From 2f922279ef65667a9511304a689049ce59abc9a7 Mon Sep 17 00:00:00 2001
From: HaimKortovich <haimkortovich88@gmail.com>
Date: Fri, 8 May 2026 09:30:48 -0500
Subject: [PATCH] add more gpu and change model name served

---
 clusters/k3s-dgx/nim-service/qwen.yaml | 26 ++------------------------
 1 file changed, 2 insertions(+), 24 deletions(-)

diff --git a/clusters/k3s-dgx/nim-service/qwen.yaml b/clusters/k3s-dgx/nim-service/qwen.yaml
index 1d628c1..6e655da 100644
--- a/clusters/k3s-dgx/nim-service/qwen.yaml
+++ b/clusters/k3s-dgx/nim-service/qwen.yaml
@@ -1,27 +1,5 @@
 apiVersion: apps.nvidia.com/v1alpha1
 kind: NIMCache
-metadata:
-  name: qwen3-32b-instruct
-  namespace: nim-service
-spec:
-  source:
-    ngc:
-      modelPuller: nvcr.io/nim/qwen/qwen3-32b-dgx-spark:1.1.0-variant
-      pullSecret: ngc-secret
-      authSecret: ngc-api-secret
-      model:
-        engine: "vllm"
-        tensorParallelism: "1"
-        profiles:
-          - c4f105d92c72ab56200884dfacde9d2128b139755c06b9c883eeb3e287b7408a
-  storage:
-    pvc:
-      create: true
-      size: "100Gi"
-      volumeAccessMode: ReadWriteOnce
----
-apiVersion: apps.nvidia.com/v1alpha1
-kind: NIMCache
 metadata:
   name: qwen36-27b-fp8
   namespace: nim-service
@@ -65,9 +43,9 @@ spec:
     - --host
     - 0.0.0.0
     - --served-model-name
-    - qwen36
+    - Qwen/Qwen3.6-27B-FP8
     - --gpu-memory-utilization
-    - "0.85"
+    - "0.90"
     - --max-model-len
     - "256000"
     - --language-model-only