SteamOS on Mac: How the FX Steam Launcher Mod Works
While Apple continues to expand gaming capabilities on macOS, third-party developers are finding ways to bypass ecosystem limitations. A new project called FX Steam Launcher allows Windows games to run on Mac devices through SteamOS. This solution utilizes the ARM implementation of SteamOS designed for the Steam Frame VR headset and operates within a virtual machine, requiring no major modifications to the host system.
Unlike many other implementations that attempt to adapt the handheld version of SteamOS, FX Steam Launcher leverages the ARM64 architecture. This is a crucial distinction for Apple Silicon Mac owners. The system runs inside a lightweight virtual machine based on Hypervisor.framework (via libkrun) with GPU acceleration provided by Venus. This setup allows users to run Steam games, including Windows titles through Proton and FEX translation layers, while maintaining full access to macOS.
Technical Details and Compatibility
The project's mechanics rely on translating graphics instructions. For DirectX 9, 10, and 11 games, it uses DXVK, while DirectX 12 titles rely on VKD3D-Proton. Graphics instructions are passed through Mesa Venus, MoltenVK, or the alternative KosmicKrisp driver to Apple's Metal API. However, there are notable limitations: games utilizing anti-cheat systems that block virtual machines will not function. Additionally, games requiring DirectX Feature Level 12_1 or Shader Model 6.6 remain unsupported.
Currently, the project requires Apple Silicon Macs running macOS 15 or newer. Notably, using the KosmicKrisp driver (required for DirectX 12_0 games) will necessitate macOS 26 or later.
Performance and Resource Usage
Initial tests on a MacBook Pro equipped with an M4 Max chip have shown promising but inconsistent results. While some games fail to launch, those that do run perform quite well. According to benchmarks by ETA Prime, performance in several titles was measured as follows:
Memory overhead is a significant factor. On a test machine with 36GB of unified memory, total system usage reached 27GB, with approximately 19GB allocated specifically to the VM. This highlights that users with limited RAM may struggle to run this setup effectively.
Гика
