NumPy 1.20.0 Release Notes#
This NumPy release is the largest so made to date, some 684 PRs contributed by 184 people have been merged. See the list of highlights below for more details. The Python versions supported for this release are 3.7-3.9, support for Python 3.6 has been dropped. Highlights are
Annotations for NumPy functions. This work is ongoing and improvements can be expected pending feedback from users.
Wider use of SIMD to increase execution speed of ufuncs. Much work has been done in introducing universal functions that will ease use of modern features across different hardware platforms. This work is ongoing.
Preliminary work in changing the dtype and casting implementations in order to provide an easier path to extending dtypes. This work is ongoing but enough has been done to allow experimentation and feedback.
Extensive documentation improvements comprising some 185 PR merges. This work is ongoing and part of the larger project to improve NumPy’s online presence and usefulness to new users.
Further cleanups related to removing Python 2.7. This improves code readability and removes technical debt.
Preliminary support for the upcoming Cython 3.0.
New functions#
The random.Generator class has a new permuted
function.#
The new function differs from shuffle
and permutation
in that the
subarrays indexed by an axis are permuted rather than the axis being treated as
a separate 1-D array for every combination of the other indexes. For example,
it is now possible to permute the rows or columns of a 2-D array.
(gh-15121)
sliding_window_view
provides a sliding window view for numpy arrays#
numpy.lib.stride_tricks.sliding_window_view
constructs views on numpy
arrays that offer a sliding or moving window access to the array. This allows
for the simple implementation of certain algorithms, such as running means.
(gh-17394)
numpy.broadcast_shapes
is a new user-facing function#
broadcast_shapes
gets the resulting shape from
broadcasting the given shape tuples against each other.
>>> np.broadcast_shapes((1, 2), (3, 1))
(3, 2)
>>> np.broadcast_shapes(2, (3, 1))
(3, 2)
>>> np.broadcast_shapes((6, 7), (5, 6, 1), (7,), (5, 1, 7))
(5, 6, 7)
(gh-17535)
Deprecations#
Using the aliases of builtin types like np.int
is deprecated#
For a long time, np.int
has been an alias of the builtin int
. This is
repeatedly a cause of confusion for newcomers, and existed mainly for historic
reasons.
These aliases have been deprecated. The table below shows the full list of deprecated aliases, along with their exact meaning. Replacing uses of items in the first column with the contents of the second column will work identically and silence the deprecation warning.
The third column lists alternative NumPy names which may occasionally be preferential. See also Data types for additional details.
Deprecated name |
Identical to |
NumPy scalar type names |
---|---|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
numpy.unicode_ |
To give a clear guideline for the vast majority of cases, for the types
bool
, object
, str
(and unicode
) using the plain version
is shorter and clear, and generally a good replacement.
For float
and complex
you can use float64
and complex128
if you wish to be more explicit about the precision.
For np.int
a direct replacement with np.int_
or int
is also
good and will not change behavior, but the precision will continue to depend
on the computer and operating system.
If you want to be more explicit and review the current use, you have the
following alternatives:
np.int64
ornp.int32
to specify the precision exactly. This ensures that results cannot depend on the computer or operating system.np.int_
orint
(the default), but be aware that it depends on the computer and operating system.The C types:
np.cint
(int),np.int_
(long),np.longlong
.np.intp
which is 32bit on 32bit machines 64bit on 64bit machines. This can be the best type to use for indexing.
When used with np.dtype(...)
or dtype=...
changing it to the
NumPy name as mentioned above will have no effect on the output.
If used as a scalar with:
np.float(123)
changing it can subtly change the result. In this case, the Python version
float(123)
or int(12.)
is normally preferable, although the NumPy
version may be useful for consistency with NumPy arrays (for example,
NumPy behaves differently for things like division by zero).
(gh-14882)
Passing shape=None
to functions with a non-optional shape argument is deprecated#
Previously, this was an alias for passing shape=()
.
This deprecation is emitted by PyArray_IntpConverter in the C API. If your
API is intended to support passing None
, then you should check for None
prior to invoking the converter, so as to be able to distinguish None
and
()
.
