Letting this die. New dev and user threads created.

Status
Not open for further replies.
Search This thread
By:
Advanced…
S

Stevethegreat

Senior Member
Nov 28, 2010
1,199
327
About hardware accelerarion: What exactly do you mean by it? I think games (for example) are already accelerated on ICS, so maybe you mean "hardware decoding"?

Also if you do so, do you mean the same that can be found in stock (.MP4, Netflix), which is limited at best or do you mean the whole pack (most popular filetypes), like -say- some Chinese tablets already have (for that we'd need codecs of course)?
 
S

southbird

Senior Member
Feb 12, 2010
249
100
Hardware acceleration in the context we're using it is the way that video information can be quickly moved around for high performance situations such as video playback. Google Chrome for Android also uses this for its general rendering. Lack of this hardware assistance breaks video players such as Netflix and Flash which depend on it to get performance.

It really has nothing to do with codecs, otherwise they would be calling it hardware decoding. You'd like to think a forum full of kernel geeks would know the right terms to use. :)
 
  • Like
Reactions: tofffff68
tduermeier

tduermeier

Senior Member
Oct 8, 2010
81
8
Boulder, CO
I've hit the thanks button, but I just want to express my thanks further. I'm infinitely impressed at your diligence, Hashcode. With the progress as slow as it is a lesser man would have quit long ago. Mad props to you, brother.

Sent from my SGH-T959
 
  • Like
Reactions: MaxRabbit
isaiahhhhh

isaiahhhhh

Senior Member
Jan 5, 2009
142
103
33
Maine
Hashcode, when you get to the point where you need testing (if you decide you need testing on multiple kindles aside from what you have already) i'm perfectly willing to lend mine to the cause.
 
A

AlexDeGruven

Senior Member
Mar 24, 2008
685
213
Ann Arbor
www.structuredthought.org
southbird said:
You'd like to think a forum full of kernel geeks would know the right terms to use. :)
Click to expand...
Click to collapse

Ha, except a solid half of the posters in this particular dev forum barely have enough brain cells to maintain respiration.

There should be a test for posting privileges in dev forums (general and Q/A, whatever). Only the geeks should be allowed here, and everything else goes to the other areas.

Related: This is awesome. I'll have to laugh if it takes something ridiculous like 30 sec of tweaking to get HWA working once the screen gets going properly.

Also: I wonder how much control we'll have over the LED color once that driver is up and running properly. It does green and orange already, so that means there's at least green and red LEDs. If there's a (unused) blue one in there too, and we can access it, that would be entirely awesome.
 
  • Like
Reactions: treUse and webstas
verkerria

verkerria

Senior Member
Mar 13, 2011
112
10
I was noticing as I browsed the kernel init on the blog, and where hash said the boonanim went to hell, it mentions the emif0. Im wondering if it has something to do with it. The only place this item successfully does anything prior to bootanim is irq assignment, it failed the mpu load. I dont know if this is relevant, just trying to help :)

Sent from my DROID3 using XDA
 
S

Stevethegreat

Senior Member
Nov 28, 2010
1,199
327
southbird said:
Hardware acceleration in the context we're using it is the way that video information can be quickly moved around for high performance situations such as video playback. Google Chrome for Android also uses this for its general rendering. Lack of this hardware assistance breaks video players such as Netflix and Flash which depend on it to get performance.

It really has nothing to do with codecs, otherwise they would be calling it hardware decoding. You'd like to think a forum full of kernel geeks would know the right terms to use. :)
Click to expand...
Click to collapse
OK then, sorry if I have offended some souls, so let me rephrase:

"Are we going to get proper hardware decoding too?"
Because as it stands the only filetype that is accelerated on KF (in general) is .MP4, which is a rather frustrating situation; for a tablet -at least- it is, if it was a phone I wouldn't mind.
 
K

kjwalker

Senior Member
Feb 5, 2010
139
40
Amazon Fire HD 8 and HD 10
Samsung Galaxy S21 Ultra
tduermeier said:
I've hit the thanks button, but I just want to express my thanks further. I'm infinitely impressed at your diligence, Hashcode. With the progress as slow as it is a lesser man would have quit long ago. Mad props to you, brother.

Sent from my SGH-T959
Click to expand...
Click to collapse

THIS. Cannot say it any better. You are a saint among us all, Hashcode.
 
  • Like
Reactions: DAHKindle
T

tex321

Member
Aug 19, 2010
47
1
Brasilia
OnePlus 8
Do any of you know of any functional implementation of the 3.3 kernel on a 4430 device running android?

