Python/ceval.c (part 91)

Source:

cpython 3.14 @ ab2d84fe1023/Python/ceval.c ↗

This annotation covers binary arithmetic specializations. See python_ceval90_detail for LOAD_GLOBAL specializations.

Map

Lines	Symbol	Role
1-80	BINARY_OP	Generic binary operation dispatcher
81-160	BINARY_OP_ADD_INT	Specialization for int + int
161-240	BINARY_OP_ADD_UNICODE	Specialization for str + str
241-320	BINARY_OP_MULTIPLY_INT	Specialization for int * int
321-500	BINARY_OP_SUBTRACT_INT / FLOAT variants	Other numeric specializations

Reading

BINARY_OP

// CPython: Python/ceval.c:5420 BINARY_OP
inst(BINARY_OP, (lhs, rhs -- res)) {
    assert(NB_ADD <= oparg && oparg <= NB_INPLACE_XOR);
    PyObject *(*binary_op)(PyObject *, PyObject *) = _PyEval_BinaryOps[oparg];
    res = binary_op(lhs, rhs);
    ERROR_IF(res == NULL, error);
    Py_DECREF(lhs);
    Py_DECREF(rhs);
}

BINARY_OP uses a dispatch table indexed by oparg (e.g., NB_ADD, NB_MULTIPLY). The dispatch table holds function pointers to PyNumber_Add, PyNumber_Multiply, etc.

BINARY_OP_ADD_INT

// CPython: Python/ceval.c:5480 BINARY_OP_ADD_INT
inst(BINARY_OP_ADD_INT, (left, right -- res)) {
    DEOPT_IF(!PyLong_CheckExact(left), BINARY_OP);
    DEOPT_IF(!PyLong_CheckExact(right), BINARY_OP);
    STAT_INC(BINARY_OP, hit);
    res = _PyLong_Add((PyLongObject *)left, (PyLongObject *)right);
    ERROR_IF(res == NULL, error);
    Py_DECREF(left);
    Py_DECREF(right);
}

BINARY_OP_ADD_INT calls _PyLong_Add directly, bypassing the PyNumber_Add dispatch. _PyLong_Add has a fast path for small integers (the compact representation).

BINARY_OP_ADD_UNICODE

// CPython: Python/ceval.c:5520 BINARY_OP_ADD_UNICODE
inst(BINARY_OP_ADD_UNICODE, (left, right -- res)) {
    DEOPT_IF(!PyUnicode_CheckExact(left), BINARY_OP);
    DEOPT_IF(!PyUnicode_CheckExact(right), BINARY_OP);
    res = PyUnicode_Concat(left, right);
    ERROR_IF(res == NULL, error);
    Py_DECREF(left);
    Py_DECREF(right);
}

String concatenation is specialized to call PyUnicode_Concat directly. PyUnicode_Concat allocates the result string and copies both parts; for repeated concatenation, use ''.join(parts) instead.

BINARY_OP_MULTIPLY_INT

// CPython: Python/ceval.c:5560 BINARY_OP_MULTIPLY_INT
inst(BINARY_OP_MULTIPLY_INT, (left, right -- res)) {
    DEOPT_IF(!PyLong_CheckExact(left), BINARY_OP);
    DEOPT_IF(!PyLong_CheckExact(right), BINARY_OP);
    res = _PyLong_Multiply((PyLongObject *)left, (PyLongObject *)right);
    ERROR_IF(res == NULL, error);
    Py_DECREF(left);
    Py_DECREF(right);
}

_PyLong_Multiply has a fast path for small integers (fitting in one ob_digit) using C integer multiply. Larger integers use Karatsuba or schoolbook multiplication.

gopy notes

BINARY_OP is in vm/eval_simple.go dispatching via a switch on oparg. BINARY_OP_ADD_INT calls objects.IntAdd. BINARY_OP_ADD_UNICODE calls objects.StrConcat. Specializations are not inline-cached yet; the generic path calls objects.BinaryOp.