1 | Efficiency ? = Speed Up S / P (Number of Processors) |
2 | Overhead fcomm = (P TP - T1) / T1 = 1/ ? - 1 |
3 | As fcomm linear in TP, overhead effects tend to be additive |
4 | In 2D Jacobi example fcomm = tcomm/(?n tfloat) |
5 | While efficiency takes approximate form ? ? 1 - tcomm/(?n tfloat) valid when overhead is small |
6 | As expected efficiency is < 1 corresponding to speedup being < P |