MI9 Fingerprint CPU Frequency bug - Xiaomi Mi 9 Questions & Answers

Hello, I have noticed that whenever I remove the charger from the phone after a full charge, the CPU stays idle at 1.04 and 1.06 for big cores instead of 576 and 710. It is fixed after a reboot.
Can anyone else confirm and report back the bug?
10.2.2.26 official global rom.

MysteriousStranger said:
Hello, I have noticed that whenever I remove the charger from the phone after a full charge, the CPU stays idle at 1.04 and 1.06 for big cores instead of 576 and 710. It is fixed after a reboot.
Can anyone else confirm and report back the bug?
10.2.2.26 official global rom.
Click to expand...
Click to collapse
I can confirm that the bug appears ONLY when I use fingerprints for unlock. When I delete all fingerprints, it is gone. (and in addition, still present in 10.2.28 EEA)

MysteriousStranger said:
I can confirm that the bug appears ONLY when I use fingerprints for unlock. When I delete all fingerprints, it is gone. (and in addition, still present in 10.2.28 EEA)
Click to expand...
Click to collapse
This explains why I got better battery hours after I deleted fingerprints! Not that it was bad before, but now it's ridiculously good!
I don’t think this a bug, it probably must be this way in order to wake up screen just by putting the finger on the scanner.

Having fingerprints registered means that the fingerprint sensor is kept active so there'll be higher cpu usage(shouldn't persist minutes after the phone has been woken) and more battery consumption. Removing all fingerprints( i.e. deactivating the sensor) should give ~10% more battery life but I think having the sensor on at all times even with all its hiccups is more than worth it.

kingbasescu said:
Having fingerprints registered means that the fingerprint sensor is kept active so there'll be higher cpu usage(shouldn't persist minutes after the phone has been woken) and more battery consumption. Removing all fingerprints( i.e. deactivating the sensor) should give ~10% more battery life but I think having the sensor on at all times even with all its hiccups is more than worth it.
Click to expand...
Click to collapse
This summer when I started to get the pocket photos, before the .27 was released (yes I was among the ones that got it before it was redrawn) I deleted the fingerprints to get rid of the pocket activity due to the non-existing pocket mode. As .27/.28 came with pocket mode I started to use fingerprint again but discovered that face unlock that I had been using was far more convenient and suited me better despite the somewhat lower security rating. I don’t use Gpay, and do not keep top secret stuff on my phone so I’m happy being rid of the fingerprint sensor. The sensitive economic things are protected by pin-code anyway.
So it was not battery time that drove me to ditch the fingerprint sensor, that extra battery time just came as added value for me!

kingbasescu said:
Having fingerprints registered means that the fingerprint sensor is kept active so there'll be higher cpu usage(shouldn't persist minutes after the phone has been woken) and more battery consumption. Removing all fingerprints( i.e. deactivating the sensor) should give ~10% more battery life but I think having the sensor on at all times even with all its hiccups is more than worth it.
Click to expand...
Click to collapse
If you notice, the CPU frequency increase occurs randomly after a period of time, not as soon as you turn on the phone, which seems like a bug to me. In addition, the frequency does not settle down again, leading to less standby time available. So I don't really understand how an active fingerprint sensor affects the CPU frequency after a while, and not since the boot.

gerhard_wa said:
This explains why I got better battery hours after I deleted fingerprints! Not that it was bad before, but now it's ridiculously good!
I don’t think this a bug, it probably must be this way in order to wake up screen just by putting the finger on the scanner.
Click to expand...
Click to collapse
If that was the way to be, the frequency should be that high even right after booting.

MysteriousStranger said:
If that was the way to be, the frequency should be that high even right after booting.
Click to expand...
Click to collapse
No, that does not have to be true. I can think of some reasons why CPU speed can't slowdown in order to instantly start the scanner, like for example the more services and apps are started the more CPU power is needed to make the scanner snappy enough. So that may explain why right after boot the CPU may be able to enter a lower idle state. It is in fact true that if you turn off the fingerprint reader you will not have the higher speed at idle state at all, I tested this.
Perhaps there will be a new firmware with a somewhat slower response to the fingerprint scanner that allows the CPU to enter the lower idle state to save battery power. Who knows!

