mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > GMP-ECM

Reply
 
Thread Tools
Old 2016-11-11, 01:35   #12
WraithX
 
WraithX's Avatar
 
Mar 2006

23×59 Posts
Default

Quote:
Originally Posted by EdH View Post
@WraithX: If I'm following correctly, s will be the same value for all iterations with the same B1. If I'm not lost yet, in my case, this would mean all my threads would be using the same s. With slight modification, could I not calculate s outside of the threads and reference it to each thread?
Actually, this brings up a good point that I had forgotten about. There is a way to save s to a file built in to GMP-ECM. The option looks like:
-bsaves <filename>

You would run that once, which would save s to a file, and then with all your other instances you can load s from the file with:
-bloads <filename>

I've never timed it to see if generating s or reading it in from a file on disk is faster, but with this you should be able to run several copies of GMP-ECM with the faster batch mode. Let us know how this works for you.
WraithX is offline   Reply With Quote
Old 2016-11-11, 03:34   #13
EdH
 
EdH's Avatar
 
"Ed Hall"
Dec 2009
Adirondack Mtns

5×727 Posts
Default

Quote:
Originally Posted by WraithX View Post
Actually, this brings up a good point that I had forgotten about. There is a way to save s to a file built in to GMP-ECM. The option looks like:
-bsaves <filename>

You would run that once, which would save s to a file, and then with all your other instances you can load s from the file with:
-bloads <filename>

I've never timed it to see if generating s or reading it in from a file on disk is faster, but with this you should be able to run several copies of GMP-ECM with the faster batch mode. Let us know how this works for you.
Thanks! I will try this with the aforementioned machine and get back to you.

edit: If my failure times are an indication of s causing the issue, then that was at about 21-26 minutes. I would think if I can generate s once and send it to all my machines to use in all the threads, it would save ~20 minutes of every run. That would be a huge savings. Of course, my faster machines would save less, but still quite a bunch. Well, I might be getting ahead of where I should be. Calculating s is probably taking that long due to buffering and possibly swapping prior to the failure.

Last fiddled with by EdH on 2016-11-11 at 03:42 Reason: Afterthought
EdH is offline   Reply With Quote
Old 2016-11-11, 04:15   #14
EdH
 
EdH's Avatar
 
"Ed Hall"
Dec 2009
Adirondack Mtns

70638 Posts
Default

After some experimentation and a little thought, I'm concluding that the save/load of s will not help. I tried to generate s on a different machine and the file wasn't generated due to me not using -param 1-4 and an apparent default of another value. Not really a problem - I should just use -param 1, 2, 3 or 4. However, if the size of s is the issue, it will be just as big via file as it is when generated. In fact, the size was given in the run I made:
Code:
Computing batch product (of 4183822616 bits) of primes up to B1=2900000000 took 476039ms
Unless I'm calculating incorrectly, that product is 522977827 bytes. If each thread is reading in that 500+MB file, it will still need to be held in each thread's memory. I think I might be able to get along with one thread using s, but for the other threads I think I will need to run the -param 0.

In reading further into the docs, it appears that -param is chosen randomly if not specified. That probably explains the varying messages interwoven with successful runs. Perhaps I should experiment with various -param(s). More playing is definitely needed. I'll have to try that route...
EdH is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
OMG, NSA Killed ECC! (You bastards!) R.D. Silverman Tales From the Crypt(o) 48 2015-11-16 22:22
A potential cause of Windows low-memory messages cheesehead Software 14 2013-05-16 00:45
Team sieve: c160 from 4788:i2715 (ABORTED) fivemack Aliquot Sequences 15 2011-10-29 10:12
bin Laden killed in groundstrike ixfd64 Soap Box 33 2011-05-12 02:00
Iran Exonerates Six Who Killed in Islam’s Name ewmayer Soap Box 6 2007-04-30 01:39

All times are UTC. The time now is 16:06.

Mon Mar 8 16:06:11 UTC 2021 up 95 days, 12:17, 1 user, load averages: 2.34, 1.64, 1.53

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, 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.