mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Conjectures 'R Us (https://www.mersenneforum.org/forumdisplay.php?f=81)
-   -   Riesel base 3 reservations/statuses/primes (https://www.mersenneforum.org/showthread.php?t=11151)

KEP 2008-04-17 15:31

Riesel base 3 reservations/statuses/primes
 
Completed and in process reservations for Riesel base 3 posted from the previous "Riesel base 3 attack" subforum":

[code]
k-range n-range status
4M-2.147G [B]500K-850K[/B] completed by BOINC
4M-2.147G [B]850K-900K[/B] in progress by BOINC
4M-10M [B]100K-500K[/B] completed by henryzz
10M-20M [B]100K-425K[/B] completed by henryzz
10M-20M [B]425K-500K[/B] completed by Puzzle-Peter
20M-6G [B]100K-500K[/B] completed by BOINC
6G-10G [B]100K-225K[/B] completed by BOINC
>63G [B]100K-500K[/B] completed by MisterBitcoin

<100M [B]25K-100K [/B]completed by mini-drive I
100M-200M [B]25K-100K [/B]completed by KEP
200M-250M [B]25K-100K [/B]completed by Mattyp101
250M-500M [B]25K-100K [/B]completed by grueny
500M-2G [B]25K-30K [/B]completed by VBCurtis
500M-2G [B]30K-100K [/B]completed by Lennart
2G-2.05G [B]25K-100K [/B]completed by VBCurtis
2.05G-2.5G [B]25K-100K [/B]completed by Puzzle-Peter
2.5G-8G [B]25K-100K [/B]completed by Lennart
8G-21G [B]25K-100K[/B] completed by KEP
21G-22G [B]25K-100K[/B] completed by Puzzle-Peter
22G-23G [B]25K-50K [/B] completed by Siemelink
22G-23G [B]50K-100K[/B] completed by BOINC
23G-25G [B]25K-100K[/B] completed by Puzzle-Peter
25G-55G [B]25K-100K[/B] completed by rogue
55G-60G [B]25K-50K[/B] completed by rogue
60G-63G [B]25K-50K[/B] completed by VBCurtis
60G-63G [B]50K-100K[/B] in progress by VBCurtis
>63G [B]25K-100K[/B] completed by MyDogBuster

<500M [B]25K[/B] completed by KEP
500M-510M [B]25K[/B] completed by michaf
510M-520M [B]25K [/B] completed by Flatlander
520M-570M [B]25K[/B] completed by michaf
570M-580M [B]25K[/B] completed by Mini-Geek
580M-600M [B]25K [/B]completed by KEP
600M-650M [B]25K [/B] completed by Flatlander
650M-660M [B]25K[/B] completed by henryzz
660M-700M [B]25K[/B] completed by Flatlander
700M-800M [B]25K[/B] completed by gd_barnes
800M-810M [B]25K[/B] completed by Mini-Geek
810M-820M [B]25K[/B] completed by gd_barnes
820M-830M [B]25K[/B] completed by henryzz
830M-2G [B]25K[/B] completed by Puzzle-Peter
2G-2.05G [B]25K [/B]completed by VBCurtis
2.05G-2.5G [B]25K[/B] completed by gd_barnes
2.5G-3G [B]25K[/B] completed by Lennart
3G-11G [B]25K[/B] completed by Puzzle-Peter
11G-12G [B]25K[/B] completed by KEP
12G-13G [B]25K[/B] completed by Puzzle-Peter
13G-14G [B]25K[/B] completed by KEP
14G-15G [B]25K[/B] completed by Puzzle-Peter
15G-60G [B]25K[/B] completed by KEP
60G-63G [B]25K[/B] completed by rogue
>63G [B]25K[/B] completed by KEP
[/code][B]Admin edit: Separated Riesel base 3 reservations/statuses/primes posts from Sierp base 3 reservations/status/primes posts.[/B]

[quote=gd_barnes;131492]
KEP, is the above what you are doing or are you doing any sieving beforehand? If so, what is it that you are using to sieve? IMHO, no sieving is needed at all below n=10K. A case could be made for sieving above n=10K but I prefer to just let PFGW run to n=25K without sieving.


