init commit

clusters/k3s-dgx/kserve/gpu-serving-runtime.yaml

@@ -0,0 +1,42 @@
+apiVersion: serving.kserve.io/v1beta1
+kind: ServingRuntime
+metadata:
+  name: kserve-gpu-runtime
+  namespace: kserve
+spec:
+  supportedModelFormats:
+    - name: tensorflow
+      version: "2"
+      autoSelect: true
+    - name: pytorch
+      version: "2"
+      autoSelect: true
+    - name: sklearn
+      version: "1"
+      autoSelect: true
+    - name: xgboost
+      version: "1"
+      autoSelect: true
+  protocol: v1
+  protocolVersions:
+    - v1
+    - v2
+  containers:
+    - name: kserve-container
+      image: kserve/sklearnserver-gpu:latest
+      resources:
+        requests:
+          cpu: "1"
+          memory: "2Gi"
+          nvidia.com/gpu: "1"
+        limits:
+          cpu: "4"
+          memory: "8Gi"
+          nvidia.com/gpu: "1"
+      env:
+        - name: NVIDIA_VISIBLE_DEVICES
+          value: "all"
+        - name: NVIDIA_DRIVER_CAPABILITIES
+          value: "compute,utility"
+  multiModel: false
+  disabled: false