ffdshow's DXVA video decoder problem

Hi guys, I'm using Avisynth v2.5.8 (official build), AvsP v2.0.2, GraphStudio and the latest build of ffdshow tryouts (2012-04-09). This is the graph made by me: The problem is that when I try to open this graph in AvsP I get this error: I am sure this decoder doesn't start when I open the graph in AvsP because I can't see its icon in the notifications area. What can I do in this case? Thank you very much! PS:

You can't use a native DXVA decoder with AviSynth.

Oh, I didn't know this... In fact this was the first time I tried to use this decoder (DXVA)... So there's nothing to do about this? I really want to use a DXVA decoder from now on, but I don't know what should I do in order to get it working... Thanks! :)

If you are looking for GPU decoding and not necessarily DXVA decoding, then there are GPU-enabled Avisynth source filters, such as DGDecNV.

After some research I found that I need a video renderer and because of this I have switched to CoreAVC 3.0.1. Now I have a problem with CoreAVC: when I try to use DXVA decoding I get this error -> If I select CUDA decoding or CPU-only decoding, this error doesn't appear, but MeGUI crashes after 1-2 minutes of work. Please help me to get this DXVA decoder (@CoreAVC) or ffdshow's DXVA decoder working. Thanks in advance!

Thanks, vivan! Everything's working fine now.

Does it mean there is a way to use ATI's GPU in Avisynth ?

Yes, with DirectShowSoruce() or DSS2() (if you have Haali Splitter) and LAV Video decoder set to DXVA-Copyback.

Great, I believed it was only possible for Nvidia cards, not ATI's :) I didn't know LAV. Do we need to set anything in ffdshow and GraphStudio as explained in the first post, or just LAV is enough ?

If you have multiple H.264 decoders you have to make sure that LAV Video decoder is preferred decoder, that it has highest priority. Other then that, no. LAV is all you need.