2015-04-05, 23:05
Yeah, did not feel to do a technical conversation with you, after your posts. And as I did not see _any_ code that was written by you our anyone of OSMC yet - I was not sure I can take you serious.
This was revisited some months ago in the deinterlacing testing thread. The video pll (pll5) is quite capable. So generating fractional pll outside the hdmi tx is our only option. The idea was that the hdmi tx would take the clock already divided by this 1.001. So at the end we should end up with just specifiing correct n/cts values in kernel.
Worked out quite nightmarish (thought the clock was perfectly calculated) ... you cannot fool a non matching audio clock too much and this you forget here. It's not alone about the video doing something, that was "calculated" by you, the chain must match.
This was revisited some months ago in the deinterlacing testing thread. The video pll (pll5) is quite capable. So generating fractional pll outside the hdmi tx is our only option. The idea was that the hdmi tx would take the clock already divided by this 1.001. So at the end we should end up with just specifiing correct n/cts values in kernel.
Worked out quite nightmarish (thought the clock was perfectly calculated) ... you cannot fool a non matching audio clock too much and this you forget here. It's not alone about the video doing something, that was "calculated" by you, the chain must match.
