Kubernetes for AMD-Xilinx Platforms

Kubernetes in industry is truly the key for future innovation. Now imagine taking that work flow of Kubernetes and stitching up 5-6 FPGAs? Today we have not only proved that you can send workloads on a secure private network through PYNQ, we can now implement Kubernetes (k3s) foundations to them as well. This is just the fundamental building blocks for you to elaborate an evolve this iteration. Understanding that you can run FPGAs on a cluster with Xilinx Platforms on the edge can give you the ultimate supercomputer for your tasks.

image

In the Hackster.io Series: Super96 Clusters Series we provide some simple examples on how to execute Xilinx platforms & tools with simple applications! We have climbed to the highest peaks of AMD<>Xilinx Land dancing with all the critical tools such as: Vivado, HLS and Vitis. This Series stitches all the modules seen in Part 1 | Part 2 | Part 3 into 1 Final Integration via distribution through PYNQ.

image

This is the first steppingstone in creating K3s Kubernetes on the edge for AMD <> Xilinx devices. The main goal is to setup a server and attach n-number of agents onto a single network. The base functionality has been planted showing a proof of concept that this technology is possible. It is up to the user to add dashboards, connect more server nodes or agent nodes to the network and establishing particular workload processes. The image below shows a quick extraction of the mission on how you can communicate from the edge to your applications.

Setting up docker containers and extrapolating information from different sources consumes too much time. PYNQ is the tool, its powerful framework invites you to interface with hardware from a high-level software approach making your life as a designer more explosive. Adding Kubernetes can be difficult, following some simple tricks and methods, you can boost the potential of your project, transcending most technology fronts on the market today.

2 Likes