mersenneforum.org Bases 2 & 4 reservations/statuses/primes
 Register FAQ Search Today's Posts Mark Forums Read

2021-11-04, 04:49   #474
VBCurtis

"Curtis"
Feb 2005
Riverside, CA

7·11·67 Posts

Quote:
 Originally Posted by Grotex I can't really guarantee this ...... It's expensive to run, and I'm using a PC, which really doesn't have a lot of computing power. I'll withdraw this reservation ...... Can you teach me a little bit about the experience of choosing reservation, preferably of a quantifiable kind?
He just did teach you, by doing the math of number of tests * hours per test.

So, before reserving a range for a new-to-you project, acquire a sieve file to see how many tests it will take to complete the reservation. If one doesn't exist, you might ask in the forum how big a job will be- Gary is quite good with these estimates. If nobody has an estimate, you can sieve for a day, count candidates left, and reduce by 10-20% to reflect what a full sieving effort would reduce out of the file.

 2021-11-04, 07:15 #475 gd_barnes     May 2007 Kansas; USA 1059610 Posts Adding a little more info. CRUS has evolved to the point where there are no easy or short reservations. All of the low-hanging fruit has been well searched. That is by design. I would suggest having at least 4 cores running full-time non-stop to make a reservation on the project. 8 cores is better. It's very easy to way underestimate the time it takes to complete a reservation. On our reservations pages there are many sieve files. All are fully sieved unless otherwise noted. Pick one at your choosing and consider a small n-range portion. Find a k/n pair that is at about 60% of the n-range in which you wish to search. So if you are choosing n=100K-200K pick a test at n=~160K. Run LLR or PFGW on that test. Multiply the test time by the number of tests in your chosen range. That will give you your total CPU time. Divide by the number of cores that you will be running on the range to give you the amount of calendar time needed. Also see our recommended list. I would suggest avoiding bases on that list unless you have > ~10 cores running full time. Those are our higher priority bases. Last fiddled with by gd_barnes on 2021-11-04 at 07:17
2021-11-04, 14:17   #476
Grotex

"Jiahao He"
May 2018
China

22·3·5 Posts

Quote:
 Originally Posted by gd_barnes Adding a little more info. CRUS has evolved to the point where there are no easy or short reservations. All of the low-hanging fruit has been well searched. That is by design. I would suggest having at least 4 cores running full-time non-stop to make a reservation on the project. 8 cores is better. It's very easy to way underestimate the time it takes to complete a reservation. On our reservations pages there are many sieve files. All are fully sieved unless otherwise noted. Pick one at your choosing and consider a small n-range portion. Find a k/n pair that is at about 60% of the n-range in which you wish to search. So if you are choosing n=100K-200K pick a test at n=~160K. Run LLR or PFGW on that test. Multiply the test time by the number of tests in your chosen range. That will give you your total CPU time. Divide by the number of cores that you will be running on the range to give you the amount of calendar time needed. Also see our recommended list. I would suggest avoiding bases on that list unless you have > ~10 cores running full time. Those are our higher priority bases.
So there is a "recommended list" ......
I would like to apologise: this is the first time I have participated in this project and many aspects were not well thought out, which turned into a very embarrassing situation.
Also, it is clear from the text alone that Mr Barnes' advice is very practical, thank you for your selflessness!

 2021-11-04, 14:19 #477 Grotex     "Jiahao He" May 2018 China 1111002 Posts Also, thanks to Mr Curtis's guidance.
2021-11-04, 16:51   #478
dannyridel

"AMD YES!"
Jan 2020
Bellevue, WA

4616 Posts

Quote:
 Originally Posted by Grotex Also, thanks to Mr Curtis's guidance.
https://www.mersenneforum.org/showpo...66&postcount=1

2021-11-04, 20:02   #479
gd_barnes

May 2007
Kansas; USA

22×3×883 Posts

Quote:
 Originally Posted by gd_barnes Also see our recommended list. I would suggest avoiding bases on that list unless you have > ~10 cores running full time. Those are our higher priority bases.
Quote:
 Originally Posted by dannyridel https://www.mersenneforum.org/showpo...66&postcount=1
I suggested that he avoid bases on the recommended list.

2021-11-04, 21:31   #480
dannyridel

"AMD YES!"
Jan 2020
Bellevue, WA

2×5×7 Posts

Quote:
 Originally Posted by gd_barnes I suggested that he avoid bases on the recommended list.
Ah, I am very sorry, misread!

 2021-11-11, 14:03 #481 Jean Penné     May 2004 FRANCE 2·3·97 Posts Sierpinski odd n's testing Hi, K=9267 is now tested up to n = 13,500,823 ; no prime found... I am continuing with this input file sent to me by Gary. Regards, Jean
 2021-11-13, 07:45 #482 Happy5214     "Alexander" Nov 2008 The Alamo City 3×7×37 Posts I don't remember if Gary said anything about this previously, but have the even k's been tested/accounted for in the R2 2nd conjecture? I did not test them when I initialized my set of data.
2021-11-13, 17:10   #483
gd_barnes

May 2007
Kansas; USA

22×3×883 Posts

Quote:
 Originally Posted by Happy5214 I don't remember if Gary said anything about this previously, but have the even k's been tested/accounted for in the R2 2nd conjecture? I did not test them when I initialized my set of data.
Yes they have. There are two even k's remaining at n=2M.

 2021-11-21, 08:53 #484 rebirther     Sep 2011 Germany 305210 Posts R2 odd/even-n tested to n=8M (6.65-8M) nothing found, 6 remain Results emailed - Base released

 Similar Threads Thread Thread Starter Forum Replies Last Post gd_barnes Conjectures 'R Us 2377 2022-01-25 01:08 Siemelink Conjectures 'R Us 1715 2022-01-23 14:23 gd_barnes Conjectures 'R Us 967 2022-01-23 14:21 KEP Conjectures 'R Us 4032 2022-01-23 14:16 gd_barnes Conjectures 'R Us 1414 2022-01-04 20:54

All times are UTC. The time now is 15:03.

Tue Jan 25 15:03:34 UTC 2022 up 186 days, 9:32, 0 users, load averages: 1.39, 1.32, 1.17