input() не работает в numba

В jit-функции numba я использую input(), и на этом все вылетает. Мне надо ускорить код, при этом input критически важен. Как мне это исправить. Мне надо сохранить ввод из консоли и оставить input.

P.S. Без numba все работает отлично, но медленно. P.P.S Это регистровая VR-машина

Код:

import numba, numpy


def smart_spl(x):
    return x.split(';')[0]


def load(text):
    strings = list(map(smart_spl, text.split('\n')))
    data = list(map(lambda x: x.split(':'), strings))
    data2 = list(map(lambda x: x[0], data))
    data = list(map(lambda x: list(map(int, x[1:])), data))
    data += data
    return data, data2


# @numba.jit(cache=True)
def exec(data=None, data2=None, memory=None):
    current_string = 0
    while current_string < len(data2):
        #print('DEBUG:', current_string, data2[current_string], data[current_string])

        if data2[current_string] == 'ifgotofwd':
            if memory[int(data[current_string][1])]:
                current_string += int(data[current_string][0])
                continue
        if data2[current_string] == 'gotoback':
            current_string -= int(data[current_string][0])
            continue
        elif data2[current_string] == 'define':
            define(int(data[current_string][0]), memory=memory)
        elif data2[current_string] == 'setval':
            setval(int(data[current_string][0]), int(data[current_string][1]), memory=memory)
        elif data2[current_string] == 'add':
            add(int(data[current_string][0]), int(data[current_string][1]),
                int(data[current_string][2]), memory=memory)
        elif data2[current_string] == 'mul':
            mul(int(data[current_string][0]), int(data[current_string][1]),
                int(data[current_string][2]), memory=memory)
        elif data2[current_string] == 'div':
            div(int(data[current_string][0]), int(data[current_string][1]),
                int(data[current_string][2]), memory=memory)
        elif data2[current_string] == 'sub':
            sub(int(data[current_string][0]), int(data[current_string][1]),
                int(data[current_string][2]), memory=memory)
        elif data2[current_string] == 'gre':
            gre(int(data[current_string][0]), int(data[current_string][1]),
                int(data[current_string][2]), memory=memory)
        elif data2[current_string] == 'eql':
            eql(int(data[current_string][0]), int(data[current_string][1]),
                int(data[current_string][2]), memory=memory)
        elif data2[current_string] == 'and':
            and_op(int(data[current_string][0]), int(data[current_string][1]),
                   int(data[current_string][2]), memory=memory)
        elif data2[current_string] == 'or':
            or_op(int(data[current_string][0]), int(data[current_string][1]),
                  int(data[current_string][2]), memory=memory)
        elif data2[current_string] == 'not':
            not_op(int(data[current_string][0]), int(data[current_string][1]), memory=memory)
        elif data2[current_string] == 'print':
            print_func(int(data[current_string][0]), memory=memory)
        elif data2[current_string] == 'input':
            # print('DEBUG:INPUT')
            input_func(int(data[current_string][0]), memory=memory)
        current_string += 1


@numba.jit(cache=True)
def input_func(adress, data=None, memory=None):
    setval(adress, int(input()), memory=memory)


@numba.njit(cache=True)
def print_func(adress, data=None, memory=None):
    print(int(memory[adress]))


@numba.njit(cache=True)
def graphic(data=None, memory=None):
    pass


@numba.njit(cache=True)
def console():
    pass


@numba.njit(cache=True)
def setpixel(x, y, r, g, b, data=None, memory=None):
    pass


@numba.njit(cache=True)
def define(adress, size=0, data=None, memory=None):
    memory[adress] = 0


@numba.njit(cache=True)
def setval(adress, value, data=None, memory=0):
    memory[adress] = value


@numba.njit(cache=True)
def add(adr1, adr2, adr_out, data=None, memory=None):
    memory[adr_out] = memory[adr1] + memory[adr2]


@numba.njit(cache=True)
def mul(adr1, adr2, adr_out, data=None, memory=None):
    memory[adr_out] = memory[adr1] * memory[adr2]


@numba.njit(cache=True)
def sub(adr1, adr2, adr_out, data=None, memory=None):
    memory[adr_out] = memory[adr1] - memory[adr2]


@numba.njit(cache=True)
def div(adr1, adr2, adr_out, data=None, memory=None):
    memory[adr_out] = memory[adr1] // memory[adr2]


@numba.njit(cache=True)
def gre(adr1, adr2, adr_out, data=None, memory=None):
    memory[adr_out] = int(memory[adr1] > memory[adr2])


@numba.njit(cache=True)
def eql(adr1, adr2, adr_out, data=None, memory=None):
    memory[adr_out] = int(memory[adr1] == memory[adr2])


@numba.njit(cache=True)
def or_op(adr1, adr2, adr_out, data=None, memory=None):
    memory[adr_out] = int(memory[adr1] or memory[adr2])


@numba.njit(cache=True)
def and_op(adr1, adr2, adr_out, data=None, memory=None):
    memory[adr_out] = int(memory[adr1] and memory[adr2])


@numba.njit(cache=True)
def not_op(adr1, adr_out, data=None, memory=None):
    memory[adr_out] = int(not bool(memory[adr1]))


if __name__ == '__main__':
    text = open('hello.roasm').read()
    data, data2 = load(text)
    memory = numpy.zeros(1024)
    exec(data, data2, memory)

Ошибка:

NumbaWarning: 
Compilation is falling back to object mode WITH looplifting enabled because Function "input_func" failed type inference due to: Untyped global name 'input': Cannot determine Numba type of <class 'builtin_function_or_method'>

File "roasm.py", line 70:
def input_func(adress, data=None, memory=None):
    setval(adress, int(input()), memory=memory)
    ^

  @numba.jit(cache=True)
C:\Users\Vasya\PycharmProjects\RobinN\venv\lib\site-packages\numba\core\object_mode_passes.py:151: NumbaWarning: Function "input_func" was compiled in object mode without forceobj=True.

File "roasm.py", line 69:
@numba.jit(cache=True)
def input_func(adress, data=None, memory=None):
^

  warnings.warn(errors.NumbaWarning(warn_msg,
C:\Users\Vasya\PycharmProjects\RobinN\venv\lib\site-packages\numba\core\object_mode_passes.py:161: NumbaDeprecationWarning: 
Fall-back from the nopython compilation path to the object mode compilation path has been detected, this is deprecated behaviour.

For more information visit https://numba.readthedocs.io/en/stable/reference/deprecation.html#deprecation-of-object-mode-fall-back-behaviour-when-using-jit

File "roasm.py", line 69:
@numba.jit(cache=True)
def input_func(adress, data=None, memory=None):
^

  warnings.warn(errors.NumbaDeprecationWarning(msg,

Файл hello.roasm:

define:10:0:0; создаем переменную
setval:1:1:0;
setval:0:0:0;
input:10:0:0;
define:11:0:0;
define:21:0:0;
define:31:0:0;
define:20:0:0;
gre:10:20:11;
eql:0:1:21;
or:21:11:31;
not:31:11:0;
not:11:21:0;
ifgotofwd:4:11:0;
add:20:1:20;
print:20:0:0;
gotoback:8:0:0;
add:1:1:1;