[Kernel][24/04] Perseus
- Thread starter AndreiLux
- Start date
Thanks for the update.
Minor issues on my end...
A setting that doesn't stick/apply in 31.2: SD CARD READ-AHEAD
Settings that don't stick/apply in 32.1: INTERNAL STORAGE READ-AHEAD, SD CARD READ-AHEAD, DMC MAXIMUM THREASHOLD. Possibly others, but I haven't checked them all.
EDIT: Deleting the /data/.perseus folder seems to have fixed it.
Minor issues on my end...
A setting that doesn't stick/apply in 31.2: SD CARD READ-AHEAD
Settings that don't stick/apply in 32.1: INTERNAL STORAGE READ-AHEAD, SD CARD READ-AHEAD, DMC MAXIMUM THREASHOLD. Possibly others, but I haven't checked them all.
EDIT: Deleting the /data/.perseus folder seems to have fixed it.
Last edited:
Amazing update!!
My current settings are standard profile, tweaked to be a little less saturated (somewhere in between natural and standard).
This was just done by trial and error, and I probably need a bit more time to perfect it.
Saturation weight: 0
Red yellow gain: -3
Red yellow curve: -9
GC gain: -12
GC curve: -4
BM gain: -5
BM curve: -3
All else stock.
Let me know what you think
EDIT:
Just realised that the labels are wrong. Each corresponds to only one color (yellow, red, cyan, Green, magenta, blue in that order)
My current settings are standard profile, tweaked to be a little less saturated (somewhere in between natural and standard).
This was just done by trial and error, and I probably need a bit more time to perfect it.
Saturation weight: 0
Red yellow gain: -3
Red yellow curve: -9
GC gain: -12
GC curve: -4
BM gain: -5
BM curve: -3
All else stock.
Let me know what you think
EDIT:
Just realised that the labels are wrong. Each corresponds to only one color (yellow, red, cyan, Green, magenta, blue in that order)
Last edited:
Re: [Kernel][30/01] Perseus
Could anyone help me out, my phone running Wanamlite v1.1 and Perseus 32 is having trouble entering deep sleep mode. I'm not an expert so haven't been able to work out what's wrong myself but I believe these screenshots might have an answer and someone wiser than me would be able to decipher them.
Sent from my GT-I9305 using xda premium
Could anyone help me out, my phone running Wanamlite v1.1 and Perseus 32 is having trouble entering deep sleep mode. I'm not an expert so haven't been able to work out what's wrong myself but I believe these screenshots might have an answer and someone wiser than me would be able to decipher them.
Sent from my GT-I9305 using xda premium
Attachments
Any Chance of having a fading LED like on the old Siyah Kernal?
Have used your kernal since day 1 on the i9305 and have to say its awesome, wasnt trying to compare features but its the only kernal i know that had that fading led feature.
Anyway just like to say excellent job and will send you some beer money when i get paid next week.
Have used your kernal since day 1 on the i9305 and have to say its awesome, wasnt trying to compare features but its the only kernal i know that had that fading led feature.
Anyway just like to say excellent job and will send you some beer money when i get paid next week.
tangcla
Senior Member
Hi,
Any chance that we can get an implementation of init.d for this that is fully supported by mount2sd?
My understanding is that the one in Perseus starts in parallel with the rest of the boot process, thus mount2sd doesn't work fully.
I've been using the script from this post http://xdaforums.com/showpost.php?p=36739521&postcount=1190 with this kernel and a standard ROM, without issues. I now have 64gb internal memory.
Last edited:
This kernel is kickass and to my amazement it keeps getting better. So glad to have this for my i9305, the standard mDNie settings would seriously give me eye cancer it's so ugly when looking at text. How Samsung did not adress this yet and how no one seems to notice it in reviews is a mystery to me 
Anyways thank you andreilux, you're the man! Beermoney on the way sir
Anyways thank you andreilux, you're the man! Beermoney on the way sir
Re: [Kernel][30/01] Perseus
Andrei is a great dev, Siyah kernel source pwns as well.
Sent from my GT-I9305 using xda premium
This kernel is kickass and to my amazement it keeps getting better. So glad to have this for my i9305, the standard mDNie settings would seriously give me eye cancer it's so ugly when looking at text. How Samsung did not adress this yet and how no one seems to notice it in reviews is a mystery to me
Anyways thank you andreilux, you're the man! Beermoney on the way sir
Andrei is a great dev, Siyah kernel source pwns as well.
Sent from my GT-I9305 using xda premium
Re: [Kernel][30/01] Perseus
Could you share your settings?
Sent from my GT-I9305 using xda premium
This kernel is kickass and to my amazement it keeps getting better. So glad to have this for my i9305, the standard mDNie settings would seriously give me eye cancer it's so ugly when looking at text. How Samsung did not adress this yet and how no one seems to notice it in reviews is a mystery to me
Anyways thank you andreilux, you're the man! Beermoney on the way sir
Could you share your settings?
