b11109 b11110
low risk
ggml-org/llama.cppSep 22, 2026 → Sep 22, 20260 days between releases
What's New
Adds sanity checks to Multi-Threaded Multi-Device (MTD) implementation to improve stability.
New Features
- Enhanced validation for Multi-Threaded Multi-Device (MTD) operations
Improvements
- Improved operational robustness of the MTD implementation
- Added various sanity checks to prevent potential runtime errors
Bug Fixes
- Restricts MUL_MAT_ID operations in Metal, CUDA, and Vulkan backends to prevent failures with F32 precision
Upgrade Notes
Upgrade to b11110 to benefit from improved stability and validation in the Multi-Threaded Multi-Device implementation. No configuration changes are required.
The Two Releases
b11109
Pre-release
b11109
Sep 22, 2026· 30 days ago
Restricts MUL_MAT_ID operations in Metal, CUDA, and Vulkan backends based on source precision to prevent backend failures.
Full release notesb11110
Pre-release
b11110
Sep 22, 2026· 30 days ago
Adds various sanity checks to the Multi-Threaded Multi-Device (MTD) implementation to improve operational robustness.
Full release notes