Range for k>300 that Peter Benson has searched - Page 4

gd_barnes · 2009-01-17, 09:28

For historical reference, I thought it'd be good to go ahead and attach the spreadsheet that has the analysis of Peter's results files.

Gary

gd_barnes · 2009-01-19, 12:32

I have now recieved all of the results that Peter has for the range of k=1005-1400 for n=200K-479264.

After extensive analysis and some additional info. from Peter, I have determined he following:

1. There are no "n-range holes" in the n=200K-300K range. There are only a few large missing results gaps outlined below.
2. There are a number of small "n-range holes" for all k's for n=335295-479264 but it's clear by looking at primes reported that the entire range has been searched.
3. k=1009 was NOT searched for n=200K-300K and hence has no results.
4. k=1201 was NOT searched for n=200K-219833 and hence has no results.
5. k=1200-1300 for n=200K-210646 was searched but the results are missing.
6. k=1300-1400 for n=300K-335294 was searched but the results are missing.

Total k/n pairs in sieve file without a corresponding result: 142,978.
Total k/n pairs in sieve file: 2,076,554.
6.89% missing.

Conclusions:
1. Pull k=1005-1400 for n=200K-350K out of our 9th drive as we did our 8th drive.
2. Load the missing k/n pairs into 2 temporary servers, one each for n=200K-350K and n=350K-479264.

#2 will be a very informal effort with no drive page nor separate thread created. I'll just send David the files and mention the servers in the news thread when I know for sure that Peter has no more results. When people have a spare core here and there, they can run it for a while.

David, since these will be temp. servers, I don't how you want to handle them on your pages. I don't want anyone to go to much work for this and so don't much how care it is handled. Of course it should count in the stats but if there's no separate hourly stats then it's not a big deal.

I have sent an Email to Karsten with my suggestions for updating ranges at Rieselprime.org. I have also Emailed all of Peter's results files as well as the missing k/n pairs files to both Max and Karsten.

With what we are doing here, we will not only confirm that the entire k=1005-1400 range is searched, we will also be searching a small fraction of what we would have otherwise and have first-pass residuals for all k/n pairs. This should help guarantee that no prime is left behind with minimal CPU effort and hopefully minimal administrative effort for everyone.

Whew...I'm glad this is done. Thank gish I wasn't out of town this week!

Gary

IronBits · 2009-01-19, 12:42

2 more ports eh?
Port 2000
Port 3000
Send me the files and I'll put them on line and add them into the works.

Do you want/need the email auto-notify function active for these two ports?

gd_barnes · 2009-01-19, 14:04

Quote:

Originally Posted by IronBits

2 more ports eh?
Port 2000
Port 3000
Send me the files and I'll put them on line and add them into the works.

Do you want/need the email auto-notify function active for these two ports?

2 more ports only temporarily. 2000 & 3000 are good choices I think.

It will be about a day yet. I want to make sure that I don't get any more results from Peter. If I do, I'll be able to yank some pairs out of the files that I'll send you.

G

henryzz · 2009-01-19, 17:59

i have results for 1009 up to n=500k

gd_barnes · 2009-01-20, 00:38

Quote:

Originally Posted by henryzz

i have results for 1009 up to n=500k

Oh, very good. Starting from what n-range? We need n=200K-300K.

Please attach them here when you get a chance. I'll then do a quickie compare vs. our sieve file. If everything is there, I'll pull it out of the file that we'll be loading into one of the temporary ports.

Mike, I saw your earlier post about having results for k=1991. It would be too messy to pull 1 or 2 k's here and there out of the drives so we'll still go ahead and search it. But for the huge range of k=1005-1400 that we now have ~95% of the results for, we definitely wanted to pull that entire range out.

To take the above a step further for k=1005-1400, in addition to what Henry has offered, if anyone has results for any of the other missing ranges, please post them here.

The main thing that I want to accomplish over the next few months is having 1st-pass results on all k=1005-1400 for n=200K-479264. Later on, Peter should forward us n>479264.

Gary

henryzz · 2009-01-20, 07:27

Quote:

Originally Posted by gd_barnes

Oh, very good. Starting from what n-range? We need n=200K-300K.

Please attach them here when you get a chance. I'll then do a quickie compare vs. our sieve file. If everything is there, I'll pull it out of the file that we'll be loading into one of the temporary ports.

Mike, I saw your earlier post about having results for k=1991. It would be too messy to pull 1 or 2 k's here and there out of the drives so we'll still go ahead and search it. But for the huge range of k=1005-1400 that we now have ~95% of the results for, we definitely wanted to pull that entire range out.

To take the above a step further for k=1005-1400, in addition to what Henry has offered, if anyone has results for any of the other missing ranges, please post them here.

