DSD over HDMI: audio data packetization

DSD over HDMI: audio data packetization

[Xuebing Wang]

Hi Community,

I am trying to implement DSD (one bit audio) over HDMI, for Intel
integrated graphics. I am referencing HDMI specification 1.3a.

Regarding audio data packetization, I am referencing spec:
5.3.4 Audio Sample Packet
7.6 Audio Data Packetization

Q1) Would you please point to me where is the source code to handle
L-PCM audio sample packetization (to implement 5.3.4 Audio Sample
Packet)?

Q2) I admit that I did not carefully read HD-Audio specification. Is
this possible that this L-PCM audio data packetization is done by
*hardware* (e.g HDMI related widgets/pins)?

Q3) Did we implement "7.2 Audio Sample Clock Capture and Regeneration"?

Thanks.
Xuebing Wang

Re: DSD over HDMI: audio data packetization

[Anssi Hannula]

31.12.2014, 15:32, Xuebing Wang kirjoitti:
> Hi Community,

Hi,

> I am trying to implement DSD (one bit audio) over HDMI, for Intel
> integrated graphics. I am referencing HDMI specification 1.3a.
> 
> Regarding audio data packetization, I am referencing spec:
> 5.3.4 Audio Sample Packet
> 7.6 Audio Data Packetization
> 
> Q1) Would you please point to me where is the source code to handle
> L-PCM audio sample packetization (to implement 5.3.4 Audio Sample
> Packet)?

It is done by the hardware.

> Q2) I admit that I did not carefully read HD-Audio specification. Is
> this possible that this L-PCM audio data packetization is done by
> *hardware* (e.g HDMI related widgets/pins)?

Yes.

> Q3) Did we implement "7.2 Audio Sample Clock Capture and Regeneration"?

That is done by hardware. With some HDA codecs the video driver
calculates the N/CTS values "manually", but AFAIK not on Intel HW.

> Thanks.
> Xuebing Wang
> 
> 


-- 
Anssi Hannula

Re: DSD over HDMI: audio data packetization

[Yang, Libin]

Hi Xuebing,


> -----Original Message-----
> From: Anssi Hannula [mailto:anssi.hannula@iki.fi]
> Sent: Wednesday, December 31, 2014 9:41 PM
> To: Xuebing Wang
> Cc: alsa-devel@alsa-project.org; Yang, Libin; Lin, Mengdong; ; Takashi Iwai; ;
> Wu Fengguang
> Subject: Re: [alsa-devel] DSD over HDMI: audio data packetization
> 
> 31.12.2014, 15:32, Xuebing Wang kirjoitti:
> > Hi Community,
> 
> Hi,
> 
> > I am trying to implement DSD (one bit audio) over HDMI, for Intel
> > integrated graphics. I am referencing HDMI specification 1.3a.
> >
> > Regarding audio data packetization, I am referencing spec:
> > 5.3.4 Audio Sample Packet
> > 7.6 Audio Data Packetization
> >
> > Q1) Would you please point to me where is the source code to handle
> > L-PCM audio sample packetization (to implement 5.3.4 Audio Sample
> > Packet)?
> 
> It is done by the hardware.
> 
> > Q2) I admit that I did not carefully read HD-Audio specification. Is
> > this possible that this L-PCM audio data packetization is done by
> > *hardware* (e.g HDMI related widgets/pins)?
> 
> Yes.
> 
> > Q3) Did we implement "7.2 Audio Sample Clock Capture and Regeneration"?
> 
> That is done by hardware. With some HDA codecs the video driver calculates
> the N/CTS values "manually", but AFAIK not on Intel HW.

As the N/CTS need fs and frequency of TMDS info, the hdmi driver does not
implement this feature.

We currently are designing an audio link between audio subsystem and other 
components, like video or USB.

Maybe we will add the N/CTS information transmission. So we can calculate
the N/CTS value in audio or gfx driver.


> 
> > Thanks.
> > Xuebing Wang
> >
> >
> 
> 
> --
> Anssi Hannula

Re: DSD over HDMI: audio data packetization

[Xuebing Wang]

Anssi, Libin,

Thank you very much for your help.

If L-PCM audio data packetization is done by hardware, is it possible to 
implement DSD over HDMI on Intel integrated graphics? Thanks again.


On 12/31/2014 10:28 PM, Yang, Libin wrote:
> Hi Xuebing,
>
>
>> -----Original Message-----
>> From: Anssi Hannula [mailto:anssi.hannula@iki.fi]
>> Sent: Wednesday, December 31, 2014 9:41 PM
>> To: Xuebing Wang
>> Cc: alsa-devel@alsa-project.org; Yang, Libin; Lin, Mengdong; ; Takashi Iwai; ;
>> Wu Fengguang
>> Subject: Re: [alsa-devel] DSD over HDMI: audio data packetization
>>
>> 31.12.2014, 15:32, Xuebing Wang kirjoitti:
>>> Hi Community,
>> Hi,
>>
>>> I am trying to implement DSD (one bit audio) over HDMI, for Intel
>>> integrated graphics. I am referencing HDMI specification 1.3a.
>>>
>>> Regarding audio data packetization, I am referencing spec:
>>> 5.3.4 Audio Sample Packet
>>> 7.6 Audio Data Packetization
>>>
>>> Q1) Would you please point to me where is the source code to handle
>>> L-PCM audio sample packetization (to implement 5.3.4 Audio Sample
>>> Packet)?
>> It is done by the hardware.
>>
>>> Q2) I admit that I did not carefully read HD-Audio specification. Is
>>> this possible that this L-PCM audio data packetization is done by
>>> *hardware* (e.g HDMI related widgets/pins)?
>> Yes.
>>
>>> Q3) Did we implement "7.2 Audio Sample Clock Capture and Regeneration"?
>> That is done by hardware. With some HDA codecs the video driver calculates
>> the N/CTS values "manually", but AFAIK not on Intel HW.
> As the N/CTS need fs and frequency of TMDS info, the hdmi driver does not
> implement this feature.
>
> We currently are designing an audio link between audio subsystem and other
> components, like video or USB.
>
> Maybe we will add the N/CTS information transmission. So we can calculate
> the N/CTS value in audio or gfx driver.
>
>
>>> Thanks.
>>> Xuebing Wang
>>>
>>>
>>
>> --
>> Anssi Hannula

-- 
Thanks,
Xuebing Wang

Re: DSD over HDMI: audio data packetization

[Anssi Hannula]

31.12.2014, 17:38, Xuebing Wang kirjoitti:
> Anssi, Libin,
> 
> Thank you very much for your help.
> 
> If L-PCM audio data packetization is done by hardware, is it possible to 
> implement DSD over HDMI on Intel integrated graphics? Thanks again.

