Hello. When working with ffmpeg(cuvid) and decoding on GPU, I encountered that the memory consumption is too high for a single video stream. During tests, I got about 200 megabytes per video stream with a resolution of 704 * 576.
Is this normal, or are there ways to optimize memory consumption?
What operations cause such a large expense?
I have attached screenshots. In the first case, I am decoding 10 video streams.
In the second case, 40 is decoded
GPU: 2080 RTX TI
Driver: 470.199.02
OS: Ubuntu 22.04.2