I tried searching the forums, but couldn't find anything on the subject. Sorry if this has been asked before.
Does Nook Color support OpenGL ES 2.0 under any rom?
Got my device (firmware v1.0.0) yesterday and rooted it. Quadrant's "system info" displays it only supports OpenGL ES 1.1, and software that has requirement for OpenGL ES 2.0 (e.g. DiceShaker 3D) doesn't show on the market app. Also running the following code in an app prints 'false':
-- clip --
ActivityManager am = (ActivityManager) getSystemService(Context.ACTIVITY_SERVICE);
ConfigurationInfo info = am.getDeviceConfigurationInfo();
Log.d(TAG, info.reqGlEsVersion >= 0x20000);
-- clip --
So it's apparent that at least the stock rom with v1.0.0 firmware doesn't support OpenGL ES 2.0. Any chance that any of the other roms (Froyo, Honeycomb) do or might in the future?
Oh, and thanks for all the awesome work on the device!
Yes, the hardware does support it. The GPU is the PowerVR SGX530, which is OpenGL ES 1.1/2.0. What I suspect is the B&N did not go all out with the drivers (since it is "just" an ebook reader after all... ), and did not provide support.
Assuming a kernel can be compiled with the correct drivers, it should do 2.o just fine.
Divine_Madcat said:
Yes, the hardware does support it. The GPU is the PowerVR SGX530, which is OpenGL ES 1.1/2.0. What I suspect is the B&N did not go all out with the drivers (since it is "just" an ebook reader after all... ), and did not provide support.
Assuming a kernel can be compiled with the correct drivers, it should do 2.o just fine.
Click to expand...
Click to collapse
OpenGL ES 2.0 works on Nook with CM7, in case anyone else is wondering.
You can play games, like Need For Speed Shift that use OpenGL.
tech_head said:
You can play games, like Need For Speed Shift that use OpenGL.
Click to expand...
Click to collapse
Yeah, but all android devices support OpenGL ES 1.1. The question was about OpenGL ES 2.0, which the original BN rom didn't support. Just wanted to drop the info here that it now works with CM7, in case anyone else was wondering.
So 3d game run smoother on cm7
Yes.
/Should/ also run better on Froyo-based ROMs, but I don't have direct experience with those.
glg
Haha. I expected smoother than froyo :]
I've just updated my firmware with the new CM7 RC4 release and Darlinger's OC Kernel (Released 3/31). According to the release log, they should support hardward decoding.
However, when I open MP4 files with RockPlayer, it still reports that this format is not supported with the system player.
Am I doing something wrong, or it still doesn't support hardware decoding?
Do these files play on other ROMs?
So far as I know, they would play with the rooted stock ROM. I'm not sure about Froyo or Honeycomb, though.
poofyhairguy said:
Do these files play on other ROMs?
Click to expand...
Click to collapse
Do a search for "nook color handbrake preset". You'll find a thread on all about how to encode mp4 videos to work on the nook. There are some options when encoding the video that when enabled will break playback on the nook color. Mainly the video resolution.
Thanks. But I don't really think that's an encoding issue. The file played perfectly with the stock rom and system video player.
woot1524 said:
Do a search for "nook color handbrake preset". You'll find a thread on all about how to encode mp4 videos to work on the nook. There are some options when encoding the video that when enabled will break playback on the nook color. Mainly the video resolution.
Click to expand...
Click to collapse
woot1524 said:
Do a search for "nook color handbrake preset". You'll find a thread on all about how to encode mp4 videos to work on the nook. There are some options when encoding the video that when enabled will break playback on the nook color. Mainly the video resolution.
Click to expand...
Click to collapse
If Nook Color' sample video cannot play by CM7 RC4 so hardware decoding is still disabled.
Here is the sample video from official Nook Color intro I extracted from latest firmware (1.10) :
http://www.mediafire.com/file/s5ka0xkdnsf4ajo/welcome_video.mp4
If default video player cannot play so hardware decoder still not enabled for CM7 RC4 and all variants...
Managing to play the video by means of diffrent 3rd party applications ( all video player that decode software and hardware based) make no sense...
Hmm I don't know what the deal is then... HW decoding is working brilliantly for me on CM7 RC-4 with latest OC kernel. I have a couple 720p mp4 movies that I re-encoded via the handbreak preset, and they played just fine.
woot1524 said:
Hmm I don't know what the deal is then... HW decoding is working brilliantly for me on CM7 RC-4 with latest OC kernel. I have a couple 720p mp4 movies that I re-encoded via the handbreak preset, and they played just fine.