I don't think so, unless:

(a) It is possible to somehow encapsulate the stream in Audio Sample
Packets and be accepted by the receiver (like e.g. AC3, DTS). I'm not
aware of such a way.

or:
(b) There is some undocumented feature in the Intel HDMI codecs allowing
the driver to switch to DSD or driver-constructed packets (instead of
ASP) in similar way as there is a way to switch to HBR packets.



> 
> On 12/31/2014 10:28 PM, Yang, Libin wrote:
>> Hi Xuebing,
>>
>>
>>> -----Original Message-----
>>> From: Anssi Hannula [mailto:anssi.hannula@iki.fi]
>>> Sent: Wednesday, December 31, 2014 9:41 PM
>>> To: Xuebing Wang
>>> Cc: alsa-devel@alsa-project.org; Yang, Libin; Lin, Mengdong; ; Takashi Iwai; ;
>>> Wu Fengguang
>>> Subject: Re: [alsa-devel] DSD over HDMI: audio data packetization
>>>
>>> 31.12.2014, 15:32, Xuebing Wang kirjoitti:
>>>> Hi Community,
>>> Hi,
>>>
>>>> I am trying to implement DSD (one bit audio) over HDMI, for Intel
>>>> integrated graphics. I am referencing HDMI specification 1.3a.
>>>>
>>>> Regarding audio data packetization, I am referencing spec:
>>>> 5.3.4 Audio Sample Packet
>>>> 7.6 Audio Data Packetization
>>>>
>>>> Q1) Would you please point to me where is the source code to handle
>>>> L-PCM audio sample packetization (to implement 5.3.4 Audio Sample
>>>> Packet)?
>>> It is done by the hardware.
>>>
>>>> Q2) I admit that I did not carefully read HD-Audio specification. Is
>>>> this possible that this L-PCM audio data packetization is done by
>>>> *hardware* (e.g HDMI related widgets/pins)?
>>> Yes.
>>>
>>>> Q3) Did we implement "7.2 Audio Sample Clock Capture and Regeneration"?
>>> That is done by hardware. With some HDA codecs the video driver calculates
>>> the N/CTS values "manually", but AFAIK not on Intel HW.
>> As the N/CTS need fs and frequency of TMDS info, the hdmi driver does not
>> implement this feature.
>>
>> We currently are designing an audio link between audio subsystem and other
>> components, like video or USB.
>>
>> Maybe we will add the N/CTS information transmission. So we can calculate
>> the N/CTS value in audio or gfx driver.
>>
>>
>>>> Thanks.
>>>> Xuebing Wang
>>>>
>>>>
>>>
>>> --
>>> Anssi Hannula
> 


-- 
Anssi Hannula

Re: DSD over HDMI: audio data packetization

[Yang, Libin]

Hi,

> -----Original Message-----
> From: Anssi Hannula [mailto:anssi.hannula@iki.fi]
> Sent: Wednesday, December 31, 2014 11:45 PM
> To: Xuebing Wang
> Cc: Yang, Libin; alsa-devel@alsa-project.org; Lin, Mengdong; ; Takashi Iwai; ;
> Wu Fengguang
> Subject: Re: [alsa-devel] DSD over HDMI: audio data packetization
> 
> 31.12.2014, 17:38, Xuebing Wang kirjoitti:
> > Anssi, Libin,
> >
> > Thank you very much for your help.
> >
> > If L-PCM audio data packetization is done by hardware, is it possible
> > to implement DSD over HDMI on Intel integrated graphics? Thanks again.
> 
> I don't think so, unless:
> 
> (a) It is possible to somehow encapsulate the stream in Audio Sample
> Packets and be accepted by the receiver (like e.g. AC3, DTS). I'm not aware of
> such a way.
> 
> or:
> (b) There is some undocumented feature in the Intel HDMI codecs allowing
> the driver to switch to DSD or driver-constructed packets (instead of
> ASP) in similar way as there is a way to switch to HBR packets.
> 

Yes, I agree. And I'm not sure whether HDMI spec supports DSD.

