grib2io_interp

Introduction

grib2io_interp is a component package to the parent package, grib2io. This package provides interfaces to the Fortran-based NCEPLIBS-ip general interpolation library. The interface supports performing scalar and vector interpolation. grib2io_interp.interpolate is a F2PY extension module that contains 2 subroutines: interpolate_scalar and interpolate_vector. These subroutines serve as wrappers to the NCEPLIBS-ip subroutines ipolates_grib2 and ipolatev_grib2 respectively.

It is recommended that you access these interpolation subroutines via grib2io.interpolate().

OpenMP Support

NCEPLIBS-ip can be built with support for OpenMP threading. This is also supported in grib2io-interp. The F2PY signatures for interpolate_scalar and interpolate_vector subroutines contain the keyword threadsafe which unlocks Python's global interpretor lock (gil) and allows for threading. grib2io-interp contains a second F2PY extension module, grib2io_interp.openmp_handler, that allows the user to get and set the number of OpenMP threads. These functions are made public by get_openmp_threads() and set_openmp_threads().

View Source

 1"""
 2Introduction
 3============
 4
 5grib2io_interp is a component package to the parent package, [grib2io](https://github.com/NOAA-MDL/grib2io).
 6This package provides interfaces to the Fortran-based [NCEPLIBS-ip](https://github.com/NOAA-EMC/NCEPLIBS-ip)
 7general interpolation library. The interface supports performing scalar and vector interpolation. `grib2io_interp.interpolate`
 8is a F2PY extension module that contains 2 subroutines: `interpolate_scalar` and `interpolate_vector`.  These
 9subroutines serve as wrappers to the NCEPLIBS-ip subroutines `ipolates_grib2` and `ipolatev_grib2` respectively.
10
11It is **recommended** that you access these interpolation subroutines via [`grib2io.interpolate()`](https://noaa-mdl.github.io/grib2io/grib2io.html#interpolate).
12
13OpenMP Support
14==============
15
16NCEPLIBS-ip can be built with support for OpenMP threading. This is also supported in grib2io-interp. The
17F2PY signatures for `interpolate_scalar` and `interpolate_vector` subroutines contain the keyword `threadsafe`
18which unlocks Python's global interpretor lock (gil) and allows for threading. grib2io-interp contains a
19second F2PY extension module, `grib2io_interp.openmp_handler`, that allows the user to get and set the number
20of OpenMP threads.  These functions are made public by `grib2io_interp.get_openmp_threads()` and 
21`grib2io_interp.set_openmp_threads()`.
22"""
23
24from .__config__ import grib2io_interp_version as __version__
25from .__config__ import has_openmp_support
26
27def get_openmp_threads():
28    """
29    Return the number of OpenMP threads.
30
31    Returns
32    -------
33    int
34        The number of OpenMP threads.
35    """
36    try:
37        from . import openmp_handler
38        return openmp_handler.get_openmp_threads()
39    except(ImportError):
40        pass
41
42def set_openmp_threads(n: int):
43    """
44    Set the number of OpenMP threads.
45
46    Parameters
47    ----------
48    n
49        The number of OpenMP threads.
50    """
51    if n < 1:
52        raise ValueError(f'Number of threads must be > 0.')
53    try:
54        from . import openmp_handler
55        openmp_handler.set_openmp_threads(n)
56    except(ImportError):
57        pass

def get_openmp_threads(): View Source

28def get_openmp_threads():
29    """
30    Return the number of OpenMP threads.
31
32    Returns
33    -------
34    int
35        The number of OpenMP threads.
36    """
37    try:
38        from . import openmp_handler
39        return openmp_handler.get_openmp_threads()
40    except(ImportError):
41        pass

Return the number of OpenMP threads.

Returns

int: The number of OpenMP threads.

def set_openmp_threads(n: int): View Source

43def set_openmp_threads(n: int):
44    """
45    Set the number of OpenMP threads.
46
47    Parameters
48    ----------
49    n
50        The number of OpenMP threads.
51    """
52    if n < 1:
53        raise ValueError(f'Number of threads must be > 0.')
54    try:
55        from . import openmp_handler
56        openmp_handler.set_openmp_threads(n)
57    except(ImportError):
58        pass

Set the number of OpenMP threads.

Parameters

n: The number of OpenMP threads.