(gh-15886)
Indexing errors will be reported even when index result is empty#
In the future, NumPy will raise an IndexError when an integer array index contains out of bound values even if a non-indexed dimension is of length 0. This will now emit a DeprecationWarning. This can happen when the array is previously empty, or an empty slice is involved:
arr1 = np.zeros((5, 0))
arr1[[20]]
arr2 = np.zeros((5, 5))
arr2[[20], :0]
Previously the non-empty index [20]
was not checked for correctness.
It will now be checked causing a deprecation warning which will be turned
into an error. This also applies to assignments.
(gh-15900)
Inexact matches for mode
and searchside
are deprecated#
Inexact and case insensitive matches for mode
and searchside
were valid
inputs earlier and will give a DeprecationWarning now. For example, below are
some example usages which are now deprecated and will give a
DeprecationWarning:
import numpy as np
arr = np.array([[3, 6, 6], [4, 5, 1]])
# mode: inexact match
np.ravel_multi_index(arr, (7, 6), mode="clap") # should be "clip"
# searchside: inexact match
np.searchsorted(arr[0], 4, side='random') # should be "right"
(gh-16056)
Deprecation of numpy.dual#
The module numpy.dual is deprecated. Instead of importing functions from numpy.dual, the functions should be imported directly from NumPy or SciPy.
(gh-16156)
outer
and ufunc.outer
deprecated for matrix#
np.matrix
use with outer
or generic ufunc outer
calls such as numpy.add.outer
. Previously, matrix was
converted to an array here. This will not be done in the future
requiring a manual conversion to arrays.
(gh-16232)
Further Numeric Style types Deprecated#
The remaining numeric-style type codes Bytes0
, Str0
,
Uint32
, Uint64
, and Datetime64
have been deprecated. The lower-case variants should be used
instead. For bytes and string "S"
and "U"
are further alternatives.
(gh-16554)
The ndincr
method of ndindex
is deprecated#
The documentation has warned against using this function since NumPy 1.8.
Use next(it)
instead of it.ndincr()
.
(gh-17233)
ArrayLike objects which do not define __len__
and __getitem__
#
Objects which define one of the protocols __array__
,
__array_interface__
, or __array_struct__
but are not sequences
(usually defined by having a __len__
and __getitem__
) will behave
differently during array-coercion in the future.
When nested inside sequences, such as np.array([array_like])
, these
were handled as a single Python object rather than an array.
In the future they will behave identically to:
np.array([np.array(array_like)])
This change should only have an effect if np.array(array_like)
is not 0-D.
The solution to this warning may depend on the object:
Some array-likes may expect the new behaviour, and users can ignore the warning. The object can choose to expose the sequence protocol to opt-in to the new behaviour.
For example,
shapely
will allow conversion to an array-like usingline.coords
rather thannp.asarray(line)
. Users may work around the warning, or use the new convention when it becomes available.
Unfortunately, using the new behaviour can only be achieved by
calling np.array(array_like)
.
If you wish to ensure that the old behaviour remains unchanged, please create an object array and then fill it explicitly, for example:
arr = np.empty(3, dtype=object)
arr[:] = [array_like1, array_like2, array_like3]
This will ensure NumPy knows to not enter the array-like and use it as a object instead.
(gh-17973)
Future Changes#
Arrays cannot be using subarray dtypes#
Array creation and casting using np.array(arr, dtype)
and arr.astype(dtype)
will use different logic when dtype
is a subarray dtype such as np.dtype("(2)i,")
.
For such a dtype
the following behaviour is true:
res = np.array(arr, dtype)
res.dtype is not dtype
res.dtype is dtype.base
res.shape == arr.shape + dtype.shape
But res
is filled using the logic:
res = np.empty(arr.shape + dtype.shape, dtype=dtype.base)
res[...] = arr
which uses incorrect broadcasting (and often leads to an error). In the future, this will instead cast each element individually, leading to the same result as:
res = np.array(arr, dtype=np.dtype(["f", dtype]))["f"]
Which can normally be used to opt-in to the new behaviour.
