ramalama-macos(7)

🦋 ramalama-macos — the Mac GPU glow-up guide

okay bestie, welcome to ramalama-macos(7) 💖 this is the page that lives in the Miscellaneous Information Manual, and it's literally just here to get your Mac's GPU poppin' off with Podman. no cap.

---

✨ Configure Podman Machine on Mac for GPU Acceleration ✨

so here's the tea 🍵 — getting GPU acceleration on a Mac with Podman means you gotta configure the libkrun machine provider. that's the whole vibe. no libkrun, no GPU, and we simply cannot have that.

you've got two ways to do it: set an environment variable, or edit the containers.conf file. pick your fighter 💅

and listen — on MacOS you'll likely need to create a new Podman machine with libkrun to actually get at the GPU.

🫠 Podman Machines you made before this? They gotta be recreated to take advantage of the libkrun provider. no reusing, bestie, sorry, them's the rules.

also!! fun fact that's very much worth knowing: you can totally run ramalama without containers on macOS and use MLX directly. deets are in MACOS-INSTALL.md. 🎀

---

🌸 Configuration Methods: 🌸

📁 containers.conf

open up the containers.conf file — it's typically chillin' at $HOME/.config/containers/containers.conf.

then add this line inside the [machine] section:

provider = "libkrun"

and boom 💥 this change persists across sessions. it's permanent, it's iconic, it's yours.

🌈 Environment Variable

set the CONTAINERS_MACHINE_PROVIDER environment variable to libkrun.

😭 heads up though: this one is a temporary change — it lasts until you restart your terminal or your session. so don't come crying to me when it ghosts you.

for example:

export CONTAINERS_MACHINE_PROVIDER=libkrun

🍓 ramalama.conf

RamaLama can also be run in a limited manner without using containers — just specify the --nocontainer option. in that case, MLX is used for native GPU acceleration, which will result in better performance than the podman krunkit solution. lowkey a slay.

to do this, open the ramalama.conf file — typically located at $HOME/.config/ramalama/ramalama.conf.

add the following line within the [machine] section:

container = false

this change persists across sessions ✨

for more details on this setup, read MACOS-INSTALL.md. 🧚

---

👑 Podman Desktop 👑

Creating a Podman Machine with libkrun (MacOS):

1. Go to Settings > Resources in Podman Desktop.
2. In the Podman tile, click Create new.
3. In the Create a Podman machine screen, you can configure the machine's resources — CPU, Memory, Disk size — and enable Machine with root privileges if needed.

⚠️ to use libkrun, make sure the environment variable is set or the containers.conf file is configured before creating the machine. timing matters, bestie!!

once the machine is created, Podman Desktop will manage the connection to the new machine. it's giving hands-off, it's giving main character energy. 💅

---

💀 Important Notes 💀

• on MacOS, libkrun is used to leverage the system's virtualization framework for running containers, and it requires a Podman machine to be created.
• refer to the Podman Desktop documentation for detailed instructions and troubleshooting tips. 📚

---

🔥 SEE ALSO 🔥

• ramalama(1)
• podman-machine(1)
• MACOS_INSTALL.md

---

📜 HISTORY 📜

• Apr 2025, Originally compiled by Dan Walsh dwalsh@redhat.com

---

and that's the whole page, bestie 🫶 you just read ramalama-macos(7) from the Miscellaneous Information Manual, version-dated Apr 2025, and honestly? you ate. go get that GPU. ✨💖🦋