Root Can you under volt on MTDEV CM9?

[RolaAddict]

I know that with liquid smooth there was a easy way to under volt some of the apps on the phone to help save battery life.Is there a way to do it on CM9 or even CM7?

 

[nyioo7]

Are you talking about the power consumption? You can't necessarily lower the voltage for the apps since they don't use voltage. They are using the processor which draws the power from source. So, to be able to lower the power consumption, you need to down clock the frequency of your processor. I'm not sure if you can set priority of each apps by limiting their processor usage. I haven't had problem with the battery on CM9 so I never bothered.

 

[dsmryder]

I know that with liquid smooth there was a easy way to under volt some of the apps on the phone to help save battery life.Is there a way to do it on CM9 or even CM7?

You can't undervolt just an app. You do it to the whole phone. I use system tuner pro on my phone. I think you can change the voltage on the non-paid version too. That works for both our CM7 and CM9 kernels.

 

[jamesd1085]

I use incredicontrol to undervolt... its free and works well...as for underclocking CM9 you can undervolt to a degree but not like miui where you can undervolt to 600 where kernel parameters in cm9 limit it to a min of 800

 

[dsmryder]

I use incredicontrol to undervolt... its free and works well...as for underclocking CM9 you can undervolt to a degree but not like miui where you can undervolt to 600 where kernel parameters in cm9 limit it to a min of 800

The hardware limits it to 800mV. If you change the MIUI voltage for ~500MHz to 875mV you should be good. If you change it to 800mV I am willing to bet that it will lockup and reboot. Maybe the ~600MHz range if you have a really good phone.

 

[jamesd1085]

Right...I get what you are saying...but I only undervolt the values my device is using as min because I use on demand gov to keep it there when idle...and on miui I kept it set so my min was at 24mhz and undervolted to 600 and never got reboots

 

[RolaAddict]

Thanks but I was wondering if the under volting can be added to the ROMs themselves as well like it is with liquid smooth.It's been awhile since I was on that ROM but I remember it having like an app pre loaded on it that would allow for the under volting.I never used it cause I knew if I under volted too much then I would get major issues.In the mean time I'll check out the apps you all talked about.Thanks again!

 

[jamesd1085]

I'm not saying it will never be added...but right now I'm pretty sure the devs main concern is camera HDMI followed by small bugs and then at the end of the line extra add ons...and good luck with your undervolting .I like incredicontrol but if dsmryder is using system tuner its probably a good app

 

[RolaAddict]

Ok cool cool.The only reason I looked into under volting is cause I heard it was a way to help save the battery cause I still have my battery being eaten alive.Granted I do use ALOT of stuff on the phone at one time I need to find a way to help save the life of the battery while I'm out and not able to charge the phone.

 

[jamesd1085]

If done right it definitely has a noticeable effect...I can't remember where but someone posted voltage charts that give safe limits for each clockspeed... it had a light medium and aggressive chart...but if you are looking to save battery life when idle changing governor to on demand and undervolting you minimum clockspeed as low as you can without getting reboots

 

[thangfish]

Right...I get what you are saying...but I only undervolt the values my device is using as min because I use on demand gov to keep it there when idle...and on miui I kept it set so my min was at 24mhz and undervolted to 600 and never got reboots

you can set all your clock speeds to 600mv and not get reboots, because 800mv is the lower limit.
I think that setting anything lower is ignored and runs the defaults.

 

[jamesd1085]

Not true...the lower limit is set by edits in the kernel...I can't remember which rom I was messing around with but it allowed undervolting to 550...I kept it at 600 because 550-575 did cause reboots, while 600 was fully stable

 

[nyioo7]

As an Electrical Engineer, this thread gives me goosebumps. Lowering the voltage that is being supplied to the components can cause serious side effects on these ICs. Power consumption on the load is only affect by current and resistance. Yes. Power consumption of the source may be affect by voltage but the load (the device) will be running like a thirsty person. I can't imagine how the components might starts to fail in a long run. Maybe I should stop reading this thread. Lol. I know that lowering the frequency of the processor also lower the power consumption. This doesn't affect the DC power supply to the components. Ehhhh. Lol.

 

[dsmryder]

Not true...the lower limit is set by edits in the kernel...I can't remember which rom I was messing around with but it allowed undervolting to 550...I kept it at 600 because 550-575 did cause reboots, while 600 was fully stable

Our kernel doesn't seem to respect the request to go lower than 800mV even though the parts to allow it are there. If I understood why it doesn't do it I would fix it. As it stands now I just leave the lower frequencies at 800 and have come up with my other voltages based on my experiance on that particular phone. My DEV phone doesn't go quite as low as my personal phone.

As an Electrical Engineer, this thread gives me goosebumps. Lowering the voltage that is being supplied to the components can cause serious side effects on these ICs. Power consumption on the load is only affect by current and resistance. Yes. Power consumption of the source may be affect by voltage but the load (the device) will be running like a thirsty person. I can't imagine how the components might starts to fail in a long run. Maybe I should stop reading this thread. Lol. I know that lowering the frequency of the processor also lower the power consumption. This doesn't affect the DC power supply to the components. Ehhhh. Lol.

Well, as an elecrical engineer (and more for who don't know), you know that it takes time to charge a capasitor to a high enough level to reach the "1" state. At lower frequencies the amout of time that the capasitor has is longer and thus it doesn't need to have the higher voltage. As far as the system bus, that voltage splits off at about 800MHz (if I'm reading the table correctly) and is largly unaffected by the undervolting stratagy. Our biggest concern is with data integrity, as a cap. might not get charged enough to be read as a "1".

Anyway. Halla

 

[jamesd1085]

Our kernel doesn't seem to respect the request to go lower than 800mV even though the parts to allow it are there. If I understood why it doesn't do it I would fix it. As it stands now I just leave the lower frequencies at 800 and have come up with my other voltages based on my experiance on that particular phone. My DEV phone doesn't go quite as low as my personal phone.



