llama.cpp Release b11550 Fixes CMake Build Issue on s390x
llama.cpp released build b11550, featuring a targeted CMake fix that disables the z17 architecture target on s390x when using unsupported compilers. This prevents build failures during compilation on certain IBM mainframe environments. This release maintains broad portability for llama.cpp by ensuring developers and enterprise users running Linux on IBM Z mainframes can reliably build the library. It prevents compiler errors caused by hardware-specific flags on older compiler toolchains. The fix was implemented in pull request #30297, explicitly checking compiler capability before enabling IBM z17 instruction optimizations. Pre-built binaries for build b11550 continue to support Linux (CPU, CUDA, Vulkan, ROCm, s390x), Windows, macOS, Android, and Snapdragon platforms.
## BACKGROUND
llama.cpp is a widely used open-source C/C++ framework for running Large Language Model (LLM) inference locally with high efficiency. s390x refers to the 64-bit Linux architecture for IBM mainframes, where IBM z17 represents the generation of mainframe processors that introduced specialized AI vector execution acceleration.