Click to expand...
Click to collapse
Stock ROM from NC can play http://www.mediafire.com/file/s5ka0xkdnsf4ajo/welcome_video.mp4 videos but CM7 RC4 still cannot plays and they claim hardware video decoder enabled!...
Offical welcome_video.mp4 is the proof of CM7 RC4 is STILL software based...
Overclocking may help only software decoding? But I want to use hardware decoder and latest version not old one used 1.1.15.2766 .. NC's offical latest video decoder is 1.1.15.3172 for our NC ( PowerVR SGX 530)
NOBODY uses latest hardware because hardware decoding is still unusable... Acceptence of this current reality and then focusing on OVERCLOCKING CPU makes no sense when you have a hardware decoder's power ...
nemir said:
Stock ROM from NC can play http://www.mediafire.com/file/s5ka0xkdnsf4ajo/welcome_video.mp4 videos but CM7 RC4 still cannot plays and they claim hardware video decoder enabled!...
Offical welcome_video.mp4 is the proof of CM7 RC4 is STILL software based...
Overclocking may help only software decoding? But I want to use hardware decoder and latest version not old one used 1.1.15.2766 .. NC's offical latest video decoder is 1.1.15.3172 for our NC ( PowerVR SGX 530)
NOBODY uses latest hardware because hardware decoding is still unusable... Acceptence of this current reality and then focusing on OVERCLOCKING CPU makes no sense when you have a hardware decoder's power ...
Click to expand...
Click to collapse
Exactly! I was trying to make sense of this mess from January in my blog fineoils.blogspot.com. With a Nookie Froyo, Nookkolor started to lose everything that was there of hardware decoding AND hardware rendering. I was laghed at by CM7 NC dev dalingrin on my suggestion that lots of HW assistance in video playback is based on SGX, especially the HW overlay. CM7 team is of opinion that mp4/H.264 decoding can be made in DSP (IVA2 in our case) audio buffers for which were that "magic video fix" at the CM7 nightly 17 level. What we have is more or less smooth YouTube 2.1.16 (less smooth than EVO's 2.0.x "HQ" YouTube for Eclair) and stuttering/losing frames Flash 10.1...10.2 ("inline YouTube), plus stuttering local file playback with more than 480p or so. I have no indication that any HW-conscious FFmpeg codecs was ever used. NEON/Stagefright framework is missing/broken for unknown reason. Only Opera Mobile is capable of using 2D hardware assisted UI scrolling/layout, any other app/Webkit/launcher/wallpaper has no idea they all can use 2D/3D hardware acceleration, as, yes, drivers are broken/falling back to SW too easily. Sure, VOME engine by VisualOn might be just promised to AOSP 2.3.3 but never made it there, however I don't understand why latest Omapzoom commits on SGX driver and a special OMAP3 Adobe Flash Plugin couldn't be used.
Sorry for the long rant, but it seems that 90 % of CM7 dev efforts are concentrated on producing an "average" ROM for 40 or so supported devices. Our Nookcolor has unexpectedly better graphics hardware than all of them -- except Droid X.
Can't you try monitoring the CPU usage on stock and cm7 (top or similar), play a video and see if the CPU is way higher on cm7? I still see stutters on cm myself, and it would be sweet to see the video tweaked. We will also have new stuff to play with soon from B&N's froyo release. I'm guessing we will have a much easier time comparing apples to apples and diagnosing these issues if stock was 2.2 . There are a ton of factors in play (no pun intended), and we can all help by testing. It seemed like the last big jump in cm was prompted by the dude that found the video worked better with his bluetooth headset. That's the kind of thing that would take one person FOREVER to find, but amongst hundreds or thousands of testers becomes apparent. Test, and post results, and try not to be too emotionally attached to the issues. If it makes sense people will notice.... problems only arise when people start guessing, ranting, etc... (xda kids stuff) The nook is awesome and is only getting better
i would disagree - this has more to do with what codecs/containers are being supported rather than hardware acceleration
nook color at 2.1 did play that video but it could not play other formats becuase it did not support the codecs
same with flash - nook color 2.1 played that mp4 file but take same mp4 file and wrap it in a flash container and it cannot play it
nemir said:
If Nook Color' sample video cannot play by CM7 RC4 so hardware decoding is still disabled.
Here is the sample video from official Nook Color intro I extracted from latest firmware (1.10) :
http://www.mediafire.com/file/s5ka0xkdnsf4ajo/welcome_video.mp4
If default video player cannot play so hardware decoder still not enabled for CM7 RC4 and all variants...
