mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > Factoring

Reply
 
Thread Tools
Old 2004-06-25, 23:55   #1
Khemikal796
 
Feb 2003

37 Posts
Default ECM in Prime95

Im wanting to start finding some factors of a certain mersenne number. I found a factor for it at 64bit level doing TF with Prime95. I was wandering how i should setup my lowm.txt and/or lowp.txt.

M( xxxxxxx )C: 123456789

Is this correct format I need to put in the low file? Which of the files would it go in? Is there anything else that needs to accompany this line? Sorry for the dumb questions, but I havent found really any help/readmes that fully explain this stuff to someone who hasnt messed around with ecm too much.
Khemikal796 is offline   Reply With Quote
Old 2004-06-26, 06:10   #2
akruppa
 
akruppa's Avatar
 
"Nancy"
Aug 2002
Alexandria

2,467 Posts
Default

The line looks right. Since you want to factor a Mersenne number (2^x-1) it should go into the lowm.txt file. If you want to factor a 2^x+1 number, the line should be added to lowp.txt and should read

P( x )C: 123456789

The exponents ( x ) of the entries must be in ascending order in both files.

Approximately how large is the exponent of the number you want to factor? Larger exponents really want a lot of memory for ECM to run efficiently.

Alex
akruppa is offline   Reply With Quote
Old 2004-06-26, 07:30   #3
Khemikal796
 
Feb 2003

37 Posts
Default

Some of the 10mil digit numbers. ie 35xxxxxx+ for p in 2^p-1. Also assuming I find an factor some day. Like I said earlier, I just insert the factor in and restart doing more curves, or do I need to do some cofactor stuff? Wish there was a better helpfile/readme somewhere so I dont have to ask tons of questions.
Khemikal796 is offline   Reply With Quote
Old 2004-06-26, 07:56   #4
akruppa
 
akruppa's Avatar
 
"Nancy"
Aug 2002
Alexandria

2,467 Posts
Default

When Prime95 finds a factor in ECM, it divides out that factor for the following curves, but will forget about it when it is stopped and restarted. So if you want to do all your curves in one long run, there's no need to add found factors to lowm.txt. However, if you stop ECM factoring and want to restart again later, add the factors found so far to lowm.txt so they won't be reported again.

For numbers this large I'd say you should give Prime95 at least half a gigabyte to let ECM run efficiently. But my recommendation would be to first run P-1 to relatively high bound. P-1 gets the exponent p=35xxxxxx for "free" as part of the factor 2kp+1, so the factor will "look" ~7.5 digits smaller to P-1 than to ECM. This gives P-1 a huge head start. Also, stage2 of P-1 uses much less memory than ECM. I think I'd do P-1, B1=10M or even 100M first, then ECM with B1=50k.

Alex
akruppa is offline   Reply With Quote
Old 2004-06-26, 15:01   #5
Khemikal796
 
Feb 2003

458 Posts
Default

So is there a way for me to have the number as small as possible if I have to restart? Only factoring the rest after finding a factor should be faster wouldnt it?
Khemikal796 is offline   Reply With Quote
Old 2004-06-26, 16:10   #6
wblipp
 
wblipp's Avatar
 
"William"
May 2003
New Haven

94316 Posts
Default

Quote:
Originally Posted by Khemikal796
So is there a way for me to have the number as small as possible if I have to restart? Only factoring the rest after finding a factor should be faster wouldnt it?
It would be in general, but the Prime95 FFTs are much faster than any other code available, and part of the code efficiency relies on special tricks related to using a full Mersenne number. So unless your unfactored composite gets very small, Prime95 is the fastest way to go. This advantage is so large that attacks on composite exponents are best managed by trying the full composite and then testing factors against the possible algebraic factors.
wblipp is offline   Reply With Quote
Old 2004-06-28, 03:59   #7
Khemikal796
 
Feb 2003

37 Posts
Default

Ok, so continual ECM is pretty much the only way to completely factor a number using P95?
Khemikal796 is offline   Reply With Quote
Old 2004-06-28, 09:42   #8
smh
 
smh's Avatar
 
"Sander"
Oct 2002
52.345322,5.52471

29×41 Posts
Default

Quote:
Originally Posted by Khemikal796
Ok, so continual ECM is pretty much the only way to completely factor a number using P95?
No, just add the known factors to lowm.txt
smh is offline   Reply With Quote
Reply

Thread Tools


All times are UTC. The time now is 20:47.


Tue Jun 28 20:47:01 UTC 2022 up 75 days, 18:48, 2 users, load averages: 1.10, 1.06, 1.14

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2022, 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.

≠ ± ∓ ÷ × · − √ ‰ ⊗ ⊕ ⊖ ⊘ ⊙ ≤ ≥ ≦ ≧ ≨ ≩ ≺ ≻ ≼ ≽ ⊏ ⊐ ⊑ ⊒ ² ³ °
∠ ∟ ° ≅ ~ ‖ ⟂ ⫛
≡ ≜ ≈ ∝ ∞ ≪ ≫ ⌊⌋ ⌈⌉ ∘ ∏ ∐ ∑ ∧ ∨ ∩ ∪ ⨀ ⊕ ⊗ 𝖕 𝖖 𝖗 ⊲ ⊳
∅ ∖ ∁ ↦ ↣ ∩ ∪ ⊆ ⊂ ⊄ ⊊ ⊇ ⊃ ⊅ ⊋ ⊖ ∈ ∉ ∋ ∌ ℕ ℤ ℚ ℝ ℂ ℵ ℶ ℷ ℸ 𝓟
¬ ∨ ∧ ⊕ → ← ⇒ ⇐ ⇔ ∀ ∃ ∄ ∴ ∵ ⊤ ⊥ ⊢ ⊨ ⫤ ⊣ … ⋯ ⋮ ⋰ ⋱
∫ ∬ ∭ ∮ ∯ ∰ ∇ ∆ δ ∂ ℱ ℒ ℓ
𝛢𝛼 𝛣𝛽 𝛤𝛾 𝛥𝛿 𝛦𝜀𝜖 𝛧𝜁 𝛨𝜂 𝛩𝜃𝜗 𝛪𝜄 𝛫𝜅 𝛬𝜆 𝛭𝜇 𝛮𝜈 𝛯𝜉 𝛰𝜊 𝛱𝜋 𝛲𝜌 𝛴𝜎𝜍 𝛵𝜏 𝛶𝜐 𝛷𝜙𝜑 𝛸𝜒 𝛹𝜓 𝛺𝜔