Doublecheck drive #1: k=3-1001 n=100K-260K - Page 21

mdettweiler · 2010-04-17, 20:15

221-239 completed by Gary, no surprises; results filed.

mdettweiler · 2010-04-27, 06:12

Reserving 261-299 for Gary.

gd_barnes · 2010-04-27, 07:27

Good idea Max and thanks for keeping up with this. BTW, are you just checking the primes vs. Karsten's k<300 page? I suppose it might also be a good idea to check them vs. RPS's k<300 page. I don't know if we know for sure that those two pages are exactly the same for n=100K-260K.

It's nice to see that all of my machines are still crunching away while I'm out of town. Thank gish for LLRnet!

kar_bon · 2010-04-27, 10:49

Quote:

Originally Posted by gd_barnes

BTW, are you just checking the primes vs. Karsten's k<300 page? I suppose it might also be a good idea to check them vs. RPS's k<300 page. I don't know if we know for sure that those two pages are exactly the same for n=100K-260K.

Yes, this would be nice because I got no results from here and so no primes found to check!

mdettweiler · 2010-04-27, 15:23

Quote:

Originally Posted by gd_barnes

Good idea Max and thanks for keeping up with this. BTW, are you just checking the primes vs. Karsten's k<300 page? I suppose it might also be a good idea to check them vs. RPS's k<300 page. I don't know if we know for sure that those two pages are exactly the same for n=100K-260K.

It's nice to see that all of my machines are still crunching away while I'm out of town. Thank gish for LLRnet!

Quote:

Originally Posted by kar_bon

Yes, this would be nice because I got no results from here and so no primes found to check!

Yes, I'm checking all these results against Karsten's k<300 page, hence the "no surprises" that I've stated for each batch of results. You're right, I do suppose it would be good to check them against RPS's k<300 page as well...I'll do that for new ranges and catch up on the old ones as soon as I get the chance.

@Karsten: do you want me to send you the results from these, or is it sufficient that I'm checking them against your pages?

kar_bon · 2010-04-27, 18:18

Quote:

Originally Posted by mdettweiler

@Karsten: do you want me to send you the results from these, or is it sufficient that I'm checking them against your pages?

Yes, I would be good to send them to me, too. The latest results from here were for k=159 (Aug.2009). I'll make an overview the next days which results i got so far.

Take it also as a backup of those results.

gd_barnes · 2010-04-27, 19:20

Karsten,

Yes, it is a good idea that at least 2 of us have results at all times if possible.

You and I have been making sure we have everything for the other drives so it's good that you and Max have all results for this drive. Also, as requested by Max, I've been putting all results, including those for manually done ranges, on Jeepford's desktop, my server machine.

Karsten, I know that Max is very busy with his personal life. If you have a somewhat automated way to check the primes against RPS's k<300 page, that would be good. -or- If you know that the primes on your page for this range exactly match what RPS's page shows, then let us know. Then I won't suggest that either of you check RPS's page.

Max,

I know you're busy but as a "medium" priority at this point, could you set up Karsten to have access to Jeepford? I think you did this for Mark at some point. Karsten mentioned this at least once before. It would be good to have someone else who could also load server ranges when I'm out of town and when you're busy. Also, with ALL of the results for most NPLB drives on Jeepford including manually done ranges, that effectively gives a comprehensive set of ALL results for him to have access to that will ultimately be loaded into the DB. Then at some point, if he wishes, he can check ALL the primes and results in the DB vs. the original sieved files.

It would be really cool to have everything cross referenced and checked in the DB for all ranges; or at least as many as possible.

On another note, I'm looking forward to doublechecking k=300-1001. There should be lots of errors there. RPS and others had done a lot of doublechecking for n<260K already on k<300 with just a few holes here and there so we effectively did a triplecheck here on a lot of k<300. But k=300-1001 has never been doublechecked and it is from the original PrimeSearch project, which had many errors in my personal doublecheck of n<100K.

Thanks,
Gary

mdettweiler · 2010-04-27, 22:07

Quote:

Originally Posted by gd_barnes

Karsten,

Yes, it is a good idea that at least 2 of us have results at all times if possible.

