mersenneforum.org  

Go Back   mersenneforum.org > Prime Search Projects > Conjectures 'R Us

Reply
 
Thread Tools
Old 2017-10-05, 17:01   #364
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

4,219 Posts
Default

Quote:
Originally Posted by gd_barnes View Post
Alternatively I suppose if it is a BOINC effort doing the sieving then the BOINC sieving folks wouldn't mind getting additional work-unit credits for sieving so, MisterBitcoin, perhaps we could attempt to sieve those to close to optimal depth.
This is exactly what I had in mind- if BOINCers are doing both the sieving and the testing, then we should make decisions that minimize the total project length (sieve computations + LLR computations). This maximizes the productivity of BOINC, and finds primes sooner!
VBCurtis is online now   Reply With Quote
Old 2017-10-05, 18:18   #365
MisterBitcoin
 
MisterBitcoin's Avatar
 
"Nuri, the dragon :P"
Jul 2016
Good old Germany

727 Posts
Default

Quote:
Originally Posted by VBCurtis View Post
This is exactly what I had in mind- if BOINCers are doing both the sieving and the testing, then we should make decisions that minimize the total project length (sieve computations + LLR computations). This maximizes the productivity of BOINC, and finds primes sooner!
So, what do we have here.
The sieve depth:

For this test I´ve choosen 50e12 because we need two important things before sieving deeper:

-Validator
-post-processing to generate a file containing factors found

Yoyo is working on the validator, it might a few days until is working.

I asked pepi if he needs deeper sieve files for his searches, I´m offering the same to anyone working on a base too. The values will be around 1-5P (depends on N).

For BOINC-Bases I suggest to sieve them to a lower value then on user Bases. Something like that should be a good starting point:

10K-25K: 1e12
25K-100K: 15e12
100K-250K: 50e12
250K-500K: 150e12
500K-1M: 500e12
1M-3M: 2e15

I also want to focus on sieve files that are undersieved, see a few posts before.

Notice: For now 1 WU sieves a range of 50e9, if the project is out of beta we will sieve a range of 500e9. Yesterday we had 339 WUs running at the same time, just awesome!
MisterBitcoin is offline   Reply With Quote
Old 2017-10-05, 19:19   #366
pepi37
 
pepi37's Avatar
 
Dec 2011
After milion nines:)

22·17·19 Posts
Default

For your calculations about sieve depth you must also consider base. It is not same sieve depth for base 9 and exponent 1M and base 1000 with exponent 1M :)

As soon you make validator is working OK I will sent you my sieve files for futher sieving and will join project with all my cores. Since you help me, I will help project :)
pepi37 is offline   Reply With Quote
Old 2017-10-05, 20:42   #367
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

4,219 Posts
Default

Quote:
Originally Posted by MisterBitcoin View Post
Notice: For now 1 WU sieves a range of 50e9, if the project is out of beta we will sieve a range of 500e9. Yesterday we had 339 WUs running at the same time, just awesome!
The time it takes to sieve 50e9 varies with sqrt(number of k values); a base with 300 k's remaining will take 10x as long as a base with 3 k's remaining. If it's OK for a WU length to vary by 10x, I suppose you don't have to care. But BOINC power allows you to sieve bases like R19 with ~1000 k's; you don't want to assign 100e9 per WU there!

The time also varies with sqrt(n-range); sieving 500k to 2M will be much slower than 25k-100k, even if the number of k's is the same. You'll struggle to make a one-size-fits-all approach for this project, if WU length matters.
VBCurtis is online now   Reply With Quote
Old 2017-10-07, 02:45   #368
wombatman
I moo ablest echo power!
 
wombatman's Avatar
 
May 2013

173710 Posts
Default S27

S27 has been sieved to P=250e12 for n=1.211-2M. I believe that was considered sufficient. There are 13,295 candidates remaining. File is attached!
Attached Files
File Type: txt s27_1.211M-2M-250e12.txt (155.8 KB, 26 views)
wombatman is offline   Reply With Quote
Old 2017-10-08, 13:39   #369
MisterBitcoin
 
MisterBitcoin's Avatar
 
"Nuri, the dragon :P"
Jul 2016
Good old Germany

727 Posts
Default

We have done testing on R667.
R667 is sieved up to 20e12, factors are checked (all ok).
Booth sieve files attached.