I noticed this as well right from the first firmware update I used (EEA). It's been fixed after upgrading to Android Q's beta modified by Xiaomi.eu, but I don't see much, if any, improvements in battery life, while also witnessing a giant storage performance regression. Hella disappointed.

gerhard_wa said:
No, that does not have to be true. I can think of some reasons why CPU speed can't slowdown in order to instantly start the scanner, like for example the more services and apps are started the more CPU power is needed to make the scanner snappy enough. So that may explain why right after boot the CPU may be able to enter a lower idle state. It is in fact true that if you turn off the fingerprint reader you will not have the higher speed at idle state at all, I tested this.
Perhaps there will be a new firmware with a somewhat slower response to the fingerprint scanner that allows the CPU to enter the lower idle state to save battery power. Who knows!
Click to expand...
Click to collapse
I get your point but the fact is that in never goes into these lower frequencies again, even after unlocking the phone and idling with screen on, which is strange. I don't get it, really. Plus, you are not able to see in real tilme what process is hooking up CPU resources, as you could normally do in older android versions by using an app.

Ok, if the frequency never locks in the high range again and the scanner starts just as snappy I guess you are right that it must be a bug.
Skickat från min MI 9 via Tapatalk

I disabled fingerprint auth, but I don't seen any improvements in battery life.

I notice it on mine as well right now, small cores sitting at 1036 MHz and big ones+super one at 1056 MHz. Can't really put my finger on what triggered it as I've been using the FP sensor all this time and I've checked many times and the cores were at normal clocks (576, 710 and 825). I killed all running apps but it still sat like that until I restarted... at least now I can understand why sometimes my overnight consumption goes from 1% up to 5% even though I always close/kill all apps.

kingbasescu said:
I notice it on mine as well right now, small cores sitting at 1036 MHz and big ones+super one at 1056 MHz. Can't really put my finger on what triggered it as I've been using the FP sensor all this time and I've checked many times and the cores were at normal clocks (576, 710 and 825). I killed all running apps but it still sat like that until I restarted... at least now I can understand why sometimes my overnight consumption goes from 1% up to 5% even though I always close/kill all apps.
Click to expand...
Click to collapse
Try rebooting your device and you will see what the actual idle frequencies should be, then monitor it for a day or two! Can't find what triggers it but It's bugged for sure.

apoklyps3 said:
I disabled fingerprint auth, but I don't seen any improvements in battery life.
Click to expand...
Click to collapse
Reboot your phone after disabling it, otherwise the frequencies stay the same

Related

Did undervolting ruin my reception??

