Agreed completely about everything you said Mike (Curtis). I could be wrong. There still might be some small gains even for > 1000 k's but if there is, I believe it is very small.

The main case that I can see for sieving 1000s of k's at once is the "set it and forget it" idea. If you have some machines that you rarely access or you just don't have time to monitor some of your machines except maybe once every few weeks, sieving a large # of k's at once makes more sense if not necessarily from a CPU perspective but from a personal time standpoint.

[QUOTE=Puzzle-Peter;368056]There are too many subranges in R3. Let's start consolidating. I'll take k=10M-20M to n=500k.[/QUOTE]

Done. 3 of 8 k values had a prime in this range.

10623044*3^456852-1
18067102*3^438604-1
19058092*3^439282-1

[QUOTE=Puzzle-Peter;369492]Done. 3 of 8 k values had a prime in this range.

10623044*3^456852-1
18067102*3^438604-1
19058092*3^439282-1[/QUOTE]
If I had realized those primes were there I would have finished it myself.:smile:

[QUOTE=gd_barnes;368122]Good idea Peter. Probably already obvious to you...next would be k=2.05G-2.5G for n=25K-100K. There's a sieve file started for that one.[/QUOTE]

Done. The results are being uploaded to sendspace right now.

1704 primes in that range.

One oddity: for 2138815474*3^34299-1 LLR aborted with "Giving up after 11 restarts", but it could be proven with PFGW. This particular k has another prime in the n=50k range, so that doesn't really help a lot.

[QUOTE=Puzzle-Peter;371158]Done. The results are being uploaded to sendspace right now.

1704 primes in that range.

One oddity: for 2138815474*3^34299-1 LLR aborted with "Giving up after 11 restarts", but it could be proven with PFGW. This particular k has another prime in the n=50k range, so that doesn't really help a lot.[/QUOTE]

Thank you Peter. I am getting an "unknown method" and "no files to extract" error when I try to extract or open the results file after downloading it from sendspace. Can you upload it again?

Can you also send me a list of primes? I'd appreciate it. It will be a big hassle for me to extract them from such a huge results file.

Uploaded again. This time I told 7zip to use a slightly less efficient but hopefully more widely known compression algorithm.

Number of primes went down to 1700 due to 2 k values with three primes each.

I just completed a 3-month effort (10 slower cores) to doublecheck all k<=1G to n=100K. I had expected there to be quite a few missing (or incorrect) small primes or additional (or missing) k's remaining as a result of the starting bases script not being available when R3 first started and the sheer number of k's that had to be searched. I was surprised by the results. I only found 5 missing primes; all for n>48K and 4 of which were from the first team drive of k<=100M for n=25K-100K. The missing primes were as follows:

26579848*3^65025-1
43857596*3^55729-1
54582146*3^48345-1
98931556*3^66585-1
987212552*3^59938-1


This means that there was not a single error for k=100M-900M for n<=100K and for all k<=1G for n<=48K. Excellent work everyone! :smile:

I have now updated the pages accordingly.

Interesting stats:
# k's <= 1G remaining at n=25K: 4903
# k's <= 1G remaining at n=100K: 1494
69.5% of k's found prime with a 4x increase in n-value.
(Equates to 45% of k's found prime with a doubling in n-value.)

Note that the # of k's remaining at n=100K is greater than what is shown on the pages since we've searched n>100K for some k-ranges.

Grueny has sent me a sieve file for k=4G-5G n=25K-100K. I have put a link to it on the reservations page.

Reserving r3 4G-5G to 100k

Lennart

starting a sieve r3 5G-6G 25k-100k

Lennart

starting a sieve r3 20M-4G 100k-500k

Lennart