> 
> 
> >
> > On 12/31/2014 10:28 PM, Yang, Libin wrote:
> >> Hi Xuebing,
> >>
> >>
> >>> -----Original Message-----
> >>> From: Anssi Hannula [mailto:anssi.hannula@iki.fi]
> >>> Sent: Wednesday, December 31, 2014 9:41 PM
> >>> To: Xuebing Wang
> >>> Cc: alsa-devel@alsa-project.org; Yang, Libin; Lin, Mengdong; ;
> >>> Takashi Iwai; ; Wu Fengguang
> >>> Subject: Re: [alsa-devel] DSD over HDMI: audio data packetization
> >>>
> >>> 31.12.2014, 15:32, Xuebing Wang kirjoitti:
> >>>> Hi Community,
> >>> Hi,
> >>>
> >>>> I am trying to implement DSD (one bit audio) over HDMI, for Intel
> >>>> integrated graphics. I am referencing HDMI specification 1.3a.
> >>>>
> >>>> Regarding audio data packetization, I am referencing spec:
> >>>> 5.3.4 Audio Sample Packet
> >>>> 7.6 Audio Data Packetization
> >>>>
> >>>> Q1) Would you please point to me where is the source code to handle
> >>>> L-PCM audio sample packetization (to implement 5.3.4 Audio Sample
> >>>> Packet)?
> >>> It is done by the hardware.
> >>>
> >>>> Q2) I admit that I did not carefully read HD-Audio specification.
> >>>> Is this possible that this L-PCM audio data packetization is done
> >>>> by
> >>>> *hardware* (e.g HDMI related widgets/pins)?
> >>> Yes.
> >>>
> >>>> Q3) Did we implement "7.2 Audio Sample Clock Capture and
> Regeneration"?
> >>> That is done by hardware. With some HDA codecs the video driver
> >>> calculates the N/CTS values "manually", but AFAIK not on Intel HW.
> >> As the N/CTS need fs and frequency of TMDS info, the hdmi driver does
> >> not implement this feature.
> >>
> >> We currently are designing an audio link between audio subsystem and
> >> other components, like video or USB.
> >>
> >> Maybe we will add the N/CTS information transmission. So we can
> >> calculate the N/CTS value in audio or gfx driver.
> >>
> >>
> >>>> Thanks.
> >>>> Xuebing Wang
> >>>>
> >>>>
> >>>
> >>> --
> >>> Anssi Hannula
> >
> 
> 
> --
> Anssi Hannula

Re: DSD over HDMI: audio data packetization

[Anssi Hannula]

01.01.2015, 04:07, Yang, Libin kirjoitti:
> Hi,
> 
>> -----Original Message-----
>> From: Anssi Hannula [mailto:anssi.hannula@iki.fi]
>> Sent: Wednesday, December 31, 2014 11:45 PM
>> To: Xuebing Wang
>> Cc: Yang, Libin; alsa-devel@alsa-project.org; Lin, Mengdong; ; Takashi Iwai; ;
>> Wu Fengguang
>> Subject: Re: [alsa-devel] DSD over HDMI: audio data packetization
>>
>> 31.12.2014, 17:38, Xuebing Wang kirjoitti:
>>> Anssi, Libin,
>>>
>>> Thank you very much for your help.
>>>
>>> If L-PCM audio data packetization is done by hardware, is it possible
>>> to implement DSD over HDMI on Intel integrated graphics? Thanks again.
>>
>> I don't think so, unless:
>>
>> (a) It is possible to somehow encapsulate the stream in Audio Sample
>> Packets and be accepted by the receiver (like e.g. AC3, DTS). I'm not aware of
>> such a way.
>>
>> or:
>> (b) There is some undocumented feature in the Intel HDMI codecs allowing
>> the driver to switch to DSD or driver-constructed packets (instead of
>> ASP) in similar way as there is a way to switch to HBR packets.
>>
> 
> Yes, I agree. And I'm not sure whether HDMI spec supports DSD.

It does, via the "One Bit Audio" and "DST" formats (spec 5.3.9, 5.3.10
in 1.3a/1.4a).

>>
>>
>>>
>>> On 12/31/2014 10:28 PM, Yang, Libin wrote:
>>>> Hi Xuebing,
>>>>
>>>>
>>>>> -----Original Message-----
>>>>> From: Anssi Hannula [mailto:anssi.hannula@iki.fi]
>>>>> Sent: Wednesday, December 31, 2014 9:41 PM
>>>>> To: Xuebing Wang
>>>>> Cc: alsa-devel@alsa-project.org; Yang, Libin; Lin, Mengdong; ;
>>>>> Takashi Iwai; ; Wu Fengguang
>>>>> Subject: Re: [alsa-devel] DSD over HDMI: audio data packetization
>>>>>
>>>>> 31.12.2014, 15:32, Xuebing Wang kirjoitti:
>>>>>> Hi Community,
>>>>> Hi,
>>>>>
>>>>>> I am trying to implement DSD (one bit audio) over HDMI, for Intel
>>>>>> integrated graphics. I am referencing HDMI specification 1.3a.
>>>>>>
>>>>>> Regarding audio data packetization, I am referencing spec:
>>>>>> 5.3.4 Audio Sample Packet
>>>>>> 7.6 Audio Data Packetization
>>>>>>
>>>>>> Q1) Would you please point to me where is the source code to handle
>>>>>> L-PCM audio sample packetization (to implement 5.3.4 Audio Sample
>>>>>> Packet)?
>>>>> It is done by the hardware.
>>>>>
>>>>>> Q2) I admit that I did not carefully read HD-Audio specification.
>>>>>> Is this possible that this L-PCM audio data packetization is done
>>>>>> by
>>>>>> *hardware* (e.g HDMI related widgets/pins)?
>>>>> Yes.
>>>>>
>>>>>> Q3) Did we implement "7.2 Audio Sample Clock Capture and
>> Regeneration"?
>>>>> That is done by hardware. With some HDA codecs the video driver
>>>>> calculates the N/CTS values "manually", but AFAIK not on Intel HW.
>>>> As the N/CTS need fs and frequency of TMDS info, the hdmi driver does
>>>> not implement this feature.
>>>>
>>>> We currently are designing an audio link between audio subsystem and
>>>> other components, like video or USB.
>>>>
>>>> Maybe we will add the N/CTS information transmission. So we can
>>>> calculate the N/CTS value in audio or gfx driver.
>>>>
>>>>
>>>>>> Thanks.
>>>>>> Xuebing Wang
>>>>>>
>>>>>>
>>>>>
>>>>> --
>>>>> Anssi Hannula
>>>
>>
>>
>> --
>> Anssi Hannula


-- 
Anssi Hannula

Re: DSD over HDMI: audio data packetization

[Xuebing Wang]

Anssi, Libin,

Thanks for your help.

After a little bit of research, I do think it is possible to stream DSD
over HDMI (the mechanism is as Anssi mentioned, similar way as HBR).

