Ok so I learned allot of shell script today...but I have results ...the results I wrote a script to control cpu1 while the screen is on to basically inhibit it below certain cpu utilization levels or at Low clock speeds..
Wrote it 45 minute s ago so far so good and its highly observable ...but I'm gonna give it 32 ish hours if it is working good I will post it
Sent from my SPH-L710 using Tapatalk 2
freecharlesmanson said:
Ok so I learned allot of shell script today...but I have results ...the results I wrote a script to control cpu1 while the screen is on to basically inhibit it below certain cpu utilization levels or at Low clock speeds..
Wrote it 45 minute s ago so far so good and its highly observable ...but I'm gonna give it 32 ish hours if it is working good I will post it
Sent from my SPH-L710 using Tapatalk 2
Click to expand...
Click to collapse
I wouldn't mess with the default Samsung implementation of multicore cpu handling.
It already does everything automatically. Turn off cores, slow them down, all depending on load and temperature. Unless of course you just want to save the battery at the expense of performance. Then you might as well just disable the second core completely at all times.
All these scripts floating around, which are supposed to save battery by handling the second/third/fourth core, are just snake oil. No one has ever conclusively shown that they did any good with sgs3 soc.
I don't know if you monitor the core behavior but coming from a gs2 I can tell you its different and definitely not right....and if you want proof I can post screens later of it misbehaving.
So far in benchmarks or to the naked eye they're has been no I'll effects and I have1 hour screen with 88% battery....so I'd say it works
Sent from my SPH-L710 using Tapatalk 2
And I believe the drivers are wwritten by Qualcomm not Samsung
Sent from my SPH-L710 using Tapatalk 2
Related
I don't know if it is normal for my CPU speed to be so low, I know that it lowers to save battery but it is always so low and I can see the effects it has. Anyone else get something like this. (quadrant standard edition)
yes it's underclocking to use lower battery when not needed ,it's perfectly normal.
don't worry , it'll fire full throttle (1400 MHZ) when you run a game or something demanding.
Use cpuspy with it you can check the times that CPU spends on each frequency.
Sent from my GT-N7000 using Tapatalk 2
Thanks. I was a bit worried because on other devices I tested the processor was always at maximum.
Sent from my GT-N7000 using XDA
Did you at least readed it?
It says Minimum 200Mhz and Max 1400Mhz.
Its how note works. From 200Mhz to 1400Mhz. It scales the CPU speed from 200 to 1400Mhz when you need power.
It also can disable one of the two cores (CPU1 always) to save battery.
It also says that on the galaxy s 2 but that did not go below 1200MHz when i tested it so I was not sure.
Sent from my GT-N7000 using XDA
ZacDerbyshire said:
It also says that on the galaxy s 2 but that did not go below 1200MHz when i tested it so I was not sure.
Sent from my GT-N7000 using XDA
Click to expand...
Click to collapse
Something was wrong with your GS2.
It says 500MHz because that's the current frequency it used when you took that screenshot. If it had said 1400MHz, then it would mean that it used all the power to take a screenshot, which wouldn't be normal.
The CPU isn't working at 1.4GHz 24/7. If it had done that, your battery would have lasted for 1 hour. It uses as much as it needs to, and if an app doesn't require much power, it won't use much power. Simple.
LordManhattan said:
It says 500MHz because that's the current frequency it used when you took that screenshot. If it had said 1400MHz, then it would mean that it used all the power to take a screenshot, which wouldn't be normal.
The CPU isn't working at 1.4GHz 24/7. If it had done that, your battery would have lasted for 1 hour. It uses as much as it needs to, and if an app doesn't require much power, it won't use much power. Simple.
Click to expand...
Click to collapse
use the force choke on him
Guys, check out the new kernel here:
http://forum.xda-developers.com/showthread.php?t=1870259
It is written for the N8000 but I have run it on my 8013 without issues. The prior version of this caused flaky wifi but so far the newer version has been stable for me. If you do get a wifi crash, just turn wifi off then back on and it will reconnect easily. Up to you whether than small inconvenience is worth it for the extra speed.
