Call for testers: HDMI pass-through

[highrez]

It works. I hear audio - just I never see the little red light show the codec on the front of my onkyo - like I do with AC3/DTS passthrough (non HD). Also the HDMI light blinks, which I think is a sort of indication of syncing?

Also, only -D hdmitest3 works - which I think makes sense as 2,9 was the only device that ever worked for me.

Awesome work! I can't wait for this to get integrated to ffmpeg so that I can bug the myth devs to sync ffmpeg into myth.

[Anssi]

Originally Posted by highrez

It works. I hear audio - just I never see the little red light show the codec on the front of my onkyo - like I do with AC3/DTS passthrough (non HD).

Did you already try with the longer sample(s)?

Originally Posted by highrez

Also the HDMI light blinks, which I think is a sort of indication of syncing?

On my receiver (TX-SR607) it is. However, I never hear any sound when it blinks, only when it stops blinking. (I get the red light showing the codec)

If it does the same thing with the longer sample, I guess this would support that something may be wrong with either the NVIDIA HDMI hardware (non-Fermi) or the driver, causing incompatibilities with some receivers.

Originally Posted by highrez

Also, only -D hdmitest3 works - which I think makes sense as 2,9 was the only device that ever worked for me.

Yep, that is expected.

Originally Posted by highrez

Awesome work! I can't wait for this to get integrated to ffmpeg so that I can bug the myth devs to sync ffmpeg into myth.

[highrez]

Yeah, I had already tried with the longer sample (awesome sample, btw - first time I heard uncompressed multi channel audio on my system). Exact same thing, hdmi light blinks, I hear the sound, but the red-light codec indicator doesn't show anything.

[alanw]

Hi All !

I see everyone is testing on Nvidia.

I tried to do a test on a Core i3 Intel integrated motherboard.

I tried Ubuntu Natty, with 2.6.37 kernel. I have not reached to a point to actually test the passtrough, because,

speaker-test -c8 runs fine, but only outputs the two front channels. Do you guys have any idea what am i doing wrong ? Xorg is running in 1080p and i use an Onkyo TX-SR606, over a hdmi cable. Strange is that when i use 8 channel 192khz the amp receives it in 192khz stereo. So high bitrates are working but only the two front channels.

Can anyone help so that i can try to make the test for you ?

[Stephen Warren]

I'll file a bug to investigate why this works in some cases and not others. Unfortunately, I don't have an array of receivers to test with, so this may be tricky to repro/debug. Perhaps my receiver at home will repro this.

[Stephen Warren]

Some thoughts:

1) I believe these files are so-called "HBR" audio (High BitRate audio, name per HDMI spec)? If so, looking at our Windows driver, only a limited subset of our GPUs support this feature, based on audio codec ID:

0x000B (perhaps, based on silicon rev)
0x0011
0x0012
0x0013
0x0014
0x0018
0x0019
0x001A
0x001B
0x001C
(some of those IDs don't actually exist in production)

2) I don't recall what state the kernel driver is in for HBR support on NVIDIA codecs/controllers.

3) There are some AES control bits that I think indicate the format of the audio being sent to the receiver; I can't remember if these are bits that asound.conf needs to set or the application needs to specify when opening the ALSA device, or if they're something that's simply encoded into the SPDIF stream. Perhaps some receivers are better than others at interpreting audio if those bits aren't set correctly?

[Anssi]

Stephen, thanks a lot for your reply

Originally Posted by Stephen Warren

Some thoughts:

1) I believe these files are so-called "HBR" audio (High BitRate audio, name per HDMI spec)?

Yes.

Originally Posted by Stephen Warren

If so, looking at our Windows driver, only a limited subset of our GPUs support this feature, based on audio codec ID:

0x000B (perhaps, based on silicon rev)
0x0011
0x0012
0x0013
0x0014
0x0018
0x0019
0x001A
0x001B
0x001C
(some of those IDs don't actually exist in production)

Carl Eugen (non-working) and I (working) have both 0x000b codecs. Does "(perhaps, based on silicon rev)" mean that only a subset is supported? Both of the cards do have the HBR capability flag set, though.

Originally Posted by Stephen Warren

2) I don't recall what state the kernel driver is in for HBR support on NVIDIA codecs/controllers.

I added HBR mode support into ALSA several months ago, and the support is in 2.6.36. However, it is always possible I missed something (I just set the HBR flag in the pin (IIRC) and the non-pcm flag in the HDA stream format).

Originally Posted by Stephen Warren

3) There are some AES control bits that I think indicate the format of the audio being sent to the receiver; I can't remember if these are bits that asound.conf needs to set or the application needs to specify when opening the ALSA device, or if they're something that's simply encoded into the SPDIF stream. Perhaps some receivers are better than others at interpreting audio if those bits aren't set correctly?

ALSA configuration defines default AES bits, but they can be changed by using the device string or by manipulating mixer controls.
AFAICS the bits are set correctly, but I guess it wouldn't hurt to try some different values on affected systems. I believe most receivers ignore the AES bits, though.

I think the only relevant AES bit is the 0x02 (non-audio) bit. The AES bits are indeed encoded into the IEC958 stream, however that stream is produced in-hardware, while only the data payload (the IEC61937 stream) is sent to ALSA from userspace (+ the AES bits).

BTW, I got another success from a user with the following configuration:

• NVidia GT240
• 260.19.21
• self-compiled alsa on 2.6.32-26-generic
• Integra DTC-9.8

[Stephen Warren]

In /proc/asound/card0/codec#0 or equivalent, what's the revision ID. Also, if you grep rev /var/log/Xorg.0.log, what's there?

