Sieving for CRUS

[rebirther]

Originally Posted by gd_barnes

I feel like n=100K-200K for the base 3's would probably work for BOINC as long as we limit the file size to ~2000-3000 k's. For example there are 1852 k's remaining for R3 k=48G-49G at n=100K after the recent work done by Mark. Such k=1G ranges at n=100K may be something that BOINC could work with going forward. But...see below.

I think n=10K-25K would not work for BOINC for any base. It's just too inefficient to put them into most any server including individual PRPnet servers. There's quite a bit of overhead associated with sending each test out.

Reb's servers do not automatically remove k's as primes are found. He has to remove them himself. This makes work on small n with lots of k's remaining inefficient and more labor intensive.

I think this comes down to general project management and direction. I'm not fond of having BOINC work on small tests like this. They are much better suited for medium and huge tests...namely knocking bases down to 1, 2, or 3 k's remaining or proving them. :-)

Reb, can you weigh in on this again?

All bases with <100k left would be fine. You can also add all to the recommended list as now. I need a good balance with low, medium, long runs to keep up the available work.

[gd_barnes]

Originally Posted by rebirther

All bases with <100k left would be fine. You can also add all to the recommended list as now. I need a good balance with low, medium, long runs to keep up the available work.

All bases with <400k left have been tested to n=100K. That was one of our goals a few years ago.

A bunch of 11k and 12k bases tested to n=100K are being sieved right now. I'll add them to the recommended as the sieve files become available.

I feel like you have a good batch of reservations for medium and long tests right now especially with those long R2 tests on a big file.

No real short tests though...unless you count n=100K as short tests. This begs the question:

Would BOINC like any testing starting from n=25K or 50K in the future? That would require testing more than 400 k's. A way around that would be to break them up into pieces like we do with base 3 but in much smaller pieces. For instance for a base with 500 k's, we could send you 5 files in pieces of 100 k's each.

I only mention this because the n=25K-100K testing that BOINC did several years ago seemed to be fairly popular because the bases were being reserved and finished fairly quickly.

[rebirther]

Originally Posted by gd_barnes

All bases with <400k left have been tested to n=100K. That was one of our goals a few years ago.

A bunch of 11k and 12k bases tested to n=100K are being sieved right now. I'll add them to the recommended as the sieve files become available.

I feel like you have a good batch of reservations for medium and long tests right now especially with those long R2 tests on a big file.

No real short tests though...unless you count n=100K as short tests. This begs the question:

Would BOINC like any testing starting from n=25K or 50K in the future? That would require testing more than 400 k's. A way around that would be to break them up into pieces like we do with base 3 but in much smaller pieces. For instance for a base with 500 k's, we could send you 5 files in pieces of 100 k's each.

I only mention this because the n=25K-100K testing that BOINC did several years ago seemed to be fairly popular because the bases were being reserved and finished fairly quickly.

I mean 100 factors left :)

[gd_barnes]

Originally Posted by rebirther

I mean 100 factors left :)

What do you mean by factors left? Do you mean k's remaining?

If you mean k's remaining; all of our bases with < 400 k's remaining have already been tested to a limit of n=1e5. Then what I said in my previous post is correct. We could not be able to provide you with bases that have a test limit of n<1e5. In other words no short tests.

[MisterBitcoin]

I would be ready to make a few sieve files for n=10K-25K. I would go for:


R861
R448
R997
S763
S745
S448


Can you send me a file with there k´s remaining?

Anyone can work on them as soon as the files are done, it should only take 1 week per file.
For n=25K-100K (or 50K) i will wait til you got an agreement.

[KEP]

Originally Posted by gd_barnes

What do you mean by factors left? Do you mean k's remaining?

If you mean k's remaining; all of our bases with < 400 k's remaining have already been tested to a limit of n=1e5. Then what I said in my previous post is correct. We could not be able to provide you with bases that have a test limit of n<1e5. In other words no short tests.

Having communicated a lot with Reb when BOINC was set up to support CRUS, I'm 100% sure that "factors left" is the same as "k's left".

A suggestion from here, is to maybe find in stead of a lower n that BOINC would be able to handle or desire to do, then maybe establish a minimum bit length?

If n=100K for base 3 is the desired starting point, then we know that a minimum bit size per test of 158,496 bit is the desired starting point.

Have a nice one.

KEP

[gd_barnes]

Originally Posted by MisterBitcoin

I would be ready to make a few sieve files for n=10K-25K. I would go for:
<snip>

Can you send me a file with their k´s remaining?

Anyone can work on them as soon as the files are done, it should only take 1 week per file.
For n=25K-100K (or 50K) i will wait til you got an agreement.

Let's add base R295 with ~1300 k's remaining to that list. Sieving n=10K-25K for individuals to test is a good idea. Attached are the files of k's remaining for all of those bases plus R295. Thanks for offering to do that! Maybe we can make testing some of them for n=10K-25K a goal for 2022. :-)

n=25K-50K (or 100K) was heavily tested by BOINC a few years ago. With all bases in that area having > 400 k's remaining I'd still like to hold off on doing much with it yet. They are frequently a much bigger effort than they look. Of course people are welcome to work on them on their own.

Originally Posted by KEP

Having communicated a lot with Reb when BOINC was set up to support CRUS, I'm 100% sure that "factors left" is the same as "k's left".
<snip>
KEP

Thanks for the clarification KEP! :-)

We will stick with only sending bases with <=100 k's remaining to BOINC.

[MisterBitcoin]

Sieve file for S685 attached.


Sieve file for S448 is attached, sieved up to p=150e9; factor every ~32 seconds. LLR test takes about 24 seconds for me (n=21K; 8 threads).

[MisterBitcoin]

Sieve files for R295 and R997 can be found as attachments.

[rogue]

I had an idea. Could a column be added to http://www.noprimeleftbehind.net/cru...e-reserves.htm and the Riesel equivalent that indicates if the remaining k on the conjecture are "not sieved", "needs more sieving" and "sieving complete"? That column could then be used to feed a new column on the Unproven page. That column could be used by searchers to determine if one can go straight into testing or if they need to sieve first. And if you were to split the "Sieve files/general info" column into two columns, one linking the sieve file and one with the other info, then we could add a link to the sieve files directly to the Unproven page.

I don't know how many people would benefit from such additions, outside of me, but it would be nice.

[Edit] As I think about it whether or not a conjecture is sieved deeply enough is dependent upon the program and hardware used for sieving. Maybe the sieve depth has more value as "sieving complete" is more subjective. The other two are less subjective.

[gd_barnes]

My goal on sieving now is to have all of the currently posted sieve files for n>=100K fully sieved. As you found on your S1004 reservation some are under-sieved. There are 98 files that are under-sieved from Yoyo's perspective. 64 of them are for n<=1M.

I broke the 64 under-sieved files for n<=1M into 4 priorities. Priority #1 is for reserved and recommended bases. There are 7 bases in priority #1 and Yoyo is currently working on them. Mark, this includes your S1004. You can see the status of all Yoyo sieving here: https://www.rechenkraft.net/yoyo/y_status_sieve.php.

Yoyo now has tremendous resources for sieving so this should only take a few months.

Note that a high percentage of all posted sieve files for n>=100K are already fully sieved. If the sieve depth is P>=5e14 then the file is mostly likely fully sieved so you can tell by looking at their sieve depth.

Once this is complete all posted files will be fully sieved unless otherwise noted as on huge n=2500-10K files.

I'd rather not add any more detail to the pages. Once done the above effort will clean everything up.