summaryrefslogtreecommitdiff
path: root/numpy
Commit message (Collapse)AuthorAgeFilesLines
* Merge pull request #18836 from seberg/refactor-ufunc-fastpathMatti Picus2021-05-063-239/+157
|\ | | | | MAINT: Generalize and shorten the ufunc "trivially iterable" path
| * BUG: Add self-overlap check and testsSebastian Berg2021-05-042-0/+11
| | | | | | | | | | | | | | | | | | Previsouly we unnecessarily required contiguous inputs for 1-D "trivially iterable". That requirement was unnecessary, but also guarded against certain self-overlaps. This was not visible because a self-overlapping array is rarely aligned (but for complex this happens typically).
| * Update numpy/core/src/umath/ufunc_object.cSebastian Berg2021-04-231-1/+1
| | | | | | Co-authored-by: Ross Barnowski <rossbar@berkeley.edu>
| * MAINT: Generalize and shorten the ufunc "trivially iterable" pathSebastian Berg2021-04-223-239/+146
| | | | | | | | | | | | | | | | | | | | | | | | | | | | This avoids the large macros designed for specific number of arguments and generalizes it to all ufuncs with a single output (Instead of only 1 and 2 operand input ufuncs.) That is not a big generalization, but the code actually shortens and while it is slightly more complex in umath itself, it avoids the fairly complex macros as well. The `==` fixes a case where the fast-path was previously not used even though it would be valid to use it. These are still used outside the ufunc machinery.
* | Merge pull request #18864 from seberg/delete-nonsense-code-from-gufuncsMatti Picus2021-05-062-29/+11
|\ \ | | | | | | MAINT: Remove dead codepath in generalized ufuncs
| * | TST: Add test for non-broadcastibility of the gufunc outputSebastian Berg2021-04-271-0/+10
| | | | | | | | | | | | | | | | | | | | | This ensures that it is correct to not initize the output array if the outer iteration has size 0. The broadcast result would always have size 0 in that case. If the iterator rejects such a broadcast it cannot happen.
| * | MAINT: Remove dead codepath in generalized ufuncsSebastian Berg2021-04-271-29/+1
| | | | | | | | | | | | | | | | | | | | | If the iterator has size zero, then that means that the outer iteration has size. Since the outer iteration is unaffected by core dimensions, in that case the output always has size zero and the code does nothing but unnecessary checking the array sizes.
* | | Merge pull request #18909 from BvB93/placeholder_main3Charles Harris2021-05-0513-232/+1237
|\ \ \ | | | | | | | | ENH: Improve the placeholder annotations within sub-modules (part 3)
| * | | ENH: Add improved placeholder annotations for `np.polynomial`Bas van Beek2021-05-059-8/+370
| | | |
| * | | ENH: Add improved placeholder annotations for `np.ma`Bas van Beek2021-05-054-224/+867
| | | |
* | | | Merge pull request #18910 from dongkeun-oh/fix_threadCharles Harris2021-05-051-1/+3
|\ \ \ \ | | | | | | | | | | BUG : for MINGW, threads.h existence test requires GLIBC > 2.12
| * | | | BUG FIX for MINGW : threads.h existence test requires GLIBC > 2.12Dong Keun Oh2021-05-051-1/+3
| | | | |
* | | | | Merge pull request #18920 from r-devulap/exp-underflowCharles Harris2021-05-052-0/+18
|\ \ \ \ \ | | | | | | | | | | | | BUG: Report underflow condition in AVX implementation of np.exp
| * | | | | BUG: Detect and report underflow condition in AVX implementation of np.expRaghuveer Devulapalli2021-05-051-0/+12
| | | | | |
| * | | | | TST: Add test to verify underflow flag is raised for np.expRaghuveer Devulapalli2021-05-051-0/+6
| | | | | |
* | | | | | MAINT: Add placeholder annotations for two missing `np.testing` objectsBas van Beek2021-05-051-0/+6
|/ / / / / | | | | | | | | | | | | | | | Adds placeholders for two `np.testing` objects previously missed in https://github.com/numpy/numpy/pull/18842.
* | | | | Merge pull request #18911 from seiko2plus/issue_18892_p1Charles Harris2021-05-053-26/+54
|\ \ \ \ \ | | | | | | | | | | | | BLD, BUG: Fix bdist_wheel duplicate building
| * | | | | BLD, BUG: Fix bdist_wheel duplicate buildingSayed Adel2021-05-053-26/+54
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The bug can occur only if the build option `build` was passed before the option `bdist_wheel`. You may still realize a duplicate printing for the compiler optimization report in the build log, which is normal due to multiple calling of command `build` by setuptools.
