This report outlines the deployment of the **Ollama LLM runtime** on **Arch Linux** specifically tailored for the **AMD Ryzen AI Max+ 395 APU**. The primary focus is optimizing performance by leveraging the integrated **Radeon 8060S iGPU** through the **Vulkan** backend, and considering the potential of the **XDNA 2 NPU** for heterogeneous acceleration.

## 🚀 Overview of the Target Architecture

The AMD Ryzen AI Max+ 395 APU is a highly-integrated system-on-a-chip (SoC) featuring a heterogeneous architecture critical for efficient AI workloads:

  * **Zen 5 CPU Cores:** Provide general-purpose processing and host for the Ollama service.
  * **Radeon 8060S iGPU (RDNA 3.5):** The primary accelerator for LLM inference, utilizing shared system memory as **Unified Memory (VRAM)**.
  * **XDNA 2 Neural Processing Unit (NPU):** Designed for high TOPS AI acceleration, with potential for hybrid execution (prefill/context ingestion on NPU, decoding on iGPU/CPU).

-----

## 🛠️ Arch Linux Prerequisite Setup

Before deploying Ollama, the base Arch Linux installation must have the correct drivers and utilities to fully expose the APU's capabilities, especially for Vulkan and unified memory management.

### 1\. Kernel and Firmware

Ensure the system is running a recent kernel (e.g., $6.10+$ or later) for optimal Zen 5 and RDNA 3.5 support.

```bash
# Update system and install a recent kernel if not already running
sudo pacman -Syu linux linux-headers
```

### 2\. Graphics and Compute Drivers (Vulkan)

The Vulkan backend relies on the open-source **Mesa** stack via the **RADV** driver.

```bash
# Install Mesa with Vulkan support for AMDGPU
sudo pacman -S mesa vulkan-radeon lib32-vulkan-radeon
```

### 3\. ROCm (Optional but Recommended)

While the objective is Vulkan, installing the **ROCm** stack is often necessary for complete AMD GPU compute support and may be leveraged by other frameworks or future Ollama features. The Ryzen AI Max+ 395 is based on the `gfx1150` target, which has improved support in recent ROCm versions (e.g., $6.4+$).

```bash
# Install essential ROCm packages
sudo pacman -S rocm-core
# Install rocm-hip-sdk if developing or using other tools
# sudo pacman -S rocm-hip-sdk
```

### 4\. Memory Configuration

The iGPU uses **shared system RAM (Unified Memory)**. For optimal LLM performance, a large dedicated memory pool for graphics (iGPU/VRAM) is essential.

  * **BIOS/UEFI Configuration:** Set the **UMA Frame Buffer Size** (or equivalent) in the BIOS/UEFI to the maximum supported value (e.g., **16GB** or more, depending on total system RAM). This is the most crucial step for allocating dedicated VRAM.
  * **Kernel Parameters (Optional):** For fine-tuning, verify kernel parameters if necessary, but the BIOS setting is typically sufficient.

-----

## 🧠 Ollama Deployment with Vulkan Acceleration

Ollama relies on an underlying LLM engine (typically a fork of `llama.cpp`). Recent Ollama releases (e.g., $0.12.6+$) include experimental support for Vulkan acceleration, which must be explicitly enabled via an environment variable.

### 1\. Installation

The easiest method is using the Arch package manager or downloading the official binary.

```bash
# Install Ollama from the Arch Linux repository (or AUR)
sudo pacman -S ollama
# OR
# curl -L https://ollama.com/download/install.sh | sh
```

### 2\. Enabling Vulkan Backend

To force Ollama to utilize the more performant Vulkan backend on the AMD iGPU, the `OLLAMA_VULKAN` environment variable must be set when running the service.

#### **Systemd Service Configuration (Recommended)**

Modify the Ollama systemd service unit to include the required environment variable.

1.  **Create an override directory:**

    ```bash
    sudo mkdir -p /etc/systemd/system/ollama.service.d/
    ```

2.  **Create the override file** (`/etc/systemd/system/ollama.service.d/vulkan_override.conf`):

    ```ini
    [Service]
    Environment="OLLAMA_VULKAN=1"
    # Optional: Set the number of threads for the CPU fallback/host operations
    # Environment="OLLAMA_NUM_THREADS=16" 
    ```

3.  **Reload the daemon and restart the service:**

    ```bash
    sudo systemctl daemon-reload
    sudo systemctl restart ollama
    ```

#### **Verification**

Check the service logs for confirmation that Vulkan initialization was successful:

```bash
sudo journalctl -u ollama -f
```

Look for messages indicating Vulkan/GGML initialization on the iGPU. Ollama may also log which accelerator is being used when a model is run.

