MediaPortal Audio renderer - better video playback quality (5 Viewers)

[robyf]

Tested latest binaries on my secondary HTPC, ION330 with NVIDIA (audio through HDMI).

As stated I had to disable WASAPI exclusive mode to make it work.

Everything seems to work smooth, even HD TV. So even an ATOM processor seems to manage the load.

Unfortunately I'm not able to run stats without frame drops (here the ATOM or the 9400 gives up) so no screenshot.

I post audiorenderer.log.

This is a 25fps MKV/H.264 at 50Hhz.

 

[tourettes]

Tested latest binaries on my secondary HTPC, ION330 with NVIDIA (audio through HDMI).

As stated I had to disable WASAPI exclusive mode to make it work.

Everything seems to work smooth, even HD TV. So even an ATOM processor seems to manage the load.

Unfortunately I'm not able to run stats without frame drops (here the ATOM or the 9400 gives up) so no screenshot.

I post audiorenderer.log.

This is a 25fps MKV/H.264 at 50Hhz.

Could you make sure that the audio renderer is indeed used? Looks like it wouldn't be used since the WASAPI device is not accepting the format.

Also I think there is a bug in the current binary that causes the WASAPI shared / non-exclusive mode to fail on all devices (it should be already fixed in SVN)

09-08-2010 23:30:38.475 [ 76c] CreateAudioClient
09-08-2010 23:30:38.476 [ 76c] CreateAudioClient success
09-08-2010 23:30:38.476 [ 76c] CheckMediaType WASAPI client refused the format, used mix format:
09-08-2010 23:30:38.476 [ 76c] WAVEFORMATEX:
09-08-2010 23:30:38.476 [ 76c] nAvgBytesPerSec 384000
09-08-2010 23:30:38.476 [ 76c] nBlockAlign 8
09-08-2010 23:30:38.476 [ 76c] nChannels 2
09-08-2010 23:30:38.476 [ 76c] nSamplesPerSec 48000
09-08-2010 23:30:38.476 [ 76c] wBitsPerSample 32
09-08-2010 23:30:38.476 [ 76c] wFormatTag 65534
09-08-2010 23:30:38.476 [ 76c] WAVE_FORMAT_EXTENSIBLE
09-08-2010 23:30:38.476 [ 76c] dwChannelMask 3
09-08-2010 23:30:38.476 [ 76c] GUID {00000003-0000-0010-8000-00AA00389B71}

 

[tourettes]

I took a play with Matthew van Eerde's exclusive WASAPI code on the Nvidia HDMI drivers and I have to admit defeat - I can't get anything to go past the initialization in exclusive mode (0x80070057). The drivers do provide proper errors e.g. 0x88890008 for formats which aren't allowed. When I try the same tests on an intel and AMD based chipset I do get my test sound outs so it doesn't look like anything I've done is wrong - and that's always a real possibility. So I suspect the drivers (at least v1.0.15.0) don't work particularly well and it may be a lost cause trying to make it work without Nvidia fixing some of the bugs in the driver.

I'm 99% sure the WASAPI exclusive mode bug is in NVIDIA's drivers. I tried to contact them, but I havent heard anything back (and probably wont).

So, does anyone have any good contacts to NVIDIA people?

 

[tourettes]

Try using the registry setting to force the DirectSound into use. It could be that there is some WASAPI related code that is tried to be run on XP (unintentionaly).

Thanks, that did the trick

Test results:
Video and audio are perfectly in sync and video is played smoothly. Some tearing is visible and Audiorenderer.log is full of messages 'receiveSamples: Output sample queue was empty!'. If I disable timestretching tearing and those messages are gone (but so is perfect play too of course). Logs attached...

Tearing is not related to the audio renderer. Most likely DirectX Exclusive mode is not enabled in MP config.

'receiveSamples: Output sample queue was empty!

Known issue. I'll add it to the known errors list.

 

[robyf]

Tested latest binaries on my secondary HTPC, ION330 with NVIDIA (audio through HDMI).

As stated I had to disable WASAPI exclusive mode to make it work.

Everything seems to work smooth, even HD TV. So even an ATOM processor seems to manage the load.

Unfortunately I'm not able to run stats without frame drops (here the ATOM or the 9400 gives up) so no screenshot.

I post audiorenderer.log.

This is a 25fps MKV/H.264 at 50Hhz.

Could you make sure that the audio renderer is indeed used? Looks like it wouldn't be used since the WASAPI device is not accepting the format.

Also I think there is a bug in the current binary that causes the WASAPI shared / non-exclusive mode to fail on all devices (it should be already fixed in SVN)

Should I check the graph to be sure it's used?

 

[tourettes]

Should I check the graph to be sure it's used?

Yes. But since there is the known issue with WASAPI shared mode I already know what the graph is going to look - audio renderer is not connected

 

[davidf]

I took a play with Matthew van Eerde's exclusive WASAPI code on the Nvidia HDMI drivers and I have to admit defeat - I can't get anything to go past the initialization in exclusive mode (0x80070057). The drivers do provide proper errors e.g. 0x88890008 for formats which aren't allowed. When I try the same tests on an intel and AMD based chipset I do get my test sound outs so it doesn't look like anything I've done is wrong - and that's always a real possibility. So I suspect the drivers (at least v1.0.15.0) don't work particularly well and it may be a lost cause trying to make it work without Nvidia fixing some of the bugs in the driver.

I'm 99% sure the WASAPI exclusive mode bug is in NVIDIA's drivers. I tried to contact them, but I havent heard anything back (and probably wont).

So, does anyone have any good contacts to NVIDIA people?

I have quite a few in MS, it might be worth seeing if they will take any notice - although the ADC rep is really going to wonder why I am asking about WASAPI (although it won't be the first Mediaportal query he's got from me).

 

[red5goahead]

Probably also I'll try wasapi in an other popular high -end card , the Asus Xonar DX because I'll buy it switching from current Terratec Aureon and its C-Media 8768+ .

 

[tourettes]

I have quite a few in MS, it might be worth seeing if they will take any notice - although the ADC rep is really going to wonder why I am asking about WASAPI (although it won't be the first Mediaportal query he's got from me).

It is worth a try at least Currently I'm just hoping to have a good luck when it comes to the NVIDIA's response. It would be really nice to get some good contacts for the HW manufacturers like Intel, NVIDA and ATI, after all there is quite big group of MP users that would benefit from such.

 

[tourettes]

Just some "minor" update. We are currently internaly testing multichannel, SPDIF and 24,32 & float. Seems that there are few (althou much fewer than I had expected) issues that we need to fix before we can publish the next binary for public testing.

...one quite nasty thing is that I cannot debug or try to check the lipsync with SPDIF in use since AMP will disable HDMI output to monitor at that stage (and I'm not a fan of toggling cables hundreds of times per hour).