Re: [Bug-gnubg] Simpler PRNG

[Philippe Michel]

On Thu, 15 Jan 2015, Michael Petch wrote:

> Thanks for the information. I am pretty sure I saw Philippe Michel at
> one point mention information regarding his findings with profiling. My
> understanding that the random number generation (with MT) isn't much of 
> a bottleneck within the context of rollouts.

For each move happening in a 2ply rollout, there are 2 random numbers 
generated and thousands of positions evaluated, leading to something like:

  %   cumulative   self              self     total
 time   seconds   seconds    calls   s/call   s/call  name
 30.53    146.78   146.78 296200881     0.00     0.00  EvaluateSSE
...
  0.00    480.84     0.00    19518     0.00     0.00  RolloutDice

For Oply, the discrepancy is smaller but the cost of random number 
generation is still negligible.