Script 'mail_helper' called by obssrc Hello community, here is the log from the commit of package xsimd for openSUSE:Factory checked in at 2026-07-29 18:58:22 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Comparing /work/SRC/openSUSE:Factory/xsimd (Old) and /work/SRC/openSUSE:Factory/.xsimd.new.2004 (New) ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Package is "xsimd" Wed Jul 29 18:58:22 2026 rev:16 rq:1367925 version:14.3.0 Changes: -------- --- /work/SRC/openSUSE:Factory/xsimd/xsimd.changes 2026-04-16 17:25:53.236758390 +0200 +++ /work/SRC/openSUSE:Factory/.xsimd.new.2004/xsimd.changes 2026-07-29 18:58:37.624357773 +0200 @@ -1,0 +2,72 @@ +Wed Jul 22 14:18:07 UTC 2026 - Dirk Stoecker <[email protected]> + +- update to 14.3.0 + * [API] Feat/mulhilo + * [API] Add batch::to_native + * [API] fused mul_hilo for 64-bit batches (shared 32x32->64 partials) + * [API] add runtime batch_bool mask overloads for load_masked/store_masked + * [AVX512VL] Tentative support for avx512vl extensions to 256 bit registers + * [AVX512VL] Minimal support for avx512vl + * [AVX512VL] Tentative support for avx512vl extensions to 128 bit registers + * [AVX512VL] fix: avx512vl masked load/store#1354Add efficient_bmi2 cpu + feature + * [AVX512VL] Fix avx512vl_{128,256} correctness bugs + CI default-arch typo + * [AVX512VL] perf: native EVEX int64 shift, signed rotr, compress/expand + * [AVX512VL] Add gather scatter kernels + * [AVX512VL_128] route avx512vl_128 FMA ops through fma3<avx2_128> + * [AVX512VP] Fix avx512vp cpuid + * [AVX512F] single vpermt2 for zip_lo/zip_hi + * [AVX512F] vectorize 16x16 float / 8x8 double transpose#1372Add an + explicit SSE2 path for batch_bool_constant masks where the three low + lanes are active and lane 3 is zero + * [AVX512, AVX2] plain-move lowering for int & EVEX constant-mask + load/store#1381Prefer avx512vl_256 over avxvnni for 256-bit sized batches + * [AVX2] AVX256 constant mask plain moves + * [AVXVNNI] route avxvnni FMA ops through fma3<avx2> kernels + * [AVX] Tentative support for avx extensions to 128 bit registers + * [VSX] clang build + * [NEON] Fix incorrect guard for system headers on Windows + * [NEON] under-aligned store in generic swizzle kernel + * [NEON, SSE2] Fix unreachable constant-mask load_masked kernels + * [ci] Add clang workflow for PPC + * [doc] Fix doc logo + * [doc] Consolidate build options in one place + * [doc] Fix typos + * [doc] Update RTD deps + * [doc] Add jupyterlite notebook + * [doc] Add link to deployed jupyterlite + * Add more x86 cpu feature + * Fix C++ version detection for MSVC + * Uniform header inclusion + * Resolve xtl in user cmake + * fix (#1375): under-aligned buffers in large-arg range reduction + +14.2.0 + * New architecture: IBM Z (s390x) support + * [API] New cross-platform cpu_features API for querying CPU features + available at runtime + * [API] Add xsimd::get<I>() for compile-time lane extraction + * [API] Add xsimd::stream_load, xsimd::stream_store, and xsimd::fence + for non-temporal memory transfers + * [VSX] Fix dynamic dispatch support with runtime cpu feature inspection + * [VSX] Fix rounding + * [SVE/RVV] Fix dynamic dispatch by inspecting available vector length + * [AVX2] Add native uint64/int64 multiplication kernel + * [NEON] Add support for Windows ARM + * [NEON] Simplify static dispatch of intrinsicts + * [NEON] Fix batch_bool store on ARM by replacing vst1_lane_u32 + with a full lane store followed by a memcpy + * [SVE] Fix dynamic dispatch ODR violation + * [ci] Fix emulated architecture interaction with AVX512 leading to CI + failures. Provide a cmake-level configuration switch for emulated build + * Fix build with compilers that do not support C++20 (even though we + only require C++14) + * Fix xsimd::signbit scalar overload leaking into non-scalar overload + resolution + * Fix complex batch load + * Harden fast-math reassociation barriers + * Publish the C++14 requirement through the CMake interface + +Drop add 0001-Revert-Extend-1172-approach-to-arm-store-version.patch + +------------------------------------------------------------------- Old: ---- 0001-Revert-Extend-1172-approach-to-arm-store-version.patch xsimd-14.1.0.tar.gz New: ---- xsimd-14.3.0.tar.gz ----------(Old B)---------- Old: Drop add 0001-Revert-Extend-1172-approach-to-arm-store-version.patch ----------(Old E)---------- ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Other differences: ------------------ ++++++ xsimd.spec ++++++ --- /var/tmp/diff_new_pack.mc4jtQ/_old 2026-07-29 18:58:38.688394473 +0200 +++ /var/tmp/diff_new_pack.mc4jtQ/_new 2026-07-29 18:58:38.688394473 +0200 @@ -18,14 +18,13 @@ %{?sle15_python_module_pythons} Name: xsimd -Version: 14.1.0 +Version: 14.3.0 Release: 0 Summary: C++ wrappers for SIMD intrinsics License: BSD-3-Clause Group: Development/Libraries/C and C++ URL: https://xsimd.readthedocs.io/en/latest/ Source0: https://github.com/xtensor-stack/xsimd/archive/refs/tags/%{version}.tar.gz#/%{name}-%{version}.tar.gz -Patch1: 0001-Revert-Extend-1172-approach-to-arm-store-version.patch BuildRequires: %{python_module breathe} BuildRequires: %{python_module sphinx_rtd_theme} BuildRequires: cmake ++++++ xsimd-14.1.0.tar.gz -> xsimd-14.3.0.tar.gz ++++++ ++++ 21512 lines of diff (skipped)