This is a lengthy post, but please read it. These are copies of my posts on androidforums.
Post 1:
I happily bought my Nexus One 3 weeks ago from someone on craigslist. I checked out the phone, everything worked, and it's been absolutely great. I'm in the Detroit metro area and have T-mobile. I bought it, rooted it, installed cyanogen's 5.0.5.3 ROM and installed the overclocked-undervolted kernel, and it's been heaven. I'm in love with this phone. Then I drove with my dad to Tupelo, MS, on Sunday the 11th and throughout the drive the signal would disappear and come back, repeatedly. I figured this to be because of T-Mobile's terrible coverage. While in Tupelo, I hardly EVER got a signal...even for calling, forget about data. I knew this would be rectified when I got back to Detroit. Today I flew back to Detroit, and finally got 3G again and full signal and I was ecstatic that I could finally USE my phone again! Well that didn't last, now I'm at home (where I normally only get a GPRS signal), and I'm not getting ANY signal, whereas my sister and mother who are sitting next to me, who are on the same family plan with T-Mobile with me, are getting full GPRS signals on their blackberry phones. I've rebooted the phone multiple times, wiped and re-flashed the cyanogen ROM...no luck. I've even put in my sister's SIM card into the Nexus...still no signal. Meanwhile my sister and mother are getting full signal on their phones, even if its just GPRS. I'm bummed PS > No, the phone is NOT in airplane mode.
Post 2:
I did flash the orignal stock rom, and update the radio (tried 3 different radio images)....none of it helped unfortunately. My Nexus One will work fine when I'm in an area of good signal, namely where I get 3G. But when I'm at home, no dice. And the reason this is really frustrating is because everyone else in my house, on the SAME carrier, with the SAME plan, is getting a full signal. This really sucks because not only is the phone useless as a phone, but the phone runs hot (31 degrees C at idle, instead of 24 C at idle, which it what it always was before), because it's always looking for a signal (and I know this because Call Standby, which used to be 3% of battery use, is now the highest battery drain). And until I left for the trip, this was not a problem at all. So my only conclusion is that while in Tupelo, MS, the antenna/radio fried a bit, constantly looking for a signal for 4 days straight, and so now it can only pick up signals if they're very strong. Either way, I'm gonna have to sell it to someone who wants it for repair parts, and go back to my old blackberry pearl 8100
Post3:
Sent the phone in to HTC.
This is what they told me. I could go two routes: Replacement or Repair. Naturally I'd just want them to send me another certified working phone (refurbished, they don't send new ones). However, with this option, since I wasn't sure if I was under warranty or not (having unlocked the bootloader and rooted the phone, and the customer service guy said they could only determine that once the phone was with them), if they determine that you are NOT under warranty, you will get a replacement, but the cost of any repairs that are needed to the unit you send in, you WILL pay for them, no matter what. If they determined that I am still under warranty and that they can help me free of charge, then obviously this would be the best option. But not knowing if I'd be charged, and how much I'd be charged, I took the repair option.
With the repair option, they give you the option whether you want to authorize the charges once they know how much it will cost (if they can't do it for free). Then you can decide if it's worth the money they're saying, or you can tell them to just send it back, in which case they'll just charge you 28$ for the diagnosis and shipping cost to ship it back. I took this option worried about them telling me that I'd have to pay 300$ or something crazy for the repair if I took the replacement option.
I sent it in on Friday the 16th, and I got a mail from HTC on Monday 10 AM the 19th, telling me they had received the phone. Then at 5 PM I get an email from them saying it's been shipped back. That's it, no other info. I thought, how can they fix it that quick? Or did they mistakenly not even look at it and sent it back with a batch of repaired phones? Or did they just send me a replacement?
I got the phone yesterday the 20th at noon, and opened the package, and it was in fact my phone (it still had the screen protector I'd put on it, albeit they'd completely ruined it, because it had moved and now had all these large air bubbles and dust in it).
I powered it on and the bootloader was locked again, and completely returned to stock. I was at my house at this time, where I've been getting the signal problems (while everyone else at my house, with the same carrier and same plan have been getting full GPRS signal). I pop in my SIM card and YES! It was working, and working well. Made a few calls, walked all around the house, and the calls were clear and strong.
Now at this point, it was a bittersweet feeling, because while my phone was working, I was afraid to root it again. I guess I felt that somehow rooting my phone and intalling Cyanogen's mod 5.0.5.3 must have done it (though I had that ROM on for a while before my trip, when it got messed up). But then I coudln't handle the slower speed, the lack of quicker pulse trackball notifications, wireless tether, titanium backup...and so on. So I rooted my phone, installed cyanogen's mod, downloaded all the apps I had.
Except ONE thing. The only thing I can remember doing before my phone's signal started getting weak was installing an overclocked+undervolted kernel to the phone. I know they say that when your phone is in a weak area, it requires more power to work, so maybe undervolting it is what was doing it, but this doesn't make sense, as I reflashed the cyanogen mod without the kernel, and the stock rom and neither helped. Wouldn't they overwrite the kernel with the kernel's in the ROMs? Either way, that was the last thing I did to my phone before it started messing up, so while I put on every app (including setcpu) back on my phone, I didn't put the kernel on it.
So far so good, it's been a whole day, and it's worked fine at my house. All I can say is I'm enjoying my phone again. Thanks to all of you, and I hope this thread helps out someone else in a similar predicament.
Now that my phone is working fine, and the only thing I have not put back on my phone is the undervolted+overclocked kernel, I'm wondering if that's what ruined my reception at home, which is a weak signal area.
The thing that confuses me is that I thought when you flash a ROM, it over-writes the kernel with its own, but even though I flashed the stock and cyan's ROMs after full wipes, it didn't fix the problem, so I'm assuming that kernels don't get over written?
Since you guys are the experts, what do you think happened to my phone?
I am now afraid to even TRY to do anything to the phone. I really wanna try out Modaco's desire rom to try out sense UI, but I feel like I'll just risk my phone's reception again.
No, kernels do get overwritten. But, in some cases, caches can causes issues with the system interaction with the radio. Undervolting WILL screw with your radio.
bobtentpeg said:
No, kernels do get overwritten. But, in some cases, caches can causes issues with the system interaction with the radio. Undervolting WILL screw with your radio.