Gary[/quote]

It exactly is what I'm doing. Well to avoid any confusion here is my battle plan (Using WinPFGW):

1. test 2,500,000 equal k's for being PrP up to n=25,000 (command-line: "input.txt -lxxM.txt")
2. Verify those PrP (command-line: "pfgw.log -tc")
3. List all those k's remaining, and make them publicly availeable on my google page

Hope that explained, but the short answer to your question Gary is: Yes that was exactly what I am doing regards my attack on the Base3 riesel.

Now I've a question. Is there anyway to make WinPFGW work a list of k's from e.g. n=25,001 to n=100,000?

Also to the developers of WinPFGW, it would be very nice if you could make a way to avoid save a log file with factors and such stuff, and make a way to save only the Primes, the PrP and k or n remaining. I think that such a k or n function should be asked to save only for every 10 % which in my example would mean at n:

2,500
5,000
7,500
10,000
12,500
15,000
17,500
20,000
22,500
25,000

That way a lot of storage (and writing time) will be saved and it will be easier to list the k's missing and compare the number to the amount of primes found :)

Thanks everyone and happy holidays.

Regards

KEP

KEP 2008-04-17 17:50

Everyone interested can follow the progress in the Riesel Base 3 conjecture attack on following site: (none)

Expect it to be updated no more than in the weekends or extended weekends, so in case of vacation it can be more than a week or 2 or 3 weeks before I gets to update the latest progress. Thanks for understanding. I know by the way that the webpage is crap, but I doesn't have much experience using google pages yet, and does it really matter ;), there is the most nescessary to follow the progress.

Take care everyone!

gd_barnes 2008-04-19 05:28

[quote=KEP;131785]
Now I've a question. Is there anyway to make WinPFGW work a list of k's from e.g. n=25,001 to n=100,000?Regards

KEP[/quote]


Yes, you would just tell it to do so in your script. But doing so would be very inefficient. Sieving should be done first for n-values that are so high. After sieving, you could then use LLR or PFGW (or Phrot) to do primality testing.


Gary

gd_barnes 2008-04-19 06:11

On April 13th, I got an Email from KEP listing the primes for Riesel base 3 up to k=100K. ZERO k's were remaining. The reservation update and 10 highest primes are shown on the web pages.


Gary

KEP 2008-04-25 21:43

The Riesel base 3 is tested up to k=2M and n=25k. There is a 12 remaining k's and the ranges up to 100M is in progress at once. I found that taking the k to only n=5,000 and then do sieve of the remaining k's is much more efficient (about 90% faster). Is anyony by the way knowing of a great way to sieve many different k values from n=5,001 to n=25,000? I used to use NewPGen, but it requires a great deal of manual work to prepare the sieve files. Once the sieve files is prepared Sr1sieve can be used, using a .bat file and then sieveing will be rather easy, but anyone with a great idea to speedup and reduce the amont of manual work, please consider replying to me :)

On other notice to gary: I'm not working on the base 12 k=404 (sierpinski) since it appears that my reservation did not come through, and also I feel like wanting to do more somewhere else e.g. Base 3 Riesel Conjecture :)

mdettweiler 2008-04-26 01:50

[quote=KEP;132191]The Riesel base 3 is tested up to k=2M and n=25k. There is a 12 remaining k's and the ranges up to 100M is in progress at once. I found that taking the k to only n=5,000 and then do sieve of the remaining k's is much more efficient (about 90% faster). Is anyony by the way knowing of a great way to sieve many different k values from n=5,001 to n=25,000? I used to use NewPGen, but it requires a great deal of manual work to prepare the sieve files. Once the sieve files is prepared Sr1sieve can be used, using a .bat file and then sieveing will be rather easy, but anyone with a great idea to speedup and reduce the amont of manual work, please consider replying to me :)

On other notice to gary: I'm not working on the base 12 k=404 (sierpinski) since it appears that my reservation did not come through, and also I feel like wanting to do more somewhere else e.g. Base 3 Riesel Conjecture :)[/quote]
Use srsieve to start the sieving (it can accept a file with a list of k's as input--just enter one expression, such as 123*3^n-1, per line), then when you've reached about p=0.5G or so, switch to sr2sieve. :smile: It's much faster, and easier, than using NewPGen/sr1sieve for each individual k.