Was able to use SetCPU to OC my SGN10.1 to 1704 at Performance and run some tests. Nailed a 6308 quadrant first try - wow!!! I have my SDcard OC'd to 4096 as well. Remember to always uncheck "Set at Boot" in SetCPU when testing out new speeds. You don't want to get stuck in a bootloop if you push things too far.
** One point. I found that OC to 1600 crashed my machine. Don't know why (dev bumped voltage to 1425 mv at 1704 - that's probably why), but for me anyway, 1704 seems stable for now.
I take no credit for this work (this is brieuwers doing) - simply passing the word along and letting 8013 owners no that, at least so far it has not melted my rig. Please do not thank me. Just passing along the word.
P.S., I got one bootloop after flashing with ODIN but then it booted fine.
** As always, if you brick your rig it's on you.
UPDATE - This is about 15% faster now and it feels it.
mitchellvii said:
UPDATE - This is about 15% faster now and it feels it.
Click to expand...
Click to collapse
Does that mean it's 15% warmer?
TonyBigs said:
Does that mean it's 15% warmer?
Click to expand...
Click to collapse
Honestly haven't noticed any heat increase.
Sent from my GT-N8013 using Tapatalk 2
Battery life reduction?
TheWerewolf said:
Battery life reduction?
Click to expand...
Click to collapse
Not that I've noticed but I typically use my tablet in my office where it is plugged in so hard to tell.
Sent from my SPH-D710 using Tapatalk 2
there will be a battery life decrease when using the tablet constantly at 100% cpu usage. bumping the voltage guarantees that. as for how much that is i can't tell you but under my normal use my tablet doesn't run full throttle all the time so the extra drain is negligible.
madsquabbles said:
there will be a battery life decrease when using the tablet constantly at 100% cpu usage. bumping the voltage guarantees that. as for how much that is i can't tell you but under my normal use my tablet doesn't run full throttle all the time so the extra drain is negligible.
Click to expand...
Click to collapse
What one has to wonder is does the very act of the cpu constantly throttling up and down and measuring what speed is best actually burn more fuel than just remaining at a high constant overclock? Think of it as city stop and go driving vs cruising on the highway.
If we are talking hours of difference then you have a point. But if we are just talking an extra 20 to 30 minutes of battery I would rather just have the speed. YMMV.
mitchellvii said:
What one has to wonder is does the very act of the cpu constantly throttling up and down and measuring what speed is best actually burn more fuel than just remaining at a high constant overclock? Think of it as city stop and go driving vs cruising on the highway.
If we are talking hours of difference then you have a point. But if we are just talking an extra 20 to 30 minutes of battery I would rather just have the speed. YMMV.
Click to expand...
Click to collapse
Well yeah,.going.faster means finishing quicker but generally you can expect higher levels of inefficiency at higher rigor... my somewhat educated opinion, without tossing around numbers, is that he effect on the battery life should be minimal.
Theres also the fact that with a 10 inch tablet, the screen takes up a clear majority of the current.. the processor, even while at 100% still pales in comparison.. and the processor is.not often at 100%.. you dont really flash the screen.on and off, you usually have it on for a block of time.. processor only really works during tasks and then idles.
Sent from my SGH-T959 running 4.1.1
Hi, I've lowered all the figures by 100mv, but I notice no difference in my battery life. Is it safe to lower the figures more?
Sent from my Nexus 7 using xda premium
Define what you mean by "safe".
Higher core voltages are used at higher clock rates in order to preserve logic margin in the face of more power supply noise and tighter timing. So, undervolting by definition means you are giving up margin.
Whether your SoC happens to have a worst-case timing path which is "fast" or "slow" relative to other devices cannot be deduced by anyone here on XDA or even at Nvidia. About the best you can hope for is some "feel good reports" about what others do with undervolting and completely undisciplined "testing". But even that provides no information about your individual chip.
good luck
Thanks for your reply. By "safe" I mean without damaging my device. And yes, it would be interesting to hear some undervolting stories. I'd quite like to extend my battery life. Cheers!