Click to expand...
Click to collapse
Why is that?
bobtentpeg said:
No, kernels do get overwritten. But, in some cases, caches can causes issues with the system interaction with the radio. Undervolting WILL screw with your radio.
Click to expand...
Click to collapse
What cache are you talking about?
Unfortunately because of this experience I don't feel like messing with my phone again :-(
i would also like to hear why undervolting the processor would affect radio signal, as i did not think that the power supplied to the transceiver would be any different.
Running many devices at voltages outside of optimal parameters shortens their lives even if they function normally for a while. Under volted motors can burn up, which is why power brown outs can be especially dangerous. Could low voltage impact a radio? Yep - I learned that lesson in playing with early solid state ham radios. The radio is especially susceptible if the radio voltage is already automatically adjusted to conditions.
I don't what exactly is under volted with these UV kernels. I assume the CPU is under volted, but what else? Who has physically tested voltages at various places to confirm expected results?
One thing is for sure - if nothing other than the CPU voltage is adjusted with the UV kernel, then it's highly unlikely under volting caused your problem.
But if the voltage adjusted is more systemic, then the UV kernel could absolutely have an impact on the life of any component in the phone.
Just to add to what's already been said... I have yet to see solid evidence that undervolting really makes a substantial impact on battery life. If you really want to root, don't overclock the processor or undervolt it. The extra 113mHz you're getting isn't worth it, IMO. The phone's already fast stock...so with the .32 kernel it's probably more than adequate.
uansari1 said:
Just to add to what's already been said... I have yet to see solid evidence that undervolting really makes a substantial impact on battery life. If you really want to root, don't overclock the processor or undervolt it. The extra 113mHz you're getting isn't worth it, IMO. The phone's already fast stock...so with the .32 kernel it's probably more than adequate.
Click to expand...
Click to collapse
Right.
Further, if you expect to improve battery life by lowering CPU voltage, you also need to lower the frequency of the clock. Ohms law applies - it takes X amount of watts to do any task with electricity, that amount varies from device to device. But the wattage required to do a specific job with a specific device remains constant. There are two variables that determine power (wattage) - volts and amps. Decrease voltage, and the device will draw more current (amperage) to do the task.
In the case of a CPU, in you overclock (expecting it to perform at a higher level) then you need to supply more power. If you raise the clock speed and lower the voltage, then the current MUST increase accordingly. I don't understand how lowering voltage without lowering the workload (clockspeed) could possibly improve battery life, and I don't see any valid testing to prove that whatever is being done really works.
All that said, I still have no idea how available voltage mods are impacting the mean time between failure rates of handsets they are installed on. Benefits are anecdotal IMO, until I see some valid test data and a better explanation of what's going on.
uansari1 said:
Just to add to what's already been said... I have yet to see solid evidence that undervolting really makes a substantial impact on battery life. If you really want to root, don't overclock the processor or undervolt it. The extra 113mHz you're getting isn't worth it, IMO. The phone's already fast stock...so with the .32 kernel it's probably more than adequate.
Click to expand...
Click to collapse
Is there a way to check the kernel version in the phone?
Also if I want to try out a custom ROM, how do I make sure that I keep the .32 kernel (which I'm assuming you mentioned because it is the default kernel) ?
ksc6000 said:
Is there a way to check the kernel version in the phone?
Also if I want to try out a custom ROM, how do I make sure that I keep the .32 kernel (which I'm assuming you mentioned because it is the default kernel) ?
Click to expand...
Click to collapse
settings > about phone
and, i'm running intersectraven's avs kernel. it does seem to help battery life. it adjusts voltage based on cpu demand
attn1 said:
Right.
Further, if you expect to improve battery life by lowering CPU voltage, you also need to lower the frequency of the clock. Ohms law applies - it takes X amount of watts to do any task with electricity, that amount varies from device to device. But the wattage required to do a specific job with a specific device remains constant. There are two variables that determine power (wattage) - volts and amps. Decrease voltage, and the device will draw more current (amperage) to do the task.
In the case of a CPU, in you overclock (expecting it to perform at a higher level) then you need to supply more power. If you raise the clock speed and lower the voltage, then the current MUST increase accordingly. I don't understand how lowering voltage without lowering the workload (clockspeed) could possibly improve battery life, and I don't see any valid testing to prove that whatever is being done really works.
All that said, I still have no idea how available voltage mods are impacting the mean time between failure rates of handsets they are installed on. Benefits are anecdotal IMO, until I see some valid test data and a better explanation of what's going on.
Click to expand...
Click to collapse
Sorry I saw this and felt its nessisary to comment on it.
When you overclock a processor a general rule is that you increase the voltage to the processor to provide stability. (Since more clock cycles typically require more power. Not every chip is manufactured the same, it may be the same model, but they are fabricated differently. One processor might be stable with less voltage compared to another one from a different batch of processors.)
With the N1, Google ensured stability by increasing the voltage sent to the processor. As time has shown, the roof Google put on the device for stability could be lowered and still have a stable phone. (Hence 1000mv to 800mv stable.) Underclocking or undervolting WILL NOT DAMAGE a processor in no way shape or form. It doesn't happen. ONLY when you overvolt a processor do you risk damaging it. (More voltage = more heat) (I'm not saying you can't brick phones from doing it, just simply saying you are not damaging the hardware itself.)
Typically every device inside a phone has its own regulated voltage, if the radio voltage hasn't been touched, undervolting your processor shouldn't affect the stability of the radio.
timothydonohue said:
settings >, i'm running intersectraven's avs kernel. it does seem to help battery life. it adjusts voltage based on cpu demand
Click to expand...
Click to collapse
"seem to" is sort of anecdotal. Battery life is a hard thing to measure on something like a mobile phone, because the test parameters change so much from day to day in normal operation.
In order to test battery usage from kernel to kernel properly, the phone needs to be fully charged and then run a looping task that logs with a timestamp. All other system tasks and features running as this task runs need to be documented, and the phone cannot be used otherwise. When the battery dies, it should leave a log that notes the time the task started and the time the task ended, provided the phone didn't crash mid-write.
If the clock frequency is determined by load, then the loop needs to be adjusted so that the CPU can idle longer between loop cycles, this way a matrix can be built measuring battery performance under different loads. The test should be rerun with several different loop intervals, and each result documented carefully.
Then change out the kernel and repeat each test exactly the same way. Proper testing is meticulous work, and until it's done, we have no idea what, if any impact an under volted kernel has on battery life.
This may not be the best way to test, and it surely isn't the only way, but something with more control than day to day phone use is needed to prove that battery life is actually improved.
I have yet to see any kind of test designed, and I have no clue how it could work, unless undervolting results in under clocking. Same work has the same energy requirement. If a lower voltage means a lower CPU clock, then yes, it can work, like speed stepping does on a PC. Under heavy use, an overclocked CPU is going to use more energy, regardless of how voltage is applied while it is overclocked. Since the screen uses a pretty hefty percentage of power when in use and with all the data pushes going on with these devices, I don't see overclocking helping much unless the device is throttled down and sleeping far more than it is used.
Ohms law.
archangelugp said:
Sorry I saw this and felt its nessisary to comment on it.
When you overclock a processor a general rule is that you increase the voltage to the processor to provide stability. (Since more clock cycles typically require more power. Not every chip is manufactured the same, it may be the same model, but they are fabricated differently. One processor might be stable with less voltage compared to another one from a different batch of processors.)
With the N1, Google ensured stability by increasing the voltage sent to the processor. As time has shown, the roof Google put on the device for stability could be lowered and still have a stable phone. (Hence 1000mv to 800mv stable.) Underclocking or undervolting WILL NOT DAMAGE a processor in no way shape or form. It doesn't happen. ONLY when you overvolt a processor do you risk damaging it. (More voltage = more heat) (I'm not saying you can't brick phones from doing it, just simply saying you are not damaging the hardware itself.)
Typically every device inside a phone has its own regulated voltage, if the radio voltage hasn't been touched, undervolting your processor shouldn't affect the stability of the radio.
Click to expand...
Click to collapse
As voltage decreases at a given CPU frequency, amperage must increase. An extension cord used properly should be cool. Overloaded, voltage will drop and current is increased to meet the power requirement of the load, increasing heat. So reducing heat is not as simple as lowering voltage - the load must be lowered.
Your point about each device having it's own voltage regulator is well taken, and while they will protect individual devices (like radios) from damage from voltage variances, there is still no proof that an overclocked/undervolted CPU improves battery life with moderate use, however power is applied. To lower power requirements, the workload must be lowered accordingly.
You guys seem pretty knowledgeable about all this. What do you think happened to my phone that I stopped getting reception at home but as soon as I got it back from HTC it was fine? All they did was un root the phone and re lock the boot loader. What did they do that I didn't do by re flashing the ROMs and trying 3 different radios? They didn't seem to have messed with any hardware.
ksc6000 said:
You guys seem pretty knowledgeable about all this. What do you think happened to my phone that I stopped getting reception at home but as soon as I got it back from HTC it was fine? All they did was un root the phone and re lock the boot loader. What did they do that I didn't do by re flashing the ROMs and trying 3 different radios? They didn't seem to have messed with any hardware.
Click to expand...
Click to collapse
It's very unlikely that they simply unrooted and relocked the bootloader. Is the serial number of the system board the same? I suspect that HTC determined that the radio was defective and the system board was replaced, explaining why your screen protector was dislodged. You weren't charged for a new system board because they attributed the failure to hardware and gave you a pass. With a new system board the phone was once again boot loader locked.
It was probably a bad radio module completely unrelated to anything you did. It happens, and that's my guess at this point.
i'm also running intersectRaven's undervolted kernel and have not had any issues with signal on my Nexus 1 / at&t version. it's been a solid 4-bars with "H" speeds. with the T-Mobile version I would consistently experience 3G signal drops and reconnects, but did not think it was due to the kernel. battery life was good but usually suffered due to signal issues. fyi, i created a Nandroid of my T-Mobile version and restored to my at&t version. the only difference is the radio. and i can say through my testing that my battery is lasting longer; probably due to better signal strength. i really appreciate the feedback and comments from those more knowledgeable than me on the kernels. i'll continue to monitor my device going forward.
attn1 said:
Ohms law applies - it takes X amount of watts to do any task with electricity
Click to expand...
Click to collapse
Please state your source, as that isn't ohms law at all. "Ohm's law states that the current through a conductor between two points is directly proportional to the potential difference or voltage across the two points, and inversely proportional to the resistance between them."
Or, in another form:
v=ir
If you drop the voltage, the right side of the equation must also go down. We can assume that resistance stays approximately the same (this isn't true, but it is close enough). Thus, current (i) must also go down.
The power equation is p=iv. Rmember, both current AND voltage go down. Thus, power must go down as well. Substituting for i, we get p = v^2 / r. So power is directly proportional to the square of voltage. Decreasing the voltage from 1275 mV to 800 mV actually reduces power consumed by over 60%. Wow.
Of course, you can't actually reduce voltage by that much and keep the same clock speed. However, my point was that reducing voltage reduces power, which remains true.
attn1 said:
But the wattage required to do a specific job with a specific device remains constant.
Click to expand...
Click to collapse
That's wrong. The reason is not all power goes into doing useful things like switching transistors, a lot of it is lost for charging/discharging the capacitances of on-chip wiring, I/O busses, and parasitic capacitances as any transistor has them (dynamic power, used while switching). The dissipation loss is P = C*V^2*f, i.e., it is linear in the frequency and quadratic in the voltage. See http://en.wikipedia.org/wiki/CMOS#Power:_switching_and_leakage for details. This is the main reason why undervolting saves power.
This article is a nice summary: http://en.wikipedia.org/wiki/Dynamic_voltage_scaling

[Q] CPU running at high clock most of the time

Updated section:
After clearing cache through recovery and doing a hard reset, the uAh has gone back to 3142000. Standby time seems good with the battery losing 7% overnight. It does charge quite quickly from 0-100 in about 2h. The previous heating issue remains with cpu monitors showing that the cpu jumps to high clock when I touch the screen. I don't see any change in which apps use the most battery with the screen taking up the most as before the problem started.
The heat appears to be what is draining the battery but I can't find any single app that causes it to ramp to max.
Pictures:
No-frills is set to stock settings to see how the phone handles the actions
2nd picture is when phone is idle
3rd picture is immediately after swiping through a few tabs and scrolling through some pages in the app which causes cpu clock and load to increase
Old sectionFor when battery uAh was low)
I've had this Z1 for about 11 months. I can only get around 2.5h on screen time with regular usage. The battery uAh was only 1848000 Recently it fell to the 1700s. Calibration appears to reduce the value further. The phone seems to heat up in about a minute or two no matter what app is running.
The phone is rooted on stock 4.4.2
I have tried updating to Lollipop before but received an error due to rooting, so I don't think anything changed.
bump
alantay11 said:
bump
Click to expand...
Click to collapse
sounds like it's time for you to replace your battery.
You may try to disable the "input boost frequency"
By default when you toch the screen, cpu frequency of two cores instantly jumps to around 1400 to preventing any lag but it seems to use more power and causing heat
I have disables this and since then I'm no longer have any heat issue and also the battery life is so much better now but you may exprience a little lag while scrolling on heavy web pages but thats all
Btw i'm on .157 and using Kernel Adiutor for cpu tweaks
You may disable this by the app mentioned above or by editing sys.somc.touch_perf_kick=1 from 1 to 0 in buid.prop or adding this line

Huawei p9 Overheat protection feature

i heard that the huawei p9 has their unique overheat protection feature which if the phone gets hot the phone will lower its performance level. is there any way to disable this, my phone gets slow even if its not that hot. i cant play games for too long because it lags after a few mins of playing. i hope huawei makes this feature optional
rhaegon36 said:
i heard that the huawei p9 has their unique overheat protection feature which if the phone gets hot the phone will lower its performance level. is there any way to disable this, my phone gets slow even if its not that hot. i cant play games for too long because it lags after a few mins of playing. i hope huawei makes this feature optional
Click to expand...
Click to collapse
Its not a unique feature of the P9, all the CPUs (meaning Desktops computers, Laptops computers, Phones, Tablets, etc.) and also all GPUs (Nvidia cards, AMD cards, MALI, Rouge, etc.), but in general all electronic components will lower their performance if they get too hot. This is called "Thermal Throttling". All components have a maximum working temperature, if you get close to that temperature you might experience glitches or artifacts, nothing permanent, but if you get past the Maximum temperature, you will experience heavy glitching and artifacting, crashes and heavy instability of the system, causing permanent damage too.
So to answer your question, no, you cannot disable the overheating safety feature, and there will also won't be an implementation of a switch for this feature, simply because it would kill your device.
The only solution is to better cool the phone, or reduce the heat, the first is not really a great solution, the second one would require downvolting/downclocking both the CPU and GPU, both of which require a Custom Kernel. As of right now, there are none, so you are stuck the way you are.
noki57oo said:
Its not a unique feature of the P9, all the CPUs (meaning Desktops computers, Laptops computers, Phones, Tablets, etc.) and also all GPUs (Nvidia cards, AMD cards, MALI, Rouge, etc.), but in general all electronic components will lower their performance if they get too hot. This is called "Thermal Throttling". All components have a maximum working temperature, if you get close to that temperature you might experience glitches or artifacts, nothing permanent, but if you get past the Maximum temperature, you will experience heavy glitching and artifacting, crashes and heavy instability of the system, causing permanent damage too.
So to answer your question, no, you cannot disable the overheating safety feature, and there will also won't be an implementation of a switch for this feature, simply because it would kill your device.
The only solution is to better cool the phone, or reduce the heat, the first is not really a great solution, the second one would require downvolting/downclocking both the CPU and GPU, both of which require a Custom Kernel. As of right now, there are none, so you are stuck the way you are.
Click to expand...
Click to collapse
thank you for your response, but my p9 antutu benchmark score is 99k to 65k real quick when performing the 2nd try. isnt that too much? btw when i downclock my phone does it reduce the overheating but still give the same performance?
rhaegon36 said:
thank you for your response, but my p9 antutu benchmark score is 99k to 65k real quick when performing the 2nd try. isnt that too much? btw when i downclock my phone does it reduce the overheating but still give the same performance?
Click to expand...
Click to collapse
Yes, that is a huge drop, but mine behaves similarly. When you downclock your phone you reduce the clock of the processor, which means it needs less electricity to work, which also means that it generates less heat. But reducing the clock of the processor means reducing its performance too, so no, it wouldn't keep the same performance.
noki57oo said:
Yes, that is a huge drop, but mine behaves similarly. When you downclock your phone you reduce the clock of the processor, which means it needs less electricity to work, which also means that it generates less heat. But reducing the clock of the processor means reducing its performance too, so no, it wouldn't keep the same performance.
Click to expand...
Click to collapse
i havent seen any phone that lowers its brightness when overheating, but the p9 does this and then lowers its performance sometimes even if my phone isnt that hot, it looks like a feature to me but im not sure, i hope that this is really a feature cuz if it is, theres a chance to fix it

Is there any way to improve the battery management on mi a2?

