mersenneforum.org  

Go Back   mersenneforum.org > Prime Search Projects > No Prime Left Behind

Reply
 
Thread Tools
Old 2010-11-05, 08:14   #1
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

3·2,083 Posts
Arrow Plans and goals for 2011

Hi all,

With the recent rally blowing our estimates for completion of the 5th Drive totally out of the water, Gary and I have done some brainstorming over the last couple of days regarding what NPLB's priority and direction should be in the upcoming year. (That includes what's left of this year as well, since the early completion of the 5th Drive moved things up a bit.)

What we've gathered from the level of participation on the various drives is that people seem to like pushing the lower k-ranges to higher n-values (like the 5th Drive) as opposed to small top-5000 work over large k-ranges (the 11th Drive). Our initial idea was to push the 11th Drive next (based on some possibly half-crazy plan I'd come up with to juggle all of our current drives above the constantly rising top-5000 threshold) but we weren't really keen on the idea of having to pour tons of resources into maintaining the enormous 11th Drive, which to put it bluntly can get rather boring after a while. (There's a certain treadmill effect with low top-5000 ranges...you find new primes almost as fast as they get pushed off.) Much to Gary's relief, the general preference seems to be in another direction and he doesn't have to spend much of next year with all of his machines robotically cranking away on the 11th Drive.

With this in mind, here is the general plan of action we've come up with for NPLB in 2011 and (to some degree) beyond:

1. Finish the 5th Drive (est. this coming Sunday).
2. Finish the 1st Doublecheck Drive (by year end 2010).
3. Polish off the 11th Drive (likely 1-2 weeks) and the k=3010-3200 mini-drive (sometime in December) to n=600K, and drop both at that point. The k=1400-2000 range will remain as our low top-5000 drive.
4. Focus on the 6th Drive until that finishes (target: end of Q2 2011).
5. Focus on the 7th Drive until that finishes (target: end of Q3 or early Q4 2011).
6. Throughout 2011, work on completing the fill-in of non-top-5000 ranges for k=2000-3400 (target: year end 2011).
7. Start a new drive for k=400-600, n=1M-2M sometime in mid-2011.
8. Later in 2011, start n=1M-2M drives for k=600-800 and k=800-1001. At this time, set formal goals for completion of k=300-1001 to n=2M.

Our biggest immediate priority, therefore, is the 6th Drive. Currently this is loaded into LLRnet port 5000 in the vicinity of n=~764.9K. What we will do to ensure that nobody's clients go abruptly idle when port 3000 runs out of 5th Drive work is to load a small amount of 6th Drive work in after it. I will do that sometime tomorrow. From there, anyone who wishes to continue helping out with the 6th Drive via LLRnet can transition their machines to port 5000 at their relative leisure.

Meanwhile, our plan is to keep PRPnet port 9000 as a "roving" server perpetually loaded with whatever our current priority is. The next range loaded into it will also be from the 6th Drive, and we will continue to load 6th Drive work into it until that drive finishes.

You can monitor our progress through the 6th and 7th Drives in our countdown meter thread as with the 5th. (We admins have slacked a little on updating that thread of late, but yes, it will be updated soon, really... )

Side priorities throughout 2011 will be the 10th and 12th Drives. Gary and I are still discussing how we should set targets for the 10th Drive, but the goal for the 12th is simple: close the gap below n=425K in the k=2000-3000 range by year end 2011! We'll also do a similar fill-in for k=3000-3400 somewhere in there; that will be a rather smaller effort and we'll decide how to handle that when the k=2000-3000 fill-in is closer to completion.

We also have the individual-k drive for k=300-400, n=1M-2M which has had a certain niche popularity and offers a fun alternative for those who prefer doing k's on their own. If you'd like to find some megabit primes, check it out here--you can try grabbing a k of your own if you're so inclined, or if you'd prefer a more traditional automated approach you can still find some nice megabits on LLRnet port 3500, in which we are running 6 high-weight k's as a mini-drive.

