Math::bigint and math::bigfloat allow one to declare big integers / floating point numbers as

$i = new math::bigint '123_456_789_123_456_789';
$i = new math::bigfloat '12_3.456_789_123_456_789e-2';
all operators are overloaded; operations with overloaded operators preserve the arguments.

math::bigint is also in perl core, this package provides newer versions separately.