My Sapphire Reference RX-480 8GB came.

On paper, ideally it should be roughly 42 percent faster than one of my Sapphire HD7970 OC (factory overclocked). I was expecting somewhere between 30 percent and 50 percent faster (not knowing much about the architecture). In real games and benchmarks, both the CPU and GPU have a role. So, in games I would expect less than 40 percent and down to 0 percent in some cases.

In my tests (3 game settings and 3 benchmarks -- 1440P max settings), I got 0, 29, 35, 18, 38, and 40 percent faster than HD7970 OC. Pretty much like I expected. But, I was hoping to be "pleasantly surprised" with consistently higher numbers than I actually got. I'm rarely "pleasantly surprised"

I did not try overclocking, yet. I may. But, overclocking rarely gives me "visible" improvements. I would expect about 7 to 10 percent more with my Reference RX-480 (i.e. approximately 50 percent improvement in that one test).

In a nutshell, to me it was worth buying at the price of $244 (the two HD7970 cost over $700 total). It will slightly improve my game experience on my more graphic intensive GPU limited games -- that do not benefit from CrossFire. However, it does not have "mind boggling" performance (mind boggling would be 50 percent or more faster than HD7970 for the $244 price). Since the RX480 4GB at $200 has nearly the same performance, the RX-480 4GB would approach mind boggling for the price.

This RX-480 was for fun and is a place holder until Vega 10. Then, when I get my Vega, my wife (who just got one of my HD7970) will get the RX-480.

[Edit] A very gentle overclock got that 40 percent number (noted above) to 54 percent. In that sense, RX-480 now performs as well as I could have expected (I expected 50 percent max improved graphics performance).