AMD introduces the Radeon Pro VII high-end GPU for workstations

Today, AMD introduced the Radeon Pro VII, a high-end graphics card for workstations built on a 7nm process. This is AMD’s range-topping GPU for professional workloads, and it comes with 16GB of HBM2 memory, complete with ECC support, offering memory bandwidth up to 1TB/s. Until now, AMD’s 7nm workstation GPUs were mostly aimed at the mid-range, such as in the case of the Radeon Pro W5700.

AMD is promising 26% higher performance in 8K image processing in DaVinci Resolve, and up 5.6x the peak performance in double precision (FP64) workloads compared to Nvidia’s Quadro RTX 5000. A full list of comparisons in various software, including Adobe After Effects, can be found here.

It also comes with support for technologies such as Infinity Fabric Link, which allows two GPUs to connect to each other and share memory with a low-latency connection, and PCIe 4.0, promising double the bandwidth of PCIe 3.0.

Here are the specs for the Radeon Pro VII:

Compute Units Stream processors TFLOPS Memory (Bandwidth) Memory Interface Display outputs
Radeon Pro VII 60 3840 Up to 6.5 (FP64)
Up to 13.1 (FP32)
16GB HBM2 with ECC (1 TB/s) 4096-bit 6x

AMD has also announced software updates for its Radeon Pro Software for Enterprise, with version 20.Q2 promising 14% better performance on current Radeon Pro GPUs, specifically the Radeon Pro WX4100. AMD’s Radeon ProRender has also been updated with new SideFX Houdini and unreal Engine plug-ins, plus updated plug-ins for Autodesk Maya and Blender.

The new GPU will be available in mid-June, and it will cost $1899 if you want to buy it separately. Workstations with the Radeon Pro VII will be available in the second half of the year.