As for your base 12 k=404 reservation not going through, Gary is on vacation so though he can get on mersenneforum a couple times a day, he can't update any of the web pages. So, even though it's not marked on the page yet, your reservation is still OK--though if you don't want to search it after all that's fine too. :smile:

KEP 2008-04-26 05:47

[quote=Anonymous;132200]Use srsieve to start the sieving (it can accept a file with a list of k's as input--just enter one expression, such as 123*3^n-1, per line), then when you've reached about p=0.5G or so, switch to sr2sieve. :smile: It's much faster, and easier, than using NewPGen/sr1sieve for each individual k.

As for your base 12 k=404 reservation not going through, Gary is on vacation so though he can get on mersenneforum a couple times a day, he can't update any of the web pages. So, even though it's not marked on the page yet, your reservation is still OK--though if you don't want to search it after all that's fine too. :smile:[/quote]

Thanks for your tips, but in your srsieve example, I can see that there is an option to choose a min and max n aswell as the other usefull options known from sr1sieve. Now 2 questions, can I sieve from p1 using srsieve? and can I switch directly from srsieve to sr2sieve without having to do anything or changing the file(s)?

About the base 12 attack, I feel like its better to release it, since I've a lot of testing to do to find the most efficient way to attack the base3 problem. And about Gart being on vacation (hope that I was Gary now) I didn't know that, so thanks for telling me. So to sum up, I'm releasing the Base 12 sierpinski range for others to work on, and will now only consider to find the most efficient way to attack the base 3 riesel conjecture. The most efficient way seems to be as follows:

1. Use WinPFGW to n=2,500
2. Use srsieve from n=2,501 to n=25,000
3. To start out by testing the lowest k. If it fails to prime, then move all k's that somewhere is the product of 3 of tested k, which has no n value higher than n=25,000-(the times that k has been multiplied by 3). Doing this requires some manual work, but saves up to 8 hours LLR testing per k :)

If you have any questions about this please feel free to ask them or if you feels like joining the war please consider help :)

Go to to this page: (none) to check the progress and what I'm up to with this base. On this page you can also see what battleplans (I know dont lay out your plans in the open for your enemy to see) on how to combat this k the fastest :)

Thank you and take care!

KEP

tnerual 2008-04-26 05:52

with large number of k (>300), forget about sr2sieve ! only use srsieve

i have 2000k left in my base 31 file and i can't use sr2 sieve (after 300k, i'm at 1.5 GB of memory used) the problem is the legendre symbol table that is too big ...

KEP 2008-04-26 06:33

[QUOTE=tnerual;132209]with large number of k (>300), forget about sr2sieve ! only use srsieve

i have 2000k left in my base 31 file and i can't use sr2 sieve (after 300k, i'm at 1.5 GB of memory used) the problem is the legendre symbol table that is too big ...[/QUOTE]

2 things.

1. Maybe you can PM me how you got started using srsieve, since I don't seem to be able to get it started. So would you be kind enough to tell me how you got started?... you doesn't have to PM me you can also tell me here :)

2. Also can't sr2sieve be told not to build the legendre table if it seems to take up all the memory?

Now about my first question to Mr a :), I found the answer myself, and I'm assuming that I should be able to go directly to sr2sieve.exe, if this is wrong please correct me.

mdettweiler 2008-04-26 06:56

[quote=tnerual;132209]with large number of k (>300), forget about sr2sieve ! only use srsieve

i have 2000k left in my base 31 file and i can't use sr2 sieve (after 300k, i'm at 1.5 GB of memory used) the problem is the legendre symbol table that is too big ...[/quote]
With the recent updates to sr2sieve, though, it's now faster even for lots of k's. For example, in the NPLB doublecheck team-sieve (with 500 k's) that I ran a while back, we found that sr2sieve was faster.

