AMD Radeon VII Desktop GPU Review

AMD introduced the Radeon VII at CES 2019 and based the GPU on its Vega architecture. The card is also the first to be manufactured on a 7 nm FinFET process and has 16 GB of VRAM, which should futureproof it for years to come. By contrast, the RTX 2080 Ti, which is currently the flagship GPU in the green corner, has only 11 GB of VRAM, so the Radeon VII has the edge in this regard. 

The smaller architecture should also significantly improve performance per watt and help the card run more efficiently. AMD has packed approximately 13.2 billion transistors within 331 mm², along with 3,840 shader units and a 4,096-bit memory interface.

The Radeon VII still consumes around 300 W under load, for which AMD has designed a new cooling system. The GPU has vapour chamber heatpipes and a set of aluminium fins that are cooled by three axial fans. Previously, the company has equipped its reference GPUs with radial fans, but one fan design is not necessarily more effective than the other, and it depends more on how AMD has optimised the three fans to run appropriately under load.

The Radeon VII also has three DisplayPort and one HDMI connections, which are spaced out well. The GPU requires two 8-pin power connectors too.

The Radeon VII is currently AMD’s flagship desktop GPU and has a 1,400 MHz base clock speed that can boost to 1,750 MHz when required. The card also has 60 CUs, 16 GB of HBM2 VRAM, a 4,096 memory bus and a 1,024 GB/s memory bandwidth. The GPU also has a significantly smaller die than its 14 nm predecessors, but AMD has crammed 13.2 billion transistors onto the Radeon VII compared to the 12.5 billion with which it equipped the Radeon RX Vega 64. The former also has 3,840 shader units but does not support raytracing or tensor cores like NVIDIA RTX 20 series GPUs do. It is worth keeping in mind that the Radeon VII is rated to consume 300 W, which is 5 W higher than its predecessor and 50 W more than the RTX 2080 Ti.