good reason why. People can't have cake and eat it also. Rebooted n wake up issues resolved and highest speeds possible. So someone can simply manually enable higher speeds if its not activating long enough for you.
---------- Post added at 10:01 AM ---------- Previous post was at 09:56 AM ----------
tdrussell said:
Even if it has absolutely zero impact on usability and battery life, i would still like a technical explanation from Gary as to why the frequency stepping was changed.
Keep in mind, a code change in the stepping could cause CPU spy to report incorrectly.
Click to expand...
Click to collapse
True.....we will have to see. I'm not worried as methods I described can easily override Asus change. Plus I run on overclock most of the time anyways now. People have to tax the system more for a longer period of time to really know. Play some hardcore games and then check CPU spy. People swiping homescreens n think higher speeds will activate. It will only do so when needed.
Hi guys i finally managed how to root a s2 duo the dramatic battery life.
I installed the Setcpu and underclocked my s2.
But it only makes 10 hour or even less. My brother has also a S2 and his battery life goes like 2 and a half day without root just a stock rom. He uses his s2 only to whatsapp and internet browsing like hell just like me. And his batttery life is 2 days
My question how do i get that battery life like 2 days, its really annoying charging the s2 when you want go outside with a low battery.
I got also screenshot of my S2 setting and info:
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
PS: My brothers display percent is also like 50 precent like me.
(Sorry for bad english)
so your brother uses his phone to whatsapp and browsing only?
you must be doing so much more stuff to manage to drain it in 10 hours....hell even i get 1d 12hours at least with my ICS rom
also how does he do the whatsapp thing? via 3g or wifi
we need so much more details in order to give you a decent help
the amount of apps/widgets also decrease/increase ur battery life
Download betterbatterystats and check if your phone goes in to deep sleep..
Otherwise you might have some app or something else keeping it awake all the time even if the display is off.
RorixRebel said:
so your brother uses his phone to whatsapp and browsing only?
you must be doing so much more stuff to manage to drain it in 10 hours....hell even i get 1d 12hours at least with my ICS rom
also how does he do the whatsapp thing? via 3g or wifi
we need so much more details in order to give you a decent help
the amount of apps/widgets also decrease/increase ur battery life
Click to expand...
Click to collapse
From my experience, 3G is the battery killer. When I am on 3g network I get 9 hours until the phone shuts down. THis is with email checking when it is pushed ( 20-30 messages) and about 15 minute of calls and random 4sq check ins.
When I am on GSM only, connected to EDGE rather than 3g i get a bout 1.5 days with a lot greater use.
If I am at home all day and stay connected on wi-fi, I seem to get the best battery usage.
Coming from a BlackBerry this is a bit strange as the Wi fi was the one that used to drain the battery.
My brother uses his s2 always on the wifi and outside of course the 3g signal.
semitor said:
Download betterbatterystats and check if your phone goes in to deep sleep..
Otherwise you might have some app or something else keeping it awake all the time even if the display is off.
Click to expand...
Click to collapse
I just installed betterbatterystats how can i check it if it goes too deep sleep
Check the thread in my signature. 3d battery life! Always in wifi and in 3g (or 2g if i need only whatsapp/mail).
Pako010 said:
I just installed betterbatterystats how can i check it if it goes too deep sleep
Click to expand...
Click to collapse
for checking that i recommend cpuspy much easier to verify if your phone goes to deep sleep
in BBS verify the partial wakelocks and let us know whats in the top 3 positions
and how many counts that activity contains
RorixRebel said:
for checking that i recommend cpuspy much easier to verify if your phone goes to deep sleep
in BBS verify the partial wakelocks and let us know whats in the top 3 positions
and how many counts that activity contains
Click to expand...
Click to collapse
The app is showing this infoo
Hi,
You could try a different rom aswell, there is quit some difference on battery life for the different roms out there.
Bade90 said:
Check the thread in my signature. 3d battery life! Always in wifi and in 3g (or 2g if i need only whatsapp/mail).
Click to expand...
Click to collapse
Thxx my s2 using the 100mhz more than the 500 and alsoo configurated to the info of the topic.
i use siyah kernel and voltage controle
lol the lulzcrative setting is wonderfull
Everybody thnx it sure helped to my battery drain. I flashed the kernel to the latest siyah version avaiable and then i installed voltage control and 2nd core app. And folowed the instruction topic of voltage control. Everybody thnxxx a lot.
Hi everyone
my battery never hold me a day without charging it
I use wanamlite rom 2.3.6
please how can i improve the battery life
when I am on gprs connection and the speed is slow about 10 to 25 kb/s
browsing by opera mini 2 to 4max total browsing ,usuing what's app, messaging via go sms about 100 sms send and receive,
and some othet things is that normal.
Hi everyone
my battery never hold me a day without charging it
I use wanamlite rom 2.3.6
please how can i improve the battery life
when I am on gprs connection and the speed is slow about 10 to 25 kb/s
browsing by opera mini 2 to 4max total browsing ,usuing what's app, messaging via go sms about 100 sms send and receive,
and some othet things is that normal.
Click to expand...
