+- Kodi Community Forum (https://forum.kodi.tv)
+-- Forum: Discussions (https://forum.kodi.tv/forumdisplay.php?fid=222)
+--- Forum: Hardware (https://forum.kodi.tv/forumdisplay.php?fid=112)
+--- Thread: Bay Trail-D motherboards (/showthread.php?tid=184930)
RE: Bay Trail-D motherboards - fritsch - 2014-08-12
Give that build a try: saraev.ca/OpenELEC-Generic.x86_64-devel-20140812223656-r19607-g9d78bd3.tar - make a backup. It's Helix with our new VAAPI code. Curious on your stats.
Warning: You need an OpenELEC 4.x installed - this is only for updating via /storage/.update or copying va the smb share.
RE: Bay Trail-D motherboards - Matt Devo - 2014-08-12
(2014-08-12, 21:55)fritsch Wrote: Give that build a try: saraev.ca/OpenELEC-Generic.x86_64-devel-20140812223656-r19607-g9d78bd3.tar - make a backup. It's Helix with our new VAAPI code. Curious on your stats.
Warning: You need an OpenELEC 4.x installed - this is only for updating via /storage/.update or copying va the smb share.
On the 60mpbs file, 4.1.3 is slightly more smooth - the devel version has small micro-pauses only visible during panning/rotation, but no drops or skips. It also drops a few more frames on the 100mbps file, but like you said that's not a realistic test.
RE: Bay Trail-D motherboards - fritsch - 2014-08-12
@matt Devo:
Did you have "Prefer VAAPI Render Method" enabled with that new build? Also make sure that no fancy upscalers are forced - Wondering why your 2955U should now be slower than my Baytrail :-)
RE: Bay Trail-D motherboards - topside - 2014-08-12
(2014-08-12, 21:55)fritsch Wrote: Give that build a try: saraev.ca/OpenELEC-Generic.x86_64-devel-20140812223656-r19607-g9d78bd3.tar - make a backup. It's Helix with our new VAAPI code. Curious on your stats.fps is higher, but
RE: Bay Trail-D motherboards - Matt Devo - 2014-08-13
(2014-08-12, 22:41)fritsch Wrote: @matt Devo:
Did you have "Prefer VAAPI Render Method" enabled with that new build? Also make sure that no fancy upscalers are forced - Wondering why your 2955U should now be slower than my Baytrail :-)
wasn't slower, as there were still 0 dropped/skipped frames. Just didn't appear perfectly smooth. All the VAAPI acceleration options were selected.
RE: Bay Trail-D motherboards - Stereodude - 2014-08-13
(2014-08-12, 21:44)wanthd Wrote: After reading all the 45 pages of this topic, i have a doubt. Does HD passthrough works with Windows?No, not so far.
RE: Bay Trail-D motherboards - fritsch - 2014-08-13
(2014-08-12, 23:06)topside Wrote:(2014-08-12, 21:55)fritsch Wrote: Give that build a try: saraev.ca/OpenELEC-Generic.x86_64-devel-20140812223656-r19607-g9d78bd3.tar - make a backup. It's Helix with our new VAAPI code. Curious on your stats.fps is higher, but
Difference:
J1800: GPU base: 688 MHz GPU max: 792 Mhz CPU base: 2.41 ghz burst: 2.58 ghz
Q1900: GPU base: 688 MHz max: 854 MHz CPU base: 2 ghz burst: 2.41 ghz
Edit: And your Q even has 2 cores more ...
I cannot explain that to me. Perhaps someone else with such a Q1900 can give those samples a try. Do you use Dual Channel Memory? Any bios settings that might be relevant?
Also post an xbmc.log with debugging turned on, please
Code:
cat /storage/.xbmc/temp/xbmc.log | pastebinit
sensors | pastebinitEdit:When playing that movie, please press "Enter" and click on the film reel - make sure Scaling is set to Bilinear. (but logfile will show that)
RE: Bay Trail-D motherboards - -DDD- - 2014-08-13
http://ark.intel.com/compare/78867,78866
2 vs 4 Cores
RE: Bay Trail-D motherboards - fritsch - 2014-08-13
The 2 Core works without drops .... the 4 Core not :-)
Bay Trail-D motherboards - decpvr - 2014-08-13
(2014-08-13, 07:29)fritsch Wrote: I cannot explain that to me. Perhaps someone else with such a Q1900 can give those samples a try. Do you use Dual Channel Memory? Any bios settings that might be relevant?
Do you need the testing done with the experimental OE build you linked to above?
I've got a Q1900, with single channel memory, hooked up to a 720p TV. I can test it (later) but at the moment I'm only using OE 4.0.7.
Would the TV resolution affect playback rate - if it's scaling to 1080p or 720p?
RE: Bay Trail-D motherboards - fritsch - 2014-08-13
Just set scaling to bilinear and we will see. A 4.0.7 test should also be okay.
Bay Trail-D motherboards - decpvr - 2014-08-13
(2014-08-13, 09:30)fritsch Wrote: Just set scaling to bilinear and we will see. A 4.0.7 test should also be okay.
Ok, I'll try to test it this evening.
I'm currently using software playback (for better de-interlacing). I'll switch back to hardware acceleration for the test and report back.
I've got a MSI J1900i motherboard:
http://www.msi.com/product/mb/J1900I.html#hero-overview
RE: Bay Trail-D motherboards - fritsch - 2014-08-13
You won't be able to decode the 4K sample in software - so switching back is needed.
What deinterlacing method are you using? 1080i sw decoding + yadif?
Bay Trail-D motherboards - decpvr - 2014-08-13
(2014-08-13, 09:46)fritsch Wrote: You won't be able to decode the 4K sample in software - so switching back is needed.
What deinterlacing method are you using? 1080i sw decoding + yadif?
I'm just using the software 'deinterlace' option for 576i (UK SD TV) material. It's much better quality than the 'bob' option available with hardware acceleration.
I tested a bit of 1080i material and found I had to set 'deinterlace (half)' for smooth playback with software.
I didn't think yadif was an option available to Baytrail CPU/GPU's?
RE: Bay Trail-D motherboards - topside - 2014-08-13
@fritsch
2x4GB (Dual Channel memory mode)
OE Dual Channel vs OE Single Channel vs MPC-BE Single Channel (demo, 76MB)
We will probably need more optimizations of OE/drv for smooth playback of UHD AVC video with a single channel memory mode like MPC-BE. Is it possible? Forget it.