I tried to use g6k with Python=92s multiprocessing but faced a huge difference between the run-time of different runs. I ran a version of the "find all short vectors" example and made a test in which I ran in on the same lattice many times. To be more specific:
the lattice is of dimension 60
I ran 160 instances of it
I used 40 workers
most of the workers (39) ran out of jobs after 10~12 CPU minutes
the last worker is running for nearly a CPU hour and still continues
Is it a feature of g6k having some extreme run times?
I'd suspect this instance somehow cannot reach saturation? Did you pick seeds for all instances? Could try to run just that and see if you get warnings?
Hello,
I tried to use g6k with Python=92s multiprocessing but faced a huge difference between the run-time of different runs. I ran a version of the "find all short vectors" example and made a test in which I ran in on the same lattice many times. To be more specific:
Thanks, Dotan Gazith.