Choose your own K and work on finding a top-5000 prime!

[Cruelty]

k=1515 tested till n=1.04M
k=12345 tested till n=1.04M

I am still working on the above.

k=617 tested till n=1.2M - testing suspended
k=151515 tested till n=1.1M - releasing this k
k=736320585 tested till n=860k - testing suspended

[cipher]

Reserving 432383773965*2^n-1 Nash Weight 7788


Thanks
cipher

P.S: Special Thanks to Thomas11 for helping me find a "Real Heavy K"

[cipher]

432383773965 has 100 Primes between n=0 to n=50k

432383773965 3
432383773965 4
432383773965 6
432383773965 7
432383773965 10
432383773965 11
432383773965 12
432383773965 18
432383773965 19
432383773965 20
432383773965 23
432383773965 39
432383773965 47
432383773965 61
432383773965 64
432383773965 72
432383773965 73
432383773965 84
432383773965 93
432383773965 94
432383773965 109
432383773965 123
432383773965 127
432383773965 141
432383773965 145
432383773965 161
432383773965 163
432383773965 183
432383773965 223
432383773965 227
432383773965 232
432383773965 259
432383773965 321
432383773965 322
432383773965 326
432383773965 329
432383773965 337
432383773965 340
432383773965 349
432383773965 368
432383773965 426
432383773965 501
432383773965 551
432383773965 553
432383773965 801
432383773965 802
432383773965 890
432383773965 902
432383773965 928
432383773965 1067
432383773965 1106
432383773965 1379
432383773965 1444
432383773965 1499
432383773965 1557
432383773965 1698
432383773965 1774
432383773965 2207
432383773965 2426
432383773965 2606
432383773965 3142
432383773965 3313
432383773965 3622
432383773965 3773
432383773965 3789
432383773965 4103
432383773965 4482
432383773965 4573
432383773965 5445
432383773965 5551
432383773965 5616
432383773965 6188
432383773965 6408
432383773965 7532
432383773965 7593
432383773965 8072
432383773965 8365
432383773965 8571
432383773965 8878
432383773965 9133
432383773965 9870
432383773965 9972
432383773965 12569
432383773965 13543
432383773965 16515
432383773965 17635
432383773965 19586
432383773965 21595
432383773965 21784
432383773965 22184
432383773965 26754
432383773965 29867
432383773965 30984
432383773965 32862
432383773965 36069
432383773965 37967
432383773965 38113
432383773965 40197
432383773965 45654
432383773965 48611

Next update when i reach 250k or sooner.

[Cruelty]

k=1515 tested till n=1.05M
k=12345 tested till n=1.05M

I am still working on those.

[gd_barnes]

I think this is as good a place as any to post this:

NPLB will be pulling out active reserved k's from it's upcoming k=2000-3400 efforts. I have sent PM's to 3 people about some reservations that have not had a search depth reported for 6-12 months to verify that they are still working on them.

Please see details in the applicable posting in our thread. If any problems, please let me know. Best would be to reply in that thread or to send me a PM.


Thanks,
Gary

[Svenie25]

Originally Posted by lavalamp

Personally I now start sieving with srsieve, then move onto sr1sieve. This is because NewPGen misses factors and is no longer being actively developed.

If you're on windows, you could put these commands in a batch file and run it to create a sieve:

srsieve -v -g -m 4e15 -P 1e10 -n 0 -N 101419 "15431*2^n-1"
 
ren "t17_b2_k15431.npg" "15431_sieve.txt"
 
pause

That command will sieve 15431*2^n-1 for n up to 101,419 which is an Intel CPU FFT jump point. If you have an AMD CPU under the hood, you might want to change that to 102,919 for the corresponding AMD jump point. It will sieve to a depth of 10 billion, at which point I would recommend switching to sr1sieve.


The ren command simply renames the output file to something more friendly, and the pause command holds the batch file window open when it's done so that you can read any output.

If you want to know what all of the other switches in that command line do, just run srsieve -h, and likewise sr1sieve -h for sr1sieve.

Sieving and LLRing a k up to n=101,419 (or n=102,919) will take maybe 1 - 2 weeks *, depending on how fast your CPU is and how many candidates are removed while sieving.

To get those jump points I used llrtools.

* completely wild guess

That sounds very intreting, but have you also a commandline for using sr2sieve in the second step? ;)

I have the following idea:

sr1sieve-x86_64-windows -t 2 -A 1 -A 2 -p 1e10 -P 30e12 -i 15431_sieve.txt -o 15431_sieve_2.txt

The filename is to keep your example. Would this line work?

[VBCurtis]

sr2sieve is for an entirely different sort of sieving, with many k values. To sieve a single k, the directions are as good as need be, and sr2 is of no help. Sr2 becomes useful when one wishes to sieve 3+ k values on identical ranges of n; its command line is similar to sr1, or it can be run from a sr2work file listing the ranges of n to search in billions.

I don't know what the -t and -A flags are for sr1, and I've used it many many times; what are you trying to do that's nonstandard with those?
edit- if they are for threads, I would like to know if much speed is gained over just running two instances of sr1sieve; I've always done without threads, but if it's faster....

-Curtis

[mdettweiler]

Originally Posted by VBCurtis

I don't know what the -t and -A flags are for sr1, and I've used it many many times; what are you trying to do that's nonstandard with those?
edit- if they are for threads, I would like to know if much speed is gained over just running two instances of sr1sieve; I've always done without threads, but if it's faster....

Yes, -t is for threads (linux only AFAIK). I don't know what -A is.

As far as speed differences with multithreading, generally sieving is a bit faster *without* multithreading. The overhead for communication between threads is not much, but it is present nonetheless. I don't know what the exact % figures are, though, since I rarely use multithreading myself. The main benefit of multithreading is a smaller memory footprint; this can be quite useful if dealing with a very large sieve file.

[Svenie25]

Okay thanks a lot guys. So I will use two instances instead of -t switch. (Doesn´t function under windows, too) -A is affinity.

Sorry, but I´m a "DAU". But after finishing srsieve I let run 2 instances of sr1sieve with the factors as output. How can I get the factors out of the sievefile after finishing?

[Thomas11]

Originally Posted by Svenie25

How can I get the factors out of the sievefile after finishing?

Just by using srfile (it comes with srsieve):

srfile -k factors.txt sievefile.txt

You can also use it to convert between the different file types (e.g. srsieve.out, abcd, NewPGen, etc. ...)

[Cruelty]

k=1515 tested till n=1.07M
k=12345 tested till n=1.07M

I am still working on those.