[QUOTE=Lennart;314871]I just thought about the search. In PG we search for that k right now. We will stop at 1 030 000 and move to lower k's with higher n's.

I could load that k up to 2M maybe 3M and have it done a bit faster.

Thoughts ?

Lennart


PS: I need a answer today because we will hit 1 030 000 tonight :)[/QUOTE]

That is a question for Gary.

I would post n=2M-5M whenever is convenient for you.

9267
 

Here is the first

Lennart

I've removed all even-n candidates from sieve file, so only 1101 tests remain.

k=9267 3M-4M

Lennart

I'm not sure I really get this. Does this mean we and PG are both testing the same candidates? Or did PG some sieving on these ranges but no LLR testing?

I am doing 9519*2^n-1 at n~7.5M (leading edge) and I would hate to hear I am duplicating work that has already bben done (or that somebody may duplicate my work).

[QUOTE=Puzzle-Peter;315091]I'm not sure I really get this. Does this mean we and PG are both testing the same candidates? Or did PG some sieving on these ranges but no LLR testing?

I am doing 9519*2^n-1 at n~7.5M (leading edge) and I would hate to hear I am duplicating work that has already bben done (or that somebody may duplicate my work).[/QUOTE]

There is no duplication of work. PrimeGrid is doing an extremely deep sieve for all k's < 10000 base 2 on both the Riesel and Sierp sides. I think they are sieving all n<=6M but may also be doing some n>6M.

As for testing, they are testing all (most) Sierp k<10000 base 2. (They are not testing Riesel.) In the last month, they passed n=1M. So no duplication.

If they are sieving n>6M It's possible that you may be able to benefit somewhat from their sieve if they've sieved that high.

Lennart could probably elaborate more about this.

[QUOTE=Puzzle-Peter;315091]I'm not sure I really get this. Does this mean we and PG are both testing the same candidates? Or did PG some sieving on these ranges but no LLR testing?

I am doing 9519*2^n-1 at n~7.5M (leading edge) and I would hate to hear I am duplicating work that has already bben done (or that somebody may duplicate my work).[/QUOTE]

PrimeGrid intends to test all Proth (+1) numbers for k < 10000. Their leading edge for testing is at about n=1M for the k that this search cares about.

What is key is that since PrimeGrid will be testing those k, they need to sieve them in advance. By using ppsieve and a coordinated effort, they are able to sieve much deeper and faster than any individual at CRUS.

Since PrimeGrid won't get to n=2M anytime soon for k=9267 (+1) and as they are sieving to n=9M (IIRC), we can use their sieve data to save our project from duplicating that work.

As I understand it, PrimeGrid is sieving both Proth and Riesel forms since the cost is minimal to sieve them concurrently. They can then feed their results to NPLB for testing Riesel (-1) numbers for k < 10000. This last statement could be wrong, so I'll leave that to someone else to correct me. If my statement is correct, then their sieve data could be used for k=9519 (-1), thus saving you any effort.

PrimeGrid is sieving both, Riesel and Proth sides from k = 3 to 9999.

Only LLRing the Proth side for k = 1201 to 9999.
(Note: LLR for k=5,7 to n=6M started 2 days ago).

Only some values LLRing on Riesel side: Riesel problem (55 k-values) and k=3.

Sieve files can be found:

rsp = Riesel Sieve Project:
for n = 1M to 4M [url=http://uwin.mine.nu/sieves/rsp/sievefile/]here[/url]
for n = 4M to 10M [url=http://home.comcast.net/~jimb_primegrid/rsp/]here[/url]

pps : Proth Prime Search:
for n = 1M to 4M [url=http://uwin.mine.nu/sieves/ppse/sievefiles/]here[/url]

[QUOTE=kar_bon;315147]PrimeGrid is sieving both, Riesel and Proth sides from k = 3 to 9999.

Only LLRing the Proth side for k = 1201 to 9999.
(Note: LLR for k=5,7 to n=6M started 2 days ago).

Only some values LLRing on Riesel side: Riesel problem (55 k-values) and k=3.

Sieve files can be found:

rsp = Riesel Sieve Project:
for n = 1M to 4M [URL="http://uwin.mine.nu/sieves/rsp/sievefile/"]here[/URL]
for n = 4M to 10M [URL="http://home.comcast.net/~jimb_primegrid/rsp/"]here[/URL]

pps : Proth Prime Search:
for n = 1M to 4M [URL="http://uwin.mine.nu/sieves/ppse/sievefiles/"]here[/URL][/QUOTE]

RSP -3M
RSP 3M-9M

PPS -3M

There are 3 sievefiles when we start in Boinc The first file was 0-3M and now we sieve 3M-6M next will be 6M-9M

We split them up to total of 9 files 1-1M 1M-2M 2M-3M ................................................................8M-9M

We manually sieve them to ¨1-2P before we sieve in Boinc. There we don't use any sievefile. We sieve datless.

6M-9M is now waiting to enter Boinc. It is now at 3.3P.

Lennart

OK, thanks for the clarification. So I could (and will) replace my candidates up to n=9M for k=9519 with the candidates from the PG sieve file. They have sieved much deeper than I did.