Riesel base 6 - team drive #4 - EIGHT OR BUST!

[gd_barnes]

Originally Posted by Batalov

Yes, sorry. Correct.

P.S. The k=1597 is at n>725K already. I'll extend it until 1M, if you don't mind.

That's fine with me. Do you have the file for n=750K-1M? If not, the latest sieve file is in the Riesel base 6 sieving thread.

We may be doing some additional sieving for n=750K-1M but with only 5 k's remaining now and only ~30% of the file remaining, I'm going to guess that the optimum depth for all 5 k's will be barely higher than what we are at right now. And since k=1597 is by far the lowest weight of the group, it's probably just as well to test it up to n=1M anyway and remove it from the big sieve.

[Batalov]

Exactly, that was my motivation, too. The sparse k=1597 set is less effective for sieving than other 4 k's (or is it? I may be wrong, though, simplistically thinking, because the sieve makes it less sparse in the mod 6^m space; gotta run some offline tests now; interesting).
And yes, I've generated the file from ABCD; too easy.

[gd_barnes]

Originally Posted by Batalov

Exactly, that was my motivation, too. The sparse k=1597 set is less effective for sieving than other 4 k's (or is it? I may be wrong, though, simplistically thinking, because the sieve makes it less sparse in the mod 6^m space; gotta run some offline tests now; interesting).
And yes, I've generated the file from ABCD; too easy.

Correct, k=1597 is less effective for sieving the small part of the rest of the file so if there is any efficiency lost by not sieving it further with the other 4 k's, it is likely to be quite small. That said, when we start the next big sieve for the range...likely to be something along the lines of n=1M-4M, we'll definitely want to include k=1597 in it and not break it out separately. In other words:

This:

Sieveing all 5 k's at once using sr2sieve.

Is probably more efficient than this:

Sieving 4 k's at once with sr2sieve -and-
Sieving only k=1597 with sr1sieve.

Sr1sieve is extremely fast but past testing has shown that it's only faster than sr2sieve for 1 or 2 k's. That is, it's faster to sieve 2 k's using 2 instances of sr1sieve than it is to sieve both k's together using 2 instances of sr2sieve but at different sieve depths. But for 3 k's or more, sr2sieve is always faster. That's because sr2sieve is able to gain efficiency by sieving several k's of the same base together.

That said again (lol)...I haven't tested it where one k is much lower weight than 4 other k's. When we get to that point, since the sieving effort will be so large, we should probably test it to be sure.


Gary

[Batalov]

514-520K done, 0 primes.

+2 more from 749K, while testing speed:
36772*6^749907-1 is composite: RES64: [440E594E759E87D4] (7572.9323s+0.0361s)
77743*6^749913-1 is composite: RES64: [71BC61E2807EDAD7] (7558.3327s+0.0366s)

[Batalov]

k=1597, 514-750K done, 0 primes.

[Batalov]

Alarming apathy, guys!
Take something. Now.

[Dougal]

taking 514-515k,should take a week,hopefully this will get the ball rolling again.

[Batalov]

Originally Posted by Batalov

514-520K done, 0 primes.

...see above. Please take above 520...

[gd_barnes]

I guess I should remove n=514K-520K from the posted file. I'll do that in a little bit.

Before making reservations, everyone please be sure and check the completion and reservation status in the 1st post. I can't guarantee that I'll always quickly remove the lower reserved ranges from the posted file. That's part of the reason in having a large posted file...so multiple people can reserve whatever size they want.

[Dougal]

sorry,taking 520-521

[Dougal]

520-521 complete no primes

520-521.txt