Source: crmvec
Section: libs
Priority: optional
Maintainer: anun333 <anun333@posteo.net>
Build-Depends: debhelper-compat (= 13), libsimde-dev [arm64], libmpfr-dev [amd64]
Standards-Version: 4.6.2
Homepage: https://github.com/anun333/crmvec
Rules-Requires-Root: no

Package: crmvec
Architecture: amd64 arm64
Multi-Arch: same
Depends: ${shlibs:Depends}, ${misc:Depends}
Description: correctly rounded vector math (a drop-in libmvec)
 A replacement for glibc's libmvec.so.1 whose results are the correctly
 rounded ones, bit for bit CORE-MATH's, in every rounding mode: the vector
 functions gcc and clang call when they vectorize a loop over sin, exp,
 pow and 23 more (x86-64: SSE2, AVX, AVX2, AVX-512 entry points; arm64:
 AdvSIMD and SVE). On arm64 it is also a replacement for SLEEF's
 libsleefgnuabi.so.3. Half-precision and bfloat16 arrays through
 CORE-MATH's correctly rounded functions (crmvec.h).
 .
 The libraries are installed in their own directory and replace the
 system's only for programs run with crmvec-run or linked with pkg-config
 crmvec.