This change does not affect np.array(list, dtype="(2)i,")
unless the
list
itself includes at least one array. In particular, the behaviour
is unchanged for a list of tuples.
(gh-17596)
Expired deprecations#
The deprecation of numeric style type-codes
np.dtype("Complex64")
(with upper case spelling), is expired."Complex64"
corresponded to"complex128"
and"Complex32"
corresponded to"complex64"
.The deprecation of
np.sctypeNA
andnp.typeNA
is expired. Both have been removed from the public API. Usenp.typeDict
instead.(gh-16554)
The 14-year deprecation of
np.ctypeslib.ctypes_load_library
is expired. Useload_library
instead, which is identical.(gh-17116)
Financial functions removed#
In accordance with NEP 32, the financial functions are removed
from NumPy 1.20. The functions that have been removed are fv
,
ipmt
, irr
, mirr
, nper
, npv
, pmt
, ppmt
,
pv
, and rate
. These functions are available in the
numpy_financial
library.
(gh-17067)
Compatibility notes#
isinstance(dtype, np.dtype)
and not type(dtype) is not np.dtype
#
NumPy dtypes are not direct instances of np.dtype
anymore. Code that
may have used type(dtype) is np.dtype
will always return False
and
must be updated to use the correct version isinstance(dtype, np.dtype)
.
This change also affects the C-side macro PyArray_DescrCheck
if compiled
against a NumPy older than 1.16.6. If code uses this macro and wishes to
compile against an older version of NumPy, it must replace the macro
(see also C API changes section).
Same kind casting in concatenate with axis=None
#
When concatenate
is called with axis=None
,
the flattened arrays were cast with unsafe
. Any other axis
choice uses “same kind”. That different default
has been deprecated and “same kind” casting will be used
instead. The new casting
keyword argument
can be used to retain the old behaviour.
(gh-16134)
NumPy Scalars are cast when assigned to arrays#
When creating or assigning to arrays, in all relevant cases NumPy scalars will now be cast identically to NumPy arrays. In particular this changes the behaviour in some cases which previously raised an error:
np.array([np.float64(np.nan)], dtype=np.int64)
will succeed and return an undefined result (usually the smallest possible integer). This also affects assignments:
arr[0] = np.float64(np.nan)
At this time, NumPy retains the behaviour for:
np.array(np.float64(np.nan), dtype=np.int64)
The above changes do not affect Python scalars:
np.array([float("NaN")], dtype=np.int64)
remains unaffected (np.nan
is a Python float
, not a NumPy one).
Unlike signed integers, unsigned integers do not retain this special case,
since they always behaved more like casting.
The following code stops raising an error:
np.array([np.float64(np.nan)], dtype=np.uint64)
To avoid backward compatibility issues, at this time assignment from
datetime64
scalar to strings of too short length remains supported.
This means that np.asarray(np.datetime64("2020-10-10"), dtype="S5")
succeeds now, when it failed before. In the long term this may be
deprecated or the unsafe cast may be allowed generally to make assignment
of arrays and scalars behave consistently.
Array coercion changes when Strings and other types are mixed#
When strings and other types are mixed, such as:
np.array(["string", np.float64(3.)], dtype="S")
The results will change, which may lead to string dtypes with longer strings
in some cases. In particularly, if dtype="S"
is not provided any numerical
value will lead to a string results long enough to hold all possible numerical
values. (e.g. “S32” for floats). Note that you should always provide
dtype="S"
when converting non-strings to strings.
If dtype="S"
is provided the results will be largely identical to before,
but NumPy scalars (not a Python float like 1.0
), will still enforce
a uniform string length:
np.array([np.float64(3.)], dtype="S") # gives "S32"
np.array([3.0], dtype="S") # gives "S3"
Previously the first version gave the same result as the second.
Array coercion restructure#
Array coercion has been restructured. In general, this should not affect users. In extremely rare corner cases where array-likes are nested:
np.array([array_like1])
Things will now be more consistent with:
np.array([np.array(array_like1)])
This can subtly change output for some badly defined array-likes.
