In my current setup I have a large number (50+) of preloaded movie-file-in TOPs, up to 4-6 of them will play at the same time. If they are not playing, the current frame of all 50+ movies is displayed in the UI
If I enable hardware decoding (on a quadro rtx 6000) for all movie-file-ins they will complain as it will exceed the total number of available decoder units (forgot the actual error message)
If I dynamically enable hardware decoding just for the playing movies it will drop frames upon enabling hw decoding.
Any ideas how to enable hw decoding in such a setup ?
There will always be a limit on the number of simultaneous nvdec decoders that can run.
All I can imagine is loading them offscreen for a few frames until their HW decoding is back on track, which sounds quite hacky without guarantees.
So instead of trying some convoluted back to try to work around that nvdec limit, wouldn’t it be easier to autoconvert them all to HAPQ or NotchLC? Harddisk space is cheap, and then you’ll have HW acceleration on all of them simultaneously.
I am converting the standard show to hapq, but for content that is delivered during a show I need to support h264
testing this again I currently do not run into above issue. I can enable hw decode on all movie file ins and it seems it will only use the decoder units for the movie file ins that are actually playing . At least I don’t get the error anymore.
Not sure why . I was on a different quadro GPU last time I tested this . Gotte do more tests