+- Kodi Community Forum (https://forum.kodi.tv)
+-- Forum: Support (https://forum.kodi.tv/forumdisplay.php?fid=33)
+--- Forum: General Support (https://forum.kodi.tv/forumdisplay.php?fid=111)
+---- Forum: Linux (https://forum.kodi.tv/forumdisplay.php?fid=52)
+---- Thread: Linux VAAPI: Nuc, Chromebox, HSW, IVB, Baytrail with Ubuntu 14.04 (/showthread.php?tid=165707)
Pages:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
RE: vaapi-sse4: Deinterlacing Testing - menno - 2014-11-03
I saw it, but their symptoms are different"If its enabled sometimes the picture freezeses for 2 - 3 seconds an then the playback continues in fastforwarder"
versus, a picture freeze forever, with sound continuing.
RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-11-03
Providing the logfiles requested in the first post would show immediately, especially dmesg | pastebinit after a longer hang.
RE: vaapi-sse4: Deinterlacing Testing - wo0zy - 2014-11-05
Sorry guys another dumb question.
Where does Sandy Bridge fit in to all of this? Just realised my test system has it and not IVB.
Senior moment
.Cheers,
Wo0zy.
RE: vaapi-sse4: Deinterlacing Testing - -DDD- - 2014-11-05
1st Post?
Quote:Summary:
HSW: VAAPI-MCDI
IVB: VAAPI-BOB (cause of the artifacts mentioned above)
Baytrail / SNB / In general HD2000: BOB (which is done in the renderer and costs nearly nothing, not VPP-BOB).
RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-11-05
I need to update the first page a bit. It seems that at least BT (not this one single core mini machine) is capable of doing the VPP-BOB and also VPP-MADI, VPP-MCDI. Here the driver still has a nice bug I posted earlier, but it's good enough for performance testing.
Edit: Tested with current vaapi ppa of wsnipex, mesa 10.3 and kernel 3.16.7
RE: vaapi-sse4: Deinterlacing Testing - adr3nal1n - 2014-11-05
Hi fritsch,
Have been following this thread and have a quick query for you please,
I am running Arch Linux on a H97 chipset board with Haswell Pentium G3240 with the below elements installed and I experience random GPU hangs on playback when vaapi is enabled. Noting some of your earlier comments, Do you think it will be enough for me to rollback the intel driver to version xf86-video-intel-2.99.911 or do I need to ditch Arch Linux altogether and go with Xbmcbuntu or Openelec for a more consistent and stable XBMC experience? (I had previously been running Arch Linux for years with a core2duo CPU and Nvidia G210 card with no such problems). What is your preferred platform/OS for a good XBMC experience?
Elements currently running under Arch Linux:
- Kernel 3.17.1
- libva-intel-driver 1.4.1
- libva 1.4.1
- xf86-video-intel 2.99.916
- xbmc 13.2
Thanks in advance for your advice on this.
RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-11-05
Your libva-intel-driver 1.4.1 is missing the relevant MCDI /MADI patches (see my github ppamaster branch).
There are two options you have:
a) You can try this: http://forum.kodi.tv/showthread.php?tid=199498&pid=1827074#pid1827074
This will workaround the bug by copying surfaces back to system memory
b) Try to boot with those parameters to your bootloader:
Code:
i915.semaphores=0 i915.use_mmio_flip=1 i915.enable_ppgtt=1 drm.vblankoffdelay=1It would be nice if you could test b) first and report back.
RE: vaapi-sse4: Deinterlacing Testing - adr3nal1n - 2014-11-05
Ok, will test this evening when I get home and will post the results.
Thanks again for your help.
RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-11-05
Erm! Wait a moment. You are running xbmc 13.2(!) - do my eyes see that right?
RE: vaapi-sse4: Deinterlacing Testing - adr3nal1n - 2014-11-05
Hi,
Yes, I am running 13.2 Gotham.
Under Arch Linux there is no package for the Helix beta.
Do I need to alter my approach to testing?
Thanks very much.
RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-11-05
Nope ... that just proofs that this regression was not introduced with our new VAAPI architecture. Workaround a) won't work for you cause that code is not in 13.2 Mainline.
So try b) approach please.
If this does not help, try to downgrade the kernel to the lowest version you can find >= 3.12
RE: vaapi-sse4: Deinterlacing Testing - adr3nal1n - 2014-11-05
Ah ok great, will test as you have suggested this evening.
Thanks again.
RE: vaapi-sse4: Deinterlacing Testing - adr3nal1n - 2014-11-06
Hi fritsch,
I added the kernel options you suggested and re-enabled vaapi acceleration in xbmc (was using software decoding previously due to all the gpu hangs with vaapi), I made no other changes. I played an hour of a hd movie and then a 45 minute tv episode and both played without frame drops or any gpu hangs!
Hope this information helps in some way.
Thanks very much for the tip regarding the kernel options to use.
RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-11-06
Please provide: cat /proc/cpuinfo
I need to know you exact CPU revision / version.
RE: vaapi-sse4: Deinterlacing Testing - fritsch - 2014-11-06
For others with the GPU hang (Celeron HSW, Chromebox, etc.) - please try the kernel patch linked here: https://bugs.freedesktop.org/show_bug.cgi?id=83677 I have build 3.17.2 based kernel from it: https://bugs.freedesktop.org/show_bug.cgi?id=83677#c78