This is the release note of v14.0.0a1. See here for the complete list of solved issues and merged PRs.

💬 Join the Matrix chat to talk with developers and users and ask quick questions!

🙌 Help us sustain the project by sponsoring CuPy!

✨ Highlights

This is the first alpha release of the CuPy v14 series, containing:

New type promotion rules and behaviors aligned with the NumPy 2 specification.
42 new NumPy/SciPy-compatible APIs, including cupy.concat, cupyx.scipy.interpolate.CubicSpline, cupyx.scipy.spatial.Delaunay, cupyx.scipy.ndimage.find_objects, and cupyx.scipy.special.lambertw. See the Comparison Table for the detailed coverage.

Binary packages are available for testing. Try installing now by:

$ pip install cupy-cuda12x --pre -U -f https://pip.cupy.dev/pre

🛠️ Changes without compatibility

CuPy v14’s behavior will be aligned with NumPy v2.
- Type promotion rules are now NEP50 compatible. See Changes to NumPy data type promotion.
- int is now 64-bit (int64) on Windows. See Windows default integer.
- APIs removed in NumPy v2 (see Changes to namespaces) were marked deprecated in CuPy v14. Although they are kept available in v14 for smooth migration, they are planned to be removed in the next major release (CuPy v15).
- The behavior of copy argument has been changed (#8545). See Adapting to changes in the copy keyword.
Support for Python 3.9, NumPy 1.22 and 1.23, SciPy 1.7, 1.8, and 1.9 has been dropped. (#8491)
cupy.random.choice may return different results from CuPy v13. (#8483)
Building CuPy from source code now requires Cython 3.0. (#8457)
cupyx.scipy.linalg.{tri,tril,triu} APIs were removed from CuPy to follow the latest SciPy’s specification. Use cupy.{tri,tril.triu} instead. (#8499)
NumPy fallback mode (cupyx.fallback_mode) has been removed as discussed in #8497. (#8816)
Legacy DLPack APIs (cupy.toDlpack and cupy.fromDlpack) are now marked deprecated. Use cupy.from_dlpack instead. See the documentation for the usage. (#8831)

📝 Changes

New Features

Add KDTree to cupyx.scipy.spatial (#7671)
Add neighbors option to RbfInterpolator (#7864)
ENH: cupyx/signal: add sweep_poly (#7873)
Add 2D Delaunay triangulation (#7985)
Add cupyx.signal.pulse_compression from cuSignal's non SciPy-compat API (#8022)
Add LinearNDInterpolator to cupyx.scipy.interpolate (#8035)
Add cupyx.signal.convolve1d3o from cuSignal's non SciPy-compat API (#8037)
Add cupyx.signal.{firfilter,firfilter_zi,firfilter2} (#8052)
Add cupyx.signal.{pulse_doppler, cfar_alpha} (#8057)
Add cupyx.signal.{complex_cepstrum,real_cepstrum,inverse_complex_cepstrum,minimum_phase} (#8062)
Add cupyx.signal.mvdr (#8077)
ENH: signal: add lanczos and kaiser_bessel_derived windows (#8081)
Add cupyx.signal.ca_cfar (#8087)
Add cupyx.signal.convolve1d2o (#8101)
Add cupyx.signal.freq_shift (#8128)
Add lambertw function (#8140)
Add cupyx.signal.channelize_poly (#8141)
Add cupyx.scipy.interpolate.CubicSpline (#8175)
Add apply_over_axes API (#8177)
Add cupy.put_along_axis API (#8199)
Add CloughTocher2DInterpolator to cupyx.scipy.interpolate (#8208)
Add NearestNDInterpolator to cupyx.scipy.interpolate (#8220)
Add NdBSpline to cupyx.scipy.interpolate (#8223)
ENH: cupyx/scipy/interpolate: add *UnivariateSpline for 1D smoothing splines (#8267)
Add NdBSpline based interpolation methods to RGI (#8276)
ENH: cupyx/interpolate: port interp1d from scipy (#8289)
Add batched solve_triangular (#8329)
Add Incomplete Elliptic Integrals to special (#8425)
Support system allocated memory (#8442)
Add CUDA graph debug function (#8502)
Add sici and shichi to special for sine and cosine integrals (#8620)
Update unique_xxx (nep52) (#8665)
Add cupyx.scipy.ndimage.find_objects (#8916)

Enhancements

Support for break and continue keywords in CuPy JIT (#8010)
Make cupyx.signal.radartools private (#8047)
Remove usages of numpy.float_ and numpy.complex_ (#8050)
Support cusparseLt 0.6.1 (#8074)
Add incontiguous support for cutensor functions (#8149)
Add complex support for the digamma function (#8163)
Fix expm(complex matrix) (#8206)
Add CutensorMg support (#8212)
Add cudaStreamCreateWithPriority (#8219)
Add the nearest method for percentile/quantile estimation (#8224)
Various Jitify improvements (#8235)
Support fallback algorithm for spgemm (#8252)
Bump to cuTENSOR 2.0.1 (#8282)
Preload cuTENSORMg (#8283)
Use weakref.finalize instead of __del__ for RandomState._generator destruction (#8315)
Support ROCm 6 (#8319)
cupyx: cleanup use of deprecated NumPy functionality (NumPy 2.0 compatibility) (#8320)
Add wright_bessel function to special (#8324)
MAINT: fft, linalg: add __all__ lists (#8333)
Cuda 12.5 Tests (#8337)
Add axes support in ndimage filters module (#8339)
MAINT: interpolate: update RBF to scipy 1.13 (#8343)
Make CuPy import under NumPy 2.0 (#8346)
Lazy-preload NCCL (#8360)
Fix map_coordinates recompilation condition (#8378)
Disable jitify for cub & Bump CCCL (#8412)
Use custom less instead of specializing thrust (#8446)
Port to Cython 3.0 (#8457)
Avoid using Jitify everywhere inside CuPy (#8467)
Get rid of pkg_resources (#8480)
Drop support for Python 3.9, NumPy 1.22 and 1.23, SciPy 1.7, 1.8 and 1.9 (#8491)
Remove deprecated cupyx.scipy.linalg.{tri,tril,triu} (#8499)
Use .toarray() instead of .A attribute (#8508)
Support half option in scipy.signal.minimum_phase (#8510)
Increase MAX_NDIM to 64 (#8511)
Support CUDA 12.6 (#8513)
Fallback to system headers for future CUDA 12.x versions (#8518)
Extend runtime header search logic to conda (#8519)
Support copy=None in cp.array / cp.asarray / cp.asanyarray (#8545)
Fix dtype rule of cupy.scipy.stats.entropy for SciPy 1.14 (#8547)
Support setuptools 74.0.0 or later (#8583)
Add NCCL_ERROR_REMOTE_ERROR to the set of errors from NCCL (#8662)
Replace numpy.ComplexWarning with cupy.exceptions.ComplexWarning (#8676)
ENH: Implement dlpack v1 (#8683)
Fix some NumPy 2.x CI failures (cont.) (#8695)
Bump CUDA version in cuda11x-cuda-python CI (#8737)
[ROCm 6.2.2] Conditionally define CUDA_SUCCESS only if it's not (#8793)
Remove fallback mode (#8816)
Raise user warning in both {to,from}Dlpack & Update the Interoperability page (#8831)
Use a custom Min/Max instead of specializing CUB (#8846)
Updating pylibraft pairwise_distance to cuvs (#8847)
add axes support for additional functions in cupyx.scipy.ndimage (from SciPy 1.15.0) (#8858)
Raise VisibleDeprecationWarning for wavelet functions (#8865)
Support CUDA 12.8 + Blackwell GPUs (sm_100, sm_120) (#8899)
Bump library installers for CUDA 12.8 (#8914)
Use CCCL 2.8.x branch + Use CUPY_CACHE_KEY in hash keys (#8919)
Use NVIDIA CCCL 2.8 latest w/CUDA 12.3 fix (#8924)
Use C++17 in JIT compile (#8940)
Restore CUB histogram and bincount (#8950)
Broaden usage of C++17 (#8952)
cupyx.scipy.distance: initialize output array with empty instead of zeros (#8971)
cupyx.scipy.spatial.distance.cdist remove explicit zeroing of user-provided output array (#8988)
Fix rocThrust build for ROCm 6.3 (#9022)
Allow discovering cuTENSOR using major version (#9030)
Update cutensornet accelerator based on cuquantum-python 25.03 deprecation (#9045)
Support FIPS enabled machines with MD5 hashing (#9053)
Refactor hashing (#9057)

Enhancements for NumPy & SciPy compatibility:

Fix scp.signal.{medfilt,medfilt2d} to raise ValueError for complex64 inputs (#8059)
Deprecate cupyx.scipy wavelet functions (#8061)
Fix csrmatrix.__pow__ to raise ValueError for non-int other (#8063)
Fix cupyx.scipy.special.betainc for invalid inputs (#8065)
scipy.special.{btdtr,btdtri} are deprecated since SciPy 1.12 (#8066)
Fix boxcox_llf for SciPy 1.12 changes (#8095)
NEP50 (#8323)
Resolve Ruff NPY errors - fix exception imports and asfarray usage in test code (#8455)
Fix sparse.linalg function signatures following SciPy 1.14 (#8526)
NumPy 2.0 compatibility: (partially) sync with NEP52 (#8531)
Fix dtype rule of special functions for SciPy 1.14 (#8532)
Fix cupy.histogram arg order to match NumPy (v1.24+) (#8559)
Make cupy.linalg.solve compatible with numpy v2 (#8629)
Silence FutureWarning emitted when rcond is missing (#8638)
Fix some NumPy 2.x CI failures (#8690)
Support kind arg. in sorting methods (#8708)
Fix cupy.percentile for NumPy 2.x (#8726)
Fix some NumPy 2.x CI failures (cupyx) (#8727)
Skip some tests incompatible with NumPy 2.2 (#8817)
Fix scipy.spmatrix.sign for complex dtype inputs (#8822)
Fix return type of cupy.where for scalar arguments for NumPy 2.0 (#8835)
Fix cupyx.scipy.special.logsumexp for NumPy 2.0 (#8836)
Fix cupy.cov (#8839)
Fix cupy.histogramdd for NumPy 2.x (#8873)
Raise ValueError upon attempts to create 3-dim sparse array (#8877)
Disable contiguous_check for COO/dense matmul test (#8878)
Skip a test for invalid scipy return value of invalid COO matmul (#8879)
Support empty tuple indexing for sparse matrix (#8882)
Fix fft.fht following bug fix in SciPy 1.15 (#8883)
Deprecate cupyx.scipy.linalg.kron (#8885)
Add special.sph_harm_y and deprecate special.sph_harm (#8898)
Fix boxcox_llf for SciPy 1.13 (#8907)
Update cupyx.scipy.special functions for SciPy 1.15 (#8908)
Fix dtype rule of stats.entropy for SciPy 1.15 (#8911)
Fix dtype rule of boxcox_llf for SciPy 1.15 (#8913)
Fix cupyx.scipy.stats.zscore for SciPy 1.15 (#8977)
Update cupy.testing following numpy (#9076)

Performance Improvements

Speed up cupy environment duplicate detection (#8017)
Improve efficiency of binary morphology functions by accepting a tuple for footprint (#8312)
Improving performance of cupy.random.choice (#8483)
Improve performance of cupyx.scipy.ndimage.binary_fill_holes (#8956)