Sent from my GT-I9305 using xda premium
AW: [Kernel][30/01] Perseus
Hi, i have an weird vibration issue.
If i choose an vibration pattern that is longer than the normal SMS vibration and the Screen is off, it sounds like there is something loose at the vibration.
An example for a better understanding.: Assuming the vibration of an normal SMS takes 1 second, and I have choose a pattern that takes 2 seconds, then the vibration begins to cracle after one second. But only when the Screen is off.
And i post this here because i think it is the Kernel because on 26.1 its fine but not on 31.2.
I can't remember if the Versions in between have the same issue, but I would think all versions after 26.1 have it.
Is there a better way to solve this than as choosing a short vibration pattern or using Perseus 26.1?
----------------------------------------------
Gesendet von meinem GT-I9305
ROM: | Pandoriam v6.5 | Kernel: | Perseus a31.2 |
Don't say thanks, hit Thanks!
----------------------------------------------
Hi, i have an weird vibration issue.
If i choose an vibration pattern that is longer than the normal SMS vibration and the Screen is off, it sounds like there is something loose at the vibration.
An example for a better understanding.: Assuming the vibration of an normal SMS takes 1 second, and I have choose a pattern that takes 2 seconds, then the vibration begins to cracle after one second. But only when the Screen is off.
And i post this here because i think it is the Kernel because on 26.1 its fine but not on 31.2.
I can't remember if the Versions in between have the same issue, but I would think all versions after 26.1 have it.
Is there a better way to solve this than as choosing a short vibration pattern or using Perseus 26.1?
----------------------------------------------
Gesendet von meinem GT-I9305
ROM: | Pandoriam v6.5 | Kernel: | Perseus a31.2 |
Don't say thanks, hit Thanks!
----------------------------------------------
Re: [Kernel][30/01] Perseus
Yes I was keeping it at 640 MHz for a while, and went 533 to see if there is any noticeable difference in performance / battery. It works rock stable on 640.
Sent from my GT-I9305 using Tapatalk 2
---------- Post added at 10:39 PM ---------- Previous post was at 10:38 PM ----------
Or TWRP / odin.
Sent from my GT-I9305 using Tapatalk 2
Yes I was keeping it at 640 MHz for a while, and went 533 to see if there is any noticeable difference in performance / battery. It works rock stable on 640.
Sent from my GT-I9305 using Tapatalk 2
---------- Post added at 10:39 PM ---------- Previous post was at 10:38 PM ----------
Or TWRP / odin.
Sent from my GT-I9305 using Tapatalk 2
Thanks. I will try that. Will be interesting to see what the differences on the battery will be.
Similar threads
- Locked
- Sticky
- Poll
Top Liked Posts
-
There are no posts matching your filters.
-
104Identical kernel to the i9300 one. Only difference is:
- CMA device memory allocation is gone as we don't need the hacks used in it for larger memory, 2GB with the default setting will suffice.
- You apparently have no FM radio.
The port is trivial and everything else should work flawlessly.
Please refer to the I9300 thread for changelogs, bug-reports and other issues. Kernel itself will be posted here.26
I've told you guys the i9300 will always be first to get new versions since that's what I have to develop on, test, and use. Please refrain from such posts because I will update here as well (i9305 users are first in line before the 7 different Note 2 versions...). I also use the i9300 as a testbed for fast feedback in case of bugs, and that's why you people for example never get non-hotfixed versions first because things will already have been debugged on the i9300.
So, with that in mind, uploading 30.1 now for here. Please read the changelog in the main thread.9Basic, not very scientific UV test
Hi Guys,
I finished my two days of UV and non UV testing.
TL;DR: UV seems to help a decent amount, I think it's worth the hassle.
What did I undervolt?
I spent weeks undervolting my CPU, GPU, internal, and memory components. I then went through each component, and used the ABB interface to see if I could further undervolt. On the CPU I could undervolt a little bit further, on the GPU and internal I could get a decent amount of extra UV, on the memory component I could gain no further UV with ABB. Please note I'm unsure about the GPU undervolting - there appears to be a bug with the GPU frequency or volt_table interfaces, more on that HERE and HERE.
Click HERE to see my full undervolting, but for a quick idea, here are my average undervolts for each component:
CPU: 155mV
GPU: 147mV
Internal: 175mV
Memory: 150mV
Not bad for an ASV3 chip
How did I test?
Both days I took it off the charge at midnight, and left it off the charge for the next 18 hours. Used BetterBatterStats to monitor everything, I tried to use the phone in exactly the same way both days - about 1h30m screen on time, which included about 20mins of candy crush saga, both days push syncing an exchange account from 8am - 6pm, all google services on push (google now disabled), about the same SMSs, same amount of wifi and 4G, I think I was pretty consistent.
I wish I had more time to knock off a few more tests, but my i9505 came in, and I've already left it in the box for 2 days.. I can't resist anymore
Results!
Day 1: undervolting applied.
