Producing large samples of uniform pseudo-random numbers in high dimensions is very tricky. If COINFLIPS can solve that problem then that would be very useful.

Producing large samples of uniform pseudo-random numbers in high dimensions is very tricky. If COINFLIPS can solve that problem then that would be very useful. That said, I don't know if there are many computational workloads where the basic random number generation step is the performance bottleneck.