Using leaked/rumored benchmark results for a "stock" RX580 versus my installed RX480 Reference:
RX580 stock clocks are 8% above my stock clocks. RX580 "graphics" performance in "Fire Strike Extreme" benchmark is 13% greater than my RX480. RX580 also overclocks -- but, overclocking is for fun -- not for anything you can "see" with unaided eyes.
So, RX580 is an improved RX480 (RX580 uses new Polaris 20 chip -- RX480 is Polaris 10). It provides "Bang for Buck" and completely meets minimum VR requirements (they say).
Full disclosure: I'm running Fallout 4 heavily modded (loads of grass and trees and plants where there were none in stock FO4). 1440p resolution. All graphics settings on "ultra" or equal, extra wide field of view, and new 4K graphics throughout (textures are 4096x4096 pixel textures even for small things). In the most complex scenes, FPS falls below 15 at times. So, I need Vega