mersenneforum.org  

Go Back   mersenneforum.org > Prime Search Projects > Conjectures 'R Us

Reply
 
Thread Tools
Old 2018-02-08, 03:38   #12
masser
 
masser's Avatar
 
Jul 2003
Behind BB

7·257 Posts
Default P-1 for R6

For the R6 sequence, Mr. Bitcoin, YoYo and I have completed a rather deep sieve.

On my fastest i5 processor cores, the sieve is now removing candidates at less than 1 every 14 hours.

On a slower i5 (dual core laptop) processor, P-1 factoring removed about 1 out of every 36 candidates. See attached results file. The P-1 (via mprime) work removes candidate at a rate of 1 every 11.5 hours (per laptop core).

On the fast i5, the LLR tests for these candidates take about 14 hours (again, per core).

In a few weeks, I will finish sieving and begin additional testing of P-1 and LLR on the faster i5.

P-1 on this sequence does NOT appear to be a waste of resources.

Is there an easy explanation for the factors found by P-1? Is the P-1 work economical because of the depth of the sieve or because of the size of the tests?
Attached Files
File Type: txt pm1_trial_results.txt (103.5 KB, 360 views)
masser is offline   Reply With Quote
Old 2018-06-07, 19:10   #13
masser
 
masser's Avatar
 
Jul 2003
Behind BB

7×257 Posts
Default More P-1 results for R6

Unfortunately, the (1 out of 36) numbers I posted above were too optimistic. However, P-1 continues to make sense for the time-consuming LLR tests of the R6 sequence. Near n=700,000, LLR tests on my 4 core Haswell take about 180,000 seconds. P-1 work finds a factor approximately every 160,000 seconds.

With B1=70,000, the P-1 runs took 33 minutes per candidate; on average, one factor was found per 80.6 candidates.

With B1=110,000, one factor was found every 50.3 runs. I got lucky with the first 1400 runs and that lead to the 1 out of 36 number previously cited.

Results files attached. Please let me know if you have any suggestions or questions.
Attached Files
File Type: txt R6_pm1_70000.txt (257.5 KB, 524 views)
File Type: txt R6_pm1_110000.txt (348.8 KB, 297 views)

Last fiddled with by masser on 2018-06-07 at 19:12
masser is offline   Reply With Quote
Old 2018-06-07, 22:38   #14
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

295F16 Posts
Default

Quote:
Originally Posted by masser View Post
Unfortunately, the (1 out of 36) numbers I posted above were too optimistic. However, P-1 continues to make sense for the time-consuming LLR tests of the R6 sequence. Near n=700,000, LLR tests on my 4 core Haswell take about 180,000 seconds. P-1 work finds a factor approximately every 160,000 seconds.

With B1=70,000, the P-1 runs took 33 minutes per candidate; on average, one factor was found per 80.6 candidates.

With B1=110,000, one factor was found every 50.3 runs. I got lucky with the first 1400 runs and that lead to the 1 out of 36 number previously cited.

Results files attached. Please let me know if you have any suggestions or questions.

I assume you mean near n=7,000,000.

I don't know what to do with this. I'm not familiar with P-1 testing. Is this removing candidates from testing -or- are you doing actual testing? Let me know what I should do if anything.

Last fiddled with by gd_barnes on 2018-06-07 at 22:38
gd_barnes is online now   Reply With Quote
Old 2018-06-07, 22:44   #15
masser
 
masser's Avatar
 
Jul 2003
Behind BB

7·257 Posts
Default

Doh. You're right; that should have been n=7,000,000.

There's nothing you have to do. I'm just trying to add some data to the P-1 discussion. Ultimately, for high-n bases, it might become a recommended part of the process: sieve, P-1, then test.
masser is offline   Reply With Quote
Old 2018-06-07, 22:51   #16
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

7×17×89 Posts
Default

Quote:
Originally Posted by masser View Post
Doh. You're right; that should have been n=7,000,000.

There's nothing you have to do. I'm just trying to add some data to the P-1 discussion. Ultimately, for high-n bases, it might become a recommended part of the process: sieve, P-1, then test.

Great thanks for info.


The extent of my knowledge of P-1 is that I know it needs to be done for very long tests. Obviously 180,000 seconds (>2 days) is a long test. I'll leave it up to the masses to determine when it is necessary for removing candidates before actual testing is done.
gd_barnes is online now   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
New BOINC framework for factoring jasonp Factoring 4 2011-10-09 11:24
BOINC.BE BATKrikke Teams 2 2010-03-05 18:57
Boinc Xentar Sierpinski/Riesel Base 5 4 2009-04-25 10:26
BOINC? masser Sierpinski/Riesel Base 5 1 2009-02-09 01:10
BOINC bebarce Software 3 2005-12-15 18:35

All times are UTC. The time now is 11:39.


Wed Jan 19 11:39:29 UTC 2022 up 180 days, 6:08, 0 users, load averages: 1.97, 1.54, 1.40

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