One example for this are array-like objects which are not also sequences
of matching shape.
In NumPy 1.20, a warning will be given when an array-like is not also a
sequence (but behaviour remains identical, see deprecations).
If an array like is also a sequence (defines __getitem__
and __len__
)
NumPy will now only use the result given by __array__
,
__array_interface__
, or __array_struct__
. This will result in
differences when the (nested) sequence describes a different shape.
(gh-16200)
Writing to the result of numpy.broadcast_arrays
will export readonly buffers#
In NumPy 1.17 numpy.broadcast_arrays
started warning when the resulting array
was written to. This warning was skipped when the array was used through the
buffer interface (e.g. memoryview(arr)
). The same thing will now occur for the
two protocols __array_interface__
, and __array_struct__
returning read-only
buffers instead of giving a warning.
(gh-16350)
Numeric-style type names have been removed from type dictionaries#
To stay in sync with the deprecation for np.dtype("Complex64")
and other numeric-style (capital case) types. These were removed
from np.sctypeDict
and np.typeDict
. You should use
the lower case versions instead. Note that "Complex64"
corresponds to "complex128"
and "Complex32"
corresponds
to "complex64"
. The numpy style (new) versions, denote the full
size and not the size of the real/imaginary part.
(gh-16554)
The operator.concat
function now raises TypeError for array arguments#
The previous behavior was to fall back to addition and add the two arrays, which was thought to be unexpected behavior for a concatenation function.
(gh-16570)
nickname
attribute removed from ABCPolyBase#
An abstract property nickname
has been removed from ABCPolyBase
as it
was no longer used in the derived convenience classes.
This may affect users who have derived classes from ABCPolyBase
and
overridden the methods for representation and display, e.g. __str__
,
__repr__
, _repr_latex
, etc.
(gh-16589)
float->timedelta
and uint64->timedelta
promotion will raise a TypeError#
Float and timedelta promotion consistently raises a TypeError.
np.promote_types("float32", "m8")
aligns with
np.promote_types("m8", "float32")
now and both raise a TypeError.
Previously, np.promote_types("float32", "m8")
returned "m8"
which
was considered a bug.
Uint64 and timedelta promotion consistently raises a TypeError.
np.promote_types("uint64", "m8")
aligns with
np.promote_types("m8", "uint64")
now and both raise a TypeError.
Previously, np.promote_types("uint64", "m8")
returned "m8"
which
was considered a bug.
(gh-16592)
numpy.genfromtxt
now correctly unpacks structured arrays#
Previously, numpy.genfromtxt
failed to unpack if it was called with
unpack=True
and a structured datatype was passed to the dtype
argument
(or dtype=None
was passed and a structured datatype was inferred).
For example:
>>> data = StringIO("21 58.0\n35 72.0")
>>> np.genfromtxt(data, dtype=None, unpack=True)
array([(21, 58.), (35, 72.)], dtype=[('f0', '<i8'), ('f1', '<f8')])
Structured arrays will now correctly unpack into a list of arrays, one for each column:
>>> np.genfromtxt(data, dtype=None, unpack=True)
[array([21, 35]), array([58., 72.])]
(gh-16650)
mgrid
, r_
, etc. consistently return correct outputs for non-default precision input#
Previously, np.mgrid[np.float32(0.1):np.float32(0.35):np.float32(0.1),]
and np.r_[0:10:np.complex64(3j)]
failed to return meaningful output.
This bug potentially affects mgrid
, ogrid
, r_
,
and c_
when an input with dtype other than the default
float64
and complex128
and equivalent Python types were used.
The methods have been fixed to handle varying precision correctly.
(gh-16815)
Boolean array indices with mismatching shapes now properly give IndexError
#
Previously, if a boolean array index matched the size of the indexed array but
not the shape, it was incorrectly allowed in some cases. In other cases, it
gave an error, but the error was incorrectly a ValueError
with a message
about broadcasting instead of the correct IndexError
.