Sent from my Nexus 7 using xda premium
I've been using my nexus 7 runing a AOKP pub build by mrRobinson and franco's r47 kernel. i've undervolted the entire board by 100mv it is runing pretty solid without any noticible lag or instability. Yet every chip is diferent and can or cannot this margin you'll have to test for your self. Just leave the option 'set on boot' unticked untill your sure that your device is capable of using those kind of voltages. And don't use very large steps. Just like overclocking in fact! Try it - it runs - get a little bit lower. If it crashes you shoukd be able to reboot the nexus and the settings you've changed reseted
Sent from my Nexus 7 using Tapatalk 2
antmasi said:
I've been using my nexus 7 runing a AOKP pub build by mrRobinson and franco's r47 kernel. i've undervolted the entire board by 100mv it is runing pretty solid without any noticible lag or instability. Yet every chip is diferent and can or cannot this margin you'll have to test for your self. Just leave the option 'set on boot' unticked untill your sure that your device is capable of using those kind of voltages. And don't use very large steps. Just like overclocking in fact! Try it - it runs - get a little bit lower. If it crashes you shoukd be able to reboot the nexus and the settings you've changed reseted
Sent from my Nexus 7 using Tapatalk 2
Click to expand...
Click to collapse
Thanks! Have you noticed any improvement in battery life?
Sent from my Nexus 7 using xda premium
Been testing for 2 days now! Battery seems a little bit better yet maybe placebo! But i feel it does not get has hot as runing stock voltages! It was the main reason i did the undervolt!
Sent from my Nexus 7 using Tapatalk 2
antmasi said:
Been testing for 2 days now! Battery seems a little bit better yet maybe placebo! But i feel it does not get has hot as runing stock voltages! It was the main reason i did the undervolt!
Sent from my Nexus 7 using Tapatalk 2
Click to expand...
Click to collapse
Placebo
Sent from my Nexus 7 using XDA Premium HD app
For what i've read that's the truth but if the frequencies set for the soc are getting less power shoudn't it draw less power from battery? 100mv its almost 10% of the stock voltages!
Sent from my Nexus 7 using Tapatalk 2
antmasi said:
For what i've read that's the truth but if the frequencies set for the soc are getting less power shoudn't it draw less power from battery? 100mv its almost 10% of the stock voltages!
Click to expand...
Click to collapse
For the SoC, whether it's leakage (V^2/R) or dynamic power dissipation (f*C*V^2), yes, you might expect let's say (0.9^2) = 81% battery use at the same operating frequencies.
OTOH, if 75% of the power drain normally is used by the LCD backlight (for instance in a "reading web pages" use case), then reducing the supply voltage will get you only 1/4 of that 20% savings - about a 5% improvement - because the power is being dissipated elsewhere. (Display, DRAM, 3G radio, WiFi radio, etc)
I agree with what you've pointed! In fact i've just taken out the undervolt because Inhad a reboot under heavy multitasking (torrent download, xbmc opened, chrome also downloading a smal file) i'm not certain that it was caused by the uv but it's possible. The Nexus started to lag and then froze completly! From time to tima i've the need of heavy multitask and it wasn't up to the task!
Sent from my Nexus 7 using Tapatalk 2
I just recently got my nexus 4 and I love it there is only one gripe I have which is the CPU temperature ... I ran various tests using stability control test and just gaming . I ran stability test for 30 minutes and got the following results ( below in pictures ) .When I game for an hour it hits 48-52 C . Is this normal ? Will it affect my CPU in anyway ?
Sent from my Nexus 4 using xda app-developers app
The heat causes thermal throttling. That will reduce performance. Undervolting helps with this a lot.
dia_naji said:
I just recently got my nexus 4 and I love it there is only one gripe I have which is the CPU temperature ... I ran various tests using stability control test and just gaming . I ran stability test for 30 minutes and got the following results ( below in pictures ) .When I game for an hour it hits 48-52 C . Is this normal ? Will it affect my CPU in anyway ?
Sent from my Nexus 4 using xda app-developers app
Click to expand...
Click to collapse
theres nothing wrong there, its all normal.