The main thing that I want to accomplish over the next few months is having 1st-pass results on all k=1005-1400 for n=200K-479264. Later on, Peter should forward us n>479264.

Gary

i have n=1 to n=500k
amphoria might have more higher up
i will post the file this evening

gd_barnes · 2009-01-20, 13:13

Quote:

Originally Posted by henryzz

i have n=1 to n=500k
amphoria might have more higher up
i will post the file this evening

I had previously mentioned that we only needed n=200k-300k. After looking a little closer at what we have, we could also use it all the way up to n=500k. So if you can post your entire results file for n=1 to 500k, that would help.

Thanks,
Gary

henryzz · 2009-01-20, 16:18

Quote:

Originally Posted by gd_barnes

I had previously mentioned that we only needed n=200k-300k. After looking a little closer at what we have, we could also use it all the way up to n=500k. So if you can post your entire results file for n=1 to 500k, that would help.

Thanks,
Gary

here it is
only just small enough

gd_barnes · 2009-01-20, 17:01

Quote:

Originally Posted by henryzz

here it is
only just small enough

Excellent! David, after I verify these vs. our sieve file for the 9th drive later today, I'll send you files to load up temp. ports 2000 and 3000. Then, I THINK we'll finally be set for a week or more on both the 8th and 9th drives!

Thanks, Henry, for digging up this old stuff. I know how things have a tendency to just disappear after a while.

Gary

2009-01-19, 12:42	#36
IronBits I ♥ BOINC! Oct 2002 Glendale, AZ. (USA) 3·7·53 Posts	2 more ports eh? Port 2000 Port 3000 Send me the files and I'll put them on line and add them into the works. Do you want/need the email auto-notify function active for these two ports? Last fiddled with by IronBits on 2009-01-19 at 12:56

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Peter Montgomery's Thesis	mickfrancis	Computer Science & Computational Number Theory	3	2015-06-25 14:32
Peter Montgomery (IMPORTANT)	R.D. Silverman	Factoring	8	2014-06-07 18:43
Types of primes searched for.	3.14159	Miscellaneous Math	2	2010-12-04 13:09
Benson's prime search	Kosmaj	Riesel Prime Search	67	2009-01-18 21:59
k's/n-ranges not searched for team drives	gd_barnes	No Prime Left Behind	20	2008-12-26 08:13

2009-01-19, 12:32	#35
gd_barnes May 2007 Kansas; USA 3³×5×7×11 Posts	I have now recieved all of the results that Peter has for the range of k=1005-1400 for n=200K-479264. After extensive analysis and some additional info. from Peter, I have determined he following: 1. There are no "n-range holes" in the n=200K-300K range. There are only a few large missing results gaps outlined below. 2. There are a number of small "n-range holes" for all k's for n=335295-479264 but it's clear by looking at primes reported that the entire range has been searched. 3. k=1009 was NOT searched for n=200K-300K and hence has no results. 4. k=1201 was NOT searched for n=200K-219833 and hence has no results. 5. k=1200-1300 for n=200K-210646 was searched but the results are missing. 6. k=1300-1400 for n=300K-335294 was searched but the results are missing. Total k/n pairs in sieve file without a corresponding result: 142,978. Total k/n pairs in sieve file: 2,076,554. 6.89% missing. Conclusions: 1. Pull k=1005-1400 for n=200K-350K out of our 9th drive as we did our 8th drive. 2. Load the missing k/n pairs into 2 temporary servers, one each for n=200K-350K and n=350K-479264. #2 will be a very informal effort with no drive page nor separate thread created. I'll just send David the files and mention the servers in the news thread when I know for sure that Peter has no more results. When people have a spare core here and there, they can run it for a while. David, since these will be temp. servers, I don't how you want to handle them on your pages. I don't want anyone to go to much work for this and so don't much how care it is handled. Of course it should count in the stats but if there's no separate hourly stats then it's not a big deal. I have sent an Email to Karsten with my suggestions for updating ranges at Rieselprime.org. I have also Emailed all of Peter's results files as well as the missing k/n pairs files to both Max and Karsten. With what we are doing here, we will not only confirm that the entire k=1005-1400 range is searched, we will also be searching a small fraction of what we would have otherwise and have first-pass residuals for all k/n pairs. This should help guarantee that no prime is left behind with minimal CPU effort and hopefully minimal administrative effort for everyone. Whew...I'm glad this is done. Thank gish I wasn't out of town this week! Gary

2009-01-19, 17:59	#38
henryzz Just call me Henry "David" Sep 2007 Cambridge (GMT/BST) 2³·3·5·7² Posts	i have results for 1009 up to n=500k