For example, the following used to incorrectly give ValueError: operands
could not be broadcast together with shapes (2,2) (1,4)
:
np.empty((2, 2))[np.array([[True, False, False, False]])]
And the following used to incorrectly return array([], dtype=float64)
:
np.empty((2, 2))[np.array([[False, False, False, False]])]
Both now correctly give IndexError: boolean index did not match indexed
array along dimension 0; dimension is 2 but corresponding boolean dimension is
1
.
(gh-17010)
Casting errors interrupt Iteration#
When iterating while casting values, an error may stop the iteration
earlier than before. In any case, a failed casting operation always
returned undefined, partial results. Those may now be even more
undefined and partial.
For users of the NpyIter
C-API such cast errors will now
cause the iternext() function to return 0 and thus abort
iteration.
Currently, there is no API to detect such an error directly.
It is necessary to check PyErr_Occurred()
, which
may be problematic in combination with NpyIter_Reset
.
These issues always existed, but new API could be added
if required by users.
(gh-17029)
f2py generated code may return unicode instead of byte strings#
Some byte strings previously returned by f2py generated code may now be unicode strings. This results from the ongoing Python2 -> Python3 cleanup.
(gh-17068)
The first element of the __array_interface__["data"]
tuple must be an integer#
This has been the documented interface for many years, but there was still code that would accept a byte string representation of the pointer address. That code has been removed, passing the address as a byte string will now raise an error.
(gh-17241)
poly1d respects the dtype of all-zero argument#
Previously, constructing an instance of poly1d
with all-zero
coefficients would cast the coefficients to np.float64
.
This affected the output dtype of methods which construct
poly1d
instances internally, such as np.polymul
.
(gh-17577)
The numpy.i file for swig is Python 3 only.#
Uses of Python 2.7 C-API functions have been updated to Python 3 only. Users who need the old version should take it from an older version of NumPy.
(gh-17580)
Void dtype discovery in np.array
#
In calls using np.array(..., dtype="V")
, arr.astype("V")
,
and similar a TypeError will now be correctly raised unless all
elements have the identical void length. An example for this is:
np.array([b"1", b"12"], dtype="V")
Which previously returned an array with dtype "V2"
which
cannot represent b"1"
faithfully.
(gh-17706)
C API changes#
The PyArray_DescrCheck
macro is modified#
The PyArray_DescrCheck
macro has been updated since NumPy 1.16.6 to be:
#define PyArray_DescrCheck(op) PyObject_TypeCheck(op, &PyArrayDescr_Type)
Starting with NumPy 1.20 code that is compiled against an earlier version will be API incompatible with NumPy 1.20. The fix is to either compile against 1.16.6 (if the NumPy 1.16 release is the oldest release you wish to support), or manually inline the macro by replacing it with the new definition:
PyObject_TypeCheck(op, &PyArrayDescr_Type)
which is compatible with all NumPy versions.
Size of np.ndarray
and np.void_
changed#
The size of the PyArrayObject
and PyVoidScalarObject
structures have changed. The following header definition has been
removed:
#define NPY_SIZEOF_PYARRAYOBJECT (sizeof(PyArrayObject_fields))
since the size must not be considered a compile time constant: it will change for different runtime versions of NumPy.
The most likely relevant use are potential subclasses written in C which
will have to be recompiled and should be updated. Please see the
documentation for PyArrayObject
for more details and contact
the NumPy developers if you are affected by this change.
NumPy will attempt to give a graceful error but a program expecting a fixed structure size may have undefined behaviour and likely crash.
(gh-16938)
New Features#
where
keyword argument for numpy.all
and numpy.any
functions#
The keyword argument where
is added and allows to only consider specified
elements or subaxes from an array in the Boolean evaluation of all
and
any
. This new keyword is available to the functions all
and any
both via numpy
directly or in the methods of numpy.ndarray
.
Any broadcastable Boolean array or a scalar can be set as where
. It
defaults to True
to evaluate the functions for all elements in an array if
where
is not set by the user. Examples are given in the documentation of
the functions.
where
keyword argument for numpy
functions mean
, std
, var
#
The keyword argument where
is added and allows to limit the scope in the
calculation of mean
, std
and var
to only a subset of elements. It
is available both via numpy
directly or in the methods of
numpy.ndarray
.
