Script 'mail_helper' called by obssrc Hello community, here is the log from the commit of package llamacpp for openSUSE:Factory checked in at 2026-06-12 19:27:57 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Comparing /work/SRC/openSUSE:Factory/llamacpp (Old) and /work/SRC/openSUSE:Factory/.llamacpp.new.1981 (New) ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Package is "llamacpp" Fri Jun 12 19:27:57 2026 rev:29 rq:1358796 version:9553 Changes: -------- --- /work/SRC/openSUSE:Factory/llamacpp/llamacpp.changes 2026-06-05 15:02:04.256701461 +0200 +++ /work/SRC/openSUSE:Factory/.llamacpp.new.1981/llamacpp.changes 2026-06-12 19:29:07.354202914 +0200 @@ -1,0 +2,38 @@ +Sun Jun 7 22:27:41 UTC 2026 - Robert Munteanu <[email protected]> + +- Update to version 9553: + * Added new model and multimodal support, including Granite 4 + Vision, Gemma 4 MTP, and Qwen-VL video handling via frame-merge + support. + * Improved accelerator backends with faster CUDA MTP and MoE paths, + SYCL multi-column MMVQ, Vulkan Intel FWHT support with safer + feature detection, OpenCL kernel tuning, and WASM SIMD128 + vectorization. + * Extended server and API behavior with multimodal token counting, + a new `*/input_tokens` endpoint, reduced unnecessary checkpoint + restores, and disabled on-device speculative checkpoints. + * Updated build and packaging bits: CMake now skips CPU-only tools + when the CPU backend is disabled, XCFramework header and + module-map generation was fixed, Docker CUDA 13 moved to 13.3.0, + and MSYS CI dependencies were slimmed down. + * Improved downloader and CLI behavior by avoiding duplicate MTP + downloads, skipping mmproj downloads when explicitly supplied, + and fixing CLI model parameters not being propagated correctly. + * Fixed multiple conversion and model-compatibility issues, + including Gemma 4 unified and no-audio cases, Gemma 4 audio + projector sizing, and Mistral-Medium-3.5-128B GGUF conversion. + * Fixed GPU, layer, cache, and speculative decoding stability + issues, including `n_gpu_layers` handling, vocab compatibility + checks, and KV-cache sharing and copy assertions. + * Refreshed the WebUI with single-line reasoning preview, improved + keyboard accessibility, fixed tool selector state and counting, + pinned packages for reproducible installs, and enabled npm + install safeguards with `ignore-scripts=true`. + * Improved chat and tool-call interoperability by fixing LFM2 and + LFM2.5 tool parsing and reasoning round-trip handling, and by + relaxing sampler name matching to accept common alias forms. + * Full commit log: + https://github.com/ggml-org/llama.cpp/compare/b9500...b9553 +- Include the web ui assets in the RPM package + +------------------------------------------------------------------- Old: ---- llamacpp-9500.tar.gz New: ---- llama-b9553-ui.tar.gz llamacpp-9553.tar.gz ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Other differences: ------------------ ++++++ llamacpp.spec ++++++ --- /var/tmp/diff_new_pack.JjvEBp/_old 2026-06-12 19:29:08.490250448 +0200 +++ /var/tmp/diff_new_pack.JjvEBp/_new 2026-06-12 19:29:08.494250616 +0200 @@ -29,12 +29,13 @@ %global ggml_sover_suffix 0 Name: llamacpp -Version: 9500 +Version: 9553 Release: 0 Summary: Inference of Meta's LLaMA model (and others) in pure C/C++ License: MIT URL: https://github.com/ggml-org/llama.cpp Source: %{URL}/archive/b%{version}/%{name}-%{version}.tar.gz +Source1: %{URL}/releases/download/b%{version}/llama-b%{version}-ui.tar.gz BuildRequires: cmake >= 3.14 BuildRequires: gcc-c++ BuildRequires: git @@ -171,6 +172,8 @@ %prep %autosetup -p1 -n llama.cpp-b%{version} +mkdir -p tools/ui/dist +tar -xzf %{SOURCE1} --strip-components=1 -C tools/ui/dist %build ++++++ llamacpp-9500.tar.gz -> llamacpp-9553.tar.gz ++++++ /work/SRC/openSUSE:Factory/llamacpp/llamacpp-9500.tar.gz /work/SRC/openSUSE:Factory/.llamacpp.new.1981/llamacpp-9553.tar.gz differ: char 15, line 1
