Android One Expanding to Bangladesh, Nepal, and Sri Lanka

At Google I/O 2013, Google proudly announced that the Android operating system was … more

Android 5.0.2 Fixes 2012 Nexus 7! Sony’s Making a Google Glass Competitor? – XDA TV

Android 5.0.2 Lollipop is available for the … more

Early Lollipop Ports for Micromax A116 and A117

It feels like it was only yesterday when Google announced the mysterious Android L.After … more

Now Gesture Tweaks Replaces Google Now with Custom Action

On most modern Android devices with software navigation keys, swiping up from the … more

Welcome to XDA

Search to go directly to your device's forum

Register an account

Unlock full posting privileges

Ask a question

No registration required
Post Reply

msm-dcvs vs ondemand CPU governors

OP shinesthru

25th July 2012, 06:27 AM   |  #1  
shinesthru's Avatar
OP Senior Member
Flag St. Paul, MN
Thanks Meter: 1,094
 
1,491 posts
Join Date:Joined: Apr 2011
More
Between these 2 governors, has anyone noticed a significant advantage as far as battery life goes?
26th July 2012, 12:00 AM   |  #2  
Senior Member
Flag Illinois
Thanks Meter: 492
 
2,009 posts
Join Date:Joined: Mar 2011
More
Quote:
Originally Posted by shinesthru

Between these 2 governors, has anyone noticed a significant advantage as far as battery life goes?

Hopefully more people respond to this, as I'd like to know more details around the msm-dcvs governor. I had it enabled in Clark's 0.0.6 ROM, and I noticed it was pretty much locked at right around 1000-1500mhz. It wouldn't scale below that. I don't know if it was something in the ROM or not, but I used Voltage Control to scale the min down to 352mhz or whatever it is.

I changed the governor to ondemand and set as boot as service (was getting errors creating the init.d script) and it seems to be functioning fine now.
The Following User Says Thank You to hayzooos For This Useful Post: [ View ]
28th July 2012, 02:58 PM   |  #3  
Senior Member
Chicago
Thanks Meter: 962
 
3,293 posts
Join Date:Joined: Oct 2010
Donate to Me
More
I've been using the msm-dcvs governor in CyanogenMod, and I've actually been having the opposite experience. It seems to max out around 1134MHz most of the time, though CPU Spy shows that higher clock speeds are being used, but rarely. The rest of the time is spent mostly around 384MHz. I haven't really tested it fully yet, let alone against ondemand. I'm going to keep using msm-dcvs since Qualcomm made it specifically to take advantage of our processors' scaling and voltage features. I haven't had any problems.
The Following 2 Users Say Thank You to EndlessDissent For This Useful Post: [ View ]
7th September 2012, 06:46 PM   |  #4  
Junior Member
Flag Elk River, MN
Thanks Meter: 1
 
24 posts
Join Date:Joined: Mar 2010
More
Quote:
Originally Posted by EndlessDissent

I've been using the msm-dcvs governor in CyanogenMod, and I've actually been having the opposite experience. It seems to max out around 1134MHz most of the time, though CPU Spy shows that higher clock speeds are being used, but rarely. The rest of the time is spent mostly around 384MHz. I haven't really tested it fully yet, let alone against ondemand. I'm going to keep using msm-dcvs since Qualcomm made it specifically to take advantage of our processors' scaling and voltage features. I haven't had any problems.

What has your battery life been like?
7th September 2012, 07:16 PM   |  #5  
Senior Member
Chicago
Thanks Meter: 962
 
3,293 posts
Join Date:Joined: Oct 2010
Donate to Me
More
Quote:
Originally Posted by jhavron

What has your battery life been like?

Well, I've actually been forced away from msm-dcvs because there is a bug in the AOSP Jelly Bean implementation, and none of the devs involved with either CM or other AOSP projects will acknowledge that the problem exists, let alone fix it. So, I've been using the Ktoonservative governor. It's working pretty well for now, but I'd prefer if msm-dcvs just got fixed instead
7th September 2012, 07:35 PM   |  #6  
Senior Member
Flag Illinois
Thanks Meter: 492
 
2,009 posts
Join Date:Joined: Mar 2011
More
Quote:
Originally Posted by EndlessDissent

Well, I've actually been forced away from msm-dcvs because there is a bug in the AOSP Jelly Bean implementation, and none of the devs involved with either CM or other AOSP projects will acknowledge that the problem exists, let alone fix it. So, I've been using the Ktoonservative governor. It's working pretty well for now, but I'd prefer if msm-dcvs just got fixed instead

Out of curiosity, what problems were you having? I have Voltage Control set to apply noop/ondemand at startup, but I just temp changed it back to msm-dcvs for kicks to see how it goes. I know it's the stock governor that came with the phone, so I'd like to stick with that if possible.

I thought I remembered seeing the speeds always be towards the higher end before, but I'll run it again now and see what happens. I'm using FreeGS3, v0.1.1 right now for what it's worth.
7th September 2012, 10:58 PM   |  #7  
Senior Member
Chicago
Thanks Meter: 962
 
3,293 posts
Join Date:Joined: Oct 2010
Donate to Me
More
Quote:
Originally Posted by hayzooos

Out of curiosity, what problems were you having? I have Voltage Control set to apply noop/ondemand at startup, but I just temp changed it back to msm-dcvs for kicks to see how it goes. I know it's the stock governor that came with the phone, so I'd like to stick with that if possible.

I thought I remembered seeing the speeds always be towards the higher end before, but I'll run it again now and see what happens. I'm using FreeGS3, v0.1.1 right now for what it's worth.

With the stock CM10 kernel, as soon as I touch the screen, the frequency jumps to 1134MHz and stays there until you turn the screen off. Then, it drops to the min. speed until you turn on and touch the screen again, at which point it jumps back to 1134MHz and locks itself there.

With the KT747 kernel, the speed more-or-less limits itself to 702MHz, and it's nearly impossible to make it scale any higher. KToonsez refuses to acknowledge the problem, so it won't be fixed anytime soon.

I've gotten CPU Spy graph screenshots to illustrate the problems (comparing the stock ROM to AOSP ROMs while doing the same task), but all parties still deny the problems exist. I don't really know what to do to get them to fix it at this point, so I just gave up. I'm hoping it's just some JB driver issue and it'll be fixed when the official JB kernel source is released.

Post Reply Subscribe to Thread
Previous Thread Next Thread
Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes