Consider that we have a random variable in a class with a given distribution. We know that across N number of randomize() calls, a simulator is expected to yield the probabilistic distribution as described by the constraint. However, what happens if we do ONE randomize() call per-run and measure distribution across N runs with random seeds? Is there any guarantee at all that the distribution will be honored?
class dist_class;
rand int m_var;
constraint dist_c { m_var dist { [0:9] := 10 }; }
endclass
I know that my question may be simulator-specific and simulator-seed dependent. For the purposes of this question, let us at least assume that across N runs, the simulator SV seed is uniformly distributed across the whole range of seeds.