feat: Add ROCm support and update Ollama configurations

- Created ansible playbooks for setting up AMD ROCm container environment.
- Added ROCm and AMD Graphics repositories in ansible tasks.
- Installed necessary ROCm packages including rocm-dkms and developer tools.
- Updated Dockerfile for Ollama to use version 0.13.4 and created a new Dockerfile for ROCm support.
- Enhanced game station scripts for better performance and added MangoHud configuration.
- Modified systemd service files to allow CPU core allocation.
- Updated playbooks for Ollama and Qwen2 to include new configurations and environment variables.
- Added Wake-on-LAN systemd service configuration.
- Updated Terraform configurations to include secret management for the web UI.
- Refactored Kubernetes deployment for open-webui to include secret handling and improved volume management.

ansible/playbook_podman_kokoro.yml

@@ -11,11 +11,20 @@
     pods:
       - name: kokoro
         repo:
-          image: ghcr.io/remsky/kokoro-fastapi-gpu
+          image: ghcr.io/remsky/kokoro-fastapi-cpu
+        env:
+          PYTHONPATH: /app:/app/api
+          # ONNX Optimization Settings for vectorized operations
+          ONNX_NUM_THREADS: 6  # Maximize core usage for vectorized ops
+          ONNX_INTER_OP_THREADS: 4  # Higher inter-op for parallel matrix operations
+          ONNX_EXECUTION_MODE: parallel
+          ONNX_OPTIMIZATION_LEVEL: all
+          ONNX_MEMORY_PATTERN: true
+          ONNX_ARENA_EXTEND_STRATEGY: kNextPowerOfTwo
         ports:
           - "0.0.0.0:{{ kokoro_port | default(8880) }}:8880/tcp"
-        device:
-          - "nvidia.com/gpu=all"
+        # device:
+        #   - "nvidia.com/gpu=all"
 
   tasks:
     - name: Setup Pods