FORUMS

XDA Recap: This Week In Android (May 17 – 23)

Another week, another recap. The Sunday tradition marches on this week with a fresh … more

Huawei To Sell Honor Smartphones Via Indian Retail Outlets

World’s fourth largest smartphone maker Huawei plans to now sell its … more

Sunday Debate: Are Smaller Bezels Better or Worth It?

Bezels have been getting smaller and smaller as the years go by, and while devices … more

LG G4 First Impressions and Unboxing – XDA TV

The latest in LG’s signature line of “G phones” has been released. … more
Post Reply Subscribe to Thread Email Thread

[KERNEL] [GPL] [GN] franco.Kernel r398

21st April 2012, 01:45 AM |#13121  
Senior Member
Flag Marburg
Thanks Meter: 28
 
Donate to Me
More
Does r141 include the patches for he lazy governor to manage the "bouncyness"? It seems to work very well

Sent from my Galaxy Nexus using XDA
 
 
21st April 2012, 01:49 AM |#13122  
franciscofranco's Avatar
OP Recognized Developer
Flag Mountain View, CA
Thanks Meter: 96,456
 
Donate to Me
More
Quote:
Originally Posted by Bavilo

Does r141 include the patches for he lazy governor to manage the "bouncyness"? It seems to work very well

Sent from my Galaxy Nexus using XDA

Which patches? I only included the original commit for Lazy, nothing more.
21st April 2012, 01:53 AM |#13123  
Senior Member
Flag Marburg
Thanks Meter: 28
 
Donate to Me
More
Quote:
Originally Posted by distINCtINC

Franco, I made some changes to the Lazy governor that I believe are better for our phones. I attached my cpufreq-lazy.c to this post if you're interested. Of course, you're free to do whatever you want .

I changed the sampling rate to 10,000 to match that of the other governors in your kernel. I also modified how "min_timeinstate" is calculated so that it isn't coming out equal to the sampling rate anymore. If sampling rate = min_timeinstate, you're not getting the benefit of preventing frequencies from quickly bouncing back and forth. In Ezekeel's original thread on lazy he had a sampling rate at 10,000 and min_timeinstate to 40,000, so I changed it so that after the calculations it comes out similar to that with a sampling rate of 10,000 with a min_timeinstate of 30,000. At least with my phone, this seems to have helped reduce the "bounce" effect.

These patches

Sent from my Galaxy Nexus using XDA
21st April 2012, 01:55 AM |#13124  
franciscofranco's Avatar
OP Recognized Developer
Flag Mountain View, CA
Thanks Meter: 96,456
 
Donate to Me
More
Quote:
Originally Posted by Bavilo

These patches

Sent from my Galaxy Nexus using XDA

Nop, I didn't. I'll see how they work out on my device and if I think it's necessary I'll include it for Milestone 3.
21st April 2012, 01:56 AM |#13125  
Member
Flag London
Thanks Meter: 9
 
More
Quote:
Originally Posted by distINCtINC

If sampling rate = min_timeinstate, you're not getting the benefit of preventing frequencies from quickly bouncing back and forth. [...] so I changed it so that after the calculations it comes out similar to that with a sampling rate of 10,000 with a min_timeinstate of 30,000. At least with my phone, this seems to have helped reduce the "bounce" effect.

This makes sense. If the sampling is the same as the mintime, you might as well not set a minimum time in the state. It's changing as soon as it samples no matter how long its been in the state.
21st April 2012, 01:57 AM |#13126  
Senior Member
Flag Marburg
Thanks Meter: 28
 
Donate to Me
More
Quote:
Originally Posted by franciscofranco

Nop, I didn't. I'll see how they work out on my device and if I think it's necessary I'll include it for Milestone 3.

Ok thanks!

Sent from my Galaxy Nexus using XDA
21st April 2012, 02:00 AM |#13127  
franciscofranco's Avatar
OP Recognized Developer
Flag Mountain View, CA
Thanks Meter: 96,456
 
Donate to Me
More
Quote:
Originally Posted by myrdog

This makes sense. If the sampling is the same as the mintime, you might as well not set a minimum time in the state. It's changing as soon as it samples no matter how long its been in the state.

Aye, I agree.
21st April 2012, 02:02 AM |#13128  
Antob125's Avatar
Senior Member
Thanks Meter: 53
 
More
Franco I know Milestone are usually stock GPU but is there any way when milestone comes out you can make a 512 version. Loving the speed.

Sent from my Galaxy Nexus
21st April 2012, 02:03 AM |#13129  
franciscofranco's Avatar
OP Recognized Developer
Flag Mountain View, CA
Thanks Meter: 96,456
 
Donate to Me
More
Quote:
Originally Posted by Antob125

Franco I know Milestone are usually stock GPU but is there any way when milestone comes out you can make a 512 version. Loving the speed.

Sent from my Galaxy Nexus

No, I can't have 231651 different versions, sorry =(
The Following 2 Users Say Thank You to franciscofranco For This Useful Post: [ View ]
21st April 2012, 02:26 AM |#13130  
bigeyes0x0's Avatar
Senior Member
Flag Saigon
Thanks Meter: 1,652
 
Donate to Me
More
So where're your other 231648 versions?

Jokes aside trying the latest hoping to see my Bluetooth bug fixed. Otherwise I hope it's fixed after M3 with kernel 3.0.28 included. I'm pretty sure that it's a bug in kernel now, as I do not have it in stock, ogdobber and faux's kernels.
Post Reply Subscribe to Thread

Tags
4.0.4, 4.1, 4.2, best kernel, best support, fast, franco, galaxy, jb 4.1, kernel, nexus, samsung
Previous Thread Next Thread
Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes