Update Ray Serve GPU Guide (#1563)

* Update Ray train samples Signed-off-by: ryanaoleary <[email protected]> * Update Ray Service gpu samples Signed-off-by: ryanaoleary <[email protected]> --------- Signed-off-by: ryanaoleary <[email protected]>
GoogleCloudPlatform · Dec 18, 2024 · 72392dc · 72392dc
1 parent e99202b
commit 72392dc
Show file tree

Hide file tree

Showing 2 changed files with 22 additions and 12 deletions.
diff --git a/ai-ml/gke-ray/rayserve/stable-diffusion/ray-cluster.yaml b/ai-ml/gke-ray/rayserve/stable-diffusion/ray-cluster.yaml
@@ -18,7 +18,7 @@ kind: RayCluster
 metadata:
   name: stable-diffusion-cluster
 spec:
-  rayVersion: '2.9.0'
+  rayVersion: '2.37.0'
   headGroupSpec:
     rayStartParams:
       dashboard-host: '0.0.0.0'
@@ -27,7 +27,7 @@ spec:
       spec:
         containers:
         - name: ray-head
-          image: rayproject/ray-ml:2.9.0
+          image: rayproject/ray:2.37.0
           ports:
           - containerPort: 6379
             name: gcs
@@ -40,10 +40,15 @@ spec:
           resources:
             limits:
               cpu: "2"
+              ephemeral-storage: "15Gi"
               memory: "8Gi"
             requests:
               cpu: "2"
+              ephemeral-storage: "15Gi"
               memory: "8Gi"
+        nodeSelector:
+          cloud.google.com/compute-class: Performance
+          cloud.google.com/machine-family: c3d
   workerGroupSpecs:
   - replicas: 1
     minReplicas: 1
@@ -54,7 +59,7 @@ spec:
       spec:
         containers:
         - name: ray-worker
-          image: rayproject/ray-ml:2.9.0
+          image: rayproject/ray:2.37.0-gpu
           resources:
             limits:
               cpu: 4

diff --git a/ai-ml/gke-ray/rayserve/stable-diffusion/ray-service.yaml b/ai-ml/gke-ray/rayserve/stable-diffusion/ray-service.yaml
@@ -24,17 +24,17 @@ spec:
         import_path: ai-ml.gke-ray.rayserve.stable-diffusion.stable_diffusion:entrypoint
         runtime_env:
           working_dir: "https://github.com/GoogleCloudPlatform/kubernetes-engine-samples/archive/main.zip"
-          pip: ["diffusers==0.12.1"]
+          pip: ["diffusers==0.12.1", "torch", "torchvision", "huggingface_hub==0.25.2", "transformers"]
   rayClusterConfig:
-    rayVersion: '2.9.0'
+    rayVersion: '2.37.0'
     headGroupSpec:
       rayStartParams:
         dashboard-host: '0.0.0.0'
       template:
         spec:
           containers:
           - name: ray-head
-            image: rayproject/ray-ml:2.9.0
+            image:  rayproject/ray:2.37.0
             ports:
             - containerPort: 6379
               name: gcs
@@ -46,11 +46,16 @@ spec:
               name: serve
             resources:
               limits:
-                cpu: "2"
-                memory: "8Gi"
-              requests:
-                cpu: "2"
-                memory: "8Gi"
+              cpu: "2"
+              ephemeral-storage: "15Gi"
+              memory: "8Gi"
+            requests:
+              cpu: "2"
+              ephemeral-storage: "15Gi"
+              memory: "8Gi"
+        nodeSelector:
+          cloud.google.com/compute-class: Performance
+          cloud.google.com/machine-family: c3d
     workerGroupSpecs:
     - replicas: 1
       minReplicas: 1
@@ -61,7 +66,7 @@ spec:
         spec:
           containers:
           - name: ray-worker
-            image: rayproject/ray-ml:2.9.0
+            image: rayproject/ray:2.37.0-gpu
             resources:
               limits:
                 cpu: 4