Consider the following issues with the cuRAND version of the π estimator shown in Listing 8.10.
a. If you run the app multiple times, do you keep getting the same answer? Is that a good thing or a bad thing for a Monte Carlo algorithm?
b. Experiment with changing the seed value on line 11. Does that produce a new result? If you run the app again with the new seed value, do you reproduce the new result?
c. Explore cuRAND to learn about states and offsets. Revise the code so that repeated runs of the app use different sets of random numbers and produce (appropriately) different outcomes.