Well, as an elecrical engineer (and more for who don't know), you know that it takes time to charge a capasitor to a high enough level to reach the "1" state. At lower frequencies the amout of time that the capasitor has is longer and thus it doesn't need to have the higher voltage. As far as the system bus, that voltage splits off at about 800MHz (if I'm reading the table correctly) and is largly unaffected by the undervolting stratagy. Our biggest concern is with data integrity, as a cap. might not get charged enough to be read as a "1".

Anyway. Halla

Yeah...the CM9 kernel seems to be much more complicated than past kernels...could be setup to keep min a 800 to keep something from breaking...still seems to have some positive effect on battery life...could be the good ole placebo effect lol

 

[thangfish]

Yeah...the CM9 kernel seems to be much more complicated than past kernels...could be setup to keep min a 800 to keep something from breaking...still seems to have some positive effect on battery life...could be the good ole placebo effect lol

Apparently it's not the table in the kernel at all... We've been through all this before, in one of the Cm7 threads.
You can make any number available in the kernel... It being accepted by the hardware is another issue entirely.
I'll post a link when I get time.

 

[dsmryder]

Yeah...the CM9 kernel seems to be much more complicated than past kernels...could be setup to keep min a 800 to keep something from breaking...still seems to have some positive effect on battery life...could be the good ole placebo effect lol

Well, CM9 is sopposed to have a 3.0 kernel and we have a 2.6ish kernel. One that's adapted from CM7, which is adapted from stock.

 

[jamesd1085]

Well, CM9 is sopposed to have a 3.0 kernel and we have a 2.6ish kernel. One that's adapted from CM7, which is adapted from stock.

Has triumph proven incapable of running 3.0 kernel or do we not have source?

 

[nyioo7]

Well, as an elecrical engineer (and more for who don't know), you know that it takes time to charge a capasitor to a high enough level to reach the "1" state. At lower frequencies the amout of time that the capasitor has is longer and thus it doesn't need to have the higher voltage. As far as the system bus, that voltage splits off at about 800MHz (if I'm reading the table correctly) and is largly unaffected by the undervolting stratagy. Our biggest concern is with data integrity, as a cap. might not get charged enough to be read as a "1".

Anyway. Halla

That's what I am worry about. Capacitor is charged by voltage and current deliver from DC source. AC does not charge capacitor. It goes straight through it. Lower the voltage means lower the current since that's what these devices are controlling. Current. Lower the current means lower the time it charges the capacitor or lower the time that current transfer throughout the circuit. That means, some parts of the circuit board will not get enough power supply at fast enough time. Capacitor will charge slowly and the time between charging and discharging will be big.

If you consider then RAM structure alone here, it requires 3 V for RAM to work efficiently. Even with 3 V, it requires pre-charged circuits in that microscopic structure because current cannot travel fast enough. Lowering this voltage is very inefficient. It may even cause MOSFETS to not work properly since the low gate voltage wouldn't fully open the circuit.

On the other hand, if you are talking about lowering a few milivolts, then it doesn't matter. It wouldn't affect as much as it would as lowering 5 V supply to let's say 2 V. Lowering the frequency will not cause any problem because that's just limiting the power consumption of the processor. In these devices, the most power drain is processor, video, data communication and backlights. So, lowering frequency of the processor, slowing down the data communication, turning off graphic intense apps and lowering backlights will use the device power way less and device will work efficiently. Lowering the voltage? It will cause damage to some parts of the device in long run.

Remember, these are nano circuits. Not only capacitor is important. Modern electronic does not work the way it did before. Characteristics changes as the size gets smaller.

 

[dsmryder]

That's what I am worry about. Capacitor is charged by voltage and current deliver from DC source. AC does not charge capacitor. It goes straight through it. Lower the voltage means lower the current since that's what these devices are controlling. Current. Lower the current means lower the time it charges the capacitor or lower the time that current transfer throughout the circuit. That means, some parts of the circuit board will not get enough power supply at fast enough time. Capacitor will charge slowly and the time between charging and discharging will be big.

If you consider then RAM structure alone here, it requires 3 V for RAM to work efficiently. Even with 3 V, it requires pre-charged circuits in that microscopic structure because current cannot travel fast enough. Lowering this voltage is very inefficient. It may even cause MOSFETS to not work properly since the low gate voltage wouldn't fully open the circuit.

On the other hand, if you are talking about lowering a few milivolts, then it doesn't matter. It wouldn't affect as much as it would as lowering 5 V supply to let's say 2 V. Lowering the frequency will not cause any problem because that's just limiting the power consumption of the processor. In these devices, the most power drain is processor, video, data communication and backlights. So, lowering frequency of the processor, slowing down the data communication, turning off graphic intense apps and lowering backlights will use the device power way less and device will work efficiently. Lowering the voltage? It will cause damage to some parts of the device in long run.

Remember, these are nano circuits. Not only capacitor is important. Modern electronic does not work the way it did before. Characteristics changes as the size gets smaller.

Exactly! I have been only able to drop the processor voltage from about 900mV to 800mV. And that's not the whole table, just most of the 900 section. See here's the kernel that is loaded with CM7. On my phone I can get 576000 down to 875 before I get any issues, but on the DEV phone I have to leave it at 900mV. If you read through it, I think you will see why I think the voltage table splits at 806400.

Has triumph proven incapable of running 3.0 kernel or do we not have source?

We don't have a full kernel. So the 3.0 kernel would have to be patched to work. We have a simular problem when we try to use a base 2.6 kernel. I took a quick shot at changing the file I posted, but it didn't boot. I think there are other files that go with it to make it work properly.