On Fri, Nov 19, 2010 at 5:21 PM, nathan binkert <[email protected]> wrote:
> > I see we have a random number generator in base/, and I could make > > use of it for a thing I'm doing for AMD. One thing that worries me, > > though, is getting stuck in a loop of n random numbers which would force > > my randomness to not be quite as random as I'd like when consuming lots > > of values. I think theoretically all determinstic random number > > generators are going to repeat eventually because the input space is > > only so big, but it would be great if the period of ours was a really > > large value. Does anybody know what it is? Is there a good technique for > > expanding it? > > The RNG that we're using is called the Mersenne Twister: > http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/emt.html<http://www.math.sci.hiroshima-u.ac.jp/%7Em-mat/MT/emt.html> > > I stole the code in random_mt.cc from there. The period is insane: > 2^19937-1. I'm pretty sure that you won't have a problem. > Could someone add this link and some discussion to a comment in the code, so in the future people with Gabe's question don't have to wonder? Thanks, Steve
_______________________________________________ m5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/m5-dev