Any broadcastable Boolean array or a scalar can be set as where
. It
defaults to True
to evaluate the functions for all elements in an array if
where
is not set by the user. Examples are given in the documentation of
the functions.
(gh-15852)
norm=backward
, forward
keyword options for numpy.fft
functions#
The keyword argument option norm=backward
is added as an alias for None
and acts as the default option; using it has the direct transforms unscaled
and the inverse transforms scaled by 1/n
.
Using the new keyword argument option norm=forward
has the direct
transforms scaled by 1/n
and the inverse transforms unscaled (i.e. exactly
opposite to the default option norm=backward
).
(gh-16476)
NumPy is now typed#
Type annotations have been added for large parts of NumPy. There is
also a new numpy.typing
module that contains useful types for
end-users. The currently available types are
ArrayLike
: for objects that can be coerced to an arrayDtypeLike
: for objects that can be coerced to a dtype
(gh-16515)
numpy.typing
is accessible at runtime#
The types in numpy.typing
can now be imported at runtime. Code
like the following will now work:
from numpy.typing import ArrayLike
x: ArrayLike = [1, 2, 3, 4]
(gh-16558)
New __f2py_numpy_version__
attribute for f2py generated modules.#
Because f2py is released together with NumPy, __f2py_numpy_version__
provides a way to track the version f2py used to generate the module.
(gh-16594)
mypy
tests can be run via runtests.py#
Currently running mypy with the NumPy stubs configured requires either:
Installing NumPy
Adding the source directory to MYPYPATH and linking to the
mypy.ini
Both options are somewhat inconvenient, so add a --mypy
option to runtests
that handles setting things up for you. This will also be useful in the future
for any typing codegen since it will ensure the project is built before type
checking.
(gh-17123)
Negation of user defined BLAS/LAPACK detection order#
distutils
allows negation of libraries when determining BLAS/LAPACK
libraries.
This may be used to remove an item from the library resolution phase, i.e.
to disallow NetLIB libraries one could do:
NPY_BLAS_ORDER='^blas' NPY_LAPACK_ORDER='^lapack' python setup.py build
That will use any of the accelerated libraries instead.
(gh-17219)
Allow passing optimizations arguments to asv build#
It is now possible to pass -j
, --cpu-baseline
, --cpu-dispatch
and
--disable-optimization
flags to ASV build when the --bench-compare
argument is used.
(gh-17284)
The NVIDIA HPC SDK nvfortran compiler is now supported#
Support for the nvfortran compiler, a version of pgfortran, has been added.
(gh-17344)
dtype
option for cov
and corrcoef
#
The dtype
option is now available for numpy.cov
and numpy.corrcoef
.
It specifies which data-type the returned result should have.
By default the functions still return a numpy.float64
result.
(gh-17456)
Improvements#
Improved string representation for polynomials (__str__
)#
The string representation (__str__
) of all six polynomial types in
numpy.polynomial
has been updated to give the polynomial as a mathematical
expression instead of an array of coefficients. Two package-wide formats for
the polynomial expressions are available - one using Unicode characters for
superscripts and subscripts, and another using only ASCII characters.
(gh-15666)
Remove the Accelerate library as a candidate LAPACK library#
Apple no longer supports Accelerate. Remove it.
(gh-15759)
Object arrays containing multi-line objects have a more readable repr
#
If elements of an object array have a repr
containing new lines, then the
wrapped lines will be aligned by column. Notably, this improves the repr
of
nested arrays:
>>> np.array([np.eye(2), np.eye(3)], dtype=object)
array([array([[1., 0.],
[0., 1.]]),
array([[1., 0., 0.],
[0., 1., 0.],
[0., 0., 1.]])], dtype=object)
(gh-15997)
Concatenate supports providing an output dtype#
Support was added to concatenate
to provide
an output dtype
and casting
using keyword
arguments. The dtype
argument cannot be provided
in conjunction with the out
one.