Bug Fixes

Fix argmax/argmin for large reduction axis (#8031)
Fix lfilter_zi and sosfilt_zi when any IIR coefficient is zero (#8034)
Fix cupyx.scipy.fft.{dst,dstn} in type 2/3 (#8058)
Update _nccl_comm.py (#8089)
Fix Flags not to allow setters (#8104)
Do not use from-import (#8109)
Fix overflow indexing ndarray generated with as_strided (#8150)
Prevent angular brackets from appearing in Jitify's cache filename (#8154)
Set -arch in the compiler options unconditionally (#8157)
Allow cupy.show_config() without CUDA (#8183)
Fix jitify warmup kernel (#8200)
Fix: always switch to the submodule dir before checking git tag/commit (#8201)
Fix: remove unnecessary include that causes deployment issue (#8202)
Fix CUB min/max initial values (#8203)
Fix build system for Thrust detection (#8229)
Fix overflow of index calculation in random generator API (#8243)
Fix Generator API parallelism (#8244)
Fix jitify warmup kernel - Cont'd (#8250)
Fix bounds issue in Clough-Tocher computation (#8277)
hipPointerGetAttributes returns error when pointer is unregistered in ROCm 5.7 (#8335)
Fix CUB build error on win-64 (#8345)
Re-enable NVTX range coloring for NVTX3. (#8353)
Delaunay: Prevent checking neighboring triangles if nTri <= 1 (#8363)
Fix ndarray.get() not honoring current stream when layout is not contiguous (#8369)
Fix spline temp container size in make_interp_spline (#8388)
MAINT: Avoid using np.compat.integer_types (#8409)
Fix type dispatcher for arm64 (#8410)
Fix RandomState.seed() for NumPy 2 compatibility (#8429)
Fix copyto for NumPy 2 compatibility (#8430)
Update compiler.py to avoid the popup of the nvcc.exe console (#8432)
Address KeyErrors from importlib_metadata (#8441)
Fix the size of temporary CUB output space to consider its alignment (#8444)
upfirdn: mode=None -> mode="constant" (#8476)
Search header files from CTK wheel (#8489)
Fix CUDA version condition to use headers from wheel (#8505)
Fix order 'K' with shape given for *_like array creation (#8525)
Fix ROCm 4.3 binary package build broken (#8528)
Fix cudart header detection for conda (#8530)
Properly allocate in RNG when specified dtype is neither float32/float64 (#8538)
Add nccl.broadcast 64-bit support (#8562)
Support building CuPy with setuptools 74 (#8574)
Guard for ROCm 6.x (#8610)
Fix HIP_VERSION unit (#8613)
Switch to using platform.machine() instead of platform.processor() (#8655)
Use platform.machine() instead of platform.processor() (#8657)
Fix sosfilt state output shape when ndim < 2 (#8677)
Fix undefined inf/nan constant in CuPy JIT (#8707)
Fix import order (#8723)
Fix bspline kernel to avoid out of bounds error (#8760)
Fix race during SoftLink initialization (#8774)
Fix nanargmin and nanargmax's parameter order and pass optional parameters (#8783)
Fix crashes of quantile and percentile (#8808)
Fix handling of pinned memory (#8845)
ndimage: fix bug in morphology axes support when structure=None (#8954)
Fix buffer protocol to raise TypeError when it is not meant to be supported (#8957)
Use /bigobj on Windows build (#8962)
Fix cupyx.scipy.spatial.distance's cdist for RAPIDS 24.12 compatibility (#8970)
JIT: Support empty return (#8989)
Cast points when computing Delaunay triangulation on dtype != float64 (#9021)
Fix compilation error of cupy.inf in fusion2 (#9034)
Include pxd files in cupyx (#9081)

