mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Software

Reply
 
Thread Tools
Old 2013-03-19, 16:51   #1
pepi37
 
pepi37's Avatar
 
Dec 2011
After milion nines:)

3×5×83 Posts
Default NewPgen question

NewPGen 2.82 is great program and under Services- Primality test on several machines you hove option that program evenly distribute one large NPG file in several smaller ( according to CPU speed). I try this option: take NPG, enter number of processors (8 CPU) with same speed ( 4000 MHz)
And got 8 new , smaller NPG files, but without any sense...
Picture is talking everything :)
I got NPG files from 380 - 565 bytes?
What is wrong....
Attached Thumbnails
Click image for larger version

Name:	Clipboard01.jpg
Views:	107
Size:	17.0 KB
ID:	9568  

Last fiddled with by pepi37 on 2013-03-19 at 16:52
pepi37 is offline   Reply With Quote
Old 2013-03-19, 16:58   #2
henryzz
Just call me Henry
 
henryzz's Avatar
 
"David"
Sep 2007
Cambridge (GMT)

10110000111002 Posts
Default

Quote:
Originally Posted by pepi37 View Post
NewPGen 2.82 is great program and under Services- Primality test on several machines you hove option that program evenly distribute one large NPG file in several smaller ( according to CPU speed). I try this option: take NPG, enter number of processors (8 CPU) with same speed ( 4000 MHz)
And got 8 new , smaller NPG files, but without any sense...
Picture is talking everything :)
I got NPG files from 380 - 565 bytes?
What is wrong....
Those are quite small files. I would guess that you don't have many candidates remaining and the lines for each candidate vary in length quite a bit. Have you tried looking in one of the files?
henryzz is offline   Reply With Quote
Old 2013-03-19, 17:51   #3
axn
 
axn's Avatar
 
Jun 2003

11F116 Posts
Default

You have provided absolutely no information to diagnose the situation. Can you prove that what NewPGen did is not correct -- i.e, the files don't have (approximately) equal amounts of Primality testing work?

EDIT:- Your "large" file contains 3493 bytes!

Last fiddled with by axn on 2013-03-19 at 17:57
axn is offline   Reply With Quote
Old 2013-03-19, 21:16   #4
pepi37
 
pepi37's Avatar
 
Dec 2011
After milion nines:)

100110111012 Posts
Default

Axn, in NewPgen says that program will give to faster cpu more results, and to slower cpu less results so time will be approx same.
I use for all 8 cores IDENTICAL speed ( 4000 MHZ)
1 file 36 candidates
2 file 36 candidates
3 file 25 candidates
4 file 31 candidates
5 file 39 candidates
6 file 28 candidates
7 file 34 candidates
8 file 22 candidates
------------- 251 candidates /8 = 31,375
So for same speed, same number of results, +/-1 not +/- 18 ( max difference)

Last fiddled with by pepi37 on 2013-03-19 at 21:40
pepi37 is offline   Reply With Quote
Old 2013-03-19, 21:38   #5
kracker
ἀβουλία
 
kracker's Avatar
 
"Mr. Meeseeks"
Jan 2012
California, USA

2×13×83 Posts
Default

4000 GHZ holy s**t!!
kracker is offline   Reply With Quote
Old 2013-03-19, 21:40   #6
pepi37
 
pepi37's Avatar
 
Dec 2011
After milion nines:)

4DD16 Posts
Default

MHz not GHz
Maybe NewPGen when splting file also consider i exponent number ( candidates with bigger exponenet calculates more on same speed), so it gives less candidates...?

Last fiddled with by pepi37 on 2013-03-19 at 22:16
pepi37 is offline   Reply With Quote
Old 2013-03-20, 03:13   #7
axn
 
axn's Avatar
 
Jun 2003

3·1,531 Posts
Default

Quote:
Originally Posted by pepi37 View Post
Maybe NewPGen when splting file also consider i exponent number ( candidates with bigger exponenet calculates more on same speed), so it gives less candidates...?
Yes. Which is why I said that you didn't give us any information. Are we supposed to guess that your sieve contained different-sized exponents? Even then there may or may not be a bug in NewPGen. Depends on the actual contents of these files.
axn is offline   Reply With Quote
Old 2013-03-20, 08:13   #8
pepi37
 
pepi37's Avatar
 
Dec 2011
After milion nines:)

3·5·83 Posts
Default

Quote:
Originally Posted by axn View Post
Yes. Which is why I said that you didn't give us any information. Are we supposed to guess that your sieve contained different-sized exponents? Even then there may or may not be a bug in NewPGen. Depends on the actual contents of these files.
No, size of exponents was same.
I made another attempt to "break" sieve from K 11 N from 4M to 5M into 8 pieces, and got similar result.No I will try to count exponents, their sizes, and will return with more data :)
pepi37 is offline   Reply With Quote
Old 2013-03-22, 19:54   #9
pepi37
 
pepi37's Avatar
 
Dec 2011
After milion nines:)

3×5×83 Posts
Default

It look like I was right. I made real work using my both CPU-s , and split sieve in 8 pieces. 4 is made with CPU at 4GHZ, and 4 is made with CPU on 4.2 Ghz.
Right now: one of cores on slower CPU have only 7 hours of work, until one core on faster CPU have 21 hours. So I dont know how NePGen is do calculation, but it is not even equal ...
pepi37 is offline   Reply With Quote
Old 2013-03-22, 20:07   #10
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

131158 Posts
Default

This is 11*2^n+1 or 11*2^n-1 right? If so, then I suggest using a combination of srsieve and sr1sieve. That would be much faster than NewPGen. In either case some pre-sieved data should be available.
rogue is offline   Reply With Quote
Old 2013-03-22, 20:41   #11
pepi37
 
pepi37's Avatar
 
Dec 2011
After milion nines:)

3·5·83 Posts
Default

Rogue it is not "problem" in sieving, I use sr1sieve, and that is ok. NewPgen have implemented option that split sieve file in number of CPU according to CPU speed , so faster CPU will get more work, and slower CPU will have less work.
But I noticed that function doesnot work as should, and that is all "problem"
pepi37 is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
NewPgen Cybertronic Factoring 0 2014-03-22 10:07
Does NewPGen have a bug? MooooMoo Riesel Prime Search 16 2008-12-11 11:46
Question: Multiple sequences in NewPGen format Xentar Conjectures 'R Us 3 2008-01-20 15:56
NewPGen k value problem roger Information & Answers 0 2007-04-04 22:38
NewPGen reliability Cruelty Riesel Prime Search 3 2006-02-15 05:15

All times are UTC. The time now is 10:28.

Sun May 31 10:28:12 UTC 2020 up 67 days, 8:01, 1 user, load averages: 1.56, 1.78, 1.83

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.