Remove OpenMP thread support from convolution.convolve()

See #7971

Signed-off-by: James Noss <jnoss@stsci.edu>

.travis.yml

@@ -73,7 +73,7 @@ matrix:
         - os: osx
           stage: Cron tests
           env: SETUP_CMD='test --remote-data=astropy'
-               CONDA_DEPENDENCIES="$CONDA_ALL_DEPENDENCIES clang llvm-openmp"
+               CONDA_DEPENDENCIES="$CONDA_ALL_DEPENDENCIES clang"
                PIP_DEPENDENCIES='jplephem'
                CCOMPILER=clang
                EVENT_TYPE='cron'

CHANGES.rst

@@ -15,7 +15,6 @@ astropy.convolution
 
 - ``convolve`` now accepts any array-like input, not just ``numpy.ndarray`` or
   lists. [#7303]
-- Thread with OpenMP support [#7293]
 
 astropy.coordinates
 ^^^^^^^^^^^^^^^^^^^
@@ -270,11 +269,6 @@ astropy.convolution
 
 - ``kernel`` can now be a tuple. [#7561]
 
-- Add ``n_threads`` parameter to control number of threads used in C computation.
-  An exception is raised for negative values.
-  A warning is issued if ``n_threads`` > total number of CPUs reported by the OS.
-  A warning is raised if ``n_threads`` > 1 and Astropy was NOT built with OpenMP support. [#7293]
-
 - Not technically an API changes, however, the doc string indicated that ``boundary=None``
   was the default when actually it is ``boundary='fill'``. The doc string has been corrected,
   however, someone may interpret this as an API change not realising that nothing has actually
@@ -455,12 +449,6 @@ astropy.convolution
   hoisting, and vectorization, etc. Compiler optimization level ``-O2`` required for
   hoisting and ``-O3`` for vectorization. [#7293]
 
-- ``convolve()``: Core computation now threaded using OpenMP. However, the root
-  setup disables this by default. To build the package with OpenMP support
-  instantiate the environment variable ``ASTROPY_SETUP_WITH_OPENMP=1``.
-  E.g. ``ASTROPY_SETUP_WITH_OPENMP=1 pip install astropy --no-cache-dir`` or
-  ``ASTROPY_SETUP_WITH_OPENMP=1 ./setup.py build``. [#7293]
-
 - ``convolve()``: ``nan_treatment=‘interpolate’`` was slow to compute irrespective of
   whether any NaN values exist within the array. The input array is now
   checked for NaN values and interpolation is disabled if non are found. This is a
@@ -623,12 +611,6 @@ Other Changes and Additions
   leading to some speed improvements, and setting the stage for allowing
   overrides with ``__array_ufunc__``. [#7502]
 
-- Building with OpenMP support is disabled by default. To build the package with
-  OpenMP support (currently only used in ``convolution.convolve``) create and
-  set the environment variable ``ASTROPY_SETUP_WITH_OPENMP=1``.
-  E.g. ``ASTROPY_SETUP_WITH_OPENMP=1 pip install astropy --no-cache-dir`` or
-  ``ASTROPY_SETUP_WITH_OPENMP=1 ./setup.py build``. [#7293]
-
 
 
 3.0.6 (unreleased)

astropy/convolution/convolve.py

@@ -19,7 +19,6 @@
 from ..modeling.core import _make_arithmetic_operator, BINARY_OPERATORS
 from ..modeling.core import _CompoundModelMeta
 from .utils import KernelSizeError, has_even_axis, raise_even_kernel_exception
-from ..utils.openmp import handle_n_threads_usage
 
 # Turn the faulthandler ON to help catch any signals, e.g. segfaults
 # or asserts. This doesn't, currently, work with Jupyter Notebook.
@@ -71,7 +70,7 @@
 @support_nddata(data='array')
 def convolve(array, kernel, boundary='fill', fill_value=0.,
              nan_treatment='interpolate', normalize_kernel=True, mask=None,
-             preserve_nan=False, normalization_zero_tol=1e-8, n_threads=0):
+             preserve_nan=False, normalization_zero_tol=1e-8):
     '''
     Convolve an array with a kernel.
 
@@ -130,12 +129,6 @@ def convolve(array, kernel, boundary='fill', fill_value=0.,
         The absolute tolerance on whether the kernel is different than zero.
         If the kernel sums to zero to within this precision, it cannot be
         normalized. Default is "1e-8".
-    n_threads : int
-        The number of threads to use (default 0). 0 => use all. There is no
-        limit, be cautious if over commmiting resources.
-        Note that an exception is raised for negative values. A warning is issued
-        if n_threads>1 and astropy was NOT built with OpenMP support.
-        With DASK, it is more performant to set ``n_threads=1``.
 
     Returns
     -------
@@ -159,7 +152,9 @@ def convolve(array, kernel, boundary='fill', fill_value=0.,
     if nan_treatment not in ('interpolate', 'fill'):
         raise ValueError("nan_treatment must be one of 'interpolate','fill'")
 
-    n_threads = handle_n_threads_usage(n_threads)
+    # OpenMP support is disabled at the C src code level, changing this will have
+    # no effect.
+    n_threads = 1
 
     # Keep refs to originals
     passed_kernel = kernel

astropy/convolution/setup_package.py

@@ -3,7 +3,6 @@
 import os
 import sys
 from distutils.extension import Extension
-from astropy_helpers.openmp_helpers import add_openmp_flags_if_available
 
 C_CONVOLVE_PKGDIR = os.path.relpath(os.path.dirname(__file__))
 
@@ -23,6 +22,4 @@ def get_extensions():
                  include_dirs=["numpy"],
                  language='c')
 
-    add_openmp_flags_if_available(lib_convolve_ext)
-
     return [lib_convolve_ext]

astropy/convolution/src/convolve.h

@@ -3,6 +3,9 @@
 
 #include <stddef.h>
 
+// Forcibly disable OpenMP support at the src level
+#undef _OPENMP
+
 #if defined(_MSC_VER)
 
 #define FORCE_INLINE  __forceinline

setup.py

@@ -22,7 +22,6 @@
 from astropy_helpers.distutils_helpers import is_distutils_display_option
 from astropy_helpers.git_helpers import get_git_devstr
 from astropy_helpers.version_helpers import generate_version_py
-from astropy_helpers.openmp_helpers import generate_openmp_enabled_py
 
 import astropy
 
@@ -46,11 +45,6 @@
 generate_version_py(NAME, VERSION, RELEASE, get_debug_option(NAME),
                     uses_git=not RELEASE)
 
-# Generate function to check OpenMP support at runtime
-# The current default is to NOT build with OpenMP support, i.e. ``disable_openmp = True``
-disable_openmp = os.environ.get('ASTROPY_SETUP_WITH_OPENMP', '0') == '0'
-generate_openmp_enabled_py(NAME, disable_openmp=disable_openmp)
-
 # Get configuration information from all of the various subpackages.
 # See the docstring for setup_helpers.update_package_files for more
 # details.