llama.cpp Release b11840 Adds Hexagon CEIL Operation
Release b11840 introduces the CEIL operation for the Qualcomm Hexagon backend in both f32 and f16 variants, along with comprehensive multi-platform binary distributions. This update improves hardware-accelerated large language model inference on Qualcomm Snapdragon mobile and edge devices, expanding functional coverage for the Hexagon NPU backend. The update specifically implements the hex-ceil operation for f32 and f16 data types, and updates the supported operations documentation for the Hexagon backend.
## BACKGROUND
llama.cpp is an open-source library that enables efficient inference for large language models on local hardware. The Qualcomm Hexagon is a digital signal processor and neural processing unit architecture designed to deliver high-performance tensor operations with low power consumption on Snapdragon devices.