GNOME Bugzilla – Bug 796736
nvdec: Check device codec capabilities before creating video parser
Last modified: 2018-11-03 14:27:10 UTC
Created attachment 372913 [details] [review] nvdec: Check device codec capabilities before creating video Some NVIDIA video decoder not supports HEVC or VPX decoding. So we need check device codec capabilities before creating video. See https://developer.nvidia.com/nvidia-video-codec-sdk#NVDECFeatures
Created attachment 372915 [details] [review] nvdec: Add support VC1 decoding NVIDIA video decoder supports VC1 decoding e.q. ) gst-launch-1.0 filesrc location=(video file path) ! avidemux ! nvdec ! glimagesink sample video. ) http://samples.mplayerhq.hu/V-codecs/WVC1/glitch-ffvc1.avi
Created attachment 372916 [details] [review] nvdec: Add support VC1 decoding NVIDIA video decoder supports VC1 decoding e.q. ) gst-launch-1.0 filesrc location=(video file path) ! avidemux ! nvdec ! glimagesink sample video. ) http://samples.mplayerhq.hu/V-codecs/WVC1/glitch-ffvc1.avi
Can we check hardware capabilities at run-time?
(In reply to Tim-Philipp Müller from comment #3) > Can we check hardware capabilities at run-time? I think, it's hard to check capabilities in run-time because header file is referenced at build time, and cuda API does not seem to be the latest structure to support dynamic linking. (e.q. cuda 8.x) But it seems possible if we load object file dynamically with g_module_open(), and find the module's symbols with g_module_symbol().
Use cuvidGetDecoderCaps() for querying what's supported at runtime. I would prefer avoiding the #ifdef hell if at all possible and delegating all this to runtime checks.
-- GitLab Migration Automatic Message -- This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/issues/743.