Still important to use square power of two textures?

Decades ago it was drummed into me to always use square power of two textures for efficient VRAM utilisation. I gather it’s less critical now, but should I even be concerned about it at all? Is there any benefit?

As far as I know non-square textures are generally OK as well, as long as both their dimensions are powers of two.
There was the thread related - Is texture size required to be power of two for the best performance / quality?
Power of two textures may be important if you use ETC1S compression.