bench-mr
bench-mr copied to clipboard
Environment seed behavior
I think the environment seed behavior is a bit unusual right now.
Looks as if you generate 50 environments with seed = 0 and then 50 environments with seed = 1, 49 of the environments will be the same (since the seed is incremented by one for each environment and then passed to srand for each generated environment).
Maybe we could change the code to use a C++11 generator used only for environment generation that is seeded once with the provided seed.