pass C++ double array to embedded python efficiently without NumPy -
i extending c++ embedded python. how efficiently pass array of doubles?
currently create tuple:
std::vector<double> a(len); ... pyobject* pinput = pytuple_new(len); (int i=0; i<len; ++i) pytuple_setitem(pinput, i, pyfloat_fromdouble(a[i]));
is there more efficient way large arrays (e.g. len > 1000000
)? not want use numpy. datatype use on python side then? possible construct object of type array.array
(binary python array)? or other (python) type should construct in c++ code, converted array.array
on python side.
Comments
Post a Comment