• 1
  • 81
  • 82
  • 83(current)
  • 84
  • 85
  • 244
Linux Radeon OSS with vdpau (howto)
(2013-12-06, 21:48)locomot1f Wrote: starting to wonder if it's the Richland APUs. I may be wrong. but it seems people with Trinity APUs haven't had the problem.
i'm running a6-6400 Richland APU.

i am tempted to trying the debian disto -- hehehe Smile

No, while I myself have a Richland in the bug report there are a number of Trinity APU's having the issue.

However, a larger than not number of us are using Black Edition APU's. Not everyone in the bug report has specified, I don't think, whether they are all using a BE or not.. I also have no idea why a BE would matter if it's the same chip/GPU combination.

The OP in the Report had sets of the same APU, one set non-BE and the other BE, and the Non-BE APU's worked fine.

There is also another post that talks about someone testing several variations of APU's and discovering that a significant number of "GB" labeled APU's are experiencing the problem. But I haven't taken my system apart to check the country code on mine.

In any case.. it's definitely not just Richland. I also would be extremely strongly suspicious that is not Distro specific either. There a few different distro's in the Bug Report with the issue, and while I am using the Arch AUR, I have examined the PKGBUILD's and I am, in effect, doing exactly what gradinaruvasile did - which is compile everything directly from GIT. The biggest difference would be kernel - even from git, depending on what make config options you select and what, if any, distro-specific patches you apply.
Just for the record, I built nothing from git in Debian (Jessie).

I used the newly released Mesa 10.0. I simply copied the debian/control files created from patching the 9.2.2 mesa sources with the patch from post 220, on bug 656719 (because I'm awful at using "official" methods to create the debian packages) over to the mesa-10.0.0 sources. I tweaked a few files by hand to get it to finish building successfully then installed the deb's it created.

The only other sw I needed was the 12.0+ kernel linked back quite a few pages ago in this thread.

I can say that my APU is not a Black Edition.

-Luck
(2013-12-02, 09:24)fritsch Wrote: @metheos: No clue. I don't see anything in the logs. Never seen something like this in that combination.

@deadite66: That will be in 3.13-rc3, hopefully it works.

When it appears here: http://cgit.freedesktop.org/~danvet/drm-...el-nightly you should have it in ubuntu mainline

I haven't made any progress figuring this out, but I was comparing log files from the previous xvba setup to see if there were any clues.
I don't know if this matters, but the ALSA device list that xbmc got from when I was running the xvba builds is different from what I get now.

xbmc-xvba-log
xbmc-oss-log

I made a backup before I wiped the system. If you think it would be helpful in figuring out why the DTS and AC3 passthrough don't work with the new setup I could restore it and capture any logs required.

Thanks!

Edit: I forgot to mention that TrueHD is working just fine in addition to DTS HD Master.
TrueHD and DTS HD MA have a higher bitrate than all the others in the logs at 192000
The others are 48000
@ fritsch, would it be possible for you to tag your kernel 3.12.0 official release?
i only see as far as 3.12 rc7, which does not include Temporal patch.

thanks in advance.
-------------------------
in case there are more patches for the 'artifacts' people... and i can patch myself, instead of bugging you Wink

or if there are patches for the 3.13 kernel dealing with the UVD errors that seems to be popping up...
Linux Mint 18 LTS 64-bit - Kodi 17 Beta6
Odroid-C2 - Libreelec v7.90.009
There is no temporal patch in any kernel, cause it is a mesa patch.
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
(2013-12-01, 18:20)metheos Wrote: Logfiles
dmesg
xbmc
Xorg
vdpauinfo
mesa list

The first video playback has DTS audio and the passthrough does not work.
The second video playback has DTS-HD audio and the passthrough works.
The third video playback is the same video as the second after disabling DTS as a supported passthrough format demonstrating the that DTS-HD no longer does passthrough.

I noticed that in the log for the DTS-Audio file it doesn't show the correct number of channels when it created the audio stream. Is this normal?
On line 853:
Code:
10:07:07 T:139720092796672    INFO: CAEStreamInfo::SyncDTS - dts stream detected (6 channels, 48000Hz, 16bit BE, period: 512)
10:07:07 T:139720092796672  NOTICE: Creating audio stream (codec id: 86020, channels: 2, sample rate: 48000, pass-through)

You'll notice that the DTS-HD audio stream creation shows the correct number of channels, and if I disable pass-through it shows the correct number of channels.
That's fine. DTS is coded over two real channels. It's a virtual non lossless format.
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
(2013-12-06, 21:31)Claudio.Sjo Wrote: Hi Guys,
I have an A10-5700 that is very close to your APU-
I'm still running on Lubuntu 12.10 with patched radeon drivers and XVBA.
When I read success stories, I'd wish to get details, as I'll be forced to upgrade soon.
May you please explain your installation step-by-step? Or pointing to the instruction and explain if you did exactly what is written or not.
Even via PM if you don't want to do that within the thread.

Thanks,
Claudio

Just follow the first post. Really, just follow the first post. It's all I needed to get my A10-5700 up and running. I also suggest upgrading to the 3.13-rc2 kernel (v 0.2)
(2013-12-06, 21:31)Claudio.Sjo Wrote:
(2013-12-06, 20:11)gradinaruvasile Wrote:
(2013-12-06, 17:26)Zeqad1ous Wrote: I just want to say thank you for posting such a great howto. I was able to get this working on my main HTPC last night running Debian (Jessie).

CPU: AMD A8-5500 APU
GPU: [AMD/ATI] Trinity [Radeon HD 7560D]

xbmc gotham alpha9 running vdpau on my video sources w/ no artifacts or issues at all (yet). No frame drops and uber-low cpu usage!

