H.264 4K Encoding Performance

Hello All,

I read in the Jetson Nano specification that the part is capable of encoding 1 channel of 4Kp30 to H.265 & H.264.

1x 4Kp30
2x1080p60
4x1080p30
4x720p60
9x720p30
(H.265 & H.264)

I understand that these are nominal specs, and the real results will depend on physics (movement, lighting, etc…), but I’m wondering how well the Nano has performed for people familiar with using it in 4K applications.

I’m researching kits like this one from Basler, and trying to get a picture of what a realistic gst pipeline would be from the 4k camera through the Jetson’s encoder - even if feeding into a fakesink at the end of the pipeline, similar to this:

It would be great to find out what a typical CPU load looks like (top/htop) while running a 4k stream encoding to H.264/5.

Would anyone have any experiences with a similar application that they can share?

Thank you!

M.

Hi,
For profiling system status, you can execute sudo tegrastats. Please refer to
https://docs.nvidia.com/jetson/l4t/index.html#page/Tegra%20Linux%20Driver%20Package%20Development%20Guide/AppendixTegraStats.html#

Hi @DaneLLL - thanks for responding. I don’t actually have the kit yet.

If you have a 4Kp30 capable MIPI camera connected to a Nano, would you be able to share results showing highest attainable frame rate when encoding raw —> H.264/5 via a gst pipeline like the example I linked above?

Thanks!

M.