You and I have been making sure we have everything for the other drives so it's good that you and Max have all results for this drive. Also, as requested by Max, I've been putting all results, including those for manually done ranges, on Jeepford's desktop, my server machine.

Okay, I'm uploading all my doublecheck results to a folder within jeepford's "results files" folder on its desktop. On my own computer, it's arranged like this:
-All results for this drive are within one folder.
-Each individual range that someone sent me or I processed and filed myself is contained (lresults and prime log) within an archive file. For example, the range of k=3-7 is in file 3-7.7z. (I have a rather large mix of archive types within the folder; some are .7z, some are .rar, and others are .tar.lzma. At any rate, though, all should be openable by a free archiving tool like 7-zip, or by the File Roller tool which they all open in on jeepford.)

I'm effectively mirroring my local directory on jeepford; this way I can continue to update it periodically by just SCP'ing all files in my local folder to the server folder.

As far as Karsten accessing these files, see below...

Quote:

Karsten, I know that Max is very busy with his personal life. If you have a somewhat automated way to check the primes against RPS's k<300 page, that would be good. -or- If you know that the primes on your page for this range exactly match what RPS's page shows, then let us know. Then I won't suggest that either of you check RPS's page.

Max,

I know you're busy but as a "medium" priority at this point, could you set up Karsten to have access to Jeepford? I think you did this for Mark at some point. Karsten mentioned this at least once before. It would be good to have someone else who could also load server ranges when I'm out of town and when you're busy. Also, with ALL of the results for most NPLB drives on Jeepford including manually done ranges, that effectively gives a comprehensive set of ALL results for him to have access to that will ultimately be loaded into the DB. Then at some point, if he wishes, he can check ALL the primes and results in the DB vs. the original sieved files.

Yes, I can do that...what would probably be best is to set Karsten up for VNC access into the "gary" username like I have, since that's needed in order to do anything with the servers. Just to confirm, would you be OK with that?

Quote:

It would be really cool to have everything cross referenced and checked in the DB for all ranges; or at least as many as possible.

Indeed. The idea is that in the end (hopefully sometime soon now

) we'll have a functional firstpass and doublecheck DB which, combined, will contain a full record of all pairs ever tested by NPLB. We can then write scripts with relative ease to query the database and check whether all pairs in a sieve file are present, and then check whether they all have a matching doublecheck residual. Additionally, we can write scripts to have the DB output a particular range to a text file in whatever standard format we'd like (LLR, LLRnet, PFGW, etc.), sorted however we like (by k, by n, etc.) With all that, the task of manually processing server results should be able to go bye-bye.

In fact, we may even be able to take it a step further and set it up so that anybody, not just the admins, can access a web page to download the results for whatever range they like. In keeping with our general philosophy, it would be great to have our results publicly accessible for anybody who wants to do research on them or anything like that.

Another cool idea that I've toyed with in that same vein is that of having people be able to submit manual results via the web page for direct importing into the DB, rather than having to have the admins dump them in in largish chunks. Of course, that opens a whole new can of worms in that we'd have to have some way to make sure no bogus results can be submitted and blindly accepted...anyway, we can cross that bridge when we come to it.

Quote:

On another note, I'm looking forward to doublechecking k=300-1001. There should be lots of errors there. RPS and others had done a lot of doublechecking for n<260K already on k<300 with just a few holes here and there so we effectively did a triplecheck here on a lot of k<300. But k=300-1001 has never been doublechecked and it is from the original PrimeSearch project, which had many errors in my personal doublecheck of n<100K.

Yes, I have to wonder if we'll find any errors there...as it is so far, the only errors this drive has turned up have been in the k=300-1001 range. We may well finish the k<300 range without finding any errors due to RPS's having previously touched up especially suspect areas. (But this time around, we'll have real, "tangible" residuals for it all!

)

gd_barnes · 2010-04-28, 07:19

Yes, that is perfectly fine if Karsten has VNC access to the Gary username. It's only a crunching and server machine.

Oh but wait...I have to get rid of a couple of files where I call Karsten bad names...

OK done.

NOW you can give him VNC access to the Gary username.

mdettweiler · 2010-04-28, 18:10

Quote:

Originally Posted by gd_barnes

Yes, that is perfectly fine if Karsten has VNC access to the Gary username. It's only a crunching and server machine.

