• After 15+ years, we've made a big change: Android Forums is now Early Bird Club. Learn more here.

Root [KERNEL] MiRaGe - for HTC stock ICS ROM - 3.0.101 - 10222013

After flashing .76 steam came out of my ears. My phone turned into a puddle of mercury and reconstituted as a miniature flying saucer that is now following me around the house.

Every time I try to call someone it says "resistance is futile."

I think this kernel is a great project because it creates a bridge in choices between a rooted stock ROM and a nearly stock custom ROM like the ever popular Harmonia. If there is another overclocking kernel that works on a rooted stock ROM I have not found it yet. Besides, the battery life is great.
 
What is the default behavior of the 2D and 3D GPU's. I am not a fan of kernel tuner due to some bad experiences and I hope the defaults are moderate overclock on 3D and stock on 2D for us non gamers.
 
What is the default behavior of the 2D and 3D GPU's. I am not a fan of kernel tuner due to some bad experiences and I hope the defaults are moderate overclock on 3D and stock on 2D for us non gamers.

Default behavior is for non-gamers. Actually, I have reduced the low clocks just a bit to help with battery. Overclocked speeds are only used when needed (i.e. gaming, etc.).
 
I thought I was not going to be around today. But here I am with a new build :)

- Qualcomm Crypto Engine added
- MSM_PIL_QDSP6V3 Hexagon option added
- many memory and clock fixes/updates from CAF msm-3.4 kernel

You can find the download link in the OP. Enjoy!
 
I assume you are not using stock ICS ROM as you mentioned in the other thread. If you can post a log cat, I can look at it. There must be an incompatibility. Otherwise, I can't help with this much information.

BTW: the attached is my quadrant result at 1.7 GHz.
 

Attachments

  • 2013-05-02_19-44-19.png
    2013-05-02_19-44-19.png
    36.1 KB · Views: 110
Once I had a popular battery widget that caused Antutu backup to crash every time. I don't think these benchmark programs are the most stable aps around. Besides, Quadrant is not all that reliable as a benchmark.
 
ss3.jpg



Running Harmonia 3.17 and interactive set at 1512 max / 384 min.

Not only is it running smooth, I am getting around 15-20% better battery life.:D
 
Everyone, here is a test kernel with USB OTG tested and working. I hadn't done any extensive testing with OTG. When I tested finally, I found that the stock QUP driver was constantly spamming the kernel log with errors when OTG was enabled. Spam was so much that CPU load was noticeable. It took a lot of work, to update the QUP driver as you can see in my repo. But the kernel log is clean, OTG is working smooth now. Please let me know if you try.
shooter_test_kernel_765.zip
 
mrg, you are hard at work.

Yep, there was a really hard to fix bug in USB OTG. I just couldn't give up.

Everyone,
I have uploaded a bug-fix release in the OP
- USB OTG fixed, tested. Thanks to LeslieAnn for testing.
- Reverted the low frequency bandwidth settings to original values. This will hopefully help with a few crash reports.

Thanks!
 
I was asked at the XDA thread to add kernel-based mpdecision. There are two implementations of kernel-based mpdecision.

First is used widely with almost all of the custom kernels on XDA (except MiRaGE so far :)) and implemented by Dennis Rassmann (aka showp1984). The kernel tuning apps that control mpdecision work with this code.

Second is the official Qualcomm/CAF implementation of mpdecision. This is a more complex implementation, requires the DCVS framework (MSM DCVS for CPU/GPU Frequency control) implemented first and builds on DCVS. Kernel tuner apps, I think, will not work with this code. I might also test the CAF code later if the simpler mpdecision option becomes a nuisance like the original code I deleted from my phone.

I have added Dennis Rassmann's mpdecision in the test kernel below. It seems to be working so much better than the user-space code we deleted. If you can test battery life and performance, it would be great.
shooter_test_kernel_767.zip
 
Just flashed 767 test, so all I can report is the phone rebooted normally. From mrg's description it sounds like the problem with stock mpdecision is that once the second core is shut down, it does not get turned on in a responsive manner with the result being one core running fast while the other is at zero.
 
Just flashed 767 test, so all I can report is the phone rebooted normally. From mrg's description it sounds like the problem with stock mpdecision is that once the second core is shut down, it does not get turned on in a responsive manner with the result being one core running fast while the other is at zero.

My initial testing shows that there is no perceptible performance difference unlike the original mpdecision that was dragging the performance down. If there is improvement in the battery life that would be great. Otherwise, I will not keep it in the kernel. I would appreciate your feedback in the battery life.
 
I think any improvement in battery life will have a lot to do with how the phone is used. If a lot of time is spent reading web pages that would probably be on one core with mpdecision (working properly) and two cores without. If the use involves more load either way it will be on two cores.

In any event, this will be hard to measure.

Not only that, I just flashed it on Harmonia and flashed Google Now, which tends to produce more load.
 
I think any improvement in battery life will have a lot to do with how the phone is used. If a lot of time is spent reading web pages that would probably be on one core with mpdecision (working properly) and two cores without. If the use involves more load either way it will be on two cores.

In any event, this will be hard to measure.

Not only that, I just flashed it on Harmonia and flashed Google Now, which tends to produce more load.

Sure, if someone can load both cores most of the time, MPDEC will not make much difference (maybe even worse) in battery life. When the phone sleeps, one of the cores is off with or without MPDEC anyway. Also, if you have changed your ROM, your measurements will not be consistent. Let me test it myself a couple days then. Thanks!
 
Well, in theory there should be some improvement, but I suspect it will be small.

I don't see any battery improvement so far. The difference should be during the daily use not during the standby. The performance is not as smooth either. I will complete the day with mpdecision and try the latest stable build without it tomorrow, just to have a consistent comparison. To double check, I might do another run with mpdecision after that.
 
I just flashed back to the stable Mirage build on Harmonia with G-Now and noticed an improvement in smoothness. The difference is not great, but it is visible.

BTW, I was in Austin over the weekend and the photo shooting was amazing.
 
I just flashed back to the stable Mirage build on Harmonia with G-Now and noticed an improvement in smoothness. The difference is not great, but it is visible.

BTW, I was in Austin over the weekend and the photo shooting was amazing.

Great pictures, all of them. My favorites are Pizza Girl and Red River Corner. If you could use slower shutter speed in Red River Corner to create light streaks from the passing cars that would be really nice, I thought. But you probably didn't have a tripod at the time.
 
mrg, as Red River was not posed, a tripod would have been impossible. You have a good eye because an artist friend of mine likes that one also. More to come.
 
mrg, as Red River was not posed, a tripod would have been impossible. You have a good eye because an artist friend of mine likes that one also. More to come.

I like Pizza Girl too. The intersecting straight lines and the round pizza pies, green and red contrast, and the girl's profile and black hair at the window create a really nice geometry/color composition. Reminded me Renoir or Lautrec paintings. Wonderful shot!

Just a suggestion: I would crop the left side of the window out.
 
Back
Top Bottom