mersenneforum.org  

Go Back   mersenneforum.org > New To GIMPS? Start Here! > Information & Answers

Reply
 
Thread Tools
Old 2011-02-25, 00:16   #1
agentcn
 
Feb 2011

3 Posts
Default Questions about 25.11 version

I am freshman about Mersenne Prime. I want to ask some questions. Could you explain it? Please use simple language.
First, The version is 25.11.I check the “type of work to get” in the “worker windows”. There are some choices:
1Whatever make the most sense
2World record sized numbers to test
3First time tests
4Double-check tests
5Trial factoring
6P-1 factoring
7Trial factoring to test
8ECM on small Mersenne number
9ECM on Fermat numbers
10100,000,000 digit numbers to test
So, my questions are as follow
A、 What are the exact meanings about 1 to 10?
B、 Which one is used for discovering new Mersenne Prime?Which one is used for proving new Mersenne Prime?
CWhich one is suit for common computers, high performance and older computer? Which algorithm is most fast
DHow many memory is best for the software?
Second,CPU affinity:
1Run on any CPU
2smart assignment
How to set the CPU affinity?
Third, what are the meanings of the symbols in “PrefersWorkType”? Such as GIMPS, LL, LL-WR, ECM-F, LL-10M,NF?
Fourth, what are the meanings of the symbols in “currentWorkType”?Such as S1, LL?
agentcn is offline   Reply With Quote
Old 2011-02-25, 02:06   #2
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101×103 Posts

2·53·41 Posts
Default

Here is a partial reply, copied from the Mersennewiki.org
These are variations of the above (and appear as listed in the Prime95 menus).
  • Whatever makes the most sense: This lets the PrimeNet server choose what type of assignment the machine will receive.
  • World Record sized number to test: The PrimeNet server will hand out a first time LL test that is larger than the currently known record prime.
  • First time tests: An LL primality test on a number that hasn't previously been tested for primality.
  • Double-check tests: The same test as the Lucas-Lehmer, but performed on a different machine and with a different offset. Ensures that a prime was not missed by error.
  • Trial factoring: Checking for factors of the Mersenne number. A found factor will conclusively prove that the number is composite, which elminates the need to run a primality test.
  • P-1 factoring: A second method that finds factors.
  • Trial factoring to low limits: PrimeNet assigned LMH work. Shorter duration TF work.
  • ECM on small numbers: Doing ECM factoring on Mersenne numbers with exponents below 1,000,000.
  • ECM on Fermat numbers: Doing ECM factoring on Fermat numbers. (This work type does not contribute toward GIMPS main goals.)
  • 100,000,000 digit numbers to test: Performing L-L primality testing on Mersenne numbers that have at least 100-million decimal digits. A prime number found by this worktype would eligible for the next EFF prize.
If you want to discover new Mersenne Primes choose any of the following: First time, world record, or 100,000,000.

The algorithms are all suited to the specific task and are all fast.
Older -> newer (this is only an approximate list)
TF to low limits
ECM on small numbers
TF
Double check
P-1
First time / world record (these are basically the same)
100,000,000 (only if you really want to do this, it will take 12-14 months per test.)

Somebody else can tell you about affinity (I never have changed the default settings).

Quote:
Third, what are the meanings of the symbols in “PrefersWorkType”? Such as GIMPS, LL, LL-WR, ECM-F, LL-10M,NF?
Fourth, what are the meanings of the symbols in “currentWorkType”?Such as S1, LL?
Those just show what type of work (see the explanation above) that the machine wants or uses. GIMPS=what ever makes the most sense, LL=first time, LL-10M=first time on numbers >10million digits.
Uncwilly is online now   Reply With Quote
Old 2011-02-25, 02:23   #3
Unregistered
 