Thanks for working on making Kindle fire better.
 
Last edited:
C

CUViper

Senior Member
Sep 17, 2010
127
41
AlexDeGruven said:
Highly unlikely this early in the game. I don't even know of any desktop spins (outside of individuals compiling their own kernels) using 3.3.x yet.
Click to expand...
Click to collapse

Fedora 16 is already pushing out kernel 3.3: http://koji.fedoraproject.org/koji/packageinfo?packageID=8

However, updating for x86 desktops is a whole different ballgame, pretty straightforward. That's a big reason why developers try to get their code into the kernel, so they're not continuously porting forward manually.
 
Hashcode

Hashcode

Senior Recognized Developer
Sep 3, 2011
3,424
23,730
Minor update:

I booted up into the OS with the screen driver off.

In adb, I ran some event/input tests and was able to see touchscreen events registering as well as the accel/gyro/proximity sensors. Audio is disabled

For the screen, I'm posting on the TI forums with all of the debug info to see if they can give me more info about the EMIF0 SYS error I'm seeing whenever the dmm is running in the graphics driver.

Hopefully, I'll get a response Monday.

If I get bored, I'll go ahead and try out the latest 3.2 TI kernel to see if the screen is magically working (it probably won't be).
 
  • Like
Reactions: jwish, fiddlefaddle, ObihoernchenDE and 36 others
Status
Not open for further replies.

Similar threads

bl1nkk
[ROM][24 Feb] IceCreamSandwich Development/Discussion Thread
Replies
3K
Views
1M
H
heyjoe66
twa_priv
[ROM] CM10/SGT7 for the Kindle Fire [20130117]
Replies
4K
Views
1M
sd_shadow
sd_shadow
Bandage
[GUIDE] Install Full Android Market on Kindle Fire (Compilation Master Thread)
Replies
118
Views
281K
elcidroyale
elcidroyale
BiteBlaze
  • Locked
|i'm dead| |please close| |discontinued|
Replies
1K
Views
424K
E
e495957
twa_priv
[ROM] [JB 4.2.2] CM10.1/SGT7 for the Kindle Fire [20130724]
Replies
2K
Views
632K
t3chwizard
t3chwizard

Top Liked Posts

24 Hours All time
  • There are no posts matching your filters.
  • 229
    Hashcode
    Hashcode
    LAST UPDATED: 4/2/2012

    WHERE ARE WE NOW? IS IT BOOTING? CAN I GET A NEW ROM W/ AN ICS KERNEL?

    Is it booting? Yes.

    Is there a new ROM for the kernel? Yes. It's still a work in progress and likely to continue changing as the kernel gets more developed.

    Can I test the new ROM? Right now, there's no reason to run this ROM unless you're working on the kernel and/or bug testing. It is booting into the OS now and as soon as I fix up the Wifi and SD MMC mount, I'll probably put up a test link for testers -- lots to bugfix, this isn't "usable" yet. Especially since we don't have the Ducati loading for hw codecs.

    How long will it take to fix? There is no way to know. I work on the kernel about 4-5 nights a week, and when I fix something, there's always something behind it which needs fixing as well. I am making good progress (in my opinion), but that may still mean weeks of development.

    I'm going to keep a page updated on my blog so that at any point in time if you're curious what the output of the boot up looks like you can go take a look (this will be mostly technical output and don't worry if it much of it doesn't make sense):
    http://blog.hash-of-codes.com/kindle-fire-ics/

    [NOTE: I've now highlighted items in green which are good and some red items still to be fixed.]

    2nd post has the current status in more detail
    View
    134
    Hashcode
    Hashcode
    UPDATED 4/7

    WORKING -- WILL NEED TESTING:
    :) EXT4 MMC Filesystem Access
    :D USB Android Gadget with modes like ADB and MTP (Do we want mass storage also?)
    :) Battery / Charger Driver
    :D Display driver / SGX / ION now plays nice w/ 512M memory (originally the 3.0 kernel was hard coded for 1G of memory)
    :) Touchscreen is working
    :) Sensors are working to detect tilt / movement
    :) Power button working (no colored leds yet)
    :) New Audio driver from TI is working (thanks nook_lover!)
    :) Headset output / recording
    :) Internal media partition mount
    :D Ducati binary is being loading by the system now and the OMX server is starting -- NEEDS TESTING -- probably still won't work -- but its progress towards hardware codecs

    PARTIALLY WORKING:
    :) New Backlight Driver is now partially working. Won't adjust brightness yet, but it's coming.
    :| Wireless device and driver is being recognized. Still troubleshooting.

