From c29f8ed6d85f7197b6dc564a9c0a30df5c909b72 Mon Sep 17 00:00:00 2001
From: John Garbutt <john.garbutt@stackhpc.com>
Date: Fri, 3 Oct 2025 12:29:59 +0100
Subject: [PATCH 1/2] Try to expose the GPU count

---
 charts/azimuth-chat/azimuth-ui.schema.yaml | 1 +
 charts/azimuth-llm/values.schema.json      | 8 ++++++++
 2 files changed, 9 insertions(+)

diff --git a/charts/azimuth-chat/azimuth-ui.schema.yaml b/charts/azimuth-chat/azimuth-ui.schema.yaml
index 74bd573c..a1f14887 100644
--- a/charts/azimuth-chat/azimuth-ui.schema.yaml
+++ b/charts/azimuth-chat/azimuth-ui.schema.yaml
@@ -24,6 +24,7 @@ sortOrder:
   - /azimuth-llm/ui/appSettings/model_instruction
   - /azimuth-llm/ui/appSettings/page_title
   - /azimuth-llm/api/image/version
+  - /azimuth-llm/api/gpus
   - /azimuth-llm/ui/appSettings/llm_params/temperature
   - /azimuth-llm/ui/appSettings/llm_params/max_tokens
   - /azimuth-llm/ui/appSettings/llm_params/frequency_penalty
diff --git a/charts/azimuth-llm/values.schema.json b/charts/azimuth-llm/values.schema.json
index 59e0e1b8..c13caea6 100644
--- a/charts/azimuth-llm/values.schema.json
+++ b/charts/azimuth-llm/values.schema.json
@@ -40,6 +40,14 @@
                             "default": "v0.10.2"
                         }
                     }
+                },
+                "gpus": {
+                    "type": "integer",
+                    "title": "GPU Count",
+                    "description": "The number of GPUs to allocate to the model.",
+                    "default": 1,
+                    "minimum": 1,
+                    "maximum": 8
                 }
             }
         },

From 96bd4ae847b5771aac70a32c7fb3d9e2dc819dac Mon Sep 17 00:00:00 2001
From: John Garbutt <john.garbutt@stackhpc.com>
Date: Fri, 3 Oct 2025 14:59:20 +0100
Subject: [PATCH 2/2] Try to set --tensor-parallel-size based on GPU count

---
 charts/azimuth-llm/templates/api/deployment.yml | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/charts/azimuth-llm/templates/api/deployment.yml b/charts/azimuth-llm/templates/api/deployment.yml
index 0e6206d5..00bc9af5 100644
--- a/charts/azimuth-llm/templates/api/deployment.yml
+++ b/charts/azimuth-llm/templates/api/deployment.yml
@@ -47,6 +47,10 @@ spec:
           {{- if .Values.api.extraArgs -}}
           {{- .Values.api.extraArgs | toYaml | nindent 10 }}
           {{- end -}}
+          {{- if gt .Values.api.gpus 1 }}
+          - --tensor-parallel-size
+          - {{ .Values.api.gpus }}
+          {{- end }}
         {{- if .Values.huggingface.secretName -}}
         envFrom:
         - secretRef: