As part of the RX 480 launch today at Computex, AMD has revealed a number of key improvements for their latest graphics card. One of the key takeaways was the efficiency gains that the new architecture and process node brings.
One of the dramatic reveals for the RX 480 was that it would come in at just 150 TDP. This is nearly half that of the R9 390 and 390X that have about the same theoretical performance. This would allow for a single 6pin PCIe power connector to be used but I suspect an 8pin will be provided instead in order to provide overclocking headroom.
According to AMD, the total efficiency gains come out to about 280% or 2.8x. 170% or 1.7x comes from the new 14nm FinFET process node, a denser alternative to the 16nm node from TSMC used by Nvidia. The rest of it or 190% or 1.9x comes from the new architecture and features AMD has added in. With GCN 4.0, AMD has completely revamped a lot of the component units and efficiency was a major target.
The key, of course, will be how the RX 480 does against the GTX 1070. Nvidia’s chip offers a bit more theoretical performance at the same TDP. However, we don’t know how well the RX 480 does in real world performance. Hopefully, we’ll get cards from AMD soon.
