Compiling Python code with
Numba provides several utilities for code generation, but its central
feature is the
numba.jit() decorator. Using this decorator, you can mark
a function for optimization by Numba’s JIT compiler. Various invocation
modes trigger differing compilation options and behaviours.
The recommended way to use the
@jit decorator is to let Numba decide
when and how to optimize:
from numba import jit @jit def f(x, y): # A somewhat trivial example return x + y
In this mode, compilation will be deferred until the first function
execution. Numba will infer the argument types at call time, and generate
optimized code based on this information. Numba will also be able to
compile separate specializations depending on the input types. For example,
f() function above with integer or complex numbers will
generate different code paths:
>>> f(1, 2) 3 >>> f(1j, 2) (2+1j)
You can also tell Numba the function signature you are expecting. The
f() would now look like:
from numba import jit, int32 @jit(int32(int32, int32)) def f(x, y): # A somewhat trivial example return x + y
int32(int32, int32) is the function’s signature. In this case, the
corresponding specialization will be compiled by the
and no other specialization will be allowed. This is useful if you want
fine-grained control over types chosen by the compiler (for example,
to use single-precision floats).
If you omit the return type, e.g. by writing
(int32, int32) instead of
int32(int32, int32), Numba will try to infer it for you. Function
signatures can also be strings, and you can pass several of them as a list;
numba.jit() documentation for more details.
Of course, the compiled function gives the expected results:
>>> f(1,2) 3
and if we specified
int32 as return type, the higher-order bits get
>>> f(2**31, 2**31 + 1) 1
Calling and inlining other functions
Numba-compiled functions can call other compiled functions. The function calls may even be inlined in the native code, depending on optimizer heuristics. For example:
@jit def square(x): return x ** 2 @jit def hypot(x, y): return math.sqrt(square(x) + square(y))
@jit decorator must be added to any such library function,
otherwise Numba may generate much slower code.
@jit signatures can use a number of types. Here are some
voidis the return type of functions returning nothing (which actually return
Nonewhen called from Python)
uintpare pointer-sized integers (signed and unsigned, respectively)
uintcare equivalent to C
unsigned intinteger types
uint64are fixed-width integers of the corresponding bit width (signed and unsigned)
float64are single- and double-precision floating-point numbers, respectively
complex128are single- and double-precision complex numbers, respectively
array types can be specified by indexing any numeric type, e.g.
float32[:]for a one-dimensional single-precision array or
int8[:,:]for a two-dimensional array of 8-bit integers.
A number of keyword-only arguments can be passed to the
Numba has two compilation modes: nopython mode and
object mode. The former produces much faster code, but has
limitations that can force Numba to fall back to the latter. To prevent
Numba from falling back, and instead raise an error, pass
@jit(nopython=True) def f(x, y): return x + y
Whenever Numba optimizes Python code to native code that only works on
native types and variables (rather than Python objects), it is not necessary
anymore to hold Python’s global interpreter lock (GIL).
Numba will release the GIL when entering such a compiled function if you
@jit(nogil=True) def f(x, y): return x + y
Code running with the GIL released runs concurrently with other threads executing Python or Numba code (either the same compiled function, or another one), allowing you to take advantage of multi-core systems. This will not be possible if the function is compiled in object mode.
nogil=True, you’ll have to be wary of the usual pitfalls
of multi-threaded programming (consistency, synchronization, race conditions,
To avoid compilation times each time you invoke a Python program,
you can instruct Numba to write the result of function compilation into
a file-based cache. This is done by passing
@jit(cache=True) def f(x, y): return x + y
Caching of compiled functions has several known limitations:
The caching of compiled functions is not performed on a function-by-function basis. The cached function is the the main jit function, and all secondary functions (those called by the main function) are incorporated in the cache of the main function.
Cache invalidation fails to recognize changes in functions defined in a different file. This means that when a main jit function calls functions that were imported from a different module, a change in those other modules will not be detected and the cache will not be updated. This carries the risk that “old” function code might be used in the calculations.
Global variables are treated as constants. The cache will remember the value of the global variable at compilation time. On cache load, the cached function will not rebind to the new value of the global variable.
Enables automatic parallelization (and related optimizations) for those
operations in the function known to have parallel semantics. For a list of
supported operations, see Automatic parallelization with @jit. This feature is enabled by
parallel=True and must be used in conjunction with
@jit(nopython=True, parallel=True) def f(x, y): return x + y