It's an interesting exercise to think about why the performance of Sum[i, {i, 1, 100000}] differs between Mathics and MMA. Mathics just calls down to sympy, which I think just does the sum in Python [1]. Mthematica (likely) identifies your sum as the 100000th triangular number, and computes it directly in native code, since I know Mathematica relies heavily on standard tables of summations/integrals/etc. [2]

[1] https://github.com/sympy/sympy/blob/master/sympy/concrete/su....

[2] See Sum's Method option.