### 3\. Testing Model Offload

Test a small model, ensuring the output indicates GPU/Vulkan usage. The number of layers offloaded (`--gpu-layers`) is often determined automatically by the available VRAM (shared RAM).

```bash
ollama run llama3:8b
# After the model downloads, monitor system resource usage (e.g., with htop and radeontop)
# The prompt prefill phase will typically show high iGPU usage.
```

-----

## 🌐 Heterogeneous Architecture Strategy

The Ryzen AI Max+ 395 introduces the **XDNA 2 NPU**, enabling true heterogeneous computing. While the Ollama (via `llama.cpp`) Vulkan backend accelerates the iGPU, **direct, standardized, and easily-configured NPU acceleration support within Ollama on Linux is currently limited/experimental** and often requires niche frameworks or ONNX models (e.g., for hybrid execution).

### 1\. Current State: iGPU Dominance (Vulkan/ROCm)

  * **iGPU (Vulkan):** Offers the best performance for general-purpose LLM inference, especially for larger models ($8B+$). This is the focus of the Vulkan optimization.
  * **NPU (XDNA 2):** Best leveraged by specific, often proprietary, toolchains (e.g., Lemonade, ONNXRuntime GenAI) for models converted into an optimized format (e.g., OGA/ONNX).

### 2\. Future Heterogeneous Model (Conceptual)

The optimal, long-term deployment strategy will involve splitting the LLM workload to maximize the strengths of each compute unit.

\<pre class="mermaid"\>
graph LR
subgraph Frontend
A[User Prompt] --\> B(Ollama Server/API)
end
B --\> C{Workload Scheduler (Ollama)}
C --\> |Prompt/Context Prefill (High TTFT focus)| D(XDNA 2 NPU)
C --\> |Token Decoding (High TPS focus)| E(Radeon 8060S iGPU - Vulkan)
D --\> F[NPU Results]
E --\> G[iGPU Results]
F & G --\> H(Final Token Stream)
H --\> B
\</pre\>

> **Figure: Conceptual Heterogeneous LLM Pipeline for AMD Ryzen AI Max+ 395**

### 3\. NPU Exploration (Advanced)

For the skilled Linux user aiming for NPU utilization, direct integration requires bypassing Ollama for NPU-specific frameworks:

  * **ONNX-EP:** Investigate using ONNX Runtime with the AMD EP for NPU-accelerated execution, typically requiring model conversion and separate serving.
  * **AMD Guides:** Follow AMD's specific Linux guides for the Ryzen AI 300 series to enable NPU drivers and libraries, then integrate them with a custom LLM serving solution that can utilize them (e.g., a hand-built `llama.cpp` with NPU support or a dedicated NPU LLM tool).

-----

## ⚙️ Performance Tuning Notes

| Component | Tuning Action | Rationale |
| :--- | :--- | :--- |
| **System RAM** | Maximize physical RAM (e.g., 64GB/128GB). | The iGPU uses unified memory; more RAM directly equates to more **VRAM** for larger models/context. |
| **BIOS/UEFI** | Set UMA Frame Buffer Size to max (e.g., 16GB-32GB). | Crucial for allocating a large, dedicated memory pool for GPU offload. |
| **Ollama** | Use **Q4\_K\_M** or **Q5\_K\_M** quantization. | Optimal balance of VRAM usage and inference speed/quality. Larger quantizations (Q6/Q8) may be slower or consume too much VRAM. |
| **Vulkan** | Ensure `OLLAMA_VULKAN=1` is set. | Forces the use of the Vulkan backend, which is generally reported as faster than ROCm on APUs for LLM workloads. |
| **Model Selection** | Prioritize **MoE (Mixtral, DeepSeek)** models. | The architecture excels at MoE models by efficiently leveraging the available memory bandwidth and unified memory for large models with small active experts. |

-----

Would you like a step-by-step guide on how to build Ollama from source on Arch Linux to ensure the Vulkan backend is compiled correctly?