[Schenckmeier]

argh, sorry for crying here...
XBMC was searching for "xbmc_video60" mysql-database instead of "xbmc_video58".
All is fine now, many thanks for this new release.

[gfisher]

Hi Bvan58

XFCE desktop is a good option for a fully working desktop. Originally, I tried ubuntu 11.10 32 desktop unity 2/3d but XBMC and VLC had framerate juddery problems on 11.11 ati. I installed XFCE on ubuntu and 99% of it was good! Not as pretty but worked. Defaulting XFCE was clunky on ubuntu. Got tired of formatting when errors ocurred with experimentation leading to xorg corruption. So I went Xubuntu and XFCE is default. XBMC plays much better as does VLC. Xubuntu hdmi audio stinks for me. Pulse audio volume control is needed to set output source to hdmi and disable optical, no bios switch for me. I have not tried 12.1 on gnome 3 unity so maybe it is fixed? With the latest XVBA build it is so good I hate to break it! My WAF is very good right now.

[fritsch]

Some timeline news from AMD (source: https://twitter.com/#!/CatalystCreator)

Catalyst 12.2: 5/6 of March
Catalyst 12.3: around March 21

[.:B:.]

Nothing about improved XvBA support I assume?

[konti]

I know the XvBA gpu rendering is still kind of experimental. But it looks to me there is being made allot of progress here. I've been reading through this thread, but still have some questions. As I think there is allot of knowledge here, i was hoping someone can answer them.

I'm running an ION1 system now. When watching HD channels (1080i), i need to use deinterlacing. The ION optimized option can barely do this. Fast scenes are slowing down.

Is the XvBA inplementation at a point were it can perform this deinterlacing on a E350?
Does anyone has experience with this?

[fritsch]

Good question.

VDPAU internal deinterlacing is state of the art. It is a temporal / spatial algorithm, which uses more than one frame to calculate a smooth, sharp image.

XVBA also does internal deinterlacing, but without any interaction by us. The way they are doing it is at the quality of BOB which looks compared to the nvidia version improvable.

So to conclude:
1080i will work fine deinterlaced on XVBA, but the quality is the same as within ION-1 + BOB deinterlacing.

Btw:
Fernetmenta is working on a new vdpau architecture which should allow ION-1 with the advanced internal ion deinterlacing. There are already some testimages for openelec available.

[konti]

Tnx Fritsch! It's amazing to see how far the XVBA implementation has come in such a short relative time. Do you know if the ION1 + BOB deinterlacing is being performed by ION or CPU? Then I can just compare BOB with the ION Optimized option.

It's also impressive that the full potential of the ION1 platform has also not been reached yet. This is enough for me to not start thinking of upgrading to the E350 platform just yet.

[abudabi]

New version: 2:11.0~git20120301 ??

Any clue as to a changelog?

[fritsch]

only upstream changes because of a rebase.

[wsnipex]

new packages available: 2:11.0~git20120306.2f8921a-0ubuntu1~ppa1
- xvba: correct order in clear down of gl resources. Fixes some flickering

[fat-tony]

I'm using xbmc on a Fusion motherboard for a couple of months now (openelec nightly 64 bit builds - now on Eden RC). Video performance is excellent with low cpu usage on H.264 TV. Generally audio is satisfactory with passthrough of DD 5.1 and DTS etc working over HDMI. What I can't seem to get working is multichannel (5.1) PCM. I have numerous FLAC files with 5.1 audio and they always play back as two channel. Paplayer just plays the front left and right channels. DVDplayer seems to mix down the back channels to the front channels, but only when I set the audio output to Opt/Coax instead of HDMI. Same hardware on Windows 7 will play multichannel PCM using xbmc.

Rather than start posting logs and asound conf files (I have tried lots of permutations!), I just thought I'd check to see if anyone actually has multichannel PCM working over HDMI on the Fusion platform. I'm somewhat confused by the use of the intel sound driver in the kernel, maybe this is part of the issue :confused2:

MSI E350IA-E45 motherboard. HDMI input to Onkyo SR875.

Code:

root ~ # lspci -nnk
00:00.0 Host bridge [0600]: Advanced Micro Devices [AMD] Family 14h Processor Root Complex [1022:1510]
        Subsystem: Advanced Micro Devices [AMD] Family 14h Processor Root Complex [1022:1510]
00:01.0 VGA compatible controller [0300]: Advanced Micro Devices [AMD] nee ATI Wrestler [Radeon HD 6310] [1002:9802]
        Subsystem: Micro-Star International Co., Ltd. Device [1462:7698]
        Kernel driver in use: fglrx_pci
        Kernel modules: fglrx
00:01.1 Audio device [0403]: Advanced Micro Devices [AMD] nee ATI Wrestler HDMI Audio [Radeon HD 6250/6310] [1002:1314]
        Subsystem: Micro-Star International Co., Ltd. Device [1462:7698]
        Kernel driver in use: snd_hda_intel
        Kernel modules: snd-hda-intel

[fritsch]

Sorry. No idea about the flac issue, but the xvba patches do not touch audio in any way, so this seems to be a common xbmc on linux problem then?

Don`t worry about the intel driver. The internal soundcard is intel-hda compatible, so this driver is loaded. HDMI output directly goes through the hdmi port on you graphicscard.

You can check this with aplay -L and you will see at least two devices.

[fat-tony]

HDMI works fine in stereo PCM or compressed (DD or DTS) mutichannel audio for me. It would just seem that multichannel PCM on HDMI works only on Nvidia cards, not ATI. Just waiting on someone to confirm this for sure, then I will wait for the AE development to hit linux binaries....

[fritsch]

As we are waiting now for a really long time for AMD to fix the three things still missing for a perfect htpc experience:

- H264 Level 5.1 Support
- Mpeg2 bitstream decoding support
- Advanced Deinterlacing Support

I have opened a bugreport at the catalyst bugtracker: http://ati.cchtml.com/show_bug.cgi?id=448

If you can, just give your 5 cents, provide screenshots which show the garbage in case of Level 5.1. Or which show red or purple lines at the bottom / top cause of bad deinterlacing.

You can also provide Screenshots with "o" pressed which show huge cpu usage when decoding Mpeg-2 1080p/i content.

I want AMD to see that there is a community which waits for more support by AMD. That is the reason why i do not provide these files just "alone". They shall see that a lot of people are affected by their linux driver and xvba policy.

Important: Please be polite, not shitstorm. End of March there is coming out a new Fusion APU, pretty worth buying it, if the support would be a bit better.
Important 2: Do not discuss other bugs concerning xvba implementation in this bugtracker, as AMD would easily say: you did it wrong and close this report.

[Robotica]

Great work Fritsch! Thanks for the clear status update.

Also worth mentioning: Catalyst 12.2 is released and still no support for x.org 1.12