I´ll take R93 to pmax=200e12.
Attached Files
File Type: npg t17_b667_k344.npg (57.9 KB, 24 views)
File Type: npg t17_b667_k288.npg (65.1 KB, 25 views)

Last fiddled with by MisterBitcoin on 2017-10-08 at 13:43
MisterBitcoin is offline   Reply With Quote
Old 2017-10-08, 20:08   #370
MisterBitcoin
 
MisterBitcoin's Avatar
 
"Nuri, the dragon :P"
Jul 2016
Good old Germany

2D716 Posts
Default

Quote:
Originally Posted by VBCurtis View Post
The time it takes to sieve 50e9 varies with sqrt(number of k values); a base with 300 k's remaining will take 10x as long as a base with 3 k's remaining. If it's OK for a WU length to vary by 10x, I suppose you don't have to care. But BOINC power allows you to sieve bases like R19 with ~1000 k's; you don't want to assign 100e9 per WU there!

The time also varies with sqrt(n-range); sieving 500k to 2M will be much slower than 25k-100k, even if the number of k's is the same. You'll struggle to make a one-size-fits-all approach for this project, if WU length matters.

I´m splitting each base into single sequences. That makes more WU´s, but it will reduce the size of upload and download files.


New record: 747 WU´s running overall, each contains a 500e9 range.

Taking S256 to pmax=150e12.

Last fiddled with by MisterBitcoin on 2017-10-08 at 20:14
MisterBitcoin is offline   Reply With Quote
Old 2017-10-09, 07:37   #371
MisterBitcoin
 
MisterBitcoin's Avatar
 
"Nuri, the dragon :P"
Jul 2016
Good old Germany

2D716 Posts
Default

S320 reached p=100e12, file attached.


Notice: Factors checked with check-factors. No errors found.


Reserving R800 from n=500K-1M.
Attached Files
File Type: npg t16_b320_k97.npg (102.2 KB, 24 views)

Last fiddled with by MisterBitcoin on 2017-10-09 at 07:40
MisterBitcoin is offline   Reply With Quote
Old 2017-10-09, 09:54   #372
unconnected
 
unconnected's Avatar
 
May 2009
Russia, Moscow

7·349 Posts
Default

Quote:
Originally Posted by MisterBitcoin View Post
Reserving R800 from n=500K-1M.
This base was already completed to n=1M. I've done it to n=800k and BOINC did the rest.

Last fiddled with by unconnected on 2017-10-09 at 09:55
unconnected is offline   Reply With Quote
Old 2017-10-09, 09:57   #373
MisterBitcoin
 
MisterBitcoin's Avatar
 
"Nuri, the dragon :P"
Jul 2016
Good old Germany

727 Posts
Default

Quote:
Originally Posted by unconnected View Post
This base was already completed to n=1M. I've done it to n=800k and BOINC did the rest.

It´s range n=1M-3M.

Last fiddled with by MisterBitcoin on 2017-10-09 at 09:59
MisterBitcoin is offline   Reply With Quote
Old 2017-10-12, 08:35   #374
MisterBitcoin
 
MisterBitcoin's Avatar
 
"Nuri, the dragon :P"
Jul 2016
Good old Germany

727 Posts
Default

R93 reached P=200e12.
434 factors found, new sieve file attached.

S256 reached P=150e12
387 factors found, new sieve file attached.

Reserving the next bases:

Code:
R103
R368
R1019
R1024
R1029
@Gary I didn´t reserved R102. :)
Attached Files
File Type: npg t17_b93_k424.npg (71.9 KB, 41 views)
File Type: npg t16_b256_k831.npg (92.7 KB, 30 views)

Last fiddled with by MisterBitcoin on 2017-10-12 at 08:38
MisterBitcoin is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Sieving Drive all base 2/4 k's worked by CRUS gd_barnes Conjectures 'R Us 143 2014-10-21 23:55
Some CRUS stats vmod Conjectures 'R Us 213 2014-02-28 21:23
What are your CRUS plans? rogue Conjectures 'R Us 35 2013-11-09 09:03
CRUS-like sieving challenge CRGreathouse Puzzles 24 2011-10-28 18:30
how high will CRUS go Mini-Geek Conjectures 'R Us 1 2010-11-08 20:50

All times are UTC. The time now is 16:49.

Tue Jul 14 16:49:05 UTC 2020 up 111 days, 14:22, 0 users, load averages: 2.24, 1.70, 1.66

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.