* | | | | | Merge pull request #18906 from rkern/enh/pcg64dxsmCharles Harris2021-05-0511-7/+2448
|\ \ \ \ \ \ | |_|/ / / / |/| | | | | ENH: Add PCG64DXSM BitGenerator
| * | | | | DOC: more completely describe the implemented variant.Robert Kern2021-05-051-1/+4
| | | | | |
| * | | | | BUG: Wrong logic for WindowsRobert Kern2021-05-041-1/+0
| | | | | |
| * | | | | ENH: add the emulated 128-bit math for PCG64DXSMRobert Kern2021-05-041-1/+59
| | | | | |
| * | | | | ENH: Add PCG64DXSM implementation.Robert Kern2021-05-0411-7/+2388
| | | | | |
* | | | | | MAINT: Adjust NumPy float hashing to Python's slightly changed hashSebastian Berg2021-05-043-6/+57
| |_|/ / / |/| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This is necessary, since we use the Python double hash and the semi-private function to calculate it in Python has a new signature to return the identity-hash when the value is NaN. closes gh-18833, gh-18907
* | | | | Merge pull request #18874 from Matthew-Badin/enable-accelerateCharles Harris2021-05-044-31/+72
|\ \ \ \ \ | |_|/ / / |/| | | | BLD, ENH: Enable Accelerate Framework
| * | | | BLD: Minimize diffMatthew Badin2021-05-011-14/+14
| | | | |
| * | | | BLD: Address lint issues and reviewer comments.Matthew Badin2021-04-306-37/+5
| | | | |
| * | | | BLD: Enable Accelerate FrameworkMatthew Badin2021-04-287-45/+118
| | | | |
* | | | | Merge pull request #18885 from BvB93/unsafeCharles Harris2021-05-0422-1012/+980
|\ \ \ \ \ | | | | | | | | | | | | MAINT: Remove unsafe unions and ABCs from return-annotations
| * | | | | DOC: Add a segment to the `numpy.typing` docs about 0D arraysBas van Beek2021-04-301-0/+14
| | | | | |
| * | | | | MAINT: Remove the `np.typing._ArrayOrScalar` type-aliasBas van Beek2021-04-302-3/+0
| | | | | |
| * | | | | MAINT: Remove unsafe unions from `np.core.einsumfunc`Bas van Beek2021-04-302-15/+19
| | | | | |
| * | | | | MAINT: Remove unsafe unions from `np`Bas van Beek2021-04-305-518/+519
| | | | | |
| * | | | | MAINT: Remove unsafe unions from `np.typing._callable`Bas van Beek2021-04-301-5/+5
| | | | | |
| * | | | | MAINT: Remove unsafe unions from `np.lib.ufunclike`Bas van Beek2021-04-302-18/+34
| | | | | |
| * | | | | MAINT: Remove unsafe unions from `np.lib.index_tricks`Bas van Beek2021-04-303-14/+30
| | | | | |
| * | | | | MAINT: Remove unsafe unions from `np.core.numeric`Bas van Beek2021-04-302-14/+68
| | | | | |
| * | | | | MAINT: Remove unsafe unions from `np.core.function_base`Bas van Beek2021-04-302-3/+5
| | | | | |
| * | | | | MAINT: Remove unsafe unions from `np.core.fromnumeric`Bas van Beek2021-04-303-422/+286
| | | | | |
* | | | | | MAINT: Avoid treating _SIMD module build warnings as errorsSayed Adel2021-05-041-1/+1
| | | | | |
* | | | | | Merge pull request #18900 from seiko2plus/issue_18160Matti Picus2021-05-041-4/+0
|\ \ \ \ \ \ | | | | | | | | | | | | | | BLD: remove unnecessary flag `-faltivec` on macOS
| * | | | | | BLD: remove unnecessary flag `-faltivec` on macOSSayed Adel2021-05-041-4/+0
| | |_|/ / / | |/| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Supporting PowerPC/AltiVec on macOS is no longer necessary, even if the Mac/G5 is still running having `-faltivec` or replacing it with the new AltiVec flag `-maltivec` wouldn't increase that much performance without raw SIMD. note: the flag was enabled on non-intel platforms which causes fatal build errors on macOS/arm64.
* | | | | | BLD, BUG: Fix compiler optimization log AttributeErrorSayed Adel2021-05-042-8/+9
|/ / / / / | | | | | | | | | | | | | | | | | | | | The error appears when option `build` is represented before `bdist_wheel`.
* | | | | MAINT: ssize_t -> Py_ssize_t and other fixes for Python v3.10.0 (gh-18890)Matti Picus2021-05-037-13/+44
| | | | | | | | | | | | | | | | | | | | | | | | | The 3.10 alpha 7 fails to build on windows, see below for the warnings and errors. It seems we should not be using ssize_t rather Py_ssize_t. On windows, ssize_t is not defined anymore in CPython3.10 alpha7 after python/cpython#24479 in response to issue 17717 from 2011, which was merged March 1. Fixes #18888
* | | | | MAINT: Add ".csv" to some data file names.Charles Harris2021-05-036-7/+8
| | | | | | | | | | | | | | | | | | | | It is useful to encode the data file type in the name.
* | | | | ENH: Add min values comparison for floating pointHowJmay2021-05-027-9/+119
| | | | | | | | | | | | | | | | | | | | Add the similar implementation for minimum value comparison.
* | | | | Merge pull request #18863 from HowJMay/simd-maxSayed Adel2021-04-307-0/+124
|\ \ \ \ \ | |/ / / / |/| | | | ENH: Add max values comparison for floating point
| * | | | ENH: Add max values comparison for floating pointHowJmay2021-05-017-0/+124
| | | | | | | | | | | | | | | | | | | | | | | | | SIMD intrinsics for comparing double/single precision has been added. Now only NEON and SSE ones are added.
* | | | | Merge pull request #18853 from BvB93/placeholder_main2Charles Harris2021-04-2921-335/+938
|\ \ \ \ \ | | | | | | | | | | | | ENH: Improve the placeholder annotations within sub-modules (part 2)
| * | | | | DOC: Typo fixBas van Beek2021-04-271-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | Co-Authored-By: h-vetinari <h.vetinari@gmx.com>
View Lorry Controller Status