llama.cpp Release b11269 Adds CI Build Support for zDNN Backend
llama.cpp release b11269 adds continuous integration (CI) build steps for the ggml-zdnn backend and resolves several compiler errors. The release also sets up build testing using an Ubuntu 26.04 container environment. Integrating zDNN into the CI pipeline ensures ongoing build health and code quality for running LLMs on IBM zSystems hardware. This expands llama.cpp's footprint into enterprise mainframe environments featuring dedicated AI acceleration hardware. The pull request enables build validation for the zDNN backend without running test suites yet. It also configures Bash as the default CI shell and ignores compiler warnings from vendored dependency files.
## BACKGROUND
llama.cpp is a widely used C/C++ framework for running LLM inference locally across various hardware backends. IBM zDNN (IBM Z Deep Neural Network Library) provides deep learning primitives optimized for the on-chip Integrated Accelerator for AI found in IBM zSystems mainframes.