8,363 Posts
Default

Quote:
Originally Posted by Uncwilly View Post
Here is a partial reply, copied from the Mersennewiki.org
These are variations of the above (and appear as listed in the Prime95 menus).
  • Whatever makes the most sense: This lets the PrimeNet server choose what type of assignment the machine will receive.
  • World Record sized number to test: The PrimeNet server will hand out a first time LL test that is larger than the currently known record prime.
  • First time tests: An LL primality test on a number that hasn't previously been tested for primality.
  • Double-check tests: The same test as the Lucas-Lehmer, but performed on a different machine and with a different offset. Ensures that a prime was not missed by error.
  • Trial factoring: Checking for factors of the Mersenne number. A found factor will conclusively prove that the number is composite, which elminates the need to run a primality test.
  • P-1 factoring: A second method that finds factors.
  • Trial factoring to low limits: PrimeNet assigned LMH work. Shorter duration TF work.
  • ECM on small numbers: Doing ECM factoring on Mersenne numbers with exponents below 1,000,000.
  • ECM on Fermat numbers: Doing ECM factoring on Fermat numbers. (This work type does not contribute toward GIMPS main goals.)
  • 100,000,000 digit numbers to test: Performing L-L primality testing on Mersenne numbers that have at least 100-million decimal digits. A prime number found by this worktype would eligible for the next EFF prize.
If you want to discover new Mersenne Primes choose any of the following: First time, world record, or 100,000,000.

The algorithms are all suited to the specific task and are all fast.
Older -> newer (this is only an approximate list)
TF to low limits
ECM on small numbers
TF
Double check
P-1
First time / world record (these are basically the same)
100,000,000 (only if you really want to do this, it will take 12-14 months per test.)

Somebody else can tell you about affinity (I never have changed the default settings).

Those just show what type of work (see the explanation above) that the machine wants or uses. GIMPS=what ever makes the most sense, LL=first time, LL-10M=first time on numbers >10million digits.


"World Record sized number to test" is same as "First time tests"?
  Reply With Quote
Old 2011-02-25, 03:19   #4
KingKurly
 
KingKurly's Avatar
 
Sep 2010
Annapolis, MD, USA

BD16 Posts
Default

agentcn: Consider upgrading to version 26.5, it is new and better and came out just a few days ago.

Uncwilly: I notice the Wiki said that ECM small gives out exponents lower than 1M. I don't think that's necessarily true, I think I've gotten exponents as high as 5M on machines with more RAM. (agentcn, this doesn't really affect you, but I still recommend upgrading to 26.5)
KingKurly is offline   Reply With Quote
Old 2011-02-25, 03:42   #5
cheesehead
 
cheesehead's Avatar
 
"Richard B. Woods"
Aug 2002
Wisconsin USA

22×3×641 Posts
Default

Quote:
Originally Posted by agentcn View Post
1Whatever make the most sense
In addition to what Uncwilly wrote about this type:

The PrimeNet server chooses what type to assign based on your computer's CPU type and speed, how much memory you have allocated to the GIMPS task and what worktype is most useful to the GIMPS project at that time.

If your CPU is a recent model (= "high-performance" in your question C), PrimeNet will probably assign worktype:
#2, #3 or #5 if you have only a small amount of memory allocated (PrimeNet will choose which one would be most useful to GIMPS at that time), or
#6 if you have a large amount of memory allocated.

If your CPU is not very recent, but not very old or slow either (= "common" in your question C), PrimeNet will probably assign worktype:
#4 of #5 if you have only a small amount of memory allocated (PrimeNet will choose which one would be most useful to GIMPS at that time), or
#6 if you have a large amount of memory allocated.

If your CPU is older and/or slower than most others used in GIMPS (= "older" in your question C), PrimeNet will probably assign worktype #5 or #7 (according to which one would be most useful to GIMPS at that time).

Quote:
2World record sized numbers to test
As Uncwilly wrote, PrimeNet will assign a first time LL test on an exponent that is larger than the currently known record prime.

But note that between the time that assignment is made and the time you complete the test, someone else might have found a Mersenne prime that sets a new world record higher than the record was at the time of your assignment, perhaps higher than the exponent you were assigned.

So, we can't absolutely guarantee that your assignment could find a new world record prime at the time you finish your assignment. There will an excellent chance, but not a 100% guaranteed chance.

BTW, it's quite an honor to be the discoverer of any new Mersenne prime, regardless of its size. Fewer than 50 people have ever done that, and all such discoverers are added to the list that is famous.