Managing to play the video by means of diffrent 3rd party applications ( all video player that decode software and hardware based) make no sense...
Click to expand...
Click to collapse
cyberslug23 said:
Can't you try monitoring the CPU usage on stock and cm7 (top or similar), play a video and see if the CPU is way higher on cm7? I still see stutters on cm myself, and it would be sweet to see the video tweaked. We will also have new stuff to play with soon from B&N's froyo release. I'm guessing we will have a much easier time comparing apples to apples and diagnosing these issues if stock was 2.2 . There are a ton of factors in play (no pun intended), and we can all help by testing. It seemed like the last big jump in cm was prompted by the dude that found the video worked better with his bluetooth headset. That's the kind of thing that would take one person FOREVER to find, but amongst hundreds or thousands of testers becomes apparent. Test, and post results, and try not to be too emotionally attached to the issues. If it makes sense people will notice.... problems only arise when people start guessing, ranting, etc... (xda kids stuff) The nook is awesome and is only getting better
Click to expand...
Click to collapse
Sure, customizations and many other improvements over "standard" CM7 ROM are welcome at any time, it's just an apparent lack of basic GPU/DSP functioning, as nemir stated, what might cause a somewhat critical attitudes. I repeat, the work of dalingrin is awesome, but there are still many questions/problems remain unresolved:
-- the "original" size of an audio "buffer" (if I get it right) was either zero, or small, or adaptive, plus small anyway. Wasn't it an indication that hardware decoding when functional didn't need any?
-- the addition of 32(?) of whatever units to that buffer brought a smoother video playback on YouTube, but can be also an indication that video playback can now easily fall back to software decoding;
-- switching the audio pipeline to push it via BT encoding might be originally either working better, or didn't need to be HW assisted at all. In any case, lipsyncing was missing, just like when the software decoding beyond bandwidth specs was involved
As for how many dudes it take to find a solution out of thousands of NC's inventive users, I'd rather try to find a real TI OMAP3/Zoom2 engineer or two who are NC owners and obviously ROFLing now seeing all these attempts at XDA Devs. In other words, they not only know the solution, it's offered in the open at their gits. It's our kernel level that cannot take it. CM7.x will be formidable when based on 2.6.36...38.
aludal said:
Exactly! I was trying to make sense of this mess from January in my blog fineoils.blogspot.com. With a Nookie Froyo, Nookkolor started to lose everything that was there of hardware decoding AND hardware rendering. I was laghed at by CM7 NC dev dalingrin on my suggestion that lots of HW assistance in video playback is based on SGX, especially the HW overlay. CM7 team is of opinion that mp4/H.264 decoding can be made in DSP (IVA2 in our case) audio buffers for which were that "magic video fix" at the CM7 nightly 17 level. What we have is more or less smooth YouTube 2.1.16 (less smooth than EVO's 2.0.x "HQ" YouTube for Eclair) and stuttering/losing frames Flash 10.1...10.2 ("inline YouTube), plus stuttering local file playback with more than 480p or so. I have no indication that any HW-conscious FFmpeg codecs was ever used. NEON/Stagefright framework is missing/broken for unknown reason. Only Opera Mobile is capable of using 2D hardware assisted UI scrolling/layout, any other app/Webkit/launcher/wallpaper has no idea they all can use 2D/3D hardware acceleration, as, yes, drivers are broken/falling back to SW too easily. Sure, VOME engine by VisualOn might be just promised to AOSP 2.3.3 but never made it there, however I don't understand why latest Omapzoom commits on SGX driver and a special OMAP3 Adobe Flash Plugin couldn't be used.
Sorry for the long rant, but it seems that 90 % of CM7 dev efforts are concentrated on producing an "average" ROM for 40 or so supported devices. Our Nookcolor has unexpectedly better graphics hardware than all of them -- except Droid X.
Click to expand...
Click to collapse
Yeah you are right...
We must accept the lack of full potential power of NC's 2D / 3D GPU for both custom ROMs (CM and other variants) and OFFICAL STOCK ROM...
I have Archos 5 Internet Tablet which has only 256 MB RAM and same CPU ( 800 Mhz ) and Archos 5 Internet Tablet can play 720p .mkv .avi .mp4 .wmv files without any problem...
Moreover, Nook Color has higher RAM and can be overclocked to 1 Ghz ...
Like Archos 101 Nook Color has same GPU ...
I would prefer 800 Mhz ( maybe 1 Ghz overclocked) normal CPU with fully usable GPU ...
