update: add env variable for main container(vllm) when DRA is enabled for Intel-xe (#189)

* fix: add intel-xe accelerator environment variable support

- add VLLM_WORKER_MULTIPROC_METHOD: spawn for Intel-xe
- bump version
Signed-off-by: Wen Zhou <wenzhou@redhat.com>

* update: fix schema generation

Signed-off-by: Wen Zhou <wenzhou@redhat.com>

---------

Signed-off-by: Wen Zhou <wenzhou@redhat.com>

Author: zdtsw

charts/llm-d-modelservice/values.schema.json

@@ -4,7 +4,7 @@
     "properties": {
         "accelerator": {
             "additionalProperties": false,
-            "description": " Supported types: nvidia, intel-i915, intel-xe, intel-gaudi, amd, google",
+            "description": " Supported types: nvidia, intel-i915, intel-xe, intel-gaudi, amd, google, cpu",
             "properties": {
                 "dra": {
                     "default": false,
@@ -83,6 +83,35 @@
                             "required": [],
                             "title": "intel-i915",
                             "type": "array"
+                        },
+                        "intel-xe": {
+                            "items": {
+                                "anyOf": [
+                                    {
+                                        "additionalProperties": false,
+                                        "properties": {
+                                            "name": {
+                                                "default": "VLLM_WORKER_MULTIPROC_METHOD",
+                                                "required": [],
+                                                "title": "name",
+                                                "type": "string"
+                                            },
+                                            "value": {
+                                                "default": "spawn",
+                                                "required": [],
+                                                "title": "value",
+                                                "type": "string"
+                                            }
+                                        },
+                                        "required": [],
+                                        "type": "object"
+                                    }
+                                ],
+                                "required": []
+                            },
+                            "required": [],
+                            "title": "intel-xe",
+                            "type": "array"
                         }
                     },
                     "required": [],

charts/llm-d-modelservice/values.schema.tmpl.json

@@ -4,7 +4,7 @@
   "properties": {
     "accelerator": {
       "additionalProperties": false,
-      "description": " Supported types: nvidia, intel-i915, intel-xe, intel-gaudi, amd, google",
+      "description": " Supported types: nvidia, intel-i915, intel-xe, intel-gaudi, amd, google, cpu",
       "properties": {
         "dra": {
           "default": false,
@@ -83,6 +83,35 @@
               "required": [],
               "title": "intel-i915",
               "type": "array"
+            },
+            "intel-xe": {
+              "items": {
+                "anyOf": [
+                  {
+                    "additionalProperties": false,
+                    "properties": {
+                      "name": {
+                        "default": "VLLM_WORKER_MULTIPROC_METHOD",
+                        "required": [],
+                        "title": "name",
+                        "type": "string"
+                      },
+                      "value": {
+                        "default": "spawn",
+                        "required": [],
+                        "title": "value",
+                        "type": "string"
+                      }
+                    },
+                    "required": [],
+                    "type": "object"
+                  }
+                ],
+                "required": []
+              },
+              "required": [],
+              "title": "intel-xe",
+              "type": "array"
             }
           },
           "required": [],

charts/llm-d-modelservice/values.yaml

@@ -74,7 +74,7 @@ modelArtifacts:
 multinode: false
 
 # Global accelerator configuration
-# Supported types: nvidia, intel-i915, intel-xe, intel-gaudi, amd, google
+# Supported types: nvidia, intel-i915, intel-xe, intel-gaudi, amd, google, cpu
 accelerator:
   # Type of accelerator to use
   type: nvidia
@@ -101,6 +101,9 @@ accelerator:
         value: "1"
       - name: VLLM_WORKER_MULTIPROC_METHOD
         value: "spawn"
+    intel-xe:
+      - name: VLLM_WORKER_MULTIPROC_METHOD
+        value: "spawn"
   # ResourceClaimTemplate configurations for DRA (used when dra: true)
   # Each accelerator type can have its own claim template configuration
   resourceClaimTemplates:

examples/output-xpu.yaml

@@ -93,6 +93,9 @@ spec:
           - name: HF_HOME
             value: /model-cache
 
+          - name: VLLM_WORKER_MULTIPROC_METHOD
+            value: "spawn"
+          
           ports:
           - containerPort: 8200
             protocol: TCP