DMF (-H2 option) problem when N is small

[fvella]

./mgbc -S 10 -E 16 -p 1x1 -H 2 -N 50

output

task 0 BC skipped: 130 
task 0 BC rounds: 10  
task 0 BC execution total time: 0.007300 sec
task 0 BC rounds total time: 0.007288 sec
task 0 BC rounds Computation CUDA avg time: 0.000000 sec over 1 procs
task 0 BC rounds Communication MPI avg time: 0.000000 sec over 1 procs
task 0 BC Max time: 0.003473 sec
task 0 BC Min time: 0.000381 sec
task 0 BC Mean time: 0.000729 sec
task 0 BC simulated time: 13443987080919514.000000 sec (virtual skipped 13182)
task 0 BC simulated time-2: 0.747520 sec
task 0 2-Degree reduction: 0.000021 sec.(2-degree nodes computed by 2-degree fun 0/0)

The simulation time is wrong.

[fvella]

The problem is due to N value.
DMF (Degree-2 heuristics) sorts the vertices by degree.
The first vertices have degree equal to 0.

We have to check if the first -N vertices have degree 0.
Use the following command for exact BC score

./mgbc -S 10 -E 16 -p 1x1 -H 2 -N 1024