Processor • ARM Cortex A8 at 1 GHz with DSP
• Graphic accelerator: 3D OpenGL ES 2.0
Video Playback1 • MPEG-42 HD (up to 720p, 30 [email protected])
• MPEG-42 ([email protected] AVI, up to DVD resolution, 30 [email protected])
• H.264 HD ([email protected] up to 720p, 30 [email protected])
• WMV9/VC1 (AP up to 720p 30 [email protected])
• M-JPEG (Motion JPEG Video) in VGA resolution
With optional plug-in (downloadable on www.archos.com):
• Cinema: MPEG-2 (up to DVD resolution MP/D1, 30 [email protected] Mbps)
With the above codecs, the device can play video files with the following extensions: AVI, MP4, MKV, MOV, WMV, MPG, PS, TS, VOB, FLV, RM, RMVB, ASF, 3GP
Audio Playback1 • MP3 CBR & VBR
• WMA, WMA-Pro 5.1
• WAV (PCM/ADPCM)
• AAC, AAC+ 5.13
• OGG Vorbis
• FLAC
Click to expand...
Click to collapse
http://www.archos.com/products/ta/archos_101it/specs.html?country=us&lang=en
Here is the table of CPU and GPU list
http://en.wikipedia.org/wiki/Texas_Instruments_OMAP
For Nook Color:
OMAP3621 800 MHz ARMv7 ARM Cortex-A8 PowerVR SGX530
Another information about NC
http://www.engadget.com/2010/10/28/nook-color-processor-revealed-arm-cortex-a8-based-ti-omap3621/
So Archos 101 (1024x600) can be ported to NC to utilise ...
CM7 is generic and does not utilise the power of NC's GPU ... CM7 is beneficial for bluetooth and latest android 2.3.3 ... But Arhos firmware has a better full utilised GPU driver...
Frankly, the lack of GPU can NOT be accepted. Focusing on "CPU overclock" does not make sense if you have potential of better GPU driver...
Cool story.
CM7 does have hardware accelerated video.
Also it is the DSP chipset which controls what formats/resolutions are acceptable, not the GPU. Ours is incapable of playing 720p video in hardware acceleration. Not to mention it's a 854x480 hardware accelerator, so 1280x720 video would be wasted anyway.
You guys are loud.
That welcome video won't play because the stock software was able to tell it to play sideways. We can't do thatyet . We CAN hardware accel videos such as this:
http://www.mediafire.com/?d42dmvva9vbigm2
You know, in 16:9 instead of 9:16?
aludal said:
Exactly! I was trying to make sense of this mess from January in my blog fineoils.blogspot.com. With a Nookie Froyo, Nookkolor started to lose everything that was there of hardware decoding AND hardware rendering. I was laghed at by CM7 NC dev dalingrin on my suggestion that lots of HW assistance in video playback is based on SGX, especially the HW overlay. CM7 team is of opinion that mp4/H.264 decoding can be made in DSP (IVA2 in our case) audio buffers for which were that "magic video fix" at the CM7 nightly 17 level. What we have is more or less smooth YouTube 2.1.16 (less smooth than EVO's 2.0.x "HQ" YouTube for Eclair) and stuttering/losing frames Flash 10.1...10.2 ("inline YouTube), plus stuttering local file playback with more than 480p or so. I have no indication that any HW-conscious FFmpeg codecs was ever used. NEON/Stagefright framework is missing/broken for unknown reason. Only Opera Mobile is capable of using 2D hardware assisted UI scrolling/layout, any other app/Webkit/launcher/wallpaper has no idea they all can use 2D/3D hardware acceleration, as, yes, drivers are broken/falling back to SW too easily. Sure, VOME engine by VisualOn might be just promised to AOSP 2.3.3 but never made it there, however I don't understand why latest Omapzoom commits on SGX driver and a special OMAP3 Adobe Flash Plugin couldn't be used.
Sorry for the long rant, but it seems that 90 % of CM7 dev efforts are concentrated on producing an "average" ROM for 40 or so supported devices. Our Nookcolor has unexpectedly better graphics hardware than all of them -- except Droid X.
Click to expand...
Click to collapse
What would it take to get a working patch to enable 2D/3D HW acc? I mean, if the folks working on CM7 won't take the approach you mention.
nemir said:
Here is the table of CPU and GPU list
http://en.wikipedia.org/wiki/Texas_Instruments_OMAP
Click to expand...
Click to collapse
That link tells you right there why we can't play 720p but the Archoses can:
OMAP3621 vs OMAP3630
They have a better DSP processor.
CM7 is generic and does not utilise the power of NC's GPU ... CM7 is beneficial for bluetooth and latest android 2.3.3 ... But Arhos firmware has a better full utilised GPU driver...