Code Fixes

Refactor radartools (#8072)
Refactor convolve1d3o (#8078)
MAINT: cupyx/interpolate: remove a duplicate function (#8170)
Do not use plain thrust namespace in complex clone (#8221)
Use f-strings in ndimage kernel generation code (#8311)
Upgrade pre-commit hooks to silence warnings (#8664)
Resolve import loop (#8705)
Enhance comments in percentile implementation (#8753)
Resolve uncaught type warning (#8797)
Switch from .A attribute to .toarray() method (#8810)
Add cupy/_core/numpy_allocator.h to .gitignore (#8813)
Drop unneeded bytes copy of CUPY_CACHE_KEY (#8923)
Fix typo in _cretate_frame_tree (#8942)
Fix get_typename to emit thrust::complex (#9040)

Documentation

Generate signature for ufunc documentation (#6137)
Use modern DLPack interface in torch interoperability document (#7988)
Add a note on accelerating CI/CD (#8004)
Update conda installation guide (#8129)
Add CuPy logo in SVG (#8182)
Fix pdist docstring in order to specify that the returned matrix is condensed (#8186)
Add CuPy Training Materials information to README.md (#8225)
Fixed typo in cupy.linalg.norm docs (#8264)
Replace license notice in cupyx.scipy.signal._spectral (#8268)
Update document for CUDA 12.3 and 12.4 (#8280)
Expose Delaunay in documentation (#8294)
Add comparison table for (cupyx.)scipy.sparse.*_matrix classes class methods (#8295)
Find and fix typos with codespell (#8304)
Update the version of cusparselt in the installation guide (#8362)
eigsh doc correction in _eigen.py (#8365)
Add NumPy 2.0 on document (#8368)
Update API baseline on document (#8370)
Add CUDA 12.5 to list of supported platform (#8424)
typo: coping -> copying (#8426)
Add docs about CUDA headers (#8593)
Update badges in README (#8603)
docs: update fft.rst (#8616)
Update documentation to use pre-commit (#8640)
Add tips on Windows development in Contribution Guide (#8702)
Add notice about cupy.array_api removal (#8750)
Remove "Type promotion" section from user guide (#8754)
Add CUDA 12.8 to docs (#8961)
Update list of supported versions (#8986)

Installation

Patch the build system to better support conda-build (#7603)
Skip CUDA_PATH warning in Conda installation (#8051)
Do not search for static libs (#8134)
Bump version in main branch for v14 development (#8341)
Update conda-build CUDA detection logic for Setuptools 72.2.0 (#8544)
Remove unnecessary "install" package in Docker image (#8549)
Use relative path of header files to generate cache key (#8927)
Fix minimum CUDA version check and update comments (#8928)

