From many sources in the net, it is suggested the use of random_device
to produce random numbers. All of the code instantiate random_device
, applies the proper random_engine
or distribution, and makes use of it. I'm wondering, what is the overhead of such pattern? Instantiating the random_device
will require the opening of /dev/urandom
on linux, so the overhead may be heavier than it seems to appear from that sources. Shouldn't be recommended of a monostate/singleton pattern for this instead?
Thanks in advance.
Aucun commentaire:
Enregistrer un commentaire