add nvidia gpu support

kustomize/bases/k8s-device-plugin-amd/kustomization.yaml

@@ -0,0 +1,2 @@
+resources:
+  - https://raw.githubusercontent.com/ROCm/k8s-device-plugin/master/k8s-ds-amdgpu-dp.yaml

kustomize/bases/k8s-device-plugin-nvidia/config/config0.yaml

@@ -0,0 +1,9 @@
+version: v1
+flags:
+  migStrategy: "none"
+  failOnInitError: true
+  nvidiaDriverRoot: "/"
+  plugin:
+    passDeviceSpecs: false
+    deviceListStrategy: envvar
+    deviceIDStrategy: uuid

kustomize/bases/k8s-device-plugin-nvidia/kustomization.yaml

@@ -0,0 +1,13 @@
+resources:
+  # - https://raw.githubusercontent.com/NVIDIA/k8s-device-plugin/v0.16.1/deployments/static/nvidia-device-plugin.yml
+  - nvidia-device-plugin-helmchart.yaml
+
+namespace: kube-system
+
+configMapGenerator:
+  - name: nvidia-device-plugin-config
+    files:
+      - config0=config/config0.yaml
+
+configurations:
+  - kustomizeconfig/namereference.yaml

kustomize/bases/k8s-device-plugin-nvidia/kustomizeconfig/namereference.yaml

@@ -0,0 +1,5 @@
+nameReference:
+  - kind: ConfigMap
+    fieldSpecs:
+      - kind: HelmChart
+        path: spec/set/config.name

kustomize/bases/k8s-device-plugin-nvidia/nvidia-device-plugin-helmchart.yaml

@@ -0,0 +1,18 @@
+apiVersion: helm.cattle.io/v1
+kind: HelmChart
+metadata:
+  name: nvidia-device-plugin
+  namespace: kube-system
+spec:
+  repo: https://nvidia.github.io/k8s-device-plugin
+  chart: nvidia-device-plugin
+  version: 0.16.1
+  targetNamespace: kube-system
+  set:
+    config.default: config0
+    config.name: nvidia-device-plugin-config
+    gfd.enabled: "true"
+  valuesContent: |-
+    nodeSelector:
+      kubernetes.io/arch: amd64
+    runtimeClassName: nvidia

kustomize/bases/ollama/ollama-deployment.yaml

@@ -25,10 +25,10 @@ spec:
                     values:
                       - amd64
       priorityClassName: high-priority
+      runtimeClassName: nvidia
       containers:
         - name: server
-          imagePullPolicy: Always
-          image: ollama/ollama:rocm
+          image: ollama/ollama
           # image: badjware/ollama-tweak
           # env:
             # - name: HSA_OVERRIDE_GFX_VERSION
@@ -37,12 +37,12 @@ spec:
             #   value: "0"
           resources:
             requests:
-              memory: 8Gi
-              # cpu: 2000m
+              memory: 4Gi
+              cpu: 2000m
             limits:
-              memory: 8Gi
-              # cpu: 3000m
-              amd.com/gpu: 1
+              memory: 4Gi
+              cpu: 3000m
+              nvidia.com/gpu: 1
           ports:
             - containerPort: 11434
               name: http

kustomize/env/prod/kustomization.yaml

@@ -52,6 +52,10 @@ images:
     newTag: "7.2"
   - name: bitnami/kubectl
     newTag: "1.29"
+  - name: ollama/ollama
+    newTag: 0.3.3
+    # newTag: 0.3.3-rocm
+
 
 configMapGenerator:
   - name: litellm-proxy-config

kustomize/overlays/system/kustomization.yaml

@@ -2,5 +2,5 @@ resources:
   - ../../bases/longhorn
   - ../../bases/traefik
   - ../../bases/external-secrets
-  - https://raw.githubusercontent.com/ROCm/k8s-device-plugin/master/k8s-ds-amdgpu-dp.yaml
+  - ../../bases/k8s-device-plugin-nvidia
   - clustersecretstore.yaml