Tests

Revert CI timeout changes (#7903)
Add import test without CUDA Toolkit (#7996)
Bump stable branch to v13 (#8024)
Update Compatibility Matrix (#8025)
Remove some signal.vectorstrength xfail tests (#8060)
Fix scipy.linalg not to raise DeprecationWarning for zero-size inputs (#8064)
Refactor radartools tests (#8071)
Fix scipy>=1.12 filter condition (#8088)
Fix slow test (#8115)
Support SciPy 1.12 (#8127)
Fix invalid vectorstength tests (#8142)
Fix actions versions used in workflows to avoid node 16 deprecation warning (#8189)
Add CI to test cupy.show_config() pass without CUDA installed (#8190)
BUG: cupyx/scipy/signal: fix mpmath test (#8259)
Add support for CUDA 12.3 & 12.4 (#8265)
Tentatively pin SciPy to v1.12 in CI (#8273)
ndimage: more efficient testing of binary morphology functions (#8313)
Skip some tests in aarch64 CI (#8411)
Bump NumPy/SciPy versions in cuda-example CI (#8415)
Fix CUDA 11.2 CI failure on Linux (#8436)
Decrease number of threads to avoid "system error: excessive memory usage is detected" (#8461)
CI: skip CUDA 12.1/12.2/12.3/12.4 CI on "mini" trigger (#8463)
CI: Update micro versions of Python (#8493)
Support SciPy 1.13 and 1.14 (#8500)
Relax test_firls atol (#8509)
Fix cupyx.scipy.sparse tests for SciPy 1.14 (#8527)
Skip special.logsumexp test for empty input (#8533)
Skip betaincinv test with SciPy 1.14.1 (#8546)
Revert CI timeout bump (#8564)
Add NumPy 2.x CI for Linux (#8565)
Use setuptools==73.0.1 (#8567)
Fix FFT tests for NumPy 2.0 (#8584)
Add self-hosted CI (#8612)
Add CI for ROCm 6.2 (#8624)
Update precommit (#8628)
Renegerate rocm-6-2.Dockerfile (#8632)
Skip tests if scipy is not installed (#8634)
Use with_requires rather than xfail (#8639)
Accept OverflowError in TestCopytoFromScalar for NumPy v2 (#8641)
Skip more tests if scipy is not installed (#8642)
Replace flake8 with ruff (#8671)
CI: Fix apt repository URL for Ubuntu 22.04 (#8713)
Relax tolerance of test_hilbert for NumPy 2.0 (#8740)
Temporary skip for NumPy 2.0 tests (#8741)
Remove ndarray.ptp from fallback tests (#8742)
Bump SciPy version to 1.14 in Windows CI (#8759)
Add NumPy 2.x CI for Windows (#8766)
Fix some NumPy 2.x CI failures (cont.) (#8790)
Set numpy._set_promotion_state("weak") only for test with NumPy 1.x (#8809)
Add NumPy 2.2 to CI (#8815)
CI: support "skip-ci" label (#8832)
CI: Fix FlexCI compatibility (#8840)
Support SciPy 1.15 (#8861)
Support Optuna 4 (#8862)
Fix test for csr_matrix.setdiag (#8874)
Skip some signal tests for TypeError for inputs of np.longlong dtype (#8884)
Disable contiguous_check for some signal.cont2discrete tests (#8886)
Fix test_interpnd for removed interpolate.interpnd namespace (#8887)
Add testing.shaped_linspace (#8896)
Add CI for CUDA 12.8 (#8905)
Fix splines tests to remove unexpected skips (#8918)
Minor updates for sm120 (#8920)
Add CI for Python 3.13 and mpi4py v4 (#8953)
Increase host memory in Windows CI, free GPU memory in example code (#8966)
Pass locals dict to exec (#8982)
Mark xfails in some spline tests for SciPy 1.15 (#8984)
CI: Do not run full CI on CUDA 12.0/12.1/12.2 + Windows (#8992)
TST: Add test for fusion path with constant and scalar (#9013)
CI: Pin setuptools version on Windows (#9038)
Revert "CI: Pin setuptools version on Windows" (#9050)
Add SciPy 1.15 CI for Windows (#9066)

Others

Wait before published items shown on PyPI (#8302)
Fix pull request project board workflows (#8366)
Bump version to v14.0.0a1 (#8372)
Add backport reminder (#8675)
Fix script name of backport reminder (#8685)
Add workflow_dispatch event to allow debugging workflow (#8687)
Update bug issue template (#8749)
Update pre-commit hooks (#8838)
Allow specifying no libraries when generating wheel metadata (#9078)

👥 Contributors

The CuPy Team would like to thank all those who contributed to this release!

@99991 @acc-mu3n @andfoy @arkdong @asi1024 @Azusachan @bernhardmgruber @Berrysoft @bmerry @boku13 @cclauss @cjnolet @dagardner-nv @dakofler @EarlMilktea @eltociear @emcastillo @ev-br @gdaisukesuzuki @grlee77 @hauntsaninja @hmaarrfk @HollowMan6 @jakirkham @jemiryguo @johnnynunez @kmaehashi @leofang @littlewu2508 @lowener @martinResearch @MattTheCuber @miscco @mohitreddy1996 @monzelr @mroeschke @romerojosh @rongou @seberg @so298 @steppi @swelborn @syheliel @takagi @take-cheeze @tornikeo @xuefeng-xu @yangcal @YanivDorGalron @Zhyrek

cupy 14.0.0a1 v14.0.0a1 on Python PyPI