Code:Since 17 h 59 m 12 s Level lost [%]: Bat.: -70% (87% to 17%) [3.9%/h] Deep Sleep (): 12 h 31 m 30 s (45090 s) Ratio: 68.1% Awake (): 5 h 27 m 42 s (19662 s) Ratio: 30.4% Screen On (): 1 h 32 m 30 s (5550 s) Ratio: 8.6% No or Unknown Signal (): 15 h 24 m 29 s (55469 s) Ratio: 85.7% Moderate Signal (): 3 m 28 s (208 s) Ratio: 0.3% Good Signal (): 1 m 44 s (104 s) Ratio: 0.2% Screen dark (): 1 h 8 m 44 s (4124 s) Ratio: 6.4% Screen dimmed (): 2 m 25 s (145 s) Ratio: 0.2% Screen medium (): 19 m 10 s (1150 s) Ratio: 1.7% Screen light (): 41 s (41 s) Ratio: 0.1% Screen bright (): 1 m 27 s (87 s) Ratio: 0.1% ================ Kernel Wakelocks ================ "PowerManagerService" (): 1 h 34 m 12 s (5652 s) Cnt:(c/wc/ec)6225/0/0 8.7% "mdm_hsic_pm0" (): 1 h 7 m (4020 s) Cnt:(c/wc/ec)8091/427/0 6.2% "wlan_rx_wake" (): 42 m 18 s (2538 s) Cnt:(c/wc/ec)2272/0/2273 3.9% "wlan_ctrl_wake" (): 37 m 40 s (2260 s) Cnt:(c/wc/ec)847/0/847 3.5% "alarm_rtc" (): 29 m 3 s (1743 s) Cnt:(c/wc/ec)3037/0/487 2.6% "hsicctl0" (): 16 m 34 s (994 s) Cnt:(c/wc/ec)4687/0/4687 1.5% "alarm" (): 16 m 30 s (990 s) Cnt:(c/wc/ec)4864/16/0 1.5%
Day 2: stock voltages.
Code:Since 17 h 55 m 26 s Level lost [%]: Bat.: -85% (89% to 4%) [4.7%/h] Deep Sleep (): 12 h 21 m 57 s (44517 s) Ratio: 68.2% Awake (): 5 h 33 m 28 s (20008 s) Ratio: 31.0% Screen On (): 1 h 32 m 52 s (5572 s) Ratio: 8.6% No or Unknown Signal (): 15 h 23 m 46 s (55426 s) Ratio: 85.9% Moderate Signal (): 30 s (30 s) Ratio: 0.0% Good Signal (): 4 m 38 s (278 s) Ratio: 0.4% Screen dark (): 47 m 28 s (2848 s) Ratio: 4.4% Screen dimmed (): 27 m 14 s (1634 s) Ratio: 2.5% Screen medium (): 16 m 59 s (1019 s) Ratio: 1.6% Screen bright (): 1 m 9 s (69 s) Ratio: 0.1% ================ Kernel Wakelocks ================ "PowerManagerService" (): 1 h 39 m 57 s (5997 s) Cnt:(c/wc/ec)5937/0/0 9.3% "mdm_hsic_pm0" (): 1 h 9 m 1 s (4141 s) Cnt:(c/wc/ec)8297/552/0 6.4% "wlan_rx_wake" (): 46 m 14 s (2774 s) Cnt:(c/wc/ec)1902/0/1902 4.3% "wlan_ctrl_wake" (): 32 m 40 s (1960 s) Cnt:(c/wc/ec)866/0/865 3.0% "alarm_rtc" (): 28 m 7 s (1687 s) Cnt:(c/wc/ec)3281/0/377 2.6% "hsicctl0" (): 17 m 5 s (1025 s) Cnt:(c/wc/ec)4627/0/4626 1.6% "alarm" (): 15 m 6 s (906 s) Cnt:(c/wc/ec)5053/10/0 1.4%
So, both days had the same:
- Total time on
- Screen on time
- PowerManagerService wakelocks
- mdm_hsic_pm0 wakelocks
- signal variance
The only real difference is the screen brightness, but I don't think this is enough to worry about.
Battery level, however, the UVed config used 3.9% battery per hour, whereas the stock voltage config used 4.7% battery per hour. Over the 18 hours, this meant that with UV I had an extra 15% battery than I did with no UV.
Caveat: I do think I used the phone slightly harder on the stock voltage test day, but only right at the end, i used google maps with GPS for about a minute, and I drove to the shops which I didn't do the day before. This may have made a % or two difference, but not 15%. To balance that a bit, I cut the test short on the second (no UV) test day, the overal % lost is probably about right.
Yes, this is with 36.2, but with this fix I posted to make it act like 36.3.8
Here you go. Sorry for not keeping up to date too much but some irritating people in this section and the minuscule user-base of this device makes me a bit lazy in releasing things.Hi Andrei,
Any chance we can get the mods from alpha26 in the i9300 kernel in this one?
Thanks
