> Are the video streams 4K? My guest is you are running out of memory on the >GPU.
No, the streams have HD resolution. I run my programm with debug log level and here is output: [h264 @ 0000013f85b22440] Failed setup for format cuda: hwaccel initialisation returned error [h264 @ 0000013f85b22440] Format cuda not usable, retrying get_format() without it. [h264 @ 0000013f85b22440] decode_slice_header error [h264 @ 0000013f85b22440] no frame! [h264 @ 0000013f85b22440] nal_unit_type: 1(Coded slice of a non-IDR picture), nal_ref_idc: 2 [h264 @ 0000013f85b22440] Format cuda chosen by get_format(). [h264 @ 0000013f85b22440] Format cuda requires hwaccel initialisation. [h264 @ 0000013f85b22440] Loaded lib: nvcuvid.dll [h264 @ 0000013f85b22440] Loaded sym: cuvidGetDecoderCaps [h264 @ 0000013f85b22440] Loaded sym: cuvidCreateDecoder [h264 @ 0000013f85b22440] Loaded sym: cuvidDestroyDecoder [h264 @ 0000013f85b22440] Loaded sym: cuvidDecodePicture [h264 @ 0000013f85b22440] Loaded sym: cuvidGetDecodeStatus [h264 @ 0000013f85b22440] Loaded sym: cuvidReconfigureDecoder [h264 @ 0000013f85b22440] Loaded sym: cuvidMapVideoFrame64 [h264 @ 0000013f85b22440] Loaded sym: cuvidUnmapVideoFrame64 [h264 @ 0000013f85b22440] Loaded sym: cuvidCtxLockCreate [h264 @ 0000013f85b22440] Loaded sym: cuvidCtxLockDestroy [h264 @ 0000013f85b22440] Loaded sym: cuvidCtxLock [h264 @ 0000013f85b22440] Loaded sym: cuvidCtxUnlock [h264 @ 0000013f85b22440] Loaded sym: cuvidCreateVideoSource [h264 @ 0000013f85b22440] Loaded sym: cuvidCreateVideoSourceW [h264 @ 0000013f85b22440] Loaded sym: cuvidDestroyVideoSource [h264 @ 0000013f85b22440] Loaded sym: cuvidSetVideoSourceState [h264 @ 0000013f85b22440] Loaded sym: cuvidGetVideoSourceState [h264 @ 0000013f85b22440] Loaded sym: cuvidGetSourceVideoFormat [h264 @ 0000013f85b22440] Loaded sym: cuvidGetSourceAudioFormat [h264 @ 0000013f85b22440] Loaded sym: cuvidCreateVideoParser [h264 @ 0000013f85b22440] Loaded sym: cuvidParseVideoData [h264 @ 0000013f85b22440] Loaded sym: cuvidDestroyVideoParser [h264 @ 0000013f85b22440] NVDEC capabilities: [h264 @ 0000013f85b22440] format supported: yes, max_mb_count: 65536 [h264 @ 0000013f85b22440] min_width: 48, max_width: 4096 [h264 @ 0000013f85b22440] min_height: 16, max_height: 4096 [h264 @ 0000013f85bd04c0] nal_unit_type: 1(Coded slice of a non-IDR picture), nal_ref_idc: 2 [h264 @ 0000013f1cec7580] nal_unit_type: 1(Coded slice of a non-IDR picture), nal_ref_idc: 2 [h264 @ 0000013f6efb0f40] nal_unit_type: 1(Coded slice of a non-IDR picture), nal_ref_idc: 2 I've attached the most important code snippet to this mail. Hope it will help. >Среда, 17 апреля 2019, 12:21 +03:00 от Hristo Ivanov ><hivanov.ffmail...@gmail.com>: > >Hi. > >> Is it because of NVIDIA GPUs have concurrent session limitation? > >There is a session limitation for encoding for sure, but I think there is no >such thing for decoding: >https://developer.nvidia.com/video-encode-decode-gpu-support-matrix > >> By the way, I don't understand where these messages are come from. They are >> just printed in my console. > >Those messages are from the ffmpeg log. The ffmpeg code is full with calls to >'av_log(void *avcl, int level, const char *fmt, ....)'. >Depending on the loglevel in use the messages are printed or ignored. >For the ffmpeg tool the '-loglevel' flag can be used to configure what >messages should be printed and what messages should not. >With the C-API, the ' av_log_set_level(int level)' function can be used to >configure the log level. >The different log levels can be seen here: > >https://ffmpeg.org/doxygen/trunk/group__lavu__log__constants.html > >Running your program with log level verbose or maybe even debug could give you >some clue to why the creation of the 6th decoder fails. > >Are the video streams 4K? My guest is you are running out of memory on the GPU. > >Regards. >_______________________________________________ >Libav-user mailing list >Libav-user@ffmpeg.org >https://ffmpeg.org/mailman/listinfo/libav-user > >To unsubscribe, visit link above, or email >libav-user-requ...@ffmpeg.org with subject "unsubscribe". -- С уважением, Ахмед Анам. Best regards, Akhmed Anam.
jaga-code-snippet-gpu-init.c
Description: Binary data
_______________________________________________ Libav-user mailing list Libav-user@ffmpeg.org https://ffmpeg.org/mailman/listinfo/libav-user To unsubscribe, visit link above, or email libav-user-requ...@ffmpeg.org with subject "unsubscribe".