Posts: 471
Joined: Dec 2011
Reputation:
10
Hi,
what exactly is the bug with some Intel processors not playing 23.976 'flawlessly' ?
I understand this will be fixed in chips with Intel Graphics HD4600 - but, how exactly is this problem noticeable?
Does the sound go slowly out of sync with the picture?
Does the playback stutter slightly every few minutes?
I just want to know what the actual problem is that the average person would notice when watching a movie.
Thanks
Posts: 10
Joined: Aug 2013
Reputation:
0
Tomaz
Junior Member
Posts: 10
2013-08-22, 08:45
(This post was last modified: 2013-08-23, 02:02 by Kib.)
Hi,
I have a Gigabyte Brix mini pc with Intel Grahpichs HD4000 and it still is not fluent in playback.
But I have been told that my audio settings are probably not "correctly" set or my AV receiver (Onkyo TX-NR626 is not "correctly" set)
My experience with XBMC and Gigabyte Brix mini pc is that currently my DUNE HD SMART D1 plays better in colors (but I guess I could configure that) , is fluently, no audio video sync issues.
mod edit: no need to repeat yourself
Posts: 98
Joined: Jul 2013
Reputation:
1
2013-08-22, 09:16
(This post was last modified: 2013-08-22, 09:18 by Bmul.)
The 23.976 bug results in 1 frame drop every +/- 4 minutes.
But you probably will not notice this.
madVR has smooth motion frame rate conversion algorithm.
You can use it with an external player mediaplayer (Media Player Classic - HC) in case you do notice it
Posts: 471
Joined: Dec 2011
Reputation:
10
If you drop 1 frame every 4 minutes. In a 120 minute film you will drop 30 frames.
At 24 frames a second the film could be 1 second shorter.
Does this mean that in XBMC the sound will become 1 second out of sync?
Posts: 3,610
Joined: Feb 2009
Reputation:
87
This affects Intel, Nvidia and AMD. No one does it perfect until HD4600. With Nvidia and Linux there is a way to adjust some settings to get it very close. There was a thread around here recently about it.
Posts: 171
Joined: Feb 2011
Reputation:
3
There is no "Intel-bug" , it´s a bug that affects ALL GPU makers, if the HD4600 makes it we will find out soon enough.
Posts: 6,810
Joined: Jul 2010
Reputation:
198
The "bug" is the CEA (consumer electronics association) and their closed standards. HDMI is pest for computer manufactures, they have to pay royalties for providing this interface to us. There is a lot of information we don't have access to atm. As an example a TV or AVR is capable of notifying the client of a/v sync info. That's in EEDID which we have no access to.
We only see the old EDID which may result in wrong interpretation of timing, i.e. 23.976 -> 23.971. Should drivers autocorrect rounding errors? There may be use cases which require 23.971?
It is not a big deal to configure correct timing and get 23.976. I don't lose a single frame on my systems during an entire movie.
Posts: 3,610
Joined: Feb 2009
Reputation:
87
AnandTech testing shows 23.976001 for Haswell iGPU.
Posts: 98
Joined: Jul 2013
Reputation:
1
That is still 0.000001 frame drops per second to much!
