Releases · numpy/numpy-quaddtype

Release Note Summary

This release marks the first stable version of NumPy QuadDType, providing IEEE 754 quadruple-precision (128-bit) floating-point arithmetic as a first-class NumPy dtype.

Highlights

Documentation: New documentation site at numpy.org/numpy-quaddtype with installation guide, usage examples, and comprehensive API reference
Platform support: Linux, macOS (Intel & Apple Silicon), and Windows; Python 3.11–3.14 including free-threading (GIL-free) builds; Big-endian architecture support; FMA-disable option for older x86-64 CPUs (Sandy Bridge era)
New ufuncs: logaddexp2, floor_divide, float_power, fmod, divmod, fabs, heaviside, conj/conjugate, expm1, cbrt, hypot, degrees/rad2deg, radians/deg2rad, nextafter, spacing, modf, ldexp, frexp, and all logical ufuncs
Enhanced scalar support: is_integer, as_integer_ratio, imag/real getters, buffer protocol, __version__ attribute, and support for np.bool, arbitrary-length Python ints, bytes, and sequences
Casting improvements: same_value casting rule, StringDType array casting, fixed-length string casting, bytes array casting, and improved inter-backend quad2quad fixes
Type stubs: Full static typing support
Serialization: Pickle and hash support
Array functions: argmax/argmin slots, nonzero, scanfunc/fromstr, and comparison promoter registration
Build & CI: Migrated to standalone repository, src layout restructuring, SLEEF 3.9 integration, FMA-disable option for older CPUs, and documentation site with API docs

PRs from numpy-quaddtype (after migration)

#64 Last stage fixes for v1.0.0 (@SwayamInSync)
#63 Fix elementwise min/max doc header level (@juntyr)
#62 SIMD/FMA compatibility for older CPUs (@SwayamInSync)
#61 README and documentation updates (@SwayamInSync)
#60 Register comparison promoter for QuadDType only (@SwayamInSync)
#59 CI for older CPUs without FMA support (@SwayamInSync)
#58 Add instructions for installing development version (@mhvk)
#53 Enable documentation preview on PRs (@SwayamInSync)
#51 Switch to released SLEEF 3.9 (@SwayamInSync)
#50 Adding argmax and argmin slots (@SwayamInSync)
#49 Add API docs (@juntyr)
#48 Fixing compare arrfunc (@SwayamInSync)
#45 Implement quadprec_nonzero (@SwayamInSync)
#44 Bool cast support (@SwayamInSync)

PRs from numpy-user-dtypes (before migration)

#255 Add build dependency on cmake and numpy 2.4 (@ngoldbaum)
#253 Restructuring quaddtype to use src layout (@SwayamInSync)
#251 Fix: Making pow compatible with cpython pow signature (@SwayamInSync)
#247 Refactor: General refactors + quad2quad inter-backend fixes (@SwayamInSync)
#246 Implementing same_value casting rule in quaddtype (@SwayamInSync)
#245 Implementing hash support in QuadDtype (@SwayamInSync)
#244 Adding variable length StringDType to/from QuadDType array casting (@SwayamInSync)
#240 SLEEF migration to 4.0 (@SwayamInSync)
#238 Enable GIL for bytes/unicode array casting to/from Quaddtype (@SwayamInSync)
#234 Fixing QuadPrecision scalar dtype exposure (@SwayamInSync)
#232 Adding pickle support to quaddtype (@SwayamInSync)
#228 Adding Quad to and from bytes array casting support (@SwayamInSync)
#227 Add basic documentation site (@melissawm)
#225 Adding array casting support to fixed length strings (@SwayamInSync)
#223 Adding Byte support to QuadPrecision scalar constructor (@SwayamInSync)
#221 Implementing is_integer and as_integer_ratio for QuadPrecision (@SwayamInSync)
#220 Fixing reduction logical ufuncs (@SwayamInSync)
#218 QuadPrecision <: numpy.floating (@jorenham)
#215 stub numpy_quaddtype.__version__ (@jorenham)
#214 Adding Support for scanfunc and fromstr ArrayFunc (@SwayamInSync)
#213 Support arbitrary-length Python ints in QuadPrecision constructor (@SwayamInSync)
#211 Trignometric ufunc tests (@SwayamInSync)
#210 Tests for imag and real (@SwayamInSync)
#208 imag and real getters for QuadPrecision (@SwayamInSync)
#207 Add buffer protocol support to QuadPrecision scalar (@SwayamInSync)
#203 Adding __version__ attribute to numpy_quaddtype (@SwayamInSync)
#202 Add np.bool support in QuadPrecision (@SwayamInSync)
#201 Allow passing floating, integer, and bool_ scalars to QuadPrecision.__new__ (@jorenham)
#200 frexp ufunc implementation (@SwayamInSync)
#199 Extending QuadPrecision input to support arrays, sequences and np.numbers (@SwayamInSync)
#197 Fix Round-tripping QuadPrecision through string (@SwayamInSync)
#195 ldexp ufunc implementation (@SwayamInSync)
#192 modf ufunc implementation (@SwayamInSync)
#191 spacing ufunc implementation (@SwayamInSync)
#190 nextafter ufunc implementation (@SwayamInSync)
#188 All logical ufunc implementation (@SwayamInSync)
#187 radians and deg2rad ufunc implementation (@SwayamInSync)
#186 degrees and rad2deg ufuncs implementation (@SwayamInSync)
#185 hypot ufunc implementation (@SwayamInSync)
#184 cbrt ufunc implementation (@SwayamInSync)
#183 expm1 ufunc implementation (@SwayamInSync)
#182 conj and conjugate ufunc implementation (@SwayamInSync)
#181 heaviside ufunc implementation (@SwayamInSync)
#180 fabs ufunc implementation (@SwayamInSync)
#179 divmod ufunc implementation (@SwayamInSync)
#178 fmod ufunc implementation (@SwayamInSync)
#176 float_power ufunc implementation (@SwayamInSync)
#175 floor_divide ufunc implementation (@SwayamInSync)
#174 logaddexp2 ufunc implementation (@SwayamInSync)
#172 Allow promotion of Python abstract dtype + fill implementation (@SwayamInSync)