Reply
Thread Tools
tz1's Avatar
Posts: 716 | Thanked: 236 times | Joined on Dec 2007
#21
Also, how do I use this package? Just pair with an A2DP device and it should "just work" and switch to that output when it is on?
 
Posts: 2,102 | Thanked: 1,309 times | Joined on Sep 2006
#22
Ah, are you talking about johnx's a2dp package?

In which case take a look here: http://www.internettablettalk.com/fo...highlight=a2dp

My package just moves the SBC encoding from the ARM to the DSP.
 
qwerty12's Avatar
Posts: 4,274 | Thanked: 5,358 times | Joined on Sep 2007 @ Looking at y'all and sighing
#23
Originally Posted by lardman View Post
No. But leaving the dsp task file and its conf info won't harm anything, the only thing you'll need to do is "apt-get install bluez-utils" which should download the original version from the repo (otherwise download by hand and dpkg -i it).
apt-get install --reinstall bluez-utils

alternatively, in diablo, setting tablet into red pill mode and searching for updates should list bluez-utils as an update.

hth
 
tz1's Avatar
Posts: 716 | Thanked: 236 times | Joined on Dec 2007
#24
So I need to install johnx's package too (first?) or is that part of the deb and other files and use its program to toggle things on? And only specially patched versions of certain programs work?

(sorry if I'm being dense, but it isn't clear which pieces are needed where and how).

If I have an n810 which has freshly flashed Diablo, and a pair of a2dp headphones, I download your programs, run the install script, and then?

(insert noob instructions)

... And the headphones have hifi stereo playing.

Last edited by tz1; 2008-07-07 at 20:06.
 
Posts: 2,102 | Thanked: 1,309 times | Joined on Sep 2006
#25
Install Johnx's a2dp deb (which is a bunch of scripts to save you the hassle of tweaking the config files yourself. Many thanks Johnx, I definitely use them), you will then be able to use a2dp (with the encoding done by the main ARM processor).

If you want the encoding to be done by the DSP rather than the ARM, then (i.e. as well as Johnx's a2dp deb) install my sbcenc "package" (tarball + deb + installation script) and this will install the DSP task, config data and a patched version of bluez-utils which uses the DSP rather than doing the SBC (the codec used for a2dp audio) encoding on the ARM.

Hope that makes sense
 

The Following User Says Thank You to lardman For This Useful Post:
Benson's Avatar
Posts: 4,930 | Thanked: 2,272 times | Joined on Oct 2007
#26
I don't like asking idiot questions like this... (It annoys me a little when other people ask them.)

But I'm kind of interested in acquiring some A2DP headphones; I haven't put the money out yet because of the less-than-perfect results with ARM-based encoding to date. (I'd rather deal with steady poor-quality results from HSP than high-quality with intermittent lapses from A2DP.) Does this result in what you'd consider "glitch-free" playback yet? (I'm aware that HSP or even analog outputs do have occasional glitches under heavy CPU/net load; that's where the "idiot question" label applies... I just mean to the same level of reliability as those, not perfect regardless of usage.) If not, do you have a rough estimate of how much better this is than ARM-codec A2DP?

With my new tablet PC (which speaks A2DP), it seems the investment in A2DP cans isn't a complete waste anyway; I can use them HSP with the N800 and A2DP with the PC, so I suppose any feedback on N800 quality will just alter the time until I break down and buy, but I'd still like to know.

Thanks!
 
Mara's Avatar
Posts: 1,310 | Thanked: 820 times | Joined on Mar 2006 @ Irving, TX
#27
Originally Posted by lardman View Post
If you want the encoding to be done by the DSP rather than the ARM, then (i.e. as well as Johnx's a2dp deb) install my sbcenc "package" (tarball + deb + installation script) and this will install the DSP task, config data and a patched version of bluez-utils which uses the DSP rather than doing the SBC (the codec used for a2dp audio) encoding on the ARM.

Hope that makes sense
Yes, it makes sense, especially if couple details are cleared up:

What is the benefit on doing this in DSP instead of ARM? Does it improve sound quality? (Maybe less clicks and pauses?) Is the tablet more responsive due to reduced CPU load? How does it affect battery life?

Anyone know answers to these questions...

EDIT: Benson beat me to it...
 
GeneralAntilles's Avatar
Posts: 5,478 | Thanked: 5,222 times | Joined on Jan 2006 @ St. Petersburg, FL
#28
Originally Posted by Mara View Post
What is the benefit on doing this in DSP instead of ARM? Does it improve sound quality? (Maybe less clicks and pauses?) Is the tablet more responsive due to reduced CPU load? How does it affect battery life?
Same benefits as processing your 3d graphics on a GPU.
 
Mara's Avatar
Posts: 1,310 | Thanked: 820 times | Joined on Mar 2006 @ Irving, TX
#29
Originally Posted by GeneralAntilles View Post
Same benefits as processing your 3d graphics on a GPU.
Yes, in theory. In practise it depends how well it is implemented. There may still be a lot of work left for the CPU... or if the DSP coding is not well optimized it has to work hard and use about as much power than the ARM core would...

Since it uses DSP does it also lock the ARM core speed to 330MHz? (Since with that ARM core speed the DSP can run at full clock speed.)
 
Benson's Avatar
Posts: 4,930 | Thanked: 2,272 times | Joined on Oct 2007
#30
Originally Posted by Mara View Post
Yes, it makes sense, especially if couple details are cleared up:

What is the benefit on doing this in DSP instead of ARM? Does it improve sound quality? (Maybe less clicks and pauses?) Is the tablet more responsive due to reduced CPU load? How does it affect battery life?

Anyone know answers to these questions...

EDIT: Benson beat me to it...
I'd be pretty certain it reduces battery drain (increases life) as the DSP is (AFAIK) always powered up and clocked when the CPU is running; using it in parallel with using the CPU for e.g. decoding mp3s should let you stop the clock again sooner.

Should likewise leave the tablet generally more responsive, and leave the sound cleaner from glitches, but (I assume) identical output between glitches. (Unless there are integer vs. FP + fast approximations winding up in a 1 or 2 LSB possible difference; I've done exactly 0 research into this... If so, the differences still should not be audible.) To me, it's a question of degree; how much cleaner from glitches?
 
Reply


 
Forum Jump


All times are GMT. The time now is 13:40.