Whether I use x264 or NVENC I seem to be unable to find a sweet spot for enabling consistent seeking of the resulting MKV files (have also attempted mp4). I've tried all keyframe intervals from 0-5 and changed max B-frames to 0, 2 or 4. In addition I've also tried just about all other variables (rate control, bitrate/quality/presets/profiles/etc.). The best result was to have seeking (2sec or 20sec) work instantly, as expected, for the first use but each subsequent use will take longer and longer for the seek to process. By the time I get to the 10min mark in the recording it will average 90sec to seek 2sec. If I have a 30min recording and attempt to jump directly to the 15min mark it could take as long as 5 minutes to resume. Makes no difference if its a 10min or 1hour long 10Mbps or 20Mbps bitrate recording. Multiple viewers provide the same result.
I have many thousands of clips/videos either sourced from the internet or those I've encoded or transcoded over many years and only a small handful posed this problem, usually attributable to very large keyframe intervals. What really has me stumped is the fact the seek to pause duration is progressively lengthened as you proceed further in the timeline of the recording.
I attached a couple log files that show a few of the many profiles I tested. Do note, a few of the tests were an attempt to see how far I could push CPU Preset with the OC'd (4.1GHz) 3570k when encoding to x264, veryfast seemed to be the safe 80-90% utilization limit. Based on research and initial quality testing I'm thinking NVENC could very well be the better choice for even a recording option.
I'm "seeking" a next step to my dilema. ;-)
I have many thousands of clips/videos either sourced from the internet or those I've encoded or transcoded over many years and only a small handful posed this problem, usually attributable to very large keyframe intervals. What really has me stumped is the fact the seek to pause duration is progressively lengthened as you proceed further in the timeline of the recording.
I attached a couple log files that show a few of the many profiles I tested. Do note, a few of the tests were an attempt to see how far I could push CPU Preset with the OC'd (4.1GHz) 3570k when encoding to x264, veryfast seemed to be the safe 80-90% utilization limit. Based on research and initial quality testing I'm thinking NVENC could very well be the better choice for even a recording option.
I'm "seeking" a next step to my dilema. ;-)