“AMD Navi at 7 nanometers performs better than expected”

During the autumn, Nvidia launched new graphics cards and without competition in the higher segment, the Geforce RTX series is an expensive trio. AMD has been empty of big news since the launch of the Vega architecture, which did not live up to the high expectations when it was launched in August 2017.

Next in line from AMD is the architecture Navi, which will be the sixth and probably last iteration of Graphics Core Next (GCN). In connection with this, AMD is taking the step from Globalfoundries 14 nanometers to TSMC’s 7 nanometers, where higher transistor density, energy efficiency and clock frequencies are to be expected.

Now Fudzilla reports that AMD has early copies of Navi driving internally and that the circuits are performing better than expected. The details of the Navi are still few and in addition to manufacturing technology, support for “next-generation memory technology” is the only thing AMD has officially released. This means either GDDR6 as Nvidia has already introduced with Turing or a new version of High-Bandwidth Memory (HBM).

Previous reports suggest that the focus with Navi is on the middle segment rather than the absolute top tier, not entirely unlike how the focus for the Polaris architecture has been good performance in price ranges below SEK 3,000. Flagships like the Radeon RX 580 and before that the RX 480 have competed against the Geforce GTX 1060.

The improvements expected from TSMC’s 7-nanometer technology are about three times higher transistor density and more than doubled energy efficiency, compared to both 16 and 14 nanometers from TSMC and Globalfoundries respectively. This has led many to speculate that AMD is aiming for Geforce GTX 1080 performance with Navi, but at a much lower price.

The launch of the architecture Navi will take place in 2019, apparently at the earliest this summer.

Read more about Navi and 7 nanometers: