Install Ollama with VS Code on Omarchy on an hybird Nvidia laptop

OLLAMA_VSCODE_OMARCHY.md

Local Copilot with Ollama on Omarchy (Arch Linux)

This document explains how to configure VS Code Copilot Chat to use a local LLM running on your NVIDIA GPU, optimized for Ruby on Rails development.

---

1. NVIDIA Driver Setup (Hybrid Intel + NVIDIA)

Run the following commands:

sudo pacman -S nvidia nvidia-utils nvidia-settings linux-headers

Create DRM config:

sudo vim /etc/modprobe.d/nvidia.conf

Add:

options nvidia_drm modeset=1

Rebuild and reboot:

sudo limine-mkinitcpio -P
sudo reboot

Verify:

nvidia-smi

---

2. Install CUDA-enabled Ollama

sudo pacman -S ollama-cuda
sudo systemctl enable ollama
sudo systemctl start ollama

Check version:

ollama --version

---

3. Force Ollama to Use NVIDIA GPU

On hybrid systems, explicitly bind Ollama to the discrete GPU.

Edit the service:

sudo systemctl edit ollama

Add:

[Service]
Environment=CUDA_VISIBLE_DEVICES=0

Reload and restart:

sudo systemctl daemon-reexec
sudo systemctl restart ollama

---

4. Install a Rails-Capable Model

ollama pull qwen3:14b
# or lighter
ollama pull qwen3:7b

---

5. Verify GPU Inference

Terminal A:

nvidia-smi -l 1

Terminal B:

ollama run qwen3:14b

You should see ollama using several GB of VRAM.

---

6. Install VS Code + Copilot

sudo pacman -S code

In VS Code Extensions, install:

• GitHub Copilot
• GitHub Copilot Chat

Sign in to GitHub.

---

7. Configure Copilot to Use Ollama

In VS Code:

• Open Copilot Chat
• Model Dropdown → Manage Models
• Provider: Ollama
• Model: qwen3:14b

---

Final Architecture

Component	GPU
OS / UI	Intel
Ollama / LLM	NVIDIA
Copilot Chat	Local

---

Sanity Checks

glxinfo | grep "OpenGL renderer"
nvidia-smi

Expected:

• Intel renderer for desktop
• ollama visible during inference

---

Result:

Fully local, GPU-accelerated Copilot for Ruby on Rails.