Yes, of course, I wasn't trying to say that it would beat the challenge, but that it did at least look somewhat promising. My previous implementation of the sieve of eratosthenes' memory complexity was O(n), where n is the maximum in the range, while this one's is O(n/8) (I am using the O notation correctly, am I not?), and the previous could generate primes under 10 million(Limit due to the heavy memory usage) in 10 or so secs, while the current can do so in .. 3 secs.