Frankly, the lack of GPU can NOT be accepted. Focusing on "CPU overclock" does not make sense if you have potential of better GPU driver...
Click to expand...
Click to collapse
The GPU and the DSP of the Nook Color work fine in CM7. We are able to play games just as well as rooted stock does, and videos I have encoded that maxed the rooted stock (as it had as high of a bit-rate as I could get without frameskipping) still play fine using Vital player on CM7.
As chisleu stated its not perfect, but its certainly working because the Nook Color's CPU can't even come close to playing my files.
Overclocking NC to 1 or 1.1 Ghz should give same performance as Archos's...
By the way, Motorola DEFY has 3610 GPU but it can plays 720p files by means of changing lib files ...
http://forum.xda-developers.com/showthread.php?t=935017
We can try this method to our NC to play 720p movies ...
As I said before latest stock NC has more recent builds in lib folder. CW RC4 uses old lib files...
Anyone can compare and see the difference between latest stock ROM and latest CW R4... CW R4 does not update latest official files from lib (its sub folders and files too)
hey nemir how bout taking us to solution space
nemir said:
Overclocking NC to 1 or 1.1 Ghz should give same performance as Archos's...
By the way, Motorola DEFY has 3610 GPU but it can plays 720p files by means of changing lib files ...
http://forum.xda-developers.com/showthread.php?t=935017
We can try this method to our NC to play 720p movies ...
As I said before latest stock NC has more recent builds in lib folder. CW RC4 uses old lib files...
Anyone can compare and see the difference between latest stock ROM and latest CW R4... CW R4 does not update latest official files from lib (its sub folders and files too)
Click to expand...
Click to collapse
nemir said:
Overclocking NC to 1 or 1.1 Ghz should give same performance as Archos's...
By the way, Motorola DEFY has 3610 GPU but it can plays 720p files by means of changing lib files ...
http://forum.xda-developers.com/showthread.php?t=935017
We can try this method to our NC to play 720p movies ...
As I said before latest stock NC has more recent builds in lib folder. CW RC4 uses old lib files...
Anyone can compare and see the difference between latest stock ROM and latest CW R4... CW R4 does not update latest official files from lib (its sub folders and files too)
Click to expand...
Click to collapse
Maybe you could try to copy over those lib files and see if it works? If it does then maybe cm7 devswill port it.
From 1.1 Nook Color with 1.1 ghz overclock
poofyhairguy said:
That link tells you right there why we can't play 720p but the Archoses can:
OMAP3621 vs OMAP3630
They have a better DSP processor.
Click to expand...
Click to collapse
That link mentions (quite vaguely) where variants of the same silicon (IVA2) are different:
Not highlighted in the list below is that each OMAP 3 SoC has an "Image, Video, Audio" (IVA2) accelerator. These units do not all have the same capabilities. Most devices support 12 megapixel camera images, though some support 5 or 3 megapixels. Some support HD imaging.
Click to expand...
Click to collapse
It is worth mentioning that encoding a (mp4) video is several times harder computational task than decoding it (as in playback). Sure I'm glad and happy for all owners of OMAP3 based smartphones with cameras which can record 720p, or 480p, or 360p or any medium to high bitrate video with their devices -- thanks to DSP HW acceleration of encoding process.
Now, (mp4) video decoding routine is much simpler process, and most important, it doesn't exactly use the same circuitry (the proof is in the possibility of live view while video is recorded on a given OMAP3-based device.) Sure, it doesn't conclusively prove anything, and TI engineers were not exactly helpful in detailing the 3621 IVA2 specs. But in my blog I was writing back in January/February on how smooth Evo's YouTube HQ could play clips searched out by "1080p" "Full HD" "IMAX" and the like keywords. Today, it struggles (loses frames), or refuses to play them.
The adequacy of hardware decoding of at least 480p/720p mp4 on NC was proven back then. Sure, that was Eclair back then, and AOSP 2.3 has NEON framework/hardware Stagefright pipeline broken/unimplemented since -- probably in expectation of Renderscript of Android 3.0 (or 2.4), I don't know
Out of my years (2004 or so) of struggle with mpeg2/mpeg4 decoding of 1080i satellite streams I was of conviction that in HD video rendering the major computational task is a scaling, deinterlacing overlay: once you don't have it in your hardware (shaders of GPU: SGX530 in our case), you have a stuttering, A/V desynced video -- or no video at all. Sure, it was outdated DirectX games back then, but anyway.
------------------
fineoils.blogspot.com