The only place I’ve seen it mentioned in a review is in the L1T video, but nowhere has there been any demonstration of it in action.
If it’s not a setting one can turn on and off, my proposed benchmark would be to test the difference in performance between Nvidia encoding and AMD 7900 encoding on an Intel and AM5 system. That ‘should’ show the performance benefit so long as other factors are not bottlenecking the benchmark.
I’m no expert in this, though early next year I’ll be able to see if it’s easy to get the functionality myself, but it could be that the software isn’t here to meet the hardware quite yet. Big speculation on my part mind.