[jayGlow]

based on the article the noida approach has better performance and is able to run on a GPU while the Sterling implementation must run on a cpu.

[adammarples]

Potentially though there's nothing stopping someone from writing the original three rules using cuda