Update:

ggerganov/llama.cpp b9221: b9221

GitHub·AI·Silent·5/18/2026

<details open> ggml-hexagon: add PAD op HVX kernel (#23078) * ggml-hexagon: add PAD op HVX kernel Implements GGML_OP_PAD on the Hexagon HTP backend using HVX vectorized kernels. Supports zero-padding and circular padding across all 4 tensor dimensions. * hex-ggml: remove duplicate op cases (merg

Why it matters → llama.cpp released an update. Review the changelog for relevant changes.

Who should care → Teams using llama.cpp.

llama.cpp

View original source ↗

Source payload preview

{
  "id": 324626016,
  "url": "https://api.github.com/repos/ggml-org/llama.cpp/releases/324626016",
  "body": "<details open>\n\nggml-hexagon: add PAD op HVX kernel (#23078)\n\n* ggml-hexagon: add PAD op HVX kernel\n\nImplements GGML_OP_PAD on the Hexagon HTP backend using HVX vectorized\nkernels. Supports zero-padding and circular padding across all 4 tensor\ndimensions.\n\n* hex-ggml: remove duplicate op cases (merge conflict)\n\n* hex-pad: fix editorconfig checks and macro alignment\n\n---------\n\nCo-authored-by: Max Krasnyansky <maxk@qti.qualcomm.com>\n\n</details>\n\n**macOS/iOS:**\n- [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b9221/llama-b9221-bin-macos-arm64.tar.gz)\n- [macOS Apple Silicon (arm64, KleidiAI enabled)](https://github.com/ggml
…