I don't think those temperatures mean overheating. It is indeed quite warm. It is exceptionally warm but not hot to the touch that you can't handle it.
Sent from my Nexus 4 using Tapatalk 2
So is there a fix for this ?
Where are all my computer nerds to stop all these rediculous overheating threads? 40*C in COLD for a processor. 60C is fine. It'll throttle it if it gets too hot. Come back when you hit 90C. Not trying to be mean but jeez, someone needs to open up their computer and feel how how your processor or GPU heat sink is just at idle, you'd get second degree burns touching the heatsink at full load. Its normal for processors to get hot... to a point.
username8611 said:
Where are all my computer nerds to stop all these rediculous overheating threads? 40*C in COLD for a processor. 60C is fine. It'll throttle it if it gets too hot. Come back when you hit 90C. Not trying to be mean but jeez, someone needs to open up their computer and feel how how your processor or GPU heat sink is just at idle, you'd get second degree burns touching the heatsink at full load. Its normal for processors to get hot... to a point.
Click to expand...
Click to collapse
The only reason why I asked this question is because I am still new to this device, but you can't really compare a desktop cpu or gpu to a smartphones cpu/gpu.there two distinctive hardware. PS, I work with computers and a computer science major
dia_naji said:
The only reason why I asked this question is because I am still new to this device, but you can't really compare a desktop cpu or gpu to a smartphones cpu/gpu.there two distinctive hardware. PS, I work with computers and a computer science major
Click to expand...
Click to collapse
Well for one thing they both run relativity warm.
Sent from my Nexus 4 using Tapatalk 2
dia_naji said:
The only reason why I asked this question is because I am still new to this device, but you can't really compare a desktop cpu or gpu to a smartphones cpu/gpu.there two distinctive hardware. PS, I work with computers and a computer science major
Click to expand...
Click to collapse
i hit a 100C(cpu temp) a few times, and got the safety shutdown. youll be ok.
I tried that... Now my nexus 4 won't turn on
Sent from my Nexus 4 using xda app-developers app
dia_naji said:
I tried that... Now my nexus 4 won't turn on
Sent from my Nexus 4 using xda app-developers app
Click to expand...
Click to collapse
RMA it. Don't forget to flash it back to stock first.
Sent from my Nexus 4 using Tapatalk 2
What you see in System tuner is the CPU temp and the temperature you see in stability test is the battery temp. Those temperatures are completely fine and actually really cool for this device. Now if the battery reaches 48-52 Degrees C now we have a problem. I suggest using a custom kernel and ROM to allow some undervolting for none to less throttling so your performance will not be impacted. Should give Trinity kernel a try, latest ones are USUALLY in simm's G+ page or go to the Trinity kernel thread and download it from the website. Hope I helped
Ohh no the battery's max was 40 only the CPU hits 48-52 , for me its not a problem. I just wanted to know if it was safe to continue to game.
Sent from my Nexus 4 using xda app-developers app
As above.
Daily use is very good, everything is responsive and fast.
However, benchmarks are roughly 50-70% of what they are supposed to be.
I tried to post a screenshot of CPU Spy, previously reseted timers and then ran BasemarkOS and Geekbench. (can't because less of 10 posts )
I see in CPU Spy that Max CPU state is 1555 Mhz? That's the A53 cores, right?
Is this behaviour normal?
Are you on Android N?
Please see here
http://forum.xda-developers.com/nexus-6p/help/4-cores-allways-stopped-whats-t3390543
rev_b said:
As above.
Daily use is very good, everything is responsive and fast.
However, benchmarks are roughly 50-70% of what they are supposed to be.
I tried to post a screenshot of CPU Spy, previously reseted timers and then ran BasemarkOS and Geekbench. (can't because less of 10 posts )
I see in CPU Spy that Max CPU state is 1555 Mhz? That's the A53 cores, right?
Is this behaviour normal?
Click to expand...
Click to collapse
According to the Play Store, CPU Spy hasn't been updated in 3 or 5 years. Are you using an app that supports HMP?
Sent from my Nexus 5X using Tapatalk
joemacjr said:
Are you on Android N?