Click to collapse
Hi. This is rather long and I wasn't prepared as I wrote but just simply went on with every thought. Skipping to the summary, where the bold text is, is useful for the majority of people and the entire post is intended for new people though there are some things I haven't yet seen discussed here.
I was actually reading the whole thread which dates back a bit rather slowly and wanted to give some of my experience with my battery.
First off, it seems pretty typical that when a phone is struggling to keep a connection or when there's a lot of fluctuation on 3G, then battery suffers. This was something that was really apparent for me on the SGS2 since I could more easily isolate that one issue and have a look. At the same time, though, I've found that after switching to a good solid provider in my area, who has a consistency of signal and an abundance of towers etc etc, that it made far less difference than prior if I was on 3G or 2G. I think that its actually now negligible and although I haven't tested being on 2G alone since I no longer need to, I suspect that the battery savings isn't going to effect me personally as I keep full bars or close to it most of the time now. If I go back to the previous carrier then of course it would and I'd have the same problem all over again and would be pissed off and keeping myself restricted to GSM, which sucks.
Another thing that I've found, is that every phone really must be uniquely different when it comes to undervolting, setting the governer, and even more specifically just using a custom kernel opposed to a stock kernel, or for root and CWM, one that's compiled as stock with the root goodies added by Chainfire.
I really should break this into 2 separate things.
To simplify the last paragraph, its commonplace, for example with intel or AMD processors that they might all get the same production line process but that there's such a narrow margin for error, that what's deemed acceptable for various speeds and voltages are then labeled as differant processors when they go to market, so the 3 ghz version of a processor is often the same factory result as the 1.5 ghz version and there just happened to be limitations put on the 1.5 ghz version afterward so that it simply could not go any faster as it wasn't the best batch of cookies to come out of the oven, or perhaps the L2 or L3 cache didn't turn out on one batch so part of the thing gets zapped off by a laser and marketed under a differant name.
Getting to the point, there's a lot of custom kernels and various threads about how to get the most out of these kernels. For some people this is the gold nugget but in my case it happens to be that if I want to be rooted and have the best power management while pulling adequate benchmarking numbers then the smartest thing I can do is just stick with the kernel that matches my phone that Chainfire has compiled. As I understand it he uses the same configuration file and complier options as Samsung (or gets as close as he can to it), and adds the components needed for the rooting process.
Since I'm not very smart, and have tried to save battery using Siyah various times with various settings of my own and posted by other people, only to find that my phone's personal temperment is to drain faster, I still continue to try other things and most recently the Red Pill kernel.
While trying a variety of settings with Red Pill last night and benchmarking over and over again, I found that quite simply there's a lot going on which effects the performance of the benchmark score, but even more importantly, and this took a while to notice, that undervolting to levels that didn't by any means cause crashes did not mean better battery or better benchmarks. Where it did benchmark better I found that in my case there can be a huge placebo effect because the in between scores aren't getting looked at as much, for example if I undervolt too much at 500 800 and 1000 then my phone tends to not even use those steps as much and the obvious result of this is going to be too much jumping straight from 200 to 1200 even if I'm just screwing around and flicking the home screens back and forth. This obviously will not save battery. Further to that, I felt that although the GPU can handle a larger undervolt at both settings and a larger undervolt at the lower one, that my phone's performance was much better when both of those things had less of a gap, both in speed and in voltage and went less aggressively there as well. I still have the Red Pill kernel installed and think I have reasonable values however it took quite a lot of testing, hours of it, and the jury is still out and for the most important reason as it applies to my phone which is as follows:
When I first got the phone there was a lot of talk about calibrating the battery. There's apps that will wipe a file (yes I know this is very repetitive for a lot of people but I'm writing it for those that are new) that has to do with battery calibration. You can also do it from recovery in CWM. The file will also get wiped sometimes by something you flash, for example cleaning scripts, some kernel installs might do this (I'm not sure if its wise or necissary when using a new kernel or ROM; I leave mine alone now unless something else does it).
I spent a lot of time trying to get better battery life by calibrating my battery the "proper" way, when it was completely charged, and even when it was just about to die entirely. This has always been not only fruitless, but counterproductive. I've read that the file gets deleted automatically when the battery is full and the power cable is unplugged but I'm skeptical about that as I've found that I get quite noticeably better (significant in my case) battery life when I leave it alone and let it do its own thing for at least 3 charges. That doesn't mean discharging it entirely 3 times, but that just using it as normal a few times brings it into its own all by itself. Calibrating the battery seems (to me) to actually screw this up which means another few times of use before I'm even with a clue of how a kernel (new ROM or custom kernel) is going to be on the battery.
To summarize my experience:
- Custom Kernels appear great for some people but every chip on every phone is differant. In my case sticking close to stock has been the way to go. The best I have so far done is adjusted the conservative governor to be a little snappier than default. I quite honestly got very irritated at one point at all the kernels that pop up and all the settings threads because I expected that I was going to have the same experience as other people were reporting.
