Encoding Performance Measure

With the new characteristics of DirectX 12, that they will soon add a new layer of extraction avoiding the mode Explicit Multi-Adapter, implementing a new standard for the use of combined graphic cards, a question happens unexpectedly to me.

Is it possible to use one (1) graphic card only to encode videos?

Examples of functions Bandicam has it, in which it is possible to use a 1º GTX to record and the 2º GTX to play game 4K 60fps, if DirectX 12 this improvement implements Bandicam’s function we could accomplish it with graphics of AMD and Nvidia, using for instance, the AMD to encode video in real time in H.265 to 100 MB/s in 2160p and play videogames in the Nvidia GTX 1080.

If so (and don’t I doubt that he be it), as much as performance grossly requires?

Units of Processing, Rate textures, Rate píxel, Op. floating point and Width band memory are crucial data to know if your chart can encode 4K in real time. Are there any ways or software to be able to know about this?

My cordial greetings.

This forum is for developers who are writing applications using NVIDIA SDKs.

Please post your question to the GeForce Community https://forums.geforce.com/ or ask Bandicam directly if they support your use case.