News

[Lennart]

Yep

And her is next


211195*2^3224974+1

Lennart

[Mathew]

All,

The bad sieve file for S35 as stated here, most likely came from a bad version of srsieve (v 0.7.0). These problems are gone in more recent versions.

Please ensure you do not have this older version when conducting sieving.

[gd_barnes]

Thanks for posting that Mathew.

[gd_barnes]

In case people missed the discussion, our PRPnet port 1400, which was previously running our 51 bases drive, is now running Sierp base 6 for n=700K to 1M. Come join the fun and find some huge primes!

[gd_barnes]

Kenneth (KEP) came with an excellent way to send me a huge file. He used "Google Drive". The file that he sent me was all of the primes, k's remaining, etc. for his R3 reservation for k=11G-12G to n=25K. It was over 840 MB compressed. I was able to download it in 5-10 mins. with no ads, no size restrictions, no # of download limits in a day, i.e. none of the hassles of sendspace.

If you are sending me files for any new reservation (i.e. n=1 to xxx) with a huge number of primes for bases 3, 7, or 15, and the file is larger than Google's 25 MB Email limit, I would suggest Google Drive.


Gary

[mdettweiler]

Yeah, Google Drive is very good for that - it's basically a cloud storage service similar to Dropbox, Microsoft SkyDrive, Box.com, etc. (if others are familiar with those). Generally they're intended for personal storage of files and synchronization across multiple devices; but they usually offer a "share link" option, which is what KEP used to send Gary his large file.

A similar solution I've seen a lot of people use here is to post files on a personal web space they own. I've used the noprimeleftbehind.net web site for this a couple of times in the past. Services like Google Drive and Dropbox basically just automate this in a nice user-friendly system (plus of course they're paying for the hosting).

I would note that with these services you do have a finite amount of space - so, if you're using it to transfer large files, you'll eventually need to go in and delete the files manually to free up room for future uploads. Just make sure you leave it up long enough for the person on the other end to download it - kind of similar to how Sendspace puts things up for 30 days.

[gd_barnes]

I have just completed a large update to the pages. You might check all of your statuses and reservations since about Nov. 1st to see if I missed anything.

[gd_barnes]

I am happy to announce that CRUS now has a new co-moderator. In addition to Ian (MyDogBuster) and me, Mathew has now joined our ranks. He has been a helpful and accurate contributor to the project for (I believe) 3+ years now.

Welcome to "our world" Mathew!

[gd_barnes]

There is an unplanned NPLB/CRUS server outage. See here.

Sorry everyone.

[gd_barnes]

All CRUS servers are back up. I have temporarily set the public and private PRPnet servers to 7 days for a couple of days to allow everyone to return their completed work.

I'm very sorry about the extended outage.

[gd_barnes]

If you are starting a new base, please do not use PFGW versions 3.6.0 thru 3.6.3. We have identified a problem that causes the program to output incorrect primes and miss other primes for n<=5 (or < ~40 bits).

To the best of our knowledge, these "bad" versions should still be OK for higher searches. We also believe that later and earlier versions of PFGW are OK for all n-ranges but we are still in the process of making that determination. In the mean time, whenever anyone submits work on a new base, I will ask what PFGW version was used to start the base.

Phillip previously identified bases S66 and S79 as having this issue, which alerted us to the problem. Mark has kindly done some programming that will allow us to not start the bases completely over. Ian and I have doublechecked these to n=1000 to give Mark a new starting point for determining what k's are "actually" remaining. With the "bad" primes gone, he will also merge the new smaller primes and old bigger primes together into one good master file of primes. What you will see in the future is a team drive to search a few 1000 k's for n=2.5K/5K to 25K/50K for these bases (and possibly some others). These will be k where the bad PFGW incorrectly found a small prime (and we found no prime for n<=1000 in our doublecheck) and so were never searched higher so we have to "catch them up" with the search limit for the base.

Edit: I have doublechecked R35/S35/R79 to n=100. No problems found. R51/S51 likely don't have the issue because they were started in 2010 long before PFGW 3.6.x was released but I will still doublecheck a small k-range of them.