Merge pull request #2131 from tkatila/gpu-tf-to-pt-demo

Author: mythi

deployments/gpu_pytorch_demo/demo.py

@@ -0,0 +1,29 @@
+# Copyright 2025 Intel Corporation. All Rights Reserved.
+#
+# 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
+#
+#     http://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.
+
+import torch as t
+
+if not t.xpu.is_available():
+  raise Exception("No xpu available")
+
+arr = t.tensor([1.0, 2.0, 3.0, 4.0]).to("xpu")
+
+arr *= 2
+
+print(arr)
+
+cmp = t.tensor([2.0, 4.0, 6.0, 8.0]).to("xpu")
+
+if not t.equal(arr, cmp):
+  raise Exception("Not valid result:", arr)

deployments/gpu_tensorflow_test/deployment.yaml renamed to deployments/gpu_pytorch_demo/deployment.yaml

@@ -6,12 +6,12 @@ spec:
   restartPolicy: Never
   containers:
   - name: testcontainer
-    image: intel/intel-extension-for-tensorflow:latest
+    image: intel/intel-extension-for-pytorch:xpu
     imagePullPolicy: IfNotPresent
     securityContext:
       allowPrivilegeEscalation: false
     command: ["/bin/sh", "-c"]
-    args: ["python /code/training.py"]
+    args: ["python3 /code/demo.py"]
     resources:
       limits:
         gpu.intel.com/i915: 1

deployments/gpu_tensorflow_test/kustomization.yaml renamed to deployments/gpu_pytorch_demo/kustomization.yaml

@@ -1,11 +1,11 @@
 configMapGenerator:
 - name: training-code
   files:
-  - training.py
+  - demo.py
 
 resources:
   - deployment.yaml
 
 images:
-  - name: intel/intel-extension-for-tensorflow
-    newTag: 1.2.0-gpu
+  - name: intel/intel-extension-for-pytorch
+    newTag: 2.8.10-xpu

deployments/gpu_tensorflow_test/training.py

@@ -41,8 +41,8 @@ const (
 	healthMgmtYaml      = "deployments/gpu_plugin/overlays/health/kustomization.yaml"
 	nfdRulesYaml        = "deployments/nfd/overlays/node-feature-rules/kustomization.yaml"
 	containerName       = "testcontainer"
-	tfKustomizationYaml = "deployments/gpu_tensorflow_test/kustomization.yaml"
-	tfPodName           = "training-pod"
+	ptKustomizationYaml = "deployments/gpu_pytorch_demo/kustomization.yaml"
+	ptPodName           = "training-pod"
 )
 
 func init() {
@@ -195,12 +195,12 @@ func describe() {
 			})
 		})
 
-		ginkgo.It("run a small workload on the GPU [App:tensorflow]", func(ctx context.Context) {
+		ginkgo.It("run a small workload on the GPU [App:pytorch]", func(ctx context.Context) {
 			createPluginAndVerifyExistence(f, ctx, vanillaPath, "gpu.intel.com/i915")
 
-			kustomYaml, err := utils.LocateRepoFile(tfKustomizationYaml)
+			kustomYaml, err := utils.LocateRepoFile(ptKustomizationYaml)
 			if err != nil {
-				framework.Failf("unable to locate %q: %v", tfKustomizationYaml, err)
+				framework.Failf("unable to locate %q: %v", ptKustomizationYaml, err)
 			}
 
 			ginkgo.By("submitting demo deployment")
@@ -209,8 +209,8 @@ func describe() {
 
 			ginkgo.By("waiting the pod to finish")
 
-			err = e2epod.WaitForPodSuccessInNamespaceTimeout(ctx, f.ClientSet, tfPodName, f.Namespace.Name, 300*time.Second)
-			gomega.Expect(err).To(gomega.BeNil(), utils.GetPodLogs(ctx, f, tfPodName, containerName))
+			err = e2epod.WaitForPodSuccessInNamespaceTimeout(ctx, f.ClientSet, ptPodName, f.Namespace.Name, 300*time.Second)
+			gomega.Expect(err).To(gomega.BeNil(), utils.GetPodLogs(ctx, f, ptPodName, containerName))
 
 			framework.Logf("tensorflow execution succeeded!")
 		})