From 8b7d3c0659227b25fa2853c4eb1b377519781923 Mon Sep 17 00:00:00 2001
From: chemavx <chemavx@git.chemavx.xyz>
Date: Fri, 17 Apr 2026 17:12:50 +0000
Subject: [PATCH] feat(ollama): migrate GPU from AMD ROCm to NVIDIA CUDA (RTX
 3060 via OCuLink)

Switch from ollama/ollama:rocm + amd.com/gpu to standard CUDA image + nvidia.com/gpu.
RTX 3060 (GA106, 12GB) now used via NVIDIA GPU Operator on chemavx-k8.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---
 ollama/ollama.yaml | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/ollama/ollama.yaml b/ollama/ollama.yaml
index a89ed1b..f05fd65 100644
--- a/ollama/ollama.yaml
+++ b/ollama/ollama.yaml
@@ -40,7 +40,7 @@ spec:
         kubernetes.io/hostname: chemavx-k8
       containers:
         - name: ollama
-          image: ollama/ollama:0.20.7
+          image: ollama/ollama:latest
           command: ["/usr/bin/bash", "-c"]
           args:
             - |
@@ -66,12 +66,17 @@ spec:
               value: "8192"
             - name: OLLAMA_KEEP_ALIVE
               value: "-1"
+            - name: NVIDIA_VISIBLE_DEVICES
+              value: all
+            - name: NVIDIA_DRIVER_CAPABILITIES
+              value: compute,utility
           resources:
             requests:
               memory: 8Gi
               cpu: "500m"
             limits:
               memory: 20Gi
+              nvidia.com/gpu: "1"
           volumeMounts:
             - name: ollama-data
               mountPath: /root/.ollama