I'm surprised the GT 240 worked; I don't think the Windows driver allows that chip, but anyway...

[Anssi]

Originally Posted by Stephen Warren

In /proc/asound/card0/codec#0 or equivalent, what's the revision ID. Also, if you grep rev /var/log/Xorg.0.log, what's there?

Revision Id: 0x100100
For both me and Carl Eugen.

For Xorg.0.log, I have this (note that I don't have the problem, though):

[191291.402] (--) PCI:*(0:1:0:0) 10de:0a6c:17aa:215c rev 162, Mem @ 0xcc000000/16777216, 0xd0000000/268435456, 0xce000000/33554432, I/O @ 0x00002000/128, BIOS @ 0x????????/524288

I don't have a non-working system, maybe someone (Carl Eugen?) can get you the line from one if needed.

Originally Posted by Stephen Warren

I'm surprised the GT 240 worked; I don't think the Windows driver allows that chip, but anyway...

I'll ask the reporter for his codec#0.

[highrez]

Originally Posted by Stephen Warren

In /proc/asound/card0/codec#0 or equivalent, what's the revision ID. Also, if you grep rev /var/log/Xorg.0.log, what's there?

Here's my codec:
cat /proc/asound/NVidia_1/codec#3
Codec: Nvidia GPU 0d HDMI/DP
Address: 3
AFG Function Id: 0x1 (unsol 0)
Vendor Id: 0x10de000d
Subsystem Id: 0x10de0101
Revision Id: 0x100100

And xorg:
(--) PCI:*(0:2:0:0) 10de:0ca3:10de:075b nVidia Corporation rev 162, Mem @ 0xfd000000/16777216, 0xd0000000/268435456, 0xce000000/33554432, I/O @ 0x0000ec00/128, BIOS @ 0x????????/524288

Originally Posted by Stephen Warren

I'm surprised the GT 240 worked; I don't think the Windows driver allows that chip, but anyway...

DTS-HD passthrough works on my card - it just doesn't work "right". That is, I can hear the audio output on my receiver (Onkyo TX-NR807) but the HDMI light blinks like there's a sync issue. Additionally, because it doesn't appear to like the stream, it won't allow me to apply any DSP functions (e.g. THX listening modes) to the audio.

What cards do expect will work? Is encoded audio only 'officially supported' via Fermi based cards?

[cehoyos]

Originally Posted by Stephen Warren

In /proc/asound/card0/codec#0 or equivalent, what's the revision ID. Also, if you grep rev /var/log/Xorg.0.log, what's there?

$ cat /proc/asound/card1/codec#0
Codec: Nvidia GPU 0b HDMI/DP
Address: 0
AFG Function Id: 0x1 (unsol 0)
Vendor Id: 0x10de000b
Subsystem Id: 0x10de0101
Revision Id: 0x100100
No Modem Function Group found
Default PCM:
    rates [0x0]:
    bits [0x0]:
    formats [0x0]:
Default Amp-In caps: N/A
Default Amp-Out caps: N/A
GPIO: io=0, o=0, i=0, unsolicited=0, wake=0
Node 0x04 [Audio Output] wcaps 0x72b1: 8-Channels Digital Stripe CP
  Control: name="IEC958 Playback Con Mask", index=0, device=0
  Control: name="IEC958 Playback Pro Mask", index=0, device=0
  Control: name="IEC958 Playback Default", index=0, device=0
  Control: name="IEC958 Playback Switch", index=0, device=0
  Device: name="HDMI 0", type="HDMI", device=3
  Converter: stream=0, channel=0
  Digital: Enabled
  Digital category: 0x0
  PCM:
    rates [0x7f0]: 32000 44100 48000 88200 96000 176400 192000
    bits [0xe]: 16 20 24
    formats [0x5]: PCM AC3
  Unsolicited: tag=00, enabled=0
Node 0x05 [Pin Complex] wcaps 0x407381: 8-Channels Digital CP
  Pincap 0x09000094: OUT Detect HBR HDMI DP
  Pin Default 0x18560010: [Jack] Digital Out at Int HDMI
    Conn = Digital, Color = Unknown
    DefAssociation = 0x1, Sequence = 0x0
  Pin-ctls: 0x40: OUT
  Unsolicited: tag=05, enabled=1
  Connection: 1
     0x04
$ grep rev /var/log/Xorg.0.log
(--) PCI:*(0:1:0:0) 10de:0a65:1462:2011 rev 162, Mem @ 0x92000000/16777216, 0x80000000/268435456, 0x90000000/33554432, I/O @ 0x00002000/128, BIOS @ 0x????????/524288

I'm surprised the GT 240 worked; I don't think the Windows driver allows that chip, but anyway...

Hmm...

Carl Eugen

[Anssi]

Originally Posted by Stephen Warren

I'm surprised the GT 240 worked; I don't think the Windows driver allows that chip, but anyway...

If it is of interest, that was with codec id 000d, rev 0x100100, and:
(--) PCI:*(0:2:0:0) 10de:0ca3:3842:1240 nVidia Corporation GT215 [GeForce GT 240] rev 162, Mem @ 0xdd000000/16777216, 0xb0000000/268435456, 0xce000000/33554432, I/O @ 0x0000cf00/128, BIOS @ 0x????????/524288

And I've confirmed with the tester that it works perfectly (the Integra DTC-9.8 receiver
produces audio and the TrueHD/DTS-HD logos).