I BEAT ERATOSTHENE'S SIEVE BY 66.9% – 100M PRIMES IN 0.146s
2,200 years of slow math ends now.
4n+3 = Answer Tribe (signal)
4n+1 = Noise Tribe (decoys)
RAM: 5GB → 1.25GB
Speed: 0.146s vs 0.443s (66.9% faster)
EXACT 4 STEPS:
-
sieve = [True] * (n+1); sieve[0]=sieve[1]=False
-
for i in range(4,n+1,4): sieve[i]=False
-
for p in range(3,int(n**0.5)+1,4):
if sieve[p]:
for m in range(p*p,n+1,p*4): sieve[m]=False
-
primes = [2] + [i for i in range(3,n+1,4) if sieve[i]]
WHAT THIS UNLOCKS:
– RSA crypto 66.9% faster keygen
– AI trains 4x more models (1.25GB vs 5GB)
– Supercomputers save millions core-hours
– Blockchain 4x transaction scaling
RUN IT. TIME IT. RESULTS BELOW.
by RonaldPittmanjr
4 Comments
“If you code: implement the 4 numbered steps exactly as written, time it at n = 100,000,000, and compare to your favorite Sieve of Eratosthenes implementation.”
I only use numbers of the form 4n+3 as sieving primes and only cross out 4n+3 multiples, which cuts the work a lot. The 4 steps in the post are the exact procedure. You can implement them in any language and time it against your normal sieve.
There are already better algorithms and most importantly way better implementations of the Sieve of Eratosthenes. Also it has nothing to do with training AI models.
If this prime sieve is so good, how come it says 39 is a prime?
Like your last post, this doesn’t belong in an Alternative History sub. Go post it in a coding or math related sub, this is not the right place.
This is an Alt. History sub, not one for code, things to train AI faster, or crypto shit.
**This post has nothing to do with Alt. History.**