Please see HDA spec:
3.7.1 Stream Format Structure
(If the TYPE is set to Non-PCM, the controller just pushes data over the
link and is not concerned
with formatting. The base rate, data type, and number of Words (MULT) to
send each valid frame
are specified to control the rate at which the non-PCM data is sent.)

7.3.3.8 Converter Format

What do you think? In order to implement this similiar mechanism
(similar to HBR), I still have some detailed questions that I will write
in another email. Thanks.


On Thu, 2015-01-01 at 06:37 +0200, Anssi Hannula wrote:
> 01.01.2015, 04:07, Yang, Libin kirjoitti:
> > Hi,
> > 
> >> -----Original Message-----
> >> From: Anssi Hannula [mailto:anssi.hannula@iki.fi]
> >> Sent: Wednesday, December 31, 2014 11:45 PM
> >> To: Xuebing Wang
> >> Cc: Yang, Libin; alsa-devel@alsa-project.org; Lin, Mengdong; ; Takashi Iwai; ;
> >> Wu Fengguang
> >> Subject: Re: [alsa-devel] DSD over HDMI: audio data packetization
> >>
> >> 31.12.2014, 17:38, Xuebing Wang kirjoitti:
> >>> Anssi, Libin,
> >>>
> >>> Thank you very much for your help.
> >>>
> >>> If L-PCM audio data packetization is done by hardware, is it possible
> >>> to implement DSD over HDMI on Intel integrated graphics? Thanks again.
> >>
> >> I don't think so, unless:
> >>
> >> (a) It is possible to somehow encapsulate the stream in Audio Sample
> >> Packets and be accepted by the receiver (like e.g. AC3, DTS). I'm not aware of
> >> such a way.
> >>
> >> or:
> >> (b) There is some undocumented feature in the Intel HDMI codecs allowing
> >> the driver to switch to DSD or driver-constructed packets (instead of
> >> ASP) in similar way as there is a way to switch to HBR packets.
> >>
> > 
> > Yes, I agree. And I'm not sure whether HDMI spec supports DSD.
> 
> It does, via the "One Bit Audio" and "DST" formats (spec 5.3.9, 5.3.10
> in 1.3a/1.4a).

Re: DSD over HDMI: audio data packetization

[Anssi Hannula]

04.01.2015, 05:10, Xuebing Wang kirjoitti:
> Anssi, Libin,
> 
> Thanks for your help.
> 
> After a little bit of research, I do think it is possible to stream DSD
> over HDMI (the mechanism is as Anssi mentioned, similar way as HBR).
> 
> Please see HDA spec:
> 3.7.1 Stream Format Structure
> (If the TYPE is set to Non-PCM, the controller just pushes data over the
> link and is not concerned
> with formatting. The base rate, data type, and number of Words (MULT) to
> send each valid frame
> are specified to control the rate at which the non-PCM data is sent.)
> 
> 7.3.3.8 Converter Format
> 
> What do you think? In order to implement this similiar mechanism
> (similar to HBR), I still have some detailed questions that I will write
> in another email. Thanks.

That setting will still not allow you to send One-bit audio or DST
packets. The stream will still be encapsulated in either ASP or HBR
packets, depending on the value of the EPT field (Encoded Packet Type,
7.3.3.13).


> 
> On Thu, 2015-01-01 at 06:37 +0200, Anssi Hannula wrote:
>> 01.01.2015, 04:07, Yang, Libin kirjoitti:
>>> Hi,
>>>
>>>> -----Original Message-----
>>>> From: Anssi Hannula [mailto:anssi.hannula@iki.fi]
>>>> Sent: Wednesday, December 31, 2014 11:45 PM
>>>> To: Xuebing Wang
>>>> Cc: Yang, Libin; alsa-devel@alsa-project.org; Lin, Mengdong; ; Takashi Iwai; ;
>>>> Wu Fengguang
>>>> Subject: Re: [alsa-devel] DSD over HDMI: audio data packetization
>>>>
>>>> 31.12.2014, 17:38, Xuebing Wang kirjoitti:
>>>>> Anssi, Libin,
>>>>>
>>>>> Thank you very much for your help.
>>>>>
>>>>> If L-PCM audio data packetization is done by hardware, is it possible
>>>>> to implement DSD over HDMI on Intel integrated graphics? Thanks again.
>>>>
>>>> I don't think so, unless:
>>>>
>>>> (a) It is possible to somehow encapsulate the stream in Audio Sample
>>>> Packets and be accepted by the receiver (like e.g. AC3, DTS). I'm not aware of
>>>> such a way.
>>>>
>>>> or:
>>>> (b) There is some undocumented feature in the Intel HDMI codecs allowing
>>>> the driver to switch to DSD or driver-constructed packets (instead of
>>>> ASP) in similar way as there is a way to switch to HBR packets.
>>>>
>>>
>>> Yes, I agree. And I'm not sure whether HDMI spec supports DSD.
>>
>> It does, via the "One Bit Audio" and "DST" formats (spec 5.3.9, 5.3.10
>> in 1.3a/1.4a).
> 
> 


-- 
Anssi Hannula

Re: DSD over HDMI: audio data packetization

[Anssi Hannula]

