diff --git a/charts/tfy-gpu-operator/Chart.yaml b/charts/tfy-gpu-operator/Chart.yaml index fd6c391cc..9bbc654fb 100644 --- a/charts/tfy-gpu-operator/Chart.yaml +++ b/charts/tfy-gpu-operator/Chart.yaml @@ -1,6 +1,6 @@ apiVersion: v2 name: tfy-gpu-operator -version: 0.1.21 +version: 0.1.22 description: "Truefoundry GPU Operator" maintainers: - name: truefoundry diff --git a/charts/tfy-gpu-operator/README.md b/charts/tfy-gpu-operator/README.md index 3153833c9..8844a4963 100644 --- a/charts/tfy-gpu-operator/README.md +++ b/charts/tfy-gpu-operator/README.md @@ -51,9 +51,9 @@ Tfy-gpu-operator is a Helm chart that facilitates the deployment and management | `aws-eks-gpu-operator.dcgmExporter.version` | Image tag version for DCGM Exporter. Find all tags at https://catalog.ngc.nvidia.com/orgs/nvidia/teams/k8s/containers/dcgm-exporter/tags | `3.3.8-3.6.0-ubuntu22.04` | | `aws-eks-gpu-operator.dcgmExporter.serviceMonitor.enabled` | Enable or disable ServiceMonitor for DCGM Exporter. | `false` | | `aws-eks-gpu-operator.dcgmExporter.resources.requests.cpu` | CPU request for the DCGM Exporter. | `10m` | -| `aws-eks-gpu-operator.dcgmExporter.resources.requests.memory` | Memory request for the DCGM Exporter. | `300Mi` | -| `aws-eks-gpu-operator.dcgmExporter.resources.limits.cpu` | CPU limit for the DCGM Exporter. | `50m` | -| `aws-eks-gpu-operator.dcgmExporter.resources.limits.memory` | Memory limit for the DCGM Exporter. | `400Mi` | +| `aws-eks-gpu-operator.dcgmExporter.resources.requests.memory` | Memory request for the DCGM Exporter. | `100Mi` | +| `aws-eks-gpu-operator.dcgmExporter.resources.limits.cpu` | CPU limit for the DCGM Exporter. | `100m` | +| `aws-eks-gpu-operator.dcgmExporter.resources.limits.memory` | Memory limit for the DCGM Exporter. | `1000Mi` | | `aws-eks-gpu-operator.dcgmExporter.args` | Arguments for the DCGM Exporter. | `["-c","5000"]` | ### gcp-gke-standard-driver Configuration for the GKE Standard Nvidia Drivers. This section will only be used when clusterType.gcpGkeStandard is set to true. @@ -130,15 +130,15 @@ Tfy-gpu-operator is a Helm chart that facilitates the deployment and management | `azure-aks-gpu-operator.dcgm.version` | Image tag for DCGM container. Find all image tags at https://catalog.ngc.nvidia.com/orgs/nvidia/teams/cloud-native/containers/dcgm/tags | `3.3.8-1-ubuntu22.04` | | `azure-aks-gpu-operator.dcgm.resources.requests.cpu` | CPU request for standalone DCGM container | `10m` | | `azure-aks-gpu-operator.dcgm.resources.requests.memory` | Memory request for standalone DCGM container | `100Mi` | -| `azure-aks-gpu-operator.dcgm.resources.limits.cpu` | CPU limit for standalone DCGM container | `50m` | -| `azure-aks-gpu-operator.dcgm.resources.limits.memory` | Memory limit for standalone DCGM container | `400Mi` | +| `azure-aks-gpu-operator.dcgm.resources.limits.cpu` | CPU limit for standalone DCGM container | `100m` | +| `azure-aks-gpu-operator.dcgm.resources.limits.memory` | Memory limit for standalone DCGM container | `1000Mi` | | `azure-aks-gpu-operator.dcgmExporter.enabled` | Enabled/Disable DCGM Exporter. | `true` | | `azure-aks-gpu-operator.dcgmExporter.version` | Image tag version for DCGM Exporter. Find all tags at https://catalog.ngc.nvidia.com/orgs/nvidia/teams/k8s/containers/dcgm-exporter/tags | `3.3.8-3.6.0-ubuntu22.04` | | `azure-aks-gpu-operator.dcgmExporter.serviceMonitor.enabled` | Enable or disable ServiceMonitor for DCGM Exporter. | `false` | | `azure-aks-gpu-operator.dcgmExporter.resources.requests.cpu` | CPU request for the DCGM Exporter. | `10m` | | `azure-aks-gpu-operator.dcgmExporter.resources.requests.memory` | Memory request for the DCGM Exporter. | `100Mi` | -| `azure-aks-gpu-operator.dcgmExporter.resources.limits.cpu` | CPU limit for the DCGM Exporter. | `50m` | -| `azure-aks-gpu-operator.dcgmExporter.resources.limits.memory` | Memory limit for the DCGM Exporter. | `400Mi` | +| `azure-aks-gpu-operator.dcgmExporter.resources.limits.cpu` | CPU limit for the DCGM Exporter. | `100m` | +| `azure-aks-gpu-operator.dcgmExporter.resources.limits.memory` | Memory limit for the DCGM Exporter. | `1000Mi` | | `azure-aks-gpu-operator.dcgmExporter.args` | Arguments for the DCGM Exporter. | `["-c","5000"]` | ### civo-talos-gpu-operator Configuration for the Civo Talos GPU Operator. This section will only be used when clusterType.civoTalos is set to true. @@ -178,8 +178,8 @@ Tfy-gpu-operator is a Helm chart that facilitates the deployment and management | `civo-talos-gpu-operator.dcgmExporter.serviceMonitor.enabled` | Enable or disable ServiceMonitor for DCGM Exporter. | `false` | | `civo-talos-gpu-operator.dcgmExporter.resources.requests.cpu` | CPU request for the DCGM Exporter. | `10m` | | `civo-talos-gpu-operator.dcgmExporter.resources.requests.memory` | Memory request for the DCGM Exporter. | `100Mi` | -| `civo-talos-gpu-operator.dcgmExporter.resources.limits.cpu` | CPU limit for the DCGM Exporter. | `50m` | -| `civo-talos-gpu-operator.dcgmExporter.resources.limits.memory` | Memory limit for the DCGM Exporter. | `400Mi` | +| `civo-talos-gpu-operator.dcgmExporter.resources.limits.cpu` | CPU limit for the DCGM Exporter. | `100m` | +| `civo-talos-gpu-operator.dcgmExporter.resources.limits.memory` | Memory limit for the DCGM Exporter. | `1000Mi` | | `civo-talos-gpu-operator.dcgmExporter.args` | Arguments for the DCGM Exporter. | `["-c","5000"]` | ### generic-gpu-operator Configuration for the GPU Operator. This section will only be used when clusterType.generic is set to true. diff --git a/charts/tfy-gpu-operator/values.yaml b/charts/tfy-gpu-operator/values.yaml index fe80f75be..edc923419 100644 --- a/charts/tfy-gpu-operator/values.yaml +++ b/charts/tfy-gpu-operator/values.yaml @@ -315,10 +315,10 @@ aws-eks-gpu-operator: resources: requests: cpu: 10m - memory: 300Mi + memory: 100Mi limits: - cpu: 50m - memory: 400Mi + cpu: 100m + memory: 1000Mi ## @param aws-eks-gpu-operator.dcgmExporter.args Arguments for the DCGM Exporter. args: ["-c", "5000"] @@ -731,8 +731,8 @@ azure-aks-gpu-operator: cpu: 10m memory: 100Mi limits: - cpu: 50m - memory: 400Mi + cpu: 100m + memory: 1000Mi ## DCGM Exporter configuration. dcgmExporter: @@ -762,8 +762,8 @@ azure-aks-gpu-operator: cpu: 10m memory: 100Mi limits: - cpu: 50m - memory: 400Mi + cpu: 100m + memory: 1000Mi ## @param azure-aks-gpu-operator.dcgmExporter.args Arguments for the DCGM Exporter. args: ["-c", "5000"] @@ -973,8 +973,8 @@ civo-talos-gpu-operator: cpu: 10m memory: 100Mi limits: - cpu: 50m - memory: 400Mi + cpu: 100m + memory: 1000Mi ## @param civo-talos-gpu-operator.dcgmExporter.args Arguments for the DCGM Exporter. args: ["-c", "5000"]