- changed the adaptive order algorithm to be better suited with the parallel implementation.
- ran multiple
parallel-vs-serial
tests with n-body problem with number of bodies = 200, 300, and 400. The serial time/ parallel time is always above 1. The highest ratio seen from these tests was 1.5 for the 300 bodies case at Atol = 10e-6
.
- a general pattern can be seen from all these tests. The ratio starts small for
Atol = 10e-1
, and then increases as Atol
decreases to 10e-3
. The ratio starts decreasing between Atol = 10e-3
and Atol = 10e-5
, and then continue to increase for all values of Atol < 10e-5
. This pattern happens at all different values of n bodies. this is strange, but I don’t have an idea why it happens.