06.01.2015, 06:33, Andrej Falout kirjoitti:
> Hi Anssi,

Hi,

> Just wondering how Microsoft documentation fits into this discussion:
> 
> "Representing Formats for IEC 61937 Transmissions"
> 
> http://msdn.microsoft.com/en-us/library/windows/desktop/dd316761%28v=vs.85%29.aspx
> 
> 
> "Microsoft-provided HD Audio class driver supports PCM, AC3, DTS, AAC,
> Dolby Digital Plus, WMA Pro, MAT(MLP) formats. The GUIDs for the
> compressed audio formats that are not supported by the HD audio class
> driver and can be implemented by third-party solutions are listed in the
> following table. "
> 
> CEA 861 Type	SubFormat GUID	Description
> 0x08	
> 
> 00000008-0cea-0010-8000-00aa00389b71
> 
> KSDATAFORMAT_SUBTYPE_IEC61937_ATRAC
> 
> 	Adaptive Transform Acoustic Coding (ATRAC)
> 0x09	
> 
> 00000009-0cea-0010-8000-00aa00389b71
> 
> KSDATAFORMAT_SUBTYPE_IEC61937_ONE_BIT_AUDIO
> 
> 	One-Bit Audio
> 0x0d	
> 
> 0000000d-0cea-0010-8000-00aa00389b71
> 
> KSDATAFORMAT_SUBTYPE_IEC61937_DST
> 
> 	Direct Stream Transport (DST)—lossless compressed DSD (Direct Stream
> Digital).
> 
> 
> And knowing that IEC 61937 from HDMI 1.3 also supports above 6.144 Mbps...
> 
> Is there a doubt that above information is correct, and that DST/DST can
> be carried over 61937 in Microsoft HDA class driver, in this example, by
> "implementing a third-party solution" as described in this table?


I think the One-Bit Audio and DST formats are just misnamed here and
can't be carried over IEC 61937 (ATRAC can be) - the IEC 61937
specification does not cover One-Bit Audio or DST.

I think the "third-party solution" refers to a driver implementing
support for a non-HDA-spec HW feature which allows One-Bit audio / DST /
raw packets to be transmitted (in a similar way "EPT" (Encoded Packet
Type field in Pin Widget Control 7.3.3.13) allows setting HBR or ASP
packet transmission).


> Thanks,
> Andrej Falout
> 
> On Mon, Jan 5, 2015 at 4:46 PM, Anssi Hannula <anssi.hannula@iki.fi
> <mailto:anssi.hannula@iki.fi>> wrote:
> 
>     04.01.2015, 05:10, Xuebing Wang kirjoitti:
>     > Anssi, Libin,
>     >
>     > Thanks for your help.
>     >
>     > After a little bit of research, I do think it is possible to stream DSD
>     > over HDMI (the mechanism is as Anssi mentioned, similar way as HBR).
>     >
>     > Please see HDA spec:
>     > 3.7.1 Stream Format Structure
>     > (If the TYPE is set to Non-PCM, the controller just pushes data over the
>     > link and is not concerned
>     > with formatting. The base rate, data type, and number of Words (MULT) to
>     > send each valid frame
>     > are specified to control the rate at which the non-PCM data is sent.)
>     >
>     > 7.3.3.8 Converter Format
>     >
>     > What do you think? In order to implement this similiar mechanism
>     > (similar to HBR), I still have some detailed questions that I will write
>     > in another email. Thanks.
> 
>     That setting will still not allow you to send One-bit audio or DST
>     packets. The stream will still be encapsulated in either ASP or HBR
>     packets, depending on the value of the EPT field (Encoded Packet Type,
>     7.3.3.13).
> 
> 
>     >
>     > On Thu, 2015-01-01 at 06:37 +0200, Anssi Hannula wrote:
>     >> 01.01.2015, 04:07, Yang, Libin kirjoitti:
>     >>> Hi,
>     >>>
>     >>>> -----Original Message-----
>     >>>> From: Anssi Hannula [mailto:anssi.hannula@iki.fi
>     <mailto:anssi.hannula@iki.fi>]
>     >>>> Sent: Wednesday, December 31, 2014 11:45 PM
>     >>>> To: Xuebing Wang
>     >>>> Cc: Yang, Libin; alsa-devel@alsa-project.org
>     <mailto:alsa-devel@alsa-project.org>; Lin, Mengdong; ; Takashi Iwai; ;
>     >>>> Wu Fengguang
>     >>>> Subject: Re: [alsa-devel] DSD over HDMI: audio data packetization
>     >>>>
>     >>>> 31.12.2014, 17:38, Xuebing Wang kirjoitti:
>     >>>>> Anssi, Libin,
>     >>>>>
>     >>>>> Thank you very much for your help.
>     >>>>>
>     >>>>> If L-PCM audio data packetization is done by hardware, is it
>     possible
>     >>>>> to implement DSD over HDMI on Intel integrated graphics?
>     Thanks again.
>     >>>>
>     >>>> I don't think so, unless:
>     >>>>
>     >>>> (a) It is possible to somehow encapsulate the stream in Audio
>     Sample
>     >>>> Packets and be accepted by the receiver (like e.g. AC3, DTS).
>     I'm not aware of
>     >>>> such a way.
>     >>>>
>     >>>> or:
>     >>>> (b) There is some undocumented feature in the Intel HDMI codecs
>     allowing
>     >>>> the driver to switch to DSD or driver-constructed packets
>     (instead of
>     >>>> ASP) in similar way as there is a way to switch to HBR packets.
>     >>>>
>     >>>
>     >>> Yes, I agree. And I'm not sure whether HDMI spec supports DSD.
>     >>
>     >> It does, via the "One Bit Audio" and "DST" formats (spec 5.3.9,
>     5.3.10
>     >> in 1.3a/1.4a).
>     >
>     >
> 
> 
>     --
>     Anssi Hannula
> 
> 


-- 
Anssi Hannula
_______________________________________________
Alsa-devel mailing list
Alsa-devel@alsa-project.org
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel

Re: DSD over HDMI: audio data packetization

[Xuebing Wang]

Andrej,

I have a slight doubt about
"KSDATAFORMAT_SUBTYPE_IEC61937_ONE_BIT_AUDIO".

To be concise "IEC61937_ONE_BIT_AUDIO"

I do not have all IEC 61937 specs, I can not find
"IEC61937_ONE_BIT_AUDIO" in the list below.