Of course, though, as you said, if you have too many k's in the sieve, memory usage becomes a problem. In that case, of course, srsieve is your only option.
[quote=KEP;132210]2 things.

1. Maybe you can PM me how you got started using srsieve, since I don't seem to be able to get it started. So would you be kind enough to tell me how you got started?... you doesn't have to PM me you can also tell me here :)

2. Also can't sr2sieve be told not to build the legendre table if it seems to take up all the memory?

Now about my first question to Mr a :), I found the answer myself, and I'm assuming that I should be able to go directly to sr2sieve.exe, if this is wrong please correct me.[/quote]
Okay, here's what you have to do:

1) In your srsieve folder, make a new text file called "equations.txt" or something along those lines. Open it in a text editor, and list each k you want in the sieve on its own line, like this:
[code]100*3^n-1
200*3^n-1
300*3^n-1[/code]and et cetera, for all the k's in your sieve. Save the file and close it.

2) Make another text file in your srsieve folder, named "run_srsieve.bat" or something like that. Open it and type the following line, substituting various values where specified:
[code]srsieve -m 4e9 -n [I]lower end of n-range[/I] -N [I]upper end of n-range[/I] equations.txt[/code]Save the file, and close it.

3) Double-click on run_srsieve.bat, and you'll see a DOS window pop up, wherein srsieve displays its output. It outputs status reports every minute; below 4G it won't display factors found, but once it gets past 4G it will show every factor found on screen. (Earlier on in the sieve you'll want to minimize the window when you're not watching it, so the computer won't waste CPU time trying to display the factors on screen.) To interrupt srsieve press Ctrl-C, and the DOS window will close.

4) If you interrupt srsieve, it will generate a file called srsieve.out (it also generates this file for backup purposes every 60 minutes even if you don't interrupt it). Rename this to something else, so that it won't write on top of it when you continue the sieve. (I usually name my partially-sieved files based on the date--for example, "Apr26-1.txt" if it's the first file I saved on April 26.) Then, when you want to start srsieve again and continue the sieve, open run_srsieve.bat in a text editor, and change the command line therein to the following:
[code]srsieve -m 4e9 [I]name of previous sieve file[/I][/code]Replace "name of previous sieve file" with your last copy of the sieve file. Save the file and close it, then double-click on run_srsieve.bat to run srsieve again. It will pick up where it left off and continue with the sieve.

5) Srsieve will output p/sec. and sec./factor rates onscreen, updated every minute. When you've reached a satisfactory factor rate and want to stop the sieve so you can start the PRPing, press Ctrl-C to halt srsieve as before. (There's no need to rename the output file this time, you can leave it at srsieve.out now.) Create a new text file called "srfile.bat". Open it in a text editor, and enter the following line:
[code]srfile -G srsieve.out[/code]Save and close the file. Then run srfile.bat, and srfile (srsieve's file utility) will convert srsieve.out (your final sieve file) to standard NewPGen format for PRPing. It will probably be called "t16_b3.prp" (the _b3 part will change depending on the base you're doing); rename it to whatever you want, move it to your LLR folder and proceed with the PRPing!

6) You can now go ahead and delete all your intermediate sieve files (there will be one for each time you stopped and restarted srsieve), as well as srsieve.log if you want. You may want to keep the batch files, just remember to change them to the appropriate new values for the next sieve you do. :smile:

Hope this helps! :smile:

Anon :smile:

KEP 2008-04-26 07:24

@ anon:

Thanks for your tip, it actually worked and it appears to be at least 10 times as efficient as sieving all terms individually, I'm actually quite sure that now a further speedincrease will be seen :) So on behalf of myself and those who will join me in the future (maybe rieselsieve once they complete their base 2 mission) I give you my true and deep felt thank you :)

To the rest of you, good luck on what ever part of these conjectures you are struggeling with :)


All times are UTC. The time now is 08:53.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.