Supporting the AMD/Xilinx Inference Server in KServe

[varunsh-xilinx]

/kind feature

We've been working on an official Xilinx/AMD inference server that is compatible with KServe to support AMD CPUs, GPUs and FPGAs. The work is open-sourced and under active development. Let us know what we need to do to be more officially integrated in KServe. Thanks!

[yuzisun]

Hi @varunsh-xilinx, here is a list of tasks to integrate Xilinx to kserve and add to the supported serving runtime list.

• Create Xilinx cluster serving runtime CR and add to kserve install manifests. On the cluster serving runtime CR specify the supportModelFormats and protocolVersions supported by Xilinx inference server, please don't enable the autoSelect flag.
• Add e2e test for using the Xilinx serving runtime in https://github.com/kserve/kserve/tree/master/test/e2e/predictor with a test model.
• Add Xilinx serving runtime to the serving runtime list doc https://kserve.github.io/website/master/modelserving/servingruntimes/
  https://kserve.github.io/website/master/modelserving/v1beta1/serving_runtime/
• Add Xilinx serving runtime documentation to https://github.com/kserve/website/tree/main/docs/modelserving/v1beta1