Part 1: General
Part 2: Burst-info
Part 3: Non-linear PCM bitstreams according to the AC-3 format
Part 4: Non-linear PCM bitstreams according to the MPEG audio format
Part 5: Non-linear PCM bitstreams according to the DTS (Digital Theater
Systems) format(s)
Part 6: Non-linear PCM bitstreams according to the MPEG-2 AAC and MPEG-4
AAC audio formats
Part 7: Non-linear PCM bitstreams according to the ATRAC, ATRAC2/3 and
ATRAC-X formats
Part 8: Non-linear PCM bitstreams according to the Windows Media Audio
(WMA) Professional format
Part 9: Non-linear PCM bitstreams according to the MAT format
Part 10: Non-linear PCM bitstreams according to the MPEG-4 audio
lossless coding (ALS) format - Edition 1.0
Part 11: MPEG-4 AAC and its extensions in LATM/LOAS
Part 12: Non-linear PCM bitstreams according to the DRA formats

Thanks.


On Tue, 2015-01-06 at 17:33 +1300, Andrej Falout wrote:
> Hi Anssi,
> 
> 
> Just wondering how Microsoft documentation fits into this discussion:
> 
> "Representing Formats for IEC 61937 Transmissions"
> 
> http://msdn.microsoft.com/en-us/library/windows/desktop/dd316761%
> 28v=vs.85%29.aspx
> 
> 
> "Microsoft-provided HD Audio class driver supports PCM, AC3, DTS, AAC,
> Dolby Digital Plus, WMA Pro, MAT(MLP) formats. The GUIDs for the
> compressed audio formats that are not supported by the HD audio class
> driver and can be implemented by third-party solutions are listed in
> the following table. "
> 
>      CEA 861 Type
>     SubFormat GUID
>       Description
> 0x08
> 00000008-0cea-0010-8000-00aa00389b71
> 
> KSDATAFORMAT_SUBTYPE_IEC61937_ATRAC
> 
> 
> Adaptive Transform
> Acoustic Coding (ATRAC)
> 0x09
> 00000009-0cea-0010-8000-00aa00389b71
> 
> KSDATAFORMAT_SUBTYPE_IEC61937_ONE_BIT_AUDIO
> 
> 
> One-Bit Audio
> 0x0d
> 0000000d-0cea-0010-8000-00aa00389b71
> 
> KSDATAFORMAT_SUBTYPE_IEC61937_DST
> 
> 
> Direct Stream Transport
> (DST)—lossless
> compressed DSD (Direct
> Stream Digital).
> 
> 
> And knowing that IEC 61937 from HDMI 1.3 also supports above 6.144
> Mbps...
> 
> 
> Is there a doubt that above information is correct, and that DST/DST
> can be carried over 61937 in Microsoft HDA class driver, in this
> example, by "implementing a third-party solution" as described in this
> table?
> 
> 
> Thanks,
> 
> Andrej Falout
> 
> 
> On Mon, Jan 5, 2015 at 4:46 PM, Anssi Hannula <anssi.hannula@iki.fi>
> wrote:
>         04.01.2015, 05:10, Xuebing Wang kirjoitti:
>         > Anssi, Libin,
>         >
>         > Thanks for your help.
>         >
>         > After a little bit of research, I do think it is possible to
>         stream DSD
>         > over HDMI (the mechanism is as Anssi mentioned, similar way
>         as HBR).
>         >
>         > Please see HDA spec:
>         > 3.7.1 Stream Format Structure
>         > (If the TYPE is set to Non-PCM, the controller just pushes
>         data over the
>         > link and is not concerned
>         > with formatting. The base rate, data type, and number of
>         Words (MULT) to
>         > send each valid frame
>         > are specified to control the rate at which the non-PCM data
>         is sent.)
>         >
>         > 7.3.3.8 Converter Format
>         >
>         > What do you think? In order to implement this similiar
>         mechanism
>         > (similar to HBR), I still have some detailed questions that
>         I will write
>         > in another email. Thanks.
>         
>         That setting will still not allow you to send One-bit audio or
>         DST
>         packets. The stream will still be encapsulated in either ASP
>         or HBR
>         packets, depending on the value of the EPT field (Encoded
>         Packet Type,
>         7.3.3.13).
>         
>         
>         >
>         > On Thu, 2015-01-01 at 06:37 +0200, Anssi Hannula wrote:
>         >> 01.01.2015, 04:07, Yang, Libin kirjoitti:
>         >>> Hi,
>         >>>
>         >>>> -----Original Message-----
>         >>>> From: Anssi Hannula [mailto:anssi.hannula@iki.fi]
>         >>>> Sent: Wednesday, December 31, 2014 11:45 PM
>         >>>> To: Xuebing Wang
>         >>>> Cc: Yang, Libin; alsa-devel@alsa-project.org; Lin,
>         Mengdong; ; Takashi Iwai; ;
>         >>>> Wu Fengguang
>         >>>> Subject: Re: [alsa-devel] DSD over HDMI: audio data
>         packetization
>         >>>>
>         >>>> 31.12.2014, 17:38, Xuebing Wang kirjoitti:
>         >>>>> Anssi, Libin,
>         >>>>>
>         >>>>> Thank you very much for your help.
>         >>>>>
>         >>>>> If L-PCM audio data packetization is done by hardware,
>         is it possible
>         >>>>> to implement DSD over HDMI on Intel integrated graphics?
>         Thanks again.
>         >>>>
>         >>>> I don't think so, unless:
>         >>>>
>         >>>> (a) It is possible to somehow encapsulate the stream in
>         Audio Sample
>         >>>> Packets and be accepted by the receiver (like e.g. AC3,
>         DTS). I'm not aware of
>         >>>> such a way.
>         >>>>
>         >>>> or:
>         >>>> (b) There is some undocumented feature in the Intel HDMI
>         codecs allowing
>         >>>> the driver to switch to DSD or driver-constructed packets
>         (instead of
>         >>>> ASP) in similar way as there is a way to switch to HBR
>         packets.
>         >>>>
>         >>>
>         >>> Yes, I agree. And I'm not sure whether HDMI spec supports
>         DSD.
>         >>
>         >> It does, via the "One Bit Audio" and "DST" formats (spec
>         5.3.9, 5.3.10
>         >> in 1.3a/1.4a).
>         >
>         >
>         
>         
>         
>         --
>         Anssi Hannula
> 
> 


