Script 'mail_helper' called by obssrc Hello community, here is the log from the commit of package llamacpp for openSUSE:Factory checked in at 2026-06-05 15:01:35 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Comparing /work/SRC/openSUSE:Factory/llamacpp (Old) and /work/SRC/openSUSE:Factory/.llamacpp.new.2375 (New) ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Package is "llamacpp" Fri Jun 5 15:01:35 2026 rev:28 rq:1357181 version:9500 Changes: -------- --- /work/SRC/openSUSE:Factory/llamacpp/llamacpp.changes 2026-04-25 21:43:16.072237159 +0200 +++ /work/SRC/openSUSE:Factory/.llamacpp.new.2375/llamacpp.changes 2026-06-05 15:02:04.256701461 +0200 @@ -1,0 +2,42 @@ +Thu Jun 4 14:47:32 UTC 2026 - Eyad Issa <[email protected]> + +- Split libllama-common into its own package + +------------------------------------------------------------------- +Thu Jun 4 06:55:00 UTC 2026 - Robert Munteanu <[email protected]> + +- Update to version 9500: + * Added broader model support, including DeepSeek v3.2, + DeepSeekOCR 2, EXAONE 4.5, Gemma 4 variants, Mellum, + Step3.7-Flash, and Granite multilingual embeddings R2. + * Improved multimodal inference with expanded OCR, audio, and + vision handling, including Gemma 4 unified fixes, + DeepSeek-OCR updates, Hunyuan OCR consolidation, granite-speech, + and MiniCPM-V 4.6 support. + * Enhanced CUDA and HIP acceleration with Hopper PDL support, + quantized KV-cache handling, AMD MFMA routing improvements, + Blackwell and NVFP4-related work, and faster walsh-hadamard + transforms. + * Extended Vulkan, OpenCL, SYCL, Metal, and WebGPU backends with + new operators, BF16/Flash Attention improvements, reduced + contention, broader quantization support, and better Intel, + AMD, and Adreno tuning. + * Improved CPU and GGML internals with additional RISC-V RVV/SVE + optimizations, SpacemiT support, faster quantization helpers, + tensor access improvements, and a ggml bump to 0.13.1. + * Expanded server and API support with Vertex AI compatibility, + ETag handling, `/models?reload=1`, SSE improvements, real-time + reasoning interruption, better checkpoint handling, and audio + transcription updates. + * Included important stability and security fixes, notably the + fix for server heap-buffer-overflow CVE-2026-21869, along with + multiple correctness fixes across speculative decoding, + multimodal processing, KV-cache handling, and router mode. + * Updated the application and tooling stack with the new unified + `llama` executable, additional benchmarking and quantization + tools, conversion updates, `gguf-py` 0.19.0, newer BoringSSL, + and refreshed `cpp-httplib`. + * Full commit log: + https://github.com/ggml-org/llama.cpp/compare/b8814...b9500 + +------------------------------------------------------------------- Old: ---- llamacpp-8814.tar.gz New: ---- llamacpp-9500.tar.gz ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Other differences: ------------------ ++++++ llamacpp.spec ++++++ --- /var/tmp/diff_new_pack.PduMFl/_old 2026-06-05 15:02:05.700761245 +0200 +++ /var/tmp/diff_new_pack.PduMFl/_new 2026-06-05 15:02:05.700761245 +0200 @@ -25,11 +25,11 @@ %global mtmd_sover 0.0.%{version} %global mtmd_sover_suffix 0 -%global ggml_sover 0.9.11 +%global ggml_sover 0.13.1 %global ggml_sover_suffix 0 Name: llamacpp -Version: 8814 +Version: 9500 Release: 0 Summary: Inference of Meta's LLaMA model (and others) in pure C/C++ License: MIT @@ -74,6 +74,17 @@ This package includes the shared libraries necessary for running applications that depend on libllama.so. +%package -n libllama-common%{llama_sover_suffix} +Summary: Common library for llama.cpp + +%description -n libllama-common%{llama_sover_suffix} +The llama.cpp library provides a C++ interface for running inference +with large language models (LLMs). Initially designed to support Meta's +LLaMA model, it has since been extended to work with a variety of other models. + +This package includes the shared libraries necessary for running applications +that depend on libllama-common.so. + %package -n libggml%{ggml_sover_suffix} Summary: A tensor library for C++ Requires: libggml-cpu @@ -153,6 +164,7 @@ Library to handle multimodal inputs for llama.cpp. %ldconfig_scriptlets -n libllama%{llama_sover_suffix} +%ldconfig_scriptlets -n libllama-common%{llama_sover_suffix} %ldconfig_scriptlets -n libggml%{ggml_sover_suffix} %ldconfig_scriptlets -n libggml-base%{ggml_sover_suffix} %ldconfig_scriptlets -n libmtmd%{mtmd_sover_suffix} @@ -189,13 +201,20 @@ %install %cmake_install -# dev scripts -rm %{buildroot}%{_bindir}/convert_hf_to_gguf.py - %files %doc README.md %license LICENSE +%{_bindir}/llama %{_bindir}/llama-* +# private libraries +%{_libdir}/libllama-batched-bench-impl.so +%{_libdir}/libllama-bench-impl.so +%{_libdir}/libllama-cli-impl.so +%{_libdir}/libllama-completion-impl.so +%{_libdir}/libllama-fit-params-impl.so +%{_libdir}/libllama-perplexity-impl.so +%{_libdir}/libllama-quantize-impl.so +%{_libdir}/libllama-server-impl.so %files devel %license LICENSE @@ -209,11 +228,18 @@ # libmtmd symlinks %{_libdir}/libmtmd.so %{_libdir}/libmtmd.so.0 +# libllama-common symlinks +%{_libdir}/libllama-common.so +%{_libdir}/libllama-common.so.0 %files -n libllama%{llama_sover_suffix} %license LICENSE %{_libdir}/libllama.so.%{llama_sover} +%files -n libllama-common%{llama_sover_suffix} +%license LICENSE +%{_libdir}/libllama-common.so.%{llama_sover} + %files -n libggml%{ggml_sover_suffix} %license LICENSE %{_libdir}/libggml.so.%{ggml_sover} ++++++ llamacpp-8814.tar.gz -> llamacpp-9500.tar.gz ++++++ /work/SRC/openSUSE:Factory/llamacpp/llamacpp-8814.tar.gz /work/SRC/openSUSE:Factory/.llamacpp.new.2375/llamacpp-9500.tar.gz differ: char 13, line 1
