Script 'mail_helper' called by obssrc Hello community, here is the log from the commit of package llamacpp for openSUSE:Factory checked in at 2026-08-05 17:48:38 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Comparing /work/SRC/openSUSE:Factory/llamacpp (Old) and /work/SRC/openSUSE:Factory/.llamacpp.new.16738 (New) ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Package is "llamacpp" Wed Aug 5 17:48:38 2026 rev:33 rq:1369523 version:10154 Changes: -------- --- /work/SRC/openSUSE:Factory/llamacpp/llamacpp.changes 2026-07-31 16:09:00.467978333 +0200 +++ /work/SRC/openSUSE:Factory/.llamacpp.new.16738/llamacpp.changes 2026-08-05 17:49:35.608350772 +0200 @@ -1,0 +2,53 @@ +Tue Jul 28 07:02:44 UTC 2026 - Robert Munteanu <[email protected]> + +- Update to version 10154 + * New model support: Nanbeige 4.2, MiniMax-M3 (MSA sparse attention), + GLM-5.2-Vision, Laguna XS.2 and M.1. + * CUDA: GET_ROWS for quantized types, improved NVFP4 W4A4 activation + quantization, sqrt_softplus in top-k MoE for DeepSeek v4, and a + fix for external compilation of Q1_0 MMQ. + * HIP: remove rocWMMA FlashAttention path. + * Metal: add f16 support to leaky relu. + * Hexagon: core pipeline improvements (L2, DMA, MM, FA), partial + im2col support, activation ops update, and Windows crash fix when + op_poll is enabled. + * OpenCL: cache compiled cl_program binaries on disk, fix fused RMS + norm mul view offset, and do not treat NULL-mask flash attention + as causal. + * WebGPU: add CONV_2D_DW (depthwise conv2d) kernel and fix WASM + compilation with OpenMP. + * SYCL: parallelize ocloc invocations. + * CPU/ggml: enable BF16 tiled GEMM on PowerPC, fix offloading logic + for weight backend, update PowerPC backend variants on AIX. + * Server/API: support MCP stdio tool transport, add format argument + to datetime tool, support "reasoning_effort": "none" in OAI API, + improve error handling for null context and conversation IDs, fix + stream routes for model names containing a slash, and add missing + task parameters in generation_settings. + * Common: add common_print_available_devices(), add subproc.h + wrapper, support multiple end sequences in the reasoning budget + sampler, infer speculative type from draft repo sidecars, refactor + mlock/mmap/directio into load-mode, and add -lm mlock option. + * UI: symbolic math support via nerdamer in JS sandbox, fix symbolic + math tool prompt, fix system message edit box, detect conversation + import format from file contents, reduce per-token render cost, + remove render effects, fix MCP server display name conflicts, add + "Default" option for reasoning selector, and fix context gauge + regressions. + * MTMD (multimodal): add vision support for MiniMax-M3 and + GLM-5.2-Vision, add GLM 5.2 Indexer support, fix Android build, use + align_corners for Qwen3-VL interpolation, and use RAII for + non-causal attention handling. + * Fixes: DeepSeek v4 crafted template and APE tensor op, non-MoE + NomicBert GGUF conversion, HunyuanVL XD-RoPE config, use-after-free + when loading LoRA adapter fails, and various llama bugs. + * Full commit log: + https://github.com/ggml-org/llama.cpp/compare/b10078...b10154 + +------------------------------------------------------------------- +Tue Jul 28 00:00:00 UTC 2026 - Torsten Duwe <[email protected]> + +- llama.cpp now enables SME, found in ARMv9.2. Only gcc>=16 knows about SME, + so require it on arm64. + +------------------------------------------------------------------- Old: ---- llama-b10078-ui.tar.gz llamacpp-10078.tar.gz New: ---- llama-b10154-ui.tar.gz llamacpp-10154.tar.gz ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Other differences: ------------------ ++++++ llamacpp.spec ++++++ --- /var/tmp/diff_new_pack.nJ4xso/_old 2026-08-05 17:49:36.560384802 +0200 +++ /var/tmp/diff_new_pack.nJ4xso/_new 2026-08-05 17:49:36.560384802 +0200 @@ -35,7 +35,7 @@ %endif Name: llamacpp -Version: 10078 +Version: 10154 Release: 0 Summary: Inference of Meta's LLaMA model (and others) in pure C/C++ License: MIT @@ -43,7 +43,12 @@ 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 +# Newer llama.cpp enables SME, found in ARMv9.2. Only gcc>=16 knows about SME +%ifarch aarch64 +BuildRequires: gcc16-c++ +%else BuildRequires: gcc-c++ +%endif BuildRequires: git BuildRequires: ninja BuildRequires: pkgconfig ++++++ llama-b10078-ui.tar.gz -> llama-b10154-ui.tar.gz ++++++ ++++ 7442 lines of diff (skipped) ++++++ llamacpp-10078.tar.gz -> llamacpp-10154.tar.gz ++++++ /work/SRC/openSUSE:Factory/llamacpp/llamacpp-10078.tar.gz /work/SRC/openSUSE:Factory/.llamacpp.new.16738/llamacpp-10154.tar.gz differ: char 16, line 1
