Geek Answers Handbook

Best way to handle multiple constants in a container with numba?

EDIT: This question is very outdated! numba now supports Enumand namedtupleout of the box, which provide a reasonable solution for the constant group.

I am doing some bit hipping in python and want to speed it up with numba. To do this, I have many constant integer values ​​that I should handle in a possibly well-readable way. I would like to group them together with enumeration-like objects that have all the constants in the same namespace, accessible with the attribute-get statement. And of course, I would also like numba to understand what is happening there in order to maintain high speeds using jit compilation. My first and most naive attempt, which looked like this:

class SomeConstantsContainer:
    SOME_NAME = 0x1
    SOME_OTHER_CONSTANT = 0x2
    AND_ANOTHER_CONSTANT = 0x4

Unfortunately, when I look at the annotation, it seems like numba doesn't understand that the values ​​are constant, and it always reverts to slow access of objects to python objects. This is stated in the annotation:

#   $29.2 = global(SomeConstantsContainer: <class 'constants.SomeConstantContainer'>)  :: pyobject
#   $29.3 = getattr(attr=SOME_VARIABLE, value=$29.2)  :: pyobject

I know that I can always go back to something like this:

from numpy import np
SOME_STUPID_CONSTANT = np.int64(0x1)
ANOTHER_STUPID_CONSTANT = np.int64(0x2)

jit- a) b) , . . . , , jit python . , , ? - enum, numba, ?

: Enum :

@enum.unique
class SomeConstantsContainer(enum.IntEnum):
    SOME_NAME = 0x1
    SOME_OTHER_CONSTANT = 0x2
    AND_ANOTHER_CONSTANT = 0x4

:

    #   $42.3 = global(SomeConstantsContainer: <enum 'SomeConstantsContainer'>)  :: pyobject
    #   $42.4 = getattr(attr=SOME_OTHER_CONSTANT, value=$42.3)  :: pyobject
+4
1

, , , , :

def make_numba_functions():
    SOME_NAME = 0x1
    SOME_OTHER_CONSTANT = 0x2
    AND_ANOTHER_CONSTANT = 0x4

    @jit
    def f1(x,y):
        useful code goes here

    @jit
    def f2(x,y,z):
        some more useful code goes here

    return f1, f2

f1,f2 = make_numba_functions()

, .

inspect_types,

$0.1 = freevar(A: 10.0)  :: float64

( A - ). , , . (os.environ['NUMBA_DUMP_ASSEMBLY']='1'), , , , . - nopython=True, , , .

, , , !

+2

More articles:


All Articles