Docs nvidia attribu

examples/demo_nvidia_dranet/deviceclass.yaml

@@ -19,4 +19,4 @@ metadata:
 spec:
   selectors:
     - cel:
-        expression: device.driver == "dra.net"
+        expression: device.driver == "dra.net"

examples/demo_nvidia_dranet/resourceclaims.yaml

@@ -11,11 +11,10 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-
 apiVersion: resource.k8s.io/v1beta1
 kind: ResourceClaimTemplate
 metadata:
-  name: 2-gpu
+  name: 2-gpu-nic-aligned
 spec:
   spec:
     devices:
@@ -25,26 +24,12 @@ spec:
         count: 2
         selectors:
         - cel:
-            expression: |
-                device.attributes["gpu.nvidia.com"].index < 2
-
----
-apiVersion: resource.k8s.io/v1beta1
-kind: ResourceClaimTemplate
-metadata:
-  name: 2-nic
-spec:
-  spec:
-    devices:
-      requests:
+            expression: device.attributes["gpu.nvidia.com"].index <= 2
       - name: nic
         deviceClassName: dranet
         count: 2
         selectors:
         - cel:
-            expression: device.attributes["dra.net"].rdma == true &&
-              (
-                (device.attributes["dra.net"].ifName.startsWith("gpu") &&
-                 device.attributes["dra.net"].ifName.endsWith("rdma0") &&
-                 int(device.attributes["dra.net"].ifName.substring(3, 4)) < 2)
-              )
+            expression: device.attributes["dra.net"].rdma == true
+      constraints:
+      - matchAttribute: "resource.kubernetes.io/pcieRoot"

examples/demo_nvidia_dranet/statefulset.yaml

@@ -1,17 +1,12 @@
-# Copyright 2025 Google LLC
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#    https://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
+apiVersion: v1
+kind: Service
+metadata:
+  name: nccl-gib-test
+spec:
+  selector:
+    name: nccl-gib-test
+  clusterIP: None
+---
 apiVersion: apps/v1
 kind: StatefulSet
 metadata:
@@ -36,8 +31,8 @@ spec:
           capabilities:
             add: ["IPC_LOCK"]
         volumeMounts:
-          # - name: library-dir-host
-          #   mountPath: /usr/local/nvidia
+          - name: library-dir-host
+            mountPath: /usr/local/nvidia
           - name: gib
             mountPath: /usr/local/gib
           - name: shared-memory
@@ -57,7 +52,7 @@ spec:
             sleep infinity
         resources:
           claims:
-          - name: gpu            
+          - name: gpu
       volumes:
         - name: library-dir-host
           hostPath:
@@ -71,11 +66,9 @@ spec:
             sizeLimit: 250Gi
       resourceClaims:
         - name: gpu
-          resourceClaimTemplateName: 2-gpu
-        - name: nic
-          resourceClaimTemplateName: 2-nic
+          resourceClaimTemplateName: 2-gpu-nic-aligned
       tolerations:
       - key: "nvidia.com/gpu"
         operator: "Equal"
         value: "present"
-        effect: "NoSchedule"
+        effect: "NoSchedule"

site/content/docs/concepts/references.md

@@ -4,10 +4,21 @@ date: 2024-12-19T11:20:46Z
 ---
 
 - [The Kubernetes Network Driver Model: A Composable Architecture for High-Performance Networking](/docs/kubernetes_network_driver_model_dranet_paper.pdf) - This paper introduces the Kubernetes Network Driver model and provides a detailed performance evaluation of DraNet, demonstrating significant bandwidth improvements for AI/ML workloads.
+
+{{< embed-pdf "/docs/kubernetes_network_driver_model_dranet_paper.pdf" >}}
+
+- [The Challenges of AI/ML Multi-Node Workloads in Kubernetes - Antonio Ojea, Google - Regular SIG Network Meeting for 2025-07-17](https://www.youtube.com/playlist?list=PL69nYSiGNLP2E8vmnqo5MwPOY25sDWIxb)
+
+<iframe src="https://docs.google.com/presentation/d/e/2PACX-1vSBButnm46ReLtbtgBa2b4xkmr3oXEtH5yf10xsQ4fjcqF4jSOc5MzeZQUS02Ev2j6DKFj8vQAjCIoy/pubembed?start=true&loop=true&delayms=3000" frameborder="0" width="480" height="299" allowfullscreen="true" mozallowfullscreen="true" webkitallowfullscreen="true"></iframe>
+
+- [Kubernetes Network Drivers, Antonio Ojea, Presentation](https://docs.google.com/presentation/d/1Vdr7BhbYXeWjwmLjGmqnUkvJr_eOUdU0x-JxfXWxUT8/edit?usp=sharing)
+<iframe src="https://docs.google.com/presentation/d/e/2PACX-1vRVritcaQFYkvaPuTPsxkgOt0ZfWhqYPcCjNN0UgZcEh9HR1yh3bFDXSOiPbPUayoMzbefZ_qvFoWCX/pubembed?start=true&loop=true&delayms=3000" frameborder="0" width="480" height="299" allowfullscreen="true" mozallowfullscreen="true" webkitallowfullscreen="true"></iframe>
+
 - [KEP 3063 - Dynamic Resource Allocation #306](https://github.com/kubernetes/enhancements/blob/master/keps/sig-node/3063-dynamic-resource-allocation/README.md)
 - [KEP 3695 - DRA: structured parameters #438](https://github.com/kubernetes/enhancements/issues/4381)
 - [Extend PodResources to include resources from Dynamic Resource Allocation (DRA)](https://github.com/kubernetes/enhancements/issues/3695)
 - [Working Group Device Management](https://github.com/kubernetes-sigs/wg-device-management)
-- [Kubernetes Network Drivers, Antonio Ojea, Presentation](https://docs.google.com/presentation/d/1Vdr7BhbYXeWjwmLjGmqnUkvJr_eOUdU0x-JxfXWxUT8/edit?usp=sharing)
+
+
 - [The Future of Kubernetes Networking - Antonio Ojea, Googe & Dan Winship, Red Hat - Kubernetes Contributor Summit EU 2024](https://sched.co/1aOqO)
-- [Better Together! GPU, TPU and NIC Topological Alignment with DRA - John Belamaric, Google & Patrick Ohly, Intel - Kubecon US 2024](https://sched.co/1i7pv)
+- [Better Together! GPU, TPU and NIC Topological Alignment with DRA - John Belamaric, Google & Patrick Ohly, Intel - Kubecon US 2024](https://sched.co/1i7pv)