mersenneforum.org  

Go Back   mersenneforum.org > Prime Search Projects > Riesel Prime Search

Reply
 
Thread Tools
Old 2020-02-18, 23:14   #419
storm5510
Random Account
 
storm5510's Avatar
 
Aug 2009
U.S.A.

33·41 Posts
Default

Below would be in insertion into my list:

Quote:
98475*2^257090-1 is prime! (77397 decimal digits)
I was trying to run the same k with LLR on multiple machines. No more. Doing so only served to put things out of sequence.
storm5510 is offline   Reply With Quote
Old 2020-02-19, 08:14   #420
kar_bon
 
kar_bon's Avatar
 
Mar 2006
Germany

53178 Posts
Default

There're several options to do:

1. Split your sieve file by hand/script into equal numbers of values over the n-range and run them on several machines.
-> Higher n-values will last longer for LLR so the lowest should be done first -> machines are ending with LLR on a wide range

2. Split the sieve file into several with this script with candidates spreaded over the whole n-range
-> compareable machines in speed should do the LLR almost the same time -> all done at almost same time

3. Setting up a local PRPnet server for several machines, for more info see here
kar_bon is offline   Reply With Quote
Old 2020-02-19, 12:20   #421
storm5510
Random Account
 
storm5510's Avatar
 
Aug 2009
U.S.A.

33×41 Posts
Default

Quote:
Originally Posted by kar_bon View Post
There're several options to do:

1. Split your sieve file by hand/script into equal numbers of values over the n-range and run them on several machines.
-> Higher n-values will last longer for LLR so the lowest should be done first -> machines are ending with LLR on a wide range

2. Split the sieve file into several with this script with candidates spreaded over the whole n-range
-> compareable machines in speed should do the LLR almost the same time -> all done at almost same time

3. Setting up a local PRPnet server for several machines, for more info see here

98475*2^313624-1 is prime! (94416 decimal digits)


I don't understand the script, but this has given me an idea for something comparable. Just in another language. As you say, split the sieve into two equal size files and run them concurrently. As for as I can tell, my i5 HP can run at roughly 90% of what the i7 does with the console version of LLR. No so with other things, such as Prime95 with all its GUI code. I can sieve on my laptop. The "sr" programs do not work it hard at all. It just takes a bit longer.

Last fiddled with by storm5510 on 2020-02-19 at 12:24 Reason: Additional
storm5510 is offline   Reply With Quote
Old 2020-02-22, 00:03   #422
storm5510
Random Account
 
storm5510's Avatar
 
Aug 2009
U.S.A.

33×41 Posts
Default

k = 98475. Tested to n = 400,000.
No new primes found.
Continuing for quite some time to come...
storm5510 is offline   Reply With Quote
Old 2020-02-22, 00:16   #423
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

FA416 Posts
Default

Monthly updates are plenty, or every 1M milestone.
Updating us every 100k is overkill many times over, just like posting individual primes under 1M bits in size.
VBCurtis is offline   Reply With Quote
Old 2020-02-26, 17:32   #424
storm5510
Random Account
 
storm5510's Avatar
 
Aug 2009
U.S.A.

33·41 Posts
Default

Reserving k = 741. Nash = 5029.

I did multiple searches in the forum with different criteria and found no results for this k.
storm5510 is offline   Reply With Quote
Old 2020-02-26, 18:06   #425
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

22×7×11×13 Posts
Default

Try the No Prime Left Behind forum; they coordinate most (all?) of the k's from 301 to 1999.
VBCurtis is offline   Reply With Quote
Old 2020-02-27, 13:29   #426
storm5510
Random Account
 
storm5510's Avatar
 
Aug 2009
U.S.A.

33×41 Posts
Default

Quote:
Originally Posted by storm5510 View Post
Reserving k = 741. Nash = 5029.

I did multiple searches in the forum with different criteria and found no results for this k.
This k has been reserved by NPLB since 2012. Obviously, I need to expand my search areas in the future...
storm5510 is offline   Reply With Quote
Old 2020-02-27, 13:55   #427
diep
 
diep's Avatar
 
Sep 2006
The Netherlands

2·337 Posts
Default

Quote:
Originally Posted by storm5510 View Post
This k has been reserved by NPLB since 2012. Obviously, I need to expand my search areas in the future...
I tried search exponent=733 some years ago until 4M.

If you interested you can reserve that one and continue there. If you so lucky you'll have a prime very quickly.

edit: did not finish search till 4M entirely.

Last fiddled with by diep on 2020-02-27 at 14:02
diep is offline   Reply With Quote
Old 2020-02-27, 14:15   #428
diep
 
diep's Avatar
 
Sep 2006
The Netherlands

2×337 Posts
Default

Quote:
Originally Posted by storm5510 View Post
98475*2^313624-1 is prime! (94416 decimal digits)


I don't understand the script, but this has given me an idea for something comparable. Just in another language. As you say, split the sieve into two equal size files and run them concurrently. As for as I can tell, my i5 HP can run at roughly 90% of what the i7 does with the console version of LLR. No so with other things, such as Prime95 with all its GUI code. I can sieve on my laptop. The "sr" programs do not work it hard at all. It just takes a bit longer.
Faster for you might be cllr64
i wrote a small C program to splitup sieve files converted from ABCD files into something which then writes newpgen format.

It loadbalances the sieve files.

Yet when you get to larger exponentsizes like some millions of bits you might want to use multiple threads as that trashes the cache less. For each type of CPU with its different sizes caches the number of threads that works great will be different.

Once you use multithreading you no longer want to use stand alone cllr64 as trashing the cache give bad performance for multithreading. Throughput is higher then if you keep within.

So example i'm testing now at 7.0+ M bits for k=69 with 4 threads.
If the other 4 cores i would run 4 processes cllr64 testing k=733 to finish until 4M that would be bad idea as then the cache gets trashed for the cllr64 testing k=69 with 4 threads. It suddenly is immensely slower then. From its current 28.8k seconds that could get even 132k seconds.

If i would want to finish k=733 i would need to stop all the multithreading CLLR64's and run at the 8 real cores this box has (2 socket Xeon L5420 - yeah i know old) then 8 times cllr64 to finish k=733 until 4M.

Throughput for that is faster as the Xeons have 8MB L2 cache or something like that.
diep is offline   Reply With Quote
Old 2020-02-27, 14:17   #429
diep
 
diep's Avatar
 
Sep 2006
The Netherlands

2·337 Posts
Default

Here is that splitup program. Works both in linux where i use it as well as windows where i use it too. After you compiled it of course.
Attached Files
File Type: zip splitup.zip (1.4 KB, 18 views)
diep is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
16e Post Processing Progress pinhodecarlos NFS@Home 8 2018-11-28 13:45
Sieving with powers of small primes in the Small Prime variation of the Quadratic Sieve mickfrancis Factoring 2 2016-05-06 08:13
Sierpinski/Riesel Base 5: Post Primes Here robert44444uk Sierpinski/Riesel Base 5 358 2008-12-08 16:28
Small Primes Housemouse Math 2 2008-06-04 05:23
POST PRIMES you've found here, and name or prover code TTn 15k Search 415 2006-03-02 21:17

All times are UTC. The time now is 06:47.

Mon Apr 6 06:47:55 UTC 2020 up 12 days, 4:20, 1 user, load averages: 1.23, 1.33, 1.32

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

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.