This next year should prove very fun with lots of big, juicy primes on our radar. We should be able to cap off our efforts for a number of individual subranges, producing those nice big swaths of continuously searched ranges which are the primary goal of our project. We plan to schedule our next rally sometime in January or February (once PrimeGrid's 2011 Challenge schedule comes out we'll have a better idea of what dates are open).

In the project's first year (2008) we surprised ourselves by finishing k=300-1001 to n=600K just a week in advance of our year-end deadline. This year, we beat that by about two months, accomplishing our year-end goal of getting k=400-600 to n=1M well in advance of the deadline. Can we do even better in 2011?

Max
mdettweiler is offline   Reply With Quote
Old 2010-11-05, 09:01   #2
henryzz
Just call me Henry
 
henryzz's Avatar
 
"David"
Sep 2007
Liverpool (GMT/BST)

2×5×601 Posts
Default

Why drop the 11th at 600k? We need not put any effort into it and we can eventually finish it off at our lesure perhaps when all 2011's goals are met. IMO it is better to slowly pick away at something we don't want to do than drop it all together. If the end was further than 650k I would say don't bother but we can finish it in the end. I for one would like to find a few primes at some point. I have only found one top 5000 this year I think. Once the 5th is finished I would like to give it a twirl for a bit.
henryzz is offline   Reply With Quote
Old 2010-11-05, 18:20   #3
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

32·1,283 Posts
Default

?? So, if we do what you described, we're going to "pick away" and then fully restart on the 11th drive at the end of 2011 when the 5000th place prime is likely at n=666666? How many people here want to search for non-top-5000 primes at n>600K? The drive is currently at n=~597K. Are you observing how quickly the 5000th place prime is rising? It's currently at n>570K and continuing to rise rapidly. You can feel free to search it anytime now and there's a very good chance that there are 2-3 primes left in it up to n=600K. But your prime will be gone in 2 months.

We've had virtually no participation in the 11th drive for nearly a year and the drive is huge. When we conceived of it, we never imagined that PrimeGrid would raise the 5000th place prime so quickly. It's been just me struggling to keep the primes above 4000th place, which I was able to do for about 9 months. Now a new prime there is coming in at ~4400th place and dropping off in ~2 months. Just my pulling all of my resources over to the rallies has caused it to lose ground rapidly.

The 5000th place prime will be at n>600K probably by Jan. or Feb. If you would like to find "smallish" top-5000 primes now, the 10th drive for k=1400-2000 at n=~670K is an excellent choice. If we don't drop the 11th drive, then the 10th drive would fairly quickly become in danger of dropping below top-5000 status. Yes 670K is now somewhat "smallish" but it is the best way to go for less-resourced folks who'd like to search for top-5000 primes that won't drop off very quickly. Since it is above all of the n=666666 primes, primes on that drive should stay a while.

We are not abandoning the non-top-5000 12th drive. We're going to fill in all of the non-top-5000 ranges up to n=600K. The project is no prime left behind, which mean no holes will be left in searched ranges. That will always be the case. But we also have to be prudent in adjusting our goals for searching higher n's to meet current realities.


Gary
gd_barnes is offline   Reply With Quote
Old 2010-11-05, 18:55   #4
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

186916 Posts
Default

As Gary said--the 11th Drive has become so hard to keep above the top-5000 threshold that it's not practical to keep it going. Within a few months the 10th Drive will be at the top-5000 entry level, and it can take effectively the same role as the 11th but it won't be quite as hard to maintain.
mdettweiler is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
2011's POTY Xyzzy Lounge 38 2012-06-19 12:36
Largest k*2^n-1 Primes in 2011 Kosmaj Riesel Prime Search 0 2012-01-01 16:52
End of the world May 21st, 2011? jasong Lounge 67 2011-05-30 04:15
LMH goals... Xyzzy Lone Mersenne Hunters 14 2003-08-13 23:02
goals TTn NFSNET Discussion 1 2003-06-18 17:59

All times are UTC. The time now is 18:33.


Sun Nov 27 18:33:30 UTC 2022 up 101 days, 16:02, 0 users, load averages: 0.74, 0.85, 0.86

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