(gh-16134)
Thread safe f2py callback functions#
Callback functions in f2py are now thread safe.
(gh-16519)
numpy.core.records.fromfile
now supports file-like objects#
numpy.core.records.fromfile
can now use file-like objects, for instance
io.BytesIO
(gh-16675)
RPATH support on AIX added to distutils#
This allows SciPy to be built on AIX.
(gh-16710)
Use f90 compiler specified by the command line args#
The compiler command selection for Fortran Portland Group Compiler is changed
in numpy.distutils.fcompiler
. This only affects the linking command. This
forces the use of the executable provided by the command line option (if
provided) instead of the pgfortran executable. If no executable is provided to
the command line option it defaults to the pgf90 executable, which is an alias
for pgfortran according to the PGI documentation.
(gh-16730)
Add NumPy declarations for Cython 3.0 and later#
The pxd declarations for Cython 3.0 were improved to avoid using deprecated
NumPy C-API features. Extension modules built with Cython 3.0+ that use NumPy
can now set the C macro NPY_NO_DEPRECATED_API=NPY_1_7_API_VERSION
to avoid
C compiler warnings about deprecated API usage.
(gh-16986)
Make the window functions exactly symmetric#
Make sure the window functions provided by NumPy are symmetric. There were previously small deviations from symmetry due to numerical precision that are now avoided by better arrangement of the computation.
(gh-17195)
Performance improvements and changes#
Enable multi-platform SIMD compiler optimizations#
A series of improvements for NumPy infrastructure to pave the way to NEP-38, that can be summarized as follow:
New Build Arguments
--cpu-baseline
to specify the minimal set of required optimizations, default value ismin
which provides the minimum CPU features that can safely run on a wide range of users platforms.--cpu-dispatch
to specify the dispatched set of additional optimizations, default value ismax -xop -fma4
which enables all CPU features, except for AMD legacy features.--disable-optimization
to explicitly disable the whole new improvements, It also adds a new C compiler #definition calledNPY_DISABLE_OPTIMIZATION
which it can be used as guard for any SIMD code.
Advanced CPU dispatcher
A flexible cross-architecture CPU dispatcher built on the top of Python/Numpy distutils, support all common compilers with a wide range of CPU features.
The new dispatcher requires a special file extension
*.dispatch.c
to mark the dispatch-able C sources. These sources have the ability to be compiled multiple times so that each compilation process represents certain CPU features and provides different #definitions and flags that affect the code paths.New auto-generated C header ``core/src/common/_cpu_dispatch.h``
This header is generated by the distutils module
ccompiler_opt
, and contains all the #definitions and headers of instruction sets, that had been configured through command arguments ‘–cpu-baseline’ and ‘–cpu-dispatch’.New C header ``core/src/common/npy_cpu_dispatch.h``
This header contains all utilities that required for the whole CPU dispatching process, it also can be considered as a bridge linking the new infrastructure work with NumPy CPU runtime detection.
Add new attributes to NumPy umath module(Python level)
__cpu_baseline__
a list contains the minimal set of required optimizations that supported by the compiler and platform according to the specified values to command argument ‘–cpu-baseline’.__cpu_dispatch__
a list contains the dispatched set of additional optimizations that supported by the compiler and platform according to the specified values to command argument ‘–cpu-dispatch’.
Print the supported CPU features during the run of PytestTester
(gh-13516)
Changes#
np.linspace
on integers now uses floor#
When using a int
dtype in numpy.linspace
, previously float values would
be rounded towards zero. Now numpy.floor
is used instead, which rounds toward
-inf
. This changes the results for negative values. For example, the
following would previously give:
>>> np.linspace(-3, 1, 8, dtype=int)
array([-3, -2, -1, -1, 0, 0, 0, 1])
and now results in:
>>> np.linspace(-3, 1, 8, dtype=int)
array([-3, -3, -2, -2, -1, -1, 0, 1])
The former result can still be obtained with:
>>> np.linspace(-3, 1, 8).astype(int)
array([-3, -2, -1, -1, 0, 0, 0, 1])
(gh-16841)