Quote:
3First time tests
4Double-check tests
< snip >
10100,000,000 digit numbers to test
In worktypes #3, #4 and #10, your assignment might also include a preliminary TF (trial factoring) or P-1 factoring test before it starts the main LL test. Don't be alarmed if you see that your computer has started trial factoring or P-1 factoring even though you asked for a first-time LL test or a double-check LL test.

In each case (TF or P-1), if your computer finds a factor that means you've saved yourself and GIMPS a lot of time, because then the long LL test is no longer needed! Also, once your computer reports that factor to PrimeNet, it will get another exponent assignment much sooner than if you had done the long LL test.

If your computer doesn't find a factor during TF or P-1, you will have saved someone else from doing the TF or P-1 later, and it will proceed to start the first-time LL test or double-check LL test.

Quote:
So, my questions are as follow
< snip >
Which one is used for proving new Mersenne Prime?
Once someone reports discovering a new Mersenne prime with a first-time test, two or more people will run double-checks on very fast computer systems that are not ordinarily used for GIMPS.

Quote:
CWhich one is suit for common computers, high performance and older computer?
See what I wrote in my first comment above about:

"your CPU is a recent model" = high performance

"your CPU is not very recent, but not very old or slow either" = common

"your CPU is older and/or slower than most others used in GIMPS" = older

Quote:
Which algorithm is most fast
For every worktype, GIMPS uses the fastest algorithm that is known for that type of work.

Last fiddled with by cheesehead on 2011-02-25 at 03:53
cheesehead is offline   Reply With Quote
Old 2011-02-25, 15:03   #6
Rodrigo
 
Rodrigo's Avatar
 
Jun 2010
Pennsylvania

11101010002 Posts
Thumbs up

@Uncwilly
@cheesehead

Excellent rundown of the various choices and possibilities.

Rodrigo
Rodrigo is offline   Reply With Quote
Old 2011-03-04, 19:42   #7
Mr. P-1
 
Mr. P-1's Avatar
 
Jun 2003

7·167 Posts
Default

Quote:
Originally Posted by Unregistered View Post
"World Record sized number to test" is same as "First time tests"?
No. Currently there are 1,919 exponents below M43112609 which have never had a first-time test. Most or all of these are assigned, but if they expire or are returned untested, then they will be available for assignment as first-time test, but not as World-record tests.
Mr. P-1 is offline   Reply With Quote
Old 2011-03-06, 03:56   #8
cheesehead
 
cheesehead's Avatar
 
"Richard B. Woods"
Aug 2002
Wisconsin USA

22×3×641 Posts
Default

Quote:
Originally Posted by Unregistered View Post
"World Record sized number to test" is same as "First time tests"?
Another way 2 put it:

When you ask for "First time test", your assignment could be either under or over the current world record.

When you ask for "World Record sized number to test", your assignment will be over the current world record.
cheesehead is offline   Reply With Quote
Old 2011-03-06, 04:30   #9
KingKurly
 
KingKurly's Avatar
 
Sep 2010
Annapolis, MD, USA

33×7 Posts
Default

Quote:
Originally Posted by cheesehead View Post
When you ask for "World Record sized number to test", your assignment will be over the current world record.
...at the time the assignment is made.

It'd be pretty much impossible to offer anything stronger than that, without slowing up the project dramatically.
KingKurly is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Many (gmp-)ecm questions... WraithX GMP-ECM 26 2014-04-03 15:54
A Few Questions wombatman Msieve 13 2013-08-17 16:47
llr 3.8 questions VJS Prime Sierpinski Project 17 2010-03-14 10:08
Questions joblack Math 10 2009-05-01 19:52
Some questions... OmbooHankvald PSearch 3 2005-09-17 19:29

All times are UTC. The time now is 04:15.


Tue Jan 25 04:15:19 UTC 2022 up 185 days, 22:44, 0 users, load averages: 1.89, 1.59, 1.46

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 𝛴𝜎 𝛵𝜏 𝛶𝜐 𝛷𝜙𝜑 𝛸𝜒 𝛹𝜓 𝛺𝜔