Max performance and image quality, coreAVC vs ffdshow

[heyer]

Hey there, I know that at some point in time, coreAVC was vastly superior for h.264 decoding, but with ffdshow and LAV being multitreaded now, is still the case? I also heard something about decoding artifacs in coreAVC 3.0, but I couldn't find any information on wether it had be fixed or not in 3.0.1, has it? So which decoder should I use? If it makes any diffence, I use mpc-hc with madVR and watch alot of 10-bit content, so hardware decoding capability doesn't matter.

[sneaker_ger]

From the tests I remember, LAV Video is currently the fastest 10-bit H.264 decoder. (Or madVR's internal decoders, but you cannot use filters that sit in between the decoder and the renderer, like vsfilter, then)

[heyer]

Thanks for answering :)
Does this mean there's practically no reason to use coreAVC instead of the free alternatives?

[sneaker_ger]

For almost everyone: correct.
The only exception are probably a few very old processors, like Athlon XP, where CoreAVC still might have the edge over libav based decoders and where performance actually matters.

[heyer]

Well I actually do have a reason for saving performance, since I'm using SVP with a 120hz monitor :)
Anyway, thanks for the help, was nice having some clarification on it