Sieving a Large Number of k's

[Dougal]

im doing alot of work on riesel base 79 and so far ive tested all k's less than 1,000,000 up to n=1024.im now taking all remaining k's up to n=10,000,ive been sieving in ranges of k1-50000.is there a faster way of doing this?

[gd_barnes]

Originally Posted by Dougal

im doing alot of work on riesel base 79 and so far ive tested all k's less than 1,000,000 up to n=1024.im now taking all remaining k's up to n=10,000,ive been sieving in ranges of k1-50000.is there a faster way of doing this?

That's probably close to the best way, assuming you are using srsieve. From a CPU efficiency perspective, the very fastest way is to use srsieve to sieve as many k's and as large of an n-range as you possibly can within the following constraints: The internal storage limit of srsieve, your computer's memory, and your ability to manipulate huge files reasonably across multiple machines. But that is not always the most effective way.

Srsieve is likely the only program that you will be able to use. The k's are too numerous and too large for sr2sieve. Your effort is, in effect, quite similar to our team Sierp base 63 effort, although not quite as large. In that one, Mark (rogue) set it up to sieve and test 10,000 k's at a time. It has 237,000+ k's remaining at n=1000. CPU efficiency could probably be gained there by sieving 50,000-100,000 k's at a time but the big constraint would be manipulation of huge files related to a team effort. The human/admin effort involved would become much larger. It's very difficult to post/send monsterous files back-and-forth. So the way Mark set it up is probably about as good as it could be for that particular team drive.

I personally have sieved as many as 150,000 k's at once using a fairly recent version of srsieve. That was only for n=1K-2K. The max may be less for a larger n-range since srsieve likely has an internal storage limit.


Gary