View Single Post
Old 2006-07-26, 05:35   #24
geoff's Avatar
Mar 2003
New Zealand

22058 Posts

Originally Posted by Citrix
Is there a limit on the size of the k, your program can handle? I would like to test higher values like 3^32 and 3^64 etc.
The current limit for k is 2^32-1. It is not a simple matter to increase that without affecting the sieve speed for general sequences.

It would be possible to write a seperate sieve routine that accepts k=a^b where a and b can each be up to 2^32-1. However there are other parts of the program that will limit the speed as b becomes large. In particular the prime sieve generates all primes and then checks to see if each one is of the correct form, this will become very ineffcient when b gets large, and in any case you will soon need to check factors larger than 2^62 which is the modular multiplication limit for the current code.

Have you tried one of these sieves?: GFNsieve, AthGFNsieve.

Does your program have a prover code to submit primes found using your program.
The entry is Srsieve.

Last fiddled with by geoff on 2006-07-26 at 05:38
geoff is offline   Reply With Quote