- Undervolting sounds effective in theory, but its important to look at how your phone uses various frequencies according to your habits and make sure you haven't fallen into a sitiuation similar to mine where undervolting resulted in me skipping up a frequency to quickly and thus using more power. Additionally, reducing the gap between my voltage values and the gap in the frequencies I had created with my GPU settings resulted in better performance and specifically less work for the CPU.
- Battery Calibration may or may not have its place when it comes to new ROMs or Kernels (I don't know), but Calibration in general is not something the phone requires and further to that (in my observation) just makes it drain faster for a few recharges. The longer I don't calibrate it, the better.
I was at one time not pulling a full day out of my phone. Today, admittedly with the samsung 2000 mah extendable battery though, which is 20% more juice, I can pull 24 hours (as a low number to use) with 5 hours of screen time (also a low number; its more like 6 hours screen and more than 24 hours but varies depending upon how I've used it). When I first got the thing I was calibrating a lot and 5 hours of screen time I don't think would have been possible at all. If you think your phone is a lemon, you may very well just need some patients with it. I have had my original battery in it and using it more patiently as well, and although it wasn't long enough to give out accurate numbers I knew that it would have been just fine had I had to continue using the original battery.
One last thing, I typically freeze wifi sharing and Samsung's email and so forth but have noticed in more recent ROMs that these additions are getting to be more and more negligible. I would unfreeze wifi sharing if I thought I would ever use it.
A final thought for benchmarking is to look carefully at how you use your phone first, in respect to what voltages are used most often, then adjusting each frequency individually. If I undervolt too much at my top frequency then the phone doesn't crash, but it also doesn't score as well. It may be better to find the ideal spot by doing each frequency seprately and individually, then worrying about how things scale, as previously mentioned.
I've found that I get quite noticeably better (significant in my case) battery life when I leave it alone and let it do its own thing for at least 3 charges. That doesn't mean discharging it entirely 3 times, but that just using it as normal a few times brings it into its own all by itself.
That is the correct method battery recalibrates itself .The stats play no part in the calibration process or how long a battery lasts .
https://plus.google.com/u/0/1050519...dVxPT#105051985738280261832/posts/FV3LVtdVxPT
jje
I know, the Tegra K1 gets hot. Is this a widespread issue though (some people only report minor warmth)? How hot is too hot? Mine is getting stoopid hot, and I'm pissed because I finally have a unit with almost no issues, save the over heating problem. I know the SHIELD has issues with screen cracking, and the N9 should be safer with the metal band, but something doesn't seem right...
Iboschi said:
the Tegra K1 gets hot
Click to expand...
Click to collapse
This.
combine that with a badly configured cpu governor that boosts the frequencies to 2+ ghz even for simple tasks that do not need all those cpu cycles.
In addition to the whopping 1.5ghz touchboost frequency (WTF google).
After rooting and fixing the last two points, I experience high temps only when playing games.
Blocking Ads also tremendously helps temps while web browsing.
I wouldn't consider it a problem. I also doubt you have a bad unit even though its getting hot. Chances are if you do the exact same things on 100 nexus 9's the temp will be very close. I've been looking through the kernel code for tegra throttling and doing some tests. The tests show it starts to very lightly throttle starting at 70c in my tests. I believe I saw in the kernel there is 3 throttling states basically, light, heavy, and one other I can't remember. Shutdown occurs at right above 100°c.
As far as the governor, I'm sure if it made sense gooe would have lowered it. But, if I remember right this was part of project butter to make the ui smooth, as well as some other things. I don't think touch boost is killing battery too bad, and I'm willing to sacrifice some for a smoother ui anyway.
Thisbis just a hit running CPU, no way around it and its not a defect, its just a side effect of a powerful CPU in this design. I also noticed although it heats up quick, it cools extrememly fast. Like dropping 15-20° in seconds, literally- so overall I don't think this is a huge problem, but if they can make it better, more power to them.
di11igaf said:
I wouldn't consider it a problem. I also doubt you have a bad unit even though its getting hot. Chances are if you do the exact same things on 100 nexus 9's the temp will be very close. I've been looking through the kernel code for tegra throttling and doing some tests. The tests show it starts to very lightly throttle starting at 70c in my tests. I believe I saw in the kernel there is 3 throttling states basically, light, heavy, and one other I can't remember. Shutdown occurs at right above 100°c.
As far as the governor, I'm sure if it made sense gooe would have lowered it. But, if I remember right this was part of project butter to make the ui smooth, as well as some other things. I don't think touch boost is killing battery too bad, and I'm willing to sacrifice some for a smoother ui anyway.
Thisbis just a hit running CPU, no way around it and its not a defect, its just a side effect of a powerful CPU in this design. I also noticed although it heats up quick, it cools extrememly fast. Like dropping 15-20° in seconds, literally- so overall I don't think this is a huge problem, but if they can make it better, more power to them.
Click to expand...
Click to collapse
im really happy to hear about your thought
but if the heat issue occurs many times, will it break the others hardware,
i have seen in laptop, if the graphics card heat too much, it can melt the motherboard and the others parts of the laptop,
im afraid that it will happen to this tablet eventually
i really love my nexus 9, just this heat issue make me incomfortable