Ryan Shrout at PCPer published some additional research into this topic.
For me, the biggest thing that stood out was the +12v rail current numbers. Regardless of whether it is overclocked (8.29A) or not (6.96A), the RX480 is violating the PCI-e spec of 5.5A max.
I simply cannot recommend this card to the masses. These are users that are not using enthusiast boards and may even be installing this into pre-built boxes from say an HP or Dell.
The odd part is that Ryan plays down concerns of this, but I would disagree. I certainly can’t recommend the RX480 to layman, not with the knowledge we now have on the RX480.
You saw a great deal of backpedaling and denial by marketing exec Robert Hallock - that’s not a surprise. It’s his job to professionally spin this. OTOH, I sincerely hope engineer Raja Koduri does not play with words here and actually addresses this issue head on.
It may end up being a BIOS flash that sufficiently reduces clocks, voltage, and current to safer levels - at the expense of performance. Or they provide 8-pin cards, but that requires an entirely new board. For that reason, the former is IMO the far more likely recourse here.
