mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > Factoring

Reply
 
Thread Tools
Old 2003-08-23, 05:42   #1
Xyzzy
 
Xyzzy's Avatar
 
"Mike"
Aug 2002

2·33·151 Posts
Default Memory usage of various factoring projects...

A little under a year ago I ran a few crazy ECM curves on a box with 1GB of memory...

http://www.mersenneforum.org/viewtopic.php?t=109

This year I have "graduated" to a 2GB box and I'm wondering if there is a factoring project that can make use of this additional memory... If not I suppose I'll try P-1 factoring or maybe some more ECM curves... (The ECM stuff was a lot of fun!) Any suggestions?
Xyzzy is offline   Reply With Quote
Old 2003-08-23, 06:45   #2
Axel Fox
 
Axel Fox's Avatar
 
May 2003

25·3 Posts
Default

Well, trial factoring doesn't use a lot of memory, so with that you can't take advantage of your 2GB.

P-1 and ECM benefit from it, cause they can be run with higher bounds, but I guess you're asking for something else now.

A lot of memory also allows a larger Sieve or Erathostenes, but even a sieve like that wouldn't come close to the current highest primes.

Don't know anything about NFS (Number Field Sieve), so you might check out something like that.
Axel Fox is offline   Reply With Quote
Old 2003-08-23, 11:22   #3
Wacky
 
Wacky's Avatar
 
Jun 2003
The Texas Hill Country

32×112 Posts
Default

A really large amount of memory doesn't really help all that much in the sieving phase of NFS. The inner sieving loop is divided into three levels.
The outermost loop will cover the entire sieving range (perhaps 150M).
It is done in a few passes over the amount that is stored in real memory (perhaps 50M) and that, in turn is done in sub-passes which can stay in the cache. The additional overhead in jumping from one memory pass to the next is not all that great. By keeping the entire range in real memory and doing it in one pass, you would save only a small amount (perhaps 1%). The cache size is much more significant.

We try to keep the footprint managable so that the siever can be run all the time without having significant effect on other uses of the machine. Thus, having plenty of RAM, I leave the siever running continuously in my development machines that are not being used for debug/testing. (Compile machines, this laptop, etc.) When I do something like a compile or posting this message, the only effect that I notice is that the siever stops for that time and resumes when the other task completes.

Now, the additional memory is quite valuable in the post-processing filtering. There, the initial passes need to store all the relations in memory.
Wacky is offline   Reply With Quote
Old 2003-08-23, 21:10   #4
ColdFury
 
ColdFury's Avatar
 
Aug 2002

26×5 Posts
Default

The second stage of P-1 really comes to mind.
ColdFury is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
yafu memory usage for ecm Johnatan YAFU 5 2016-06-13 11:28
Large memory usage Unregistered Information & Answers 8 2010-05-14 23:45
Prime95 v25.7 memory usage opyrt Software 1 2008-11-09 08:43
Memory usage during P-1 factoring lidocorc Software 2 2008-11-03 02:35
memory usage in stage 2 of P-1 factoring gckw Software 3 2003-09-07 06:56

All times are UTC. The time now is 02:22.

Sun May 9 02:22:31 UTC 2021 up 30 days, 21:03, 0 users, load averages: 1.17, 1.41, 1.50

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.