    STILL BROKEN:
    :/ Color Changing Power Button Leds - Old Driver needs to be updated

    FUTURE CHANGES IF NECESSARY:
    - Touchscreen Driver has a newer straight Linux driver available which would need minor Android customization (current driver seems ok for now)
    - Charger Driver has a newer straight Linux driver available which could be customized for Android (will probably do this one eventually. The current charger driver is very messy)
    - Battery Gauge has a generic 3.0 android driver available. It would need minor customization (may or may not rewrite this one eventually)
    View
    72
    Hashcode
    Hashcode
    Updated #2 post with some fixed items:
    - Audio is now working, including headset jack for output and recording. (Thanks nook_lover for getting source from TI!)
    - Internal "media" mounting is fixed
    - Wireless drivers and modules are built but hitting an error in the OS. Troubleshooting this.
    - New memory routines for handling our lower (512M) system in ICS are now working. I show 130MB free when I first boot up, and the system should dynamically allocate needed resources for the ION driver vs. The pre-allocated 200MB method that is used on 1GB systems.

    I managed to get the Archos ducati binary to load and the OMX server to start. Now I want to be very clear: this is very experimental and probably wont work right off the bat. BUT it's progress on the hardware codecs front.

    Once the wireless issues are ironed out, I'll crank up a new dev/testing thread with a CM9 v0.1 ICS ROM (running the new kernel).

    Some quick notes about the upcoming testing:
    - We'll actually have 2 threads: 1 here in the dev section for testers and devs to talk about current issues. And a 2nd "users" thread (not that the ROM should be used as a daily yet -- its not ready) in the general section. This should be where discussion about non-dev questions and/or tech support can happen.
    - No ETAs. This is still a work in progress and fixes happen as they happen.
    - Expect things to work, and then not to work, and then get fixed again.

    Hope you're all enjoying your Saturday!


    Sent from my DROID4 using XDA
    View
    57
    Hashcode
    Hashcode
    Pyrotek said:
    I built Hash's kernel from source and integrated with AOKP build 30. It boots nicely and even fixed the static wallpaper. Obviously, no wifi yet as mentioned in his progress. Haven't tried much else, as I did this late last night and have to work now. :)

    Awesome Hash, i'm learning from your hard work. Please don't bother him about this or keep asking for a test rom. If you really want one, it's not hard to build yourself.
    Click to expand...
    Click to collapse

    Pyrotek, this is awesome :) And EXACTLY the reason I'm trying to keep the new build for Kindle Fire ICS as clean as possible. We should be able to bring up the more developed AOSP-based ROMs in a day w/o little to no customisation (AOSP, CM9, AOKP, etc).

    Wifi is coming. I'm down to about 1-2 issues of debugging.

    Side note: My work has been VERY busy over the last few days so I'm trying to get time in when I can, to get this out for testing.

    Thanks for the patience.
    View
    54
    Hashcode
    Hashcode
    Quick Status update:

    I completed the latest source code merge with TI's omapzoom and the kernel is now technically version 3.0.21 (as opposed to 3.0.8). During this merge I removed quite a bit of Amazon customization from the kernel sources and will continue to do so as needed, because it makes updates about a quadrillion times easier.

    The bootup status really hasn't changed much: Kernel boots, init kicks off, the services startup and about the time we hit "surfaceflinger / bootanimation" the whole process stalls out.

    A new change, is that ADB is now up and running and that's helping troubleshoot quite a bit. It means access to logcat, shell, etc. So far it looks like the halt is due to a 2.6 backlight driver and some framebuffer configuration issues. Hoping to make some progress on those soon.

    Also, I've started the updated Android build under "ics-test" branches in the KFire-Android git for both device/amazon/otter and vendor/amazon/otter (https://github.com/KFire-Android). Here you can see what a "regular" ICS build for the Kindle Fire looks like. Almost all of the vendor files are removed (which came off the original device), and it has 3.0 kernel modules and board config. This includes removal of most of the legacy hacks that are used for OpenGL, old style Wifi config, etc.

    I'll get you more info as it happens. But, you can trust that if nothing is being updated here or on the blog (http://blog.hash-of-codes.com/kindle-fire-ics/); it's not because I'm ignoring the KFire. I'm just not making much progress.
    View

New posts