There is a new AMD GPU launching today – the next-gen Radeon Pro V620. This GPU is aimed at the data centre market, aiming to tackle the increasing need for GPU acceleration for cloud applications, 3D workloads and immersive experiences.
Based on the RDNA 2 graphics architecture, the AMD Radeon Pro V620 GPU packs a high amount of GDDR6 memory, GPU-partitioning capabilities and multi-stream hardware-accelerated encoders. Such features allow it to offer high scalability across multiple users and enough graphics power to handle AAA games with no sweat.
The GPU core uses a Navi 21 GPU with 4608 stream processors across 72 CUs clocked at 2200MHz, which allows the GPU to deliver 20.28 TFLOPs (FP32) of peak performance. As for the memory, there is 32GB of GDDR6 memory clocked at 16Gbps across a 256-bit memory bus and 128MB of Infinity Cache, which provides a maximum bandwidth of 512GB/s. The TBP is set at 300W.
Like the mainstream Radeon 6000 series GPUs, AMD’s latest data centre and cloud solution supports Vulkan, DirectX 12 Ultimate and AMD FidelityFX suite. Moreover, it also uses SR-IOV-based GPU virtualization to scale to multiple users securely, and it’s compatible with AMD ROCm software, which speeds up workloads like cloud gaming, DaaS (Desktop as a Service), WaaS (Workstation as a Service) and machine learning.
KitGuru says: It’s nice to see AMD spreading its wings across multiple markets, offering competitive solutions in other segments. If you’re into cloud gaming, there’s a good chance that some services you use will eventually be upgraded with Radeon Pro V620 GPUs.