Thanks so much!

I have the same APU and i too run Debian Jessie (64 bit). I never had artifacts or significant issues like the people around here.
But i compile everything from git myself from the kernel to mesa and xbmc too.
May be just this igp works better with oss radeon (it had issues with kernels 3.11 and up to 3.12 rc4 with random system freezing but since it was smooth sailing).
Or is the distribution ? (couldn resist Smile ).
Seriously, this APU works very very well with radeon, not just in xbmc but in games too. And it doesnt overheat even with the tiny default cooler.

Hi Guys,
I have an A10-5700 that is very close to your APU-
I'm still running on Lubuntu 12.10 with patched radeon drivers and XVBA.
When I read success stories, I'd wish to get details, as I'll be forced to upgrade soon.
May you please explain your installation step-by-step? Or pointing to the instruction and explain if you did exactly what is written or not.
Even via PM if you don't want to do that within the thread.

Thanks,
Claudio

Well i did this independently from xbmc, the idea was to get rid of fglrx and use the open drivers. The introduction of the dpm code was the perfect opportunity since the oss driver wasnt worth a damn on APUs before it.
The theory is simple: i cloned the gits for kernel, drm, mesa, glamor (optional) llvm+clang, libclc (for opencl, nothing to do with xbmc), xf86-ati then compiled them. After that i built .deb packages from them then installed.
Now, all the above while its simple in theory it can be complicated when you actually have to do it. You need tons of dependencies and if you have 64 bit system you have to compile a version for 64 bit and another for 32 for the user space components and tackle the dependency hell related to non-multiarch dev libraries. It also involves the creation of deb packages that overwrite and take ownership of libraries from other packages.
The bonus of this approach is to have a reference implementation of the oss stack.
I dont recommend doing it unless you have time to experiment, know how building from source works and how to recover a broken system from command line. I learnt all of this from bits and pieces of docs from the net. Dont expect to be bailed out by anyone else if you mess something up.
(2013-12-06, 16:40)deadite66 Wrote:
(2013-12-06, 16:34)libgradev Wrote: Hmm - the patch Alexandre Demers linked in the bug report?

yes.
3.13 gives me a dead screen.
3.12.3 no dvb-t2 tuners.
but 3.12 with the atihdmi5, mpeg2 fix and 24Hz patches has given me a usable system again.

i kept stopping and starting a mpeg2 stream about 10 times and could not get a gpu hang.

OK - last night I built 3.12.2 with the patch included. I tried this kernel in combination with mesa git, mesa from wnsipex's mesa ppa and with mesa from wnsipex's test ppa...

All combinations produced GPU resets when starting MPEG-2 recordings Sad

I guess there are multiple issues here - will update the bug report. It would be interesting to hear from the others having MPEG-2 issues, the ones that added to the bug report.

@deadite66: I'm curious as to which mesa package you are using?
latest from ppa:wsnipex/mesa
checked the last test wasn't a fluke, tried starting and stopping some mpeg2 files over 20 times and no crash.

these are the kernel packages i'm running, 3.12 source from kernel.org plus atihdmi5, mpeg2 fix and 24Hz patches.
image
headers
Intel Core i5-11500, ASRock B560M-ITX/ac, 240GB SSD + 4TB  Western Digital Blue, Arch Linux, mythtv 31+fixes + TBS 6205 QUAD FREEVIEW DVB-T2
@deadite66: You have an E350 ... that one was never affected by the mpeg-2 crash.
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
(2013-12-07, 14:21)fritsch Wrote: @deadite66: You have an E350 ... that one was never affected by the mpeg-2 crash.

hehe - yeah I did wonder :/

I have a dedicated HD 6950...
@fritsch but i was getting gpu crashes with mpeg2 (and only mpeg2) but since the patch now i don't.

the crashes happened after wsnipex ppa mesa update, 2 full lockups and 1 where i got a dmesg.
these were at the same time as others had problems.

dmesg
https://bugs.freedesktop.org/attachment.cgi?id=89679
Intel Core i5-11500, ASRock B560M-ITX/ac, 240GB SSD + 4TB  Western Digital Blue, Arch Linux, mythtv 31+fixes + TBS 6205 QUAD FREEVIEW DVB-T2
Hi!
I Got a Asus E45M1-I DELUXE (HD6320) motherboard running xbmcbuntu-12.2.Intel-AMD.iso
with the latest Xbmc monthly build.
But i got some lagging with some 720p and 1080p movies.
Wonder if this thread might be able to fix it.

Im just a newbie to linux and dont want to mess up something just because a test.

Can i safely install whats on the first post on my system to get away with the lagging or
am i already running this if i got the setup above?

And is it enough to install this two?

1.) Installation of the wsnipex repository which includes all relevant mesa patches - this is a copied version of a specific oibaf ppa version that was known to be working
Code:
sudo apt-add-repository ppa:wsnipex/mesa
sudo apt-get update
sudo apt-get dist-upgrade
sudo apt-get install libg3dvl-mesa vdpauinfo linux-firmware


2.) Installation of the repository and the new packages
Code:
sudo apt-add-repository ppa:wsnipex/xbmc-fernetmenta-master
sudo apt-get update
sudo apt-get install xbmc xbmc-bin
HTPC: Intel Nuc i3 8109u | 8Gb Ram | Windows 10 Pro
  • 1
  • 81
  • 82
  • 83(current)
  • 84
  • 85
  • 244

Logout Mark Read Team Forum Stats Members Help
Radeon OSS with vdpau (howto)26