How to Run Ollama on Linux with AMD MI50

Tested Environment: Ubuntu 22.04

Quick Start Summary

# 1. Install ROCm driver (ROCm 5.7.1 for MI50)
sudo mkdir -p /etc/apt/keyrings
wget https://repo.radeon.com/rocm/rocm.gpg.key -O - | gpg --dearmor | sudo tee /etc/apt/keyrings/rocm.gpg > /dev/null
sudo tee /etc/apt/sources.list.d/amdgpu.list <<'EOF'
deb [arch=amd64 signed-by=/etc/apt/keyrings/rocm.gpg] https://repo.radeon.com/amdgpu/5.7.1/ubuntu jammy main
EOF
sudo tee /etc/apt/sources.list.d/rocm.list <<'EOF'
deb [arch=amd64 signed-by=/etc/apt/keyrings/rocm.gpg] https://repo.radeon.com/rocm/apt/5.7.1 jammy main
EOF
echo -e 'Package: *\nPin: release o=repo.radeon.com\nPin-Priority: 600' | sudo tee /etc/apt/preferences.d/rocm-pin-600
sudo apt update
sudo apt install amdgpu-dkms rocm-hip-libraries
sudo reboot

# 2. Install Ollama
curl -LO https://ollama.com/download/ollama-linux-amd64.tgz
sudo tar -C /usr -xzf ollama-linux-amd64.tgz

# 3. Install ROCm package for Ollama
curl -L https://ollama.com/download/ollama-linux-amd64-rocm.tgz -o ollama-linux-amd64-rocm.tgz
sudo tar -C /usr -xzf ollama-linux-amd64-rocm.tgz

# 4. Run Ollama
ollama serve

Install ROCm Driver

If you want to run Ollama on Linux with an AMD MI50 GPU, you need to install the AMD graphics driver first.

I recommend using Ubuntu 22.04, as it is stable for running AMD graphics drivers and local large language models. I have tested this setup successfully.

Official ROCm installation guide: 👉 ROCm Quick Start

1. Download and Convert the Package Signing Key

sudo mkdir --parents --mode=0755 /etc/apt/keyrings
wget https://repo.radeon.com/rocm/rocm.gpg.key -O - | \
    gpg --dearmor | sudo tee /etc/apt/keyrings/rocm.gpg > /dev/null

2. Add the Repositories

sudo tee /etc/apt/sources.list.d/amdgpu.list <<'EOF'
deb [arch=amd64 signed-by=/etc/apt/keyrings/rocm.gpg] https://repo.radeon.com/amdgpu/5.7.1/ubuntu jammy main
EOF

sudo tee /etc/apt/sources.list.d/rocm.list <<'EOF'
deb [arch=amd64 signed-by=/etc/apt/keyrings/rocm.gpg] https://repo.radeon.com/rocm/apt/5.7.1 jammy main
EOF

echo -e 'Package: *\nPin: release o=repo.radeon.com\nPin-Priority: 600' \
    | sudo tee /etc/apt/preferences.d/rocm-pin-600

3. Update the Package List

sudo apt update
sudo apt install amdgpu-dkms

4. Install ROCm Runtimes

sudo apt install rocm-hip-libraries

5. Reboot the System

sudo reboot

Important: Always reboot after installing the ROCm driver.

Why Choose ROCm 5.7?

  • The AMD MI50 is an older GPU, and newer ROCm drivers no longer support it.
  • ROCm 6.4.x is not stable for this GPU.
  • Version 5.7.1 is the last stable version with official (deprecated) support.

Support Status:

The current ROCm release has limited support for this hardware. Existing features are maintained, but no new features or optimizations will be added. Support will be removed in a future release.

Install Ollama

Once the driver is installed, follow the official Ollama AMD GPU guide.

Manual Install

If upgrading from a previous version, remove old libraries first:

sudo rm -rf /usr/lib/ollama

Download and Install:

curl -LO https://ollama.com/download/ollama-linux-amd64.tgz
sudo tar -C /usr -xzf ollama-linux-amd64.tgz

Start Ollama:

ollama serve

Verify Installation:

ollama -v

Install AMD GPU Support for Ollama

curl -L https://ollama.com/download/ollama-linux-amd64-rocm.tgz -o ollama-linux-amd64-rocm.tgz
sudo tar -C /usr -xzf ollama-linux-amd64-rocm.tgz

Create a user and group:

sudo useradd -r -s /bin/false -U -m -d /usr/share/ollama ollama
sudo usermod -a -G ollama $(whoami)

Create the service file /etc/systemd/system/ollama.service:

[Unit]
Description=Ollama Service
After=network-online.target

[Service]
ExecStart=/usr/bin/ollama serve
User=ollama
Group=ollama
Restart=always
RestartSec=3
Environment="PATH=$PATH"

[Install]
WantedBy=multi-user.target

Enable and Start the Service:

sudo systemctl daemon-reload
sudo systemctl enable ollama
sudo systemctl start ollama

Customizing Ollama Settings

You can edit the service file:

sudo systemctl edit ollama

Or create an override file: /etc/systemd/system/ollama.service.d/override.conf

Example (custom settings):

[Service]
Environment="OLLAMA_DEBUG=1"
Environment="OLLAMA_HOST=0.0.0.0"
Environment="OLLAMA_ORIGINS=*"
Environment="OLLAMA_KEEP_ALIVE=24h"
Environment="OLLAMA_NUM_PARALLEL=2"
Environment="OLLAMA_MAX_LOADED_MODELS=2"

Key Parameters:

Parameter Description
OLLAMA_MODELS Model storage directory. On Windows, avoid storing on C: — use another drive.
OLLAMA_HOST Network address for Ollama to listen on. Use 0.0.0.0 for LAN access.
OLLAMA_PORT Default port is 11434. Change if needed.
OLLAMA_ORIGINS Allowed HTTP origins. Use * for no restriction.
OLLAMA_KEEP_ALIVE Time to keep a model in memory (5m default). 24h recommended for faster reuse.
OLLAMA_NUM_PARALLEL Number of parallel request handlers. Default is 1.
OLLAMA_MAX_QUEUE Request queue length (default 512). Requests beyond this limit are dropped.
OLLAMA_DEBUG Set to 1 for debug logs.
OLLAMA_MAX_LOADED_MODELS Maximum number of loaded models at once (default 1).

Updating Ollama

Re-run the install script:

curl -fsSL https://ollama.com/install.sh | sh

Or manually download and extract:

curl -L https://ollama.com/download/ollama-linux-amd64.tgz -o ollama-linux-amd64.tgz
sudo tar -C /usr -xzf ollama-linux-amd64.tgz

Viewing Logs

journalctl -e -u ollama

Uninstalling Ollama

sudo systemctl stop ollama
sudo systemctl disable ollama
sudo rm /etc/systemd/system/ollama.service
sudo rm $(which ollama)
sudo rm -r /usr/share/ollama
sudo userdel ollama
sudo groupdel ollama
sudo rm -rf /usr/local/lib/ollama