llama.cpp b11349 Released: Vulkan Logging, Platform Changes
The llama.cpp b11349 release adds Vulkan pipeline compile issue logging and disables both KleidiAI-enabled macOS Apple Silicon builds and openEuler platform support. Updated binaries are available for many platforms.
What changed?
llama.cpp version b11349 introduces logging for Vulkan pipeline compile issues, providing developers with direct feedback when pipeline compilation fails. This version also disables KleidiAI-enabled builds for macOS Apple Silicon and drops support for the openEuler platform. Pre-built binaries for a wide range of operating systems and hardware continue to be provided in this release.
Why does it matter to an everyday developer?
The new Vulkan logging helps developers who use GPU acceleration to diagnose and fix issues with pipeline compilation, potentially reducing debugging time. If you rely on KleidiAI-enabled Apple Silicon builds or openEuler platform support, you will need to adjust your workflows, as these are now unavailable. For most other users, updated binaries for macOS, Windows, Linux, Android, and iOS are still available, ensuring continued easy integration.
What can the developer do now?
- Download the updated b11349 binaries for your platform from the llama.cpp release page.
- Take advantage of new Vulkan logging if using GPU acceleration—to debug or optimize your deployment.
- Review your build and deployment process if you were relying on KleidiAI-enabled macOS Apple Silicon or openEuler support, and plan to use a supported configuration.
- Continue standard llama.cpp integration on all other supported platforms.