Oh but wait...I have to get rid of a couple of files where I call Karsten bad names...

OK done.

NOW you can give him VNC access to the Gary username.

Okay, sounds good. I'll do that as soon as I get the chance.

mdettweiler · 2010-04-30, 14:28

241-259 completed by Gary, checked against rieselprime.de and 15k.org, no surprises; results filed.

2010-04-27, 06:12	#222
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3×2,083 Posts	Reserving 261-299 for Gary. Last fiddled with by mdettweiler on 2010-04-27 at 06:13 Reason: heck, let's just take the rest of the k<300 range and get it over with... :-)

2010-04-27, 19:20	#227
gd_barnes May 2007 Kansas; USA 3³·5·7·11 Posts	Karsten, Yes, it is a good idea that at least 2 of us have results at all times if possible. You and I have been making sure we have everything for the other drives so it's good that you and Max have all results for this drive. Also, as requested by Max, I've been putting all results, including those for manually done ranges, on Jeepford's desktop, my server machine. Karsten, I know that Max is very busy with his personal life. If you have a somewhat automated way to check the primes against RPS's k<300 page, that would be good. -or- If you know that the primes on your page for this range exactly match what RPS's page shows, then let us know. Then I won't suggest that either of you check RPS's page. Max, I know you're busy but as a "medium" priority at this point, could you set up Karsten to have access to Jeepford? I think you did this for Mark at some point. Karsten mentioned this at least once before. It would be good to have someone else who could also load server ranges when I'm out of town and when you're busy. Also, with ALL of the results for most NPLB drives on Jeepford including manually done ranges, that effectively gives a comprehensive set of ALL results for him to have access to that will ultimately be loaded into the DB. Then at some point, if he wishes, he can check ALL the primes and results in the DB vs. the original sieved files. It would be really cool to have everything cross referenced and checked in the DB for all ranges; or at least as many as possible. On another note, I'm looking forward to doublechecking k=300-1001. There should be lots of errors there. RPS and others had done a lot of doublechecking for n<260K already on k<300 with just a few holes here and there so we effectively did a triplecheck here on a lot of k<300. But k=300-1001 has never been doublechecked and it is from the original PrimeSearch project, which had many errors in my personal doublecheck of n<100K. Thanks, Gary Last fiddled with by gd_barnes on 2010-04-27 at 19:22

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Team drive #14: k=600-1001 n=1M-2M	mdettweiler	No Prime Left Behind	10	2021-03-13 22:32
GPU sieving drive for k<=1001 n=1M-2M	mdettweiler	No Prime Left Behind	11	2010-10-04 22:45
Doublecheck drive #2: k=300-400 n=260K-600K	mdettweiler	No Prime Left Behind	0	2010-05-21 00:22
Team drive #3: k=300-400 n=260K-600K	gd_barnes	No Prime Left Behind	255	2008-11-12 10:43
Team drive #2: k=400-1001 n=260K-333.2K	gd_barnes	No Prime Left Behind	154	2008-03-31 02:59

2010-04-17, 20:15	#221
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3×2,083 Posts	221-239 completed by Gary, no surprises; results filed.

2010-04-27, 07:27	#223
gd_barnes May 2007 Kansas; USA 10100010011011₂ Posts	Good idea Max and thanks for keeping up with this. BTW, are you just checking the primes vs. Karsten's k<300 page? I suppose it might also be a good idea to check them vs. RPS's k<300 page. I don't know if we know for sure that those two pages are exactly the same for n=100K-260K. It's nice to see that all of my machines are still crunching away while I'm out of town. Thank gish for LLRnet!

2010-04-28, 07:19	#229
gd_barnes May 2007 Kansas; USA 3³×5×7×11 Posts	Yes, that is perfectly fine if Karsten has VNC access to the Gary username. It's only a crunching and server machine. Oh but wait...I have to get rid of a couple of files where I call Karsten bad names... OK done. NOW you can give him VNC access to the Gary username.

2010-04-30, 14:28	#231
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3·2,083 Posts	241-259 completed by Gary, checked against rieselprime.de and 15k.org, no surprises; results filed.