Please see here
http://forum.xda-developers.com/nexus-6p/help/4-cores-allways-stopped-whats-t3390543
Click to expand...
Click to collapse
No, the device is new, it's running Android 6.0.1, no mods or custom ROM whatsoever.
SlimSnoopOS said:
According to the Play Store, CPU Spy hasn't been updated in 3 or 5 years. Are you using an app that supports HMP?
Sent from my Nexus 5X using Tapatalk
Click to expand...
Click to collapse
So what is the best app to monitor cpu activity?
And why are my benchmarks so low?
For instance, I'm only getting ~25k in 3dmark Ice Storm Unlimited, roughly the same as my previous Nexus 6.
According to Anandtech I should be getting ~36k.
rev_b said:
So what is the best app to monitor cpu activity?
And why are my benchmarks so low?
Click to expand...
Click to collapse
Kernel Adiutor is a free kernel manager. I prefer using ElementalX Kernel Manager (paid app). Both work perfectly fine on Marshmallow, support HMP, and receive frequent updates.
If the phone is hot, then benchmarks will suffer. The big cluster gets throttled and goes offline whenever the temperature reaches a certain point. If it is running cool and you see low scores then idk cuz I don't do benchmarks.
Sent from my Nexus 5X using Tapatalk
If the phone is responsive and fast as you say why do benchmarks scores matter? Benchmarks are pointless, they give you a pretty useless set of numbers that don't equate to real world usage. If your real world usage is good (as you pointed out) then you're wasting time and energy worrying about something that doesn't matter.
Heisenberg said:
If the phone is responsive and fast as you say why do benchmarks scores matter? Benchmarks are pointless, they give you a pretty useless set of numbers that don't equate to real world usage. If your real world usage is good (as you pointed out) then you're wasting time and energy worrying about something that doesn't matter.
Click to expand...
Click to collapse
I don't care about benchmarks, but they can point to a problem in your phone.
And I really would like to know how a phone sent to hardware testers consistently gets much higher benchmarks than the one I bought. Something's not right.
rev_b said:
I don't care about benchmarks, but they can point to a problem in your phone.
And I really would like to know how a phone sent to hardware testers consistently gets much higher benchmarks than the one I bought. Something's not right.
Click to expand...
Click to collapse
I've had eight of these and none of them have performed well in benchmarks. Take from that what you will but I believe it's a problem that's common to the SD810, not necessarily that you have a defective unit.
rev_b said:
As above.
Daily use is very good, everything is responsive and fast.
However, benchmarks are roughly 50-70% of what they are supposed to be.
I tried to post a screenshot of CPU Spy, previously reseted timers and then ran BasemarkOS and Geekbench. (can't because less of 10 posts )
I see in CPU Spy that Max CPU state is 1555 Mhz? That's the A53 cores, right?
Is this behaviour normal?
Click to expand...
Click to collapse
The phone is thermal throttling because the 810 Is overheating and the cores are being turned off. You can thank Qualcomm for their faulty overheating 810 design.
Batfink33 said:
The phone is thermal throttling because the 810 Is overheating and the cores are being turned off. You can thank Qualcomm for their faulty overheating 810 design.
Click to expand...
Click to collapse
That still doesn't explain why phones sent to reviewers don't throttle so much...
sam here
I barely get my Bigger Cluster working
It's offline most of the time
I don;t mind the heat unless it doesn't damage my phone as I;m always using a cover
Any way I can change the throttle limits so that it throttles less?
will the CTT mod be good?
TJ_bab said:
sam here
I barely get my Bigger Cluster working
It's offline most of the time
I don;t mind the heat unless it doesn't damage my phone as I;m always using a cover
Any way I can change the throttle limits so that it throttles less?
will the CTT mod be good?
Click to expand...
Click to collapse
Ctt mod does exactly that.
feis said:
Ctt mod does exactly that.
Click to expand...
Click to collapse
Using CTT mod but still my Bigger cluster barely turns on
and also kernel Adiutor doesn't work sometimes, whenever I try to change something CPU related, it just keeps loading and loading