In recent times software-defined datacenters provide flexibility and enable operators to quickly configure and scale hardware infrastructure without having to rebuild it from scratch. As a result, multiple special-purpose accelerators for modern datacenter workloads emerged in the recent years. Among these accelerators are Mellanox’s BlueField-2 data processing units (DPUs), fully-fledged processors that can offload networking, storage, and security workloads while adding manageability.
In addition, for GPU-accelerated AI and ML workloads Nvidia offers its BlueField-2X card with Ampere GPU with an increased magnitude performance when compared to traditional AI setups. The Nvidia Mellanox BlueField-2 DPU uses a custom-designed system-on-chip and special-purpose VLIW acceleration engines to perform compute tasks for connectivity and various other data that has been stored.
The general-purpose Arm cores can run Linux operating system and offload a variety of workloads from host CPUs. The VLIW engines provide some additional acceleration. Nvidia’s is also aiming to launch BlueField-3 DPU with a GPU for AI applications that are due in 2022. Furthermore, by 2023 the company intends to launch its BlueField-4X DPU that will finally combine DPU and GPU in the same chip that increases the connectivity.