PyTorch and llama.cpp build requests

Activity in AI and ML tooling was limited to two low-volume build and packaging proposals. Contributors sought clearer support for optional runtimes and platform-specific binaries in widely used inference stacks.

NativeRT CMake support questioned in PyTorch

An open-source user opened an issue asking whether NativeRT is intended for external users after it was dropped from the CMake build options. The request seeks restoration of the option so that NativeRT remains available outside internal workflows. Readers tracking PyTorch packaging will note the uncertainty around which runtime components stay first-class in public builds.

Official Windows Arm Hexagon NPU builds requested for llama.cpp

A contributor filed a feature request asking the llama.cpp project to publish formal GitHub releases of Windows Arm binaries that target Qualcomm Hexagon NPUs. The ask centers on making those builds regularly available rather than leaving users to produce them ad hoc. Developers on Windows Arm hardware would gain simpler access to accelerated inference if the project adopts the practice.