mersenneforum.org  

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

Reply
 
Thread Tools
Old 2020-05-11, 04:03   #1
dantringuoi
 
May 2020

1 Posts
Default Tables need effort

I was trying to figure out if it was worth the effort to certify a prime and I wondered if there was a table somewhere showing the expected effort needed to do so with some program(s).

Come to think of it, this would be useful for a lot of other things as well -- effort needed to find a factor of a given size with ECM, effort needed to crack a composite with NFS, etc.

Does anyone know of a resource like that? Even just a thread?

Lee

Tks!!
dantringuoi is offline   Reply With Quote
Old 2020-05-11, 05:27   #2
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

10110110111102 Posts
Default

Cracking with GNFS is simple- time your rig for a small number like a C100, and then difficulty doubles every 5 to 5.5 digits.

Timing varies so much from machine to machine that it's vastly more accurate to run a C100 yourself than it is to trust someone else's timings. I can say that numbers under C130 can be cracked by even the impatient, while C155+ takes either some patience (like a week at 155 digits) or a fairly powerful rig.

Primality proving via LLR for the usual special forms has difficulty increasing with the square of the exponent- again, run a small test like a 100k exponent, and extrapolate. Timings of a small top-5000 number are an hour or less, a million decimal-digit number 2-4 single-thread hours.

Generating a primality certificate with e.g. Primo for a number of no special form is something I don't know the scaling for- surely someone with experience will chime in!

Edit: For ecm, just use ecm -v to get a chart of # of curves vs factor size, and calculate your own. Again, so much variance from machine to machine (not to mention that often there isn't a factor to be found- so one has to be specific about time to run a t40 or t45 versus the time to actually find a factor in a given list of numbers, or some other goal)

Last fiddled with by VBCurtis on 2020-05-11 at 05:29
VBCurtis is offline   Reply With Quote
Old 2020-05-11, 09:07   #3
paulunderwood
 
paulunderwood's Avatar
 
Sep 2002
Database er0rr

5·937 Posts
Default

Quote:
Originally Posted by VBCurtis View Post
.

Generating a primality certificate with e.g. Primo for a number of no special form is something I don't know the scaling for- surely someone with experience will chime in!
ECPP is O(ln(n)^(4+eps)) meaning a doubling in length takes 16 times as long. This can be helped by having a many-core system.

It took me 21.5 months to certify a 40000 digit number on a 48 core 2.3Ghz Opteron system. If you have time on a AMD 3990x you will be laughing!

Last fiddled with by paulunderwood on 2020-05-11 at 09:11
paulunderwood is offline   Reply With Quote
Old 2020-05-12, 04:48   #4
a1call
 
a1call's Avatar
 
"Rashid Naimi"
Oct 2015
Remote to Here/There

1001010111102 Posts
Default

Quote:
Originally Posted by paulunderwood View Post
If you have time on a AMD 3990x you will be laughing!
Probably not all the way to the bank.

https://www.amazon.com/AMD-Ryzen-Thr...1350078&sr=8-4

But thanks for the info. I was looking for a multi thread computer.
Now if I could only justify the cost. Presently I have cut back on my prime-number-crunching because I am afraid to overheat my systems which I am relying on for telecommuting. It won't be easy/safe to replace them with current pandemic in place.
BTW you had the top record in the world for quite some time my friend, so don't sell your system too short.
a1call is offline   Reply With Quote
Old 2020-05-12, 05:09   #5
paulunderwood
 
paulunderwood's Avatar
 
Sep 2002
Database er0rr

5·937 Posts
Default

Quote:
Originally Posted by a1call View Post
Probably not all the way to the bank.

https://www.amazon.com/AMD-Ryzen-Thr...1350078&sr=8-4

But thanks for the info. I was looking for a multi thread computer.
Now if I could only justify the cost. Presently I have cut back on my prime-number-crunching because I am afraid to overheat my systems which I am relying on for telecommuting. It won't be easy/safe to replace them with current pandemic in place.
BTW you had the top record in the world for quite some time my friend, so don't sell your system too short.
A 3990x will necessary to certify R49081 in a timely manner when Marcel upgrades Primo.

As for my Opteron System: It stopped working and the good parts will be sold on eBay soon.
paulunderwood is offline   Reply With Quote
Reply



Similar Threads
Thread Thread Starter Forum Replies Last Post
Tables with effort needed? CRGreathouse Software 0 2018-02-20 22:21
Attacking the 2+ and 2LM tables xilman Cunningham Tables 28 2013-02-01 21:02
Numbers in Tables wustvn Math 16 2007-11-18 08:20
New ECM Effort Tables for Cunningham Composites Ready garo Factoring 12 2005-09-06 07:53
A question about Cunningham tables T.Rex Factoring 14 2005-05-27 00:27

All times are UTC. The time now is 13:52.


Fri Jul 7 13:52:40 UTC 2023 up 323 days, 11:21, 0 users, load averages: 1.56, 1.33, 1.20

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2023, 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