_______________________________________________
Alsa-devel mailing list
Alsa-devel@alsa-project.org
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel

Re: DSD over HDMI: audio data packetization

[Andrej Falout]

Hi Anssi,

I think the One-Bit Audio and DST formats are just misnamed here and
> can't be carried over IEC 61937 (ATRAC can be) - the IEC 61937
> specification does not cover One-Bit Audio or DST.
>

Does it have to be? Given that it is defined in CEA 861, and that 61937-1
makes specific provisions for non-PCM data:

Page 16:

"In cases where the interface is used to convey non-linear PCM audio
bitstreams..."



>
> I think the "third-party solution" refers to a driver implementing
> support for a non-HDA-spec HW feature which allows One-Bit audio / DST /
> raw packets to be transmitted (in a similar way "EPT" (Encoded Packet
> Type field in Pin Widget Control 7.3.3.13) allows setting HBR or ASP
> packet transmission).
>
>
Why would an "HW feature" be required to "allows One-Bit audio / DST / raw
packets to be transmitted"?

What would such "HW feature" do? What could it possibly do?


Thanks,
Andrej Falout




>
> > Thanks,
> > Andrej Falout
> >
> > On Mon, Jan 5, 2015 at 4:46 PM, Anssi Hannula <anssi.hannula@iki.fi
> > <mailto:anssi.hannula@iki.fi>> wrote:
> >
> >     04.01.2015, 05:10, Xuebing Wang kirjoitti:
> >     > Anssi, Libin,
> >     >
> >     > Thanks for your help.
> >     >
> >     > After a little bit of research, I do think it is possible to
> stream DSD
> >     > over HDMI (the mechanism is as Anssi mentioned, similar way as
> HBR).
> >     >
> >     > Please see HDA spec:
> >     > 3.7.1 Stream Format Structure
> >     > (If the TYPE is set to Non-PCM, the controller just pushes data
> over the
> >     > link and is not concerned
> >     > with formatting. The base rate, data type, and number of Words
> (MULT) to
> >     > send each valid frame
> >     > are specified to control the rate at which the non-PCM data is
> sent.)
> >     >
> >     > 7.3.3.8 Converter Format
> >     >
> >     > What do you think? In order to implement this similiar mechanism
> >     > (similar to HBR), I still have some detailed questions that I will
> write
> >     > in another email. Thanks.
> >
> >     That setting will still not allow you to send One-bit audio or DST
> >     packets. The stream will still be encapsulated in either ASP or HBR
> >     packets, depending on the value of the EPT field (Encoded Packet
> Type,
> >     7.3.3.13).
> >
> >
> >     >
> >     > On Thu, 2015-01-01 at 06:37 +0200, Anssi Hannula wrote:
> >     >> 01.01.2015, 04:07, Yang, Libin kirjoitti:
> >     >>> Hi,
> >     >>>
> >     >>>> -----Original Message-----
> >     >>>> From: Anssi Hannula [mailto:anssi.hannula@iki.fi
> >     <mailto:anssi.hannula@iki.fi>]
> >     >>>> Sent: Wednesday, December 31, 2014 11:45 PM
> >     >>>> To: Xuebing Wang
> >     >>>> Cc: Yang, Libin; alsa-devel@alsa-project.org
> >     <mailto:alsa-devel@alsa-project.org>; Lin, Mengdong; ; Takashi
> Iwai; ;
> >     >>>> Wu Fengguang
> >     >>>> Subject: Re: [alsa-devel] DSD over HDMI: audio data
> packetization
> >     >>>>
> >     >>>> 31.12.2014, 17:38, Xuebing Wang kirjoitti:
> >     >>>>> Anssi, Libin,
> >     >>>>>
> >     >>>>> Thank you very much for your help.
> >     >>>>>
> >     >>>>> If L-PCM audio data packetization is done by hardware, is it
> >     possible
> >     >>>>> to implement DSD over HDMI on Intel integrated graphics?
> >     Thanks again.
> >     >>>>
> >     >>>> I don't think so, unless:
> >     >>>>
> >     >>>> (a) It is possible to somehow encapsulate the stream in Audio
> >     Sample
> >     >>>> Packets and be accepted by the receiver (like e.g. AC3, DTS).
> >     I'm not aware of
> >     >>>> such a way.
> >     >>>>
> >     >>>> or:
> >     >>>> (b) There is some undocumented feature in the Intel HDMI codecs
> >     allowing
> >     >>>> the driver to switch to DSD or driver-constructed packets
> >     (instead of
> >     >>>> ASP) in similar way as there is a way to switch to HBR packets.
> >     >>>>
> >     >>>
> >     >>> Yes, I agree. And I'm not sure whether HDMI spec supports DSD.
> >     >>
> >     >> It does, via the "One Bit Audio" and "DST" formats (spec 5.3.9,
> >     5.3.10
> >     >> in 1.3a/1.4a).
> >     >
> >     >
> >
> >
> >     --
> >     Anssi Hannula
> >
> >
>
>
> --
> Anssi Hannula
>

Re: DSD over HDMI: audio data packetization

[Anssi Hannula]

06.01.2015, 08:15, Andrej Falout kirjoitti:
> Hi Anssi,

Hi,

>> I think the One-Bit Audio and DST formats are just misnamed here and
>> can't be carried over IEC 61937 (ATRAC can be) - the IEC 61937
>> specification does not cover One-Bit Audio or DST.
>>
> 
> Does it have to be? Given that it is defined in CEA 861, and that 61937-1
> makes specific provisions for non-PCM data:
> 
> Page 16:
> 
> "In cases where the interface is used to convey non-linear PCM audio
> bitstreams..."

IEC 61937 *is* about transmitting non-PCM data. The specification
specifies how to encapsulate AC-3, MPEG-x, DTS, ATRAC, WMA etc for
transmission over IEC 60958 link. It does not specify anything about
One-Bit Audio or DST (see e.g. "Table 2 - Data-types" in IEC 61937-2 for
a list of supported types of which you need to select one).

>>
>> I think the "third-party solution" refers to a driver implementing
>> support for a non-HDA-spec HW feature which allows One-Bit audio / DST /
>> raw packets to be transmitted (in a similar way "EPT" (Encoded Packet
>> Type field in Pin Widget Control 7.3.3.13) allows setting HBR or ASP
>> packet transmission).
>>
>>
> Why would an "HW feature" be required to "allows One-Bit audio / DST / raw
> packets to be transmitted"?
> 
> What would such "HW feature" do? What could it possibly do?

It would allow transmission of One Bit Audio Sample Packets (HDMI spec
1.3a 5.3.9) or DST Audio Packets (HDMI spec 5.3.10), which is not
possible with HW that simply implements HDA specification. The HDA
specification only supports transmission of Audio Sample Packets (HDMI
spec 5.3.4) and High-Bitrate Audio Stream Packets (HDMI spec 5.3.11),
selectable via the EPT field (HDA spec 7.3.3.13).
The packet header (which e.g. identifies the packet type) is generated
by HW on HDA cards.

Uncompressed PCM is transmitted using ASP packets.
IEC 61937 is transmitted using ASP packets (<=6.144Mbps) or HBR packets
(>6.144Mbps).


> 
> Thanks,
> Andrej Falout
> 
> 
> 
> 
>>
>>> Thanks,
>>> Andrej Falout
>>>
>>> On Mon, Jan 5, 2015 at 4:46 PM, Anssi Hannula <anssi.hannula@iki.fi
>>> <mailto:anssi.hannula@iki.fi>> wrote:
>>>
>>>     04.01.2015, 05:10, Xuebing Wang kirjoitti:
>>>     > Anssi, Libin,
>>>     >
>>>     > Thanks for your help.
>>>     >
>>>     > After a little bit of research, I do think it is possible to
>> stream DSD
>>>     > over HDMI (the mechanism is as Anssi mentioned, similar way as
>> HBR).
>>>     >
>>>     > Please see HDA spec:
>>>     > 3.7.1 Stream Format Structure
>>>     > (If the TYPE is set to Non-PCM, the controller just pushes data
>> over the
>>>     > link and is not concerned
>>>     > with formatting. The base rate, data type, and number of Words
>> (MULT) to
>>>     > send each valid frame
>>>     > are specified to control the rate at which the non-PCM data is
>> sent.)
>>>     >
>>>     > 7.3.3.8 Converter Format
>>>     >
>>>     > What do you think? In order to implement this similiar mechanism
>>>     > (similar to HBR), I still have some detailed questions that I will
>> write
>>>     > in another email. Thanks.
>>>
>>>     That setting will still not allow you to send One-bit audio or DST
>>>     packets. The stream will still be encapsulated in either ASP or HBR
>>>     packets, depending on the value of the EPT field (Encoded Packet
>> Type,
>>>     7.3.3.13).
>>>
>>>
>>>     >
>>>     > On Thu, 2015-01-01 at 06:37 +0200, Anssi Hannula wrote:
>>>     >> 01.01.2015, 04:07, Yang, Libin kirjoitti:
>>>     >>> Hi,
>>>     >>>
>>>     >>>> -----Original Message-----
>>>     >>>> From: Anssi Hannula [mailto:anssi.hannula@iki.fi
>>>     <mailto:anssi.hannula@iki.fi>]
>>>     >>>> Sent: Wednesday, December 31, 2014 11:45 PM
>>>     >>>> To: Xuebing Wang
>>>     >>>> Cc: Yang, Libin; alsa-devel@alsa-project.org
>>>     <mailto:alsa-devel@alsa-project.org>; Lin, Mengdong; ; Takashi
>> Iwai; ;
>>>     >>>> Wu Fengguang
>>>     >>>> Subject: Re: [alsa-devel] DSD over HDMI: audio data
>> packetization
>>>     >>>>
>>>     >>>> 31.12.2014, 17:38, Xuebing Wang kirjoitti:
>>>     >>>>> Anssi, Libin,
>>>     >>>>>
>>>     >>>>> Thank you very much for your help.
>>>     >>>>>
>>>     >>>>> If L-PCM audio data packetization is done by hardware, is it
>>>     possible
>>>     >>>>> to implement DSD over HDMI on Intel integrated graphics?
>>>     Thanks again.
>>>     >>>>
>>>     >>>> I don't think so, unless:
>>>     >>>>
>>>     >>>> (a) It is possible to somehow encapsulate the stream in Audio
>>>     Sample
>>>     >>>> Packets and be accepted by the receiver (like e.g. AC3, DTS).
>>>     I'm not aware of
>>>     >>>> such a way.
>>>     >>>>
>>>     >>>> or:
>>>     >>>> (b) There is some undocumented feature in the Intel HDMI codecs
>>>     allowing
>>>     >>>> the driver to switch to DSD or driver-constructed packets
>>>     (instead of
>>>     >>>> ASP) in similar way as there is a way to switch to HBR packets.
>>>     >>>>
>>>     >>>
>>>     >>> Yes, I agree. And I'm not sure whether HDMI spec supports DSD.
>>>     >>
>>>     >> It does, via the "One Bit Audio" and "DST" formats (spec 5.3.9,
>>>     5.3.10
>>>     >> in 1.3a/1.4a).
>>>     >
>>>     >
>>>
>>>
>>>     --
>>>     Anssi Hannula
>>>
>>>
>>
>>
>> --
>> Anssi Hannula
>>
> 


-- 
Anssi Hannula