I'm curious what is actually changed when a kernel is designed as low voltage as opposed to standard voltage?
I'm curious what is actually changed when a kernel is designed as low voltage as opposed to standard voltage?
The voltage multiplier is set lower than the Stock numbers provided by Motorola.
For instance, cvpcs has his 125Mhz Standard Voltage set to 0x27 and his Low Voltage set to 0x20. While that may look like Greek, it is very easy to convert it to a Voltage. 0x27 is Hex for 39 Decimal. We then take the 39 and put into the current formula I have for determining voltage of X * 12.5 + 600mV. So 39 * 12.5 + 600mv = 1087mV or 1.087 Volts. That meas the Low Voltage is 32 * 12.5 + 600mv = 1000mV or 1 Volt. A savings of roughly 9%. That adds up if your phone spends a lot of time in sleep/standby mode.
Wow. I just read all 38 pages. First off, Skull, your attention to this thread and the people posting is incredible and commendable. I learned a lot, and confirmed a few more things. Second, there was only one "slightly irritated" post. No flame wars, no jackassery of any other sort. Very impressive everyone! That's part of why I finally joined this forum. The other part is to learn and share. So now I'll share, and hopefully from everyone's comments and recommendations, I'll learn:
SetCPU 2.0.3
Profiles (priority in this order)
125-1000 - Main - ondemand
400-500 - Screen Off - ondemand (read about lag when "waking up")
125-400 - Temp > 40.2C - powersave
125-500 - Battery < 40 - ondemand
125-800 - Battery < 60 - ondemand
125-900 - Battery < 80 - ondemand
125-800 - Charging/Full - ondemand (don't want to fry it while charging)
Advanced
30000
50
0
0
Full charge, just erased battery stats, Juice Plotter says 18 hours, but we shall see. Also have Autokiller, set to Aggressive.
I only rooted 2 days ago, and as you can see from the sig, I jumped in feet first. Everyone lemme know what you think about my settings in SetCPU, and also if you have any experience, good or bad with Autokiller. Thanks, Preston
Is there any new data or information available regarding the interactive governor? I've seen some discussion about its affect on performance and battery life, but nothing with concrete data. I tried switching to it for a few days and didn't notice a change in battery life with performance seemingly slightly faster.
But with interactive there are no advanced settings to adjust in SetCPU, so is it better to trust interactive or to use ondemand with custom values? (was using 30000 sampling rate and 50% upper threshold).
Is there any new data or information available regarding the interactive governor? I've seen some discussion about its affect on performance and battery life, but nothing with concrete data. I tried switching to it for a few days and didn't notice a change in battery life with performance seemingly slightly faster.
But with interactive there are no advanced settings to adjust in SetCPU, so is it better to trust interactive or to use ondemand with custom values? (was using 30000 sampling rate and 50% upper threshold).
After I ran the very first test and the read the actual source code for the Interactive Governor, I saw no point in testing further. My original test shows the true issue with the governor. It spends more time in the upper slots than on demand. Which of course leads to faster battery drain. Now I will say that interactive is MUCH better than the Performance Governor. Because Interactive at least has the potential to underclock about 1/5 to 1/6 of the time. Which gives your CPU a time frame to cool off. So in that respect, if you are a a die hard "must get every ounce out of my phone", then Interactive is a better alternative than Performance.
Is there any new data or information available regarding the interactive governor? I've seen some discussion about its affect on performance and battery life, but nothing with concrete data. I tried switching to it for a few days and didn't notice a change in battery life with performance seemingly slightly faster.
But with interactive there are no advanced settings to adjust in SetCPU, so is it better to trust interactive or to use ondemand with custom values? (was using 30000 sampling rate and 50% upper threshold).
After I ran the very first test and the read the actual source code for the Interactive Governor, I saw no point in testing further. My original test shows the true issue with the governor. It spends more time in the upper slots than on demand. Which of course leads to faster battery drain. Now I will say that interactive is MUCH better than the Performance Governor. Because Interactive at least has the potential to underclock about 1/5 to 1/6 of the time. Which gives your CPU a time frame to cool off. So in that respect, if you are a a die hard "must get every ounce out of my phone", then Interactive is a better alternative than Performance.
The weird thing is it should be draining my battery faster, but I haven't noticed a change in battery life at all. I'll have to continue testing it over a longer period to confirm this, but based on what I've seen so far, it's running faster with at worst similar battery life.
Note though that 8 or so of those 12 hours were spent sleeping, so my phone was idle most of that 12hrs of unplugged time.
After I ran the very first test and the read the actual source code for the Interactive Governor, I saw no point in testing further. My original test shows the true issue with the governor. It spends more time in the upper slots than on demand. Which of course leads to faster battery drain. Now I will say that interactive is MUCH better than the Performance Governor. Because Interactive at least has the potential to underclock about 1/5 to 1/6 of the time. Which gives your CPU a time frame to cool off. So in that respect, if you are a a die hard "must get every ounce out of my phone", then Interactive is a better alternative than Performance.
The weird thing is it should be draining my battery faster, but I haven't noticed a change in battery life at all. I'll have to continue testing it over a longer period to confirm this, but based on what I've seen so far, it's running faster with at worst similar battery life.
Note though that 8 or so of those 12 hours were spent sleeping, so my phone was idle most of that 12hrs of unplugged time.
Any chance the Sleep/Standby Profile is still set to ondemand, powersave or conservative? If not, then you don't have any apps that are preventing the phone from truly going into sleep mode.
I would have to uninstall at least 6 apps, of the top of my head, before my phone could actually go 100% into sleep mode.
The weird thing is it should be draining my battery faster, but I haven't noticed a change in battery life at all. I'll have to continue testing it over a longer period to confirm this, but based on what I've seen so far, it's running faster with at worst similar battery life.
Note though that 8 or so of those 12 hours were spent sleeping, so my phone was idle most of that 12hrs of unplugged time.
Any chance the Sleep/Standby Profile is still set to ondemand, powersave or conservative? If not, then you don't have any apps that are preventing the phone from truly going into sleep mode.
I would have to uninstall at least 6 apps, of the top of my head, before my phone could actually go 100% into sleep mode.
I actually have all of my profiles set to interactive right now. I thought I remembered reading that SetCPU had problems when switching governors between profiles so I've been leaving all mine the same. Is this actually necessary?
So I have two questions about setCPU:
Question 1) I have a kernel that is an LV 125-800 but I set the main profile to Min:125 Max:600. Is this okay?
Question 2) Now even with that same kernel, I were to set the main profile to let's say 250-600 but I have my <CPU Temp set to 125-125, will my phone clock to that speed EVEN though the main profile reads Min:250 Max:600??