Anything is valid. Preferentially without root.
[/COLOR]
wquayson said:
Anything is valid. Preferentially without root.
Click to expand...
Click to collapse
What is your current battery performance? Is your device affected with the fingerprint sensor bug that locks the CPU at highest frequency?
kandarpjha said:
[/COLOR]
What is your current battery performance? Is your device affected with the fingerprint sensor bug that locks the CPU at highest frequency?
Click to expand...
Click to collapse
About 6-8 hours in moderate-high use, I guess. I don't think it is, how can i know?
wquayson said:
About 6-8 hours in moderate-high use, I guess. I don't think it is, how can i know?
Click to expand...
Click to collapse
Install a hardware monitor app like CPU-z or DevCheck. If you CPU frequency is not coming down from the highest value (for 4 small cores: 1843 MHz and for 4 big cores: 2208 MHz), you are most probably affected by the bug. You can disable fingerprint sensor to conserve battery in that case.
wquayson said:
About 6-8 hours in moderate-high use, I guess. I don't think it is, how can i know?
Click to expand...
Click to collapse
If you are talking about SOT, it's quite normal.
You can't really improve SOT.
It depends mainly on two things : Brightness and Kernel settings. Kernel you cant tamper with without rooting/unlockin your phone, so the best improvement can be achieved via auto brighness.
If you are losing too much battery while phone is idle/screen locked, here are some steps you can follow:
Limit background use of an application from Settings/Battery (Even google play services, notifications still come through)
Disable background data use of apps from Data Usage (Especially facebook/messenger ones, notifications still come through via GCM - use to your own likings though)
Clear Google Play Services cache, as it somehows causes battery drain
Disable location and google location tracking
Disable "Data always on" at developer settings. No point of keeping an active connection while on WiFi
Change network to 3G Only, if your 4G coverage sucks, so it drains all your battery
kandarpjha said:
Install a hardware monitor app like CPU-z or DevCheck. If you CPU frequency is not coming down from the highest value (for 4 small cores: 1843 MHz and for 4 big cores: 2208 MHz), you are most probably affected by the bug. You can disable fingerprint sensor to conserve battery in that case.
Click to expand...
Click to collapse
When i quit the app and come back, the frequencys are at the maximum value for a milisecond then goes to 1401MHz(CPU 0-3) and 1113MHz(CPU 4-7). So, I don't think I have that bug. What do you say?
Reptant said:
If you are talking about SOT, it's quite normal.
You can't really improve SOT.
It depends mainly on two things : Brightness and Kernel settings. Kernel you cant tamper with without rooting/unlockin your phone, so the best improvement can be achieved via auto brighness.
If you are losing too much battery while phone is idle/screen locked, here are some steps you can follow:
Limit background use of an application from Settings/Battery (Even google play services, notifications still come through)
Disable background data use of apps from Data Usage (Especially facebook/messenger ones, notifications still come through via GCM - use to your own likings though)
Clear Google Play Services cache, as it somehows causes battery drain
Disable location and google location tracking
Disable "Data always on" at developer settings. No point of keeping an active connection while on WiFi
Change network to 3G Only, if your 4G coverage sucks, so it drains all your battery
Click to expand...
Click to collapse
Auto brightness? Really? I thought that was a battery killer. My brightness is in about 25% most of the time, so, I don't think that is the bigger problem, do you?
Well, thanks both of you guys for the help, and I will follow your instructions(but not the auto brightness one, I don't really like it), Reptant. I will try to remember and report the results at the end of the day.
wquayson said:
Auto brightness? Really? I thought that was a battery killer. My brightness is in about 25% most of the time, so, I don't think that is the bigger problem, do you?
Well, thanks both of you guys for the help, and I will follow your instructions(but not the auto brightness one, I don't really like it), Reptant. I will try to remember and report the results at the end of the day.
Click to expand...
Click to collapse
Yes, that's right. Disable auto brightness, and keep the brightness at around 10%. It is more than sufficient in indoor conditions. Further, disable high accuracy location settings and set it to battery saving or off. You should also turn off WiFi and Bluetooth scanning in location settings.
I personally use a rather far reaching way of force stopping every social media and other news apps after using them (I have never observed any 'misbehavior' as they warn, but it saves a great deal of battery life.). As you can see in the screenshot the battery has lost 7% of charge in 2 hours and it still expects to last for 1 full day.
wquayson said:
When i quit the app and come back, the frequencys are at the maximum value for a milisecond then goes to 1401MHz(CPU 0-3) and 1113MHz(CPU 4-7). So, I don't think I have that bug. What do you say?
Click to expand...
Click to collapse
Min freq should be 633 MHz for the little cluster,it a bug with the 2nd October update, many users are affected,and so i!
https://forum.xda-developers.com/mi-a2/help/frequency-sd660-stuck-maximum-t3832366
Hello, any better result if phone is rooted ? Or any other solution? Thanks a lot.
buythismobile said:
Hello, any better result if phone is rooted ? Or any other solution? Thanks a lot.
Click to expand...
Click to collapse
The suggestions from Reptant above are still valid.
Other than that:
1. Greenify. Set non system apps to shallow hibernation. Try Amplify if you dare.
2. Disable ads. The image/video ads may not occupy a lot of your screen, but they use much more data than text. disabling those ads will reduce the amount of data that the modem needs to download, hence reduce battery consumption. this will have big impact if you surf the web a lot using phone.
3. Change CPU governor to powersave and/or disable some CPUs, reduce GPU max clock speed (you may not like this since it will make the phone laggy)

Not removing thermal throttle, but increasing thereshold? (PerfZ bug)

A month ago or so, i had this random period of time where perfZ started to show "-" for temperature ( i assume its gpu temperature because it doesnt match with battery or cpu temp from another app) and my phone was heating up much more in games but so was the performance was way better.
Normally after thermal throttling kicks in and device reaches heat equilibrium, my battery would sit around 38-39C, cpu at 42-43C. But in that time period where it showed "-", my equilibrium was battery sitting at 44C, cpu at 47-48C. Phone was considerably hotter to touch but you could still game on it. And performance was way better, it would fall to 50fps min instead of 40fps, which is very noticable. (Dont have exact numbers maybe a bit better)
Idk how this happened as it randomly happened one day but does anyone know how to make this work? My device isnt rooted. (s10+ exy)
(Small note: Using game plugins at max performance or custom doesnt help, maybe they arent optimised for dead by daylight, because thats the only game i play)

Categories

Resources