On my Jetson TX2 (L4T 28.2), I am using a gstreamer pipeline to access 3 streams of 4K 60fps CSI cameras.
In this pipeline, I am using “nvjpegenc” to encode the frames for writing .jpeg, which is hardware accelerated. Sometime soon, I will also use (in a separate “tee” + “queue”) encode the frames in h264 with “omxh264enc”, which is also hardware accelerated.
I would like to anticipate hardware encoder access conflict using both omxh264enc and nvjpegenc. I read somewhere that the TX2’s hardware encoder can encode 3x 4K 60fps. The maximum capacities of this encoder will be used for encoding my 3 streams with omxh264enc alone.
Does nvjpegenc use the same hardware encoder, which would result in the max performances of both elements be capped ?
If not, what hardware does nvjpegenc use ?
Thanks in advance for any info/pointer you might share.