[QUOTE=VBCurtis;393757]Why not process, say, 1000 k at a time? According to Gary, there is no speed improvement beyond a few hundred k's in a single sieve.[/QUOTE]

There is some improvement, but I don't know how much since I'm using the -x option. One advantage of sr2sieve is that it doesn't write the .abcd file when it finishes, so when I ctrl-C it doesn't appear to hang writing the output file which right now has over 17M tests in it.

I had meant to imply that there is little improvement beyond several hundred k. Likely there is some but it is negligible. The main reason for doing what Mark is doing is to start one big range and forget about it for extended periods. It takes a fair amount of personal time to divide it up into multiple 1000 k chunks and manage them all when there are so many k's. As we all experience on anything related to prime searching, it's a balancing act between CPU time and personal time.

Sure, makes sense. I had in mind doing, say, 4 pieces for 2 weeks at a time, allowing BOINC to test the files while sieving continued. I would be willing to do one of those 4 sieves if Mark cares. 2 months isn't all that long to outweigh the human time of splitting it up.

Yeah splitting it into four files of ~7000 k's each seems like a good tradeoff and then BOINC could start on the first one earlier.

I'll split the file and into smaller groups of k first and see if sr2sieve can handle that on Windows.

Reserving R79 to n=100k (25-100k) for BOINC

S95
 

S95 tested n=25K-50K

98 primes found - 433 remain

Results emailed - Base released

R78
 

reserving R78 with all remaining k´s (75)
n= from 50000 to 100000

[QUOTE=gd_barnes;396143][SIZE=1][FONT=Garamond]Also R19 would be a lot easier for me to manage.[/FONT] :grin:[/SIZE][/QUOTE]

You could [B]create an srsieve.out file[/B], by sieving n=1 to n=100 to p=10 for the k's remaining in your list at the noprimeleftbehind website. Once you recieve the primes from Reb, then you [B]create a file "pfgw.log"[/B], wich contains all the primes found by BOINC in the pfgw-form: k*79^n-1. Now you can use srfile to remove all the primed sequences from your srsieve.out file, by running this in commandline window: [B]srfile -d "pfgw.log" srsieve.out[/B] - this will make srfile remove the primed sequences and leave only the unprimed k's in the srsieve.out file. Now you can [B]open a spreadsheet[/B], [B]open srsieve.out[/B] in notepad, [B]copy[/B] all the lines in the srsieve.out file to the [B]spreadsheet (column A)[/B]. In the spreadsheet, mark all entries and [B]use the split text to columns function[/B] and in [B]others[/B] you put in a [B]*[/B]. Now [B]sort the column B[/B] in ascending order, followed by column A in ascending order. By doing this last operation, you should in your spreadsheet have a complete list of the k's remaining for R79.

Hope everything is clear, else ask. One thing I can say for sure (unless a step were missed) is that this worked very well when I used it on R3 :smile:

Take care

Kenneth

[QUOTE=KEP;396161]You could [B]create an srsieve.out file[/B], by sieving n=1 to n=100 to p=10 for the k's remaining in your list at the noprimeleftbehind website. Once you recieve the primes from Reb, then you [B]create a file "pfgw.log"[/B], wich contains all the primes found by BOINC in the pfgw-form: k*79^n-1. Now you can use srfile to remove all the primed sequences from your srsieve.out file, by running this in commandline window: [B]srfile -d "pfgw.log" srsieve.out[/B] - this will make srfile remove the primed sequences and leave only the unprimed k's in the srsieve.out file. Now you can [B]open a spreadsheet[/B], [B]open srsieve.out[/B] in notepad, [B]copy[/B] all the lines in the srsieve.out file to the [B]spreadsheet (column A)[/B]. In the spreadsheet, mark all entries and [B]use the split text to columns function[/B] and in [B]others[/B] you put in a [B]*[/B]. Now [B]sort the column B[/B] in ascending order, followed by column A in ascending order. By doing this last operation, you should in your spreadsheet have a complete list of the k's remaining for R79.

Hope everything is clear, else ask. One thing I can say for sure (unless a step were missed) is that this worked very well when I used it on R3 :smile:

Take care

Kenneth[/QUOTE]

I do a variation of that already in a spreadsheet using a vertical lookup function after parsing out k's that have been primed from a file of primes. But it takes a fair amount of time to save off huge results files, run that process, update k's remaining, update the top-10 list of primes for the base, doublecheck myself, etc.

All discussions of recommendations move to the recommendations thread

.[URL]http://www.mersenneforum.org/showthread.php?t=13196[/URL]