Message boards : Number crunching : PPS (LLR) Milestones, Goals, and Progress

 Subscribe SortOldest firstNewest firstHighest rated posts first
Author Message
John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 37342 - Posted: 21 Jun 2011 | 22:25:19 UTC

PPS (LLR) Milestones, Goals, and Progress

This thread will be a general continuation of this previous effort: Getting k < 1200 to n=1M before Challenge. We are attempting to fulfill our original intentions with the Proth Prime Search which are two-fold: 1) find primes for k < 10000; 2) push lower k's higher. The effort will now search lower k's to higher n's intermixed with the 1200<k<10000 range from 670K towards 700K. The current groups are as follows:

k<10 to 3M (note k=3 is excluded since it's part of the 321 Prime Search)
10<k<32 to 2.5M
32<k<100 to 2M
100<k<500 to 1.5M
500<k<1200 to 1.3M
1200<k<4000 to 800K
4000<k<10000 to 700K

Note: Lower k's provide better opportunities to find primes with (x)GFN divisibilities and even an elusive Fermat Number divisor.

We are also doing our best to balance both the SGS (LLR) and PPS (LLR) projects as they have a direct impact on each other in regards to the Top 5000 primes list. Please see this post for more details.

The current status is: The leading edge for PPS (LLR) k=5,7,9 has reached n=2.48M (that's 150k in the last 24 hours). Testing times now are ~4.9x's longer than they were at n=1M. They will continue to increase. There are 46k candidates remaining to reach 3M.

____________

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 37361 - Posted: 22 Jun 2011 | 10:18:25 UTC

Progress Update

The leading edge for PPS (LLR) has reached n=2.6M (that's 120k in the last 24 hours). Testing times now are ~5.2x's longer than they were at n=1M. They will continue to increase. There are 36k candidates remaining to reach 3M.

The first success in this extended effort: 9*2^1807574+1 (544135 digits - ranked 111th) found by Reuben Gathright (Reuben Gathright) of the SETI.USA team. A top 30 prime at PrimeGrid. :)

[EDIT] WOW!!! The second success quickly followed: 9*2^2543551+1 (765687 digits - ranked 57th) found by Scott Brown (Scott Brown) of the Duke University team. A top 20 prime at PrimeGrid. :)

Can we bump it back up to 200k/day to finish this off in the next couple of days? Afterwards, our first intermixing will occur as 1200<k<10000 for 670k<n<672k will be inserted.
____________

Microcruncher*
Volunteer tester

Joined: 28 Jun 09
Posts: 391
ID: 42625
Credit: 45,226,534
RAC: 0

Message 37362 - Posted: 22 Jun 2011 | 11:20:46 UTC

Throwing 3 cores at the task (after trashing 24 WUs because I forgot to make the binaries executable after copying them into the project directory and modifying my app_info.xml). One core is reserved for the GPU.
____________

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 37394 - Posted: 23 Jun 2011 | 3:17:48 UTC

Prime Fermat Divisor

Scott finds one of those "elusive" prime Fermat divisors. As hoped, searching the lower k's to higher n's has paid off. After further testing of Scott's k=9 prime, it was shown to divide F2543548. See news post for more info. Testing is complete:

9*2^2543551+1 Divides F2543548
9*2^2543551+1 Divides GF(2543549,3)
9*2^2543551+1 Divides xGF(2543549,3,2)
9*2^2543551+1 Divides xGF(2543549 ,4,3)
9*2^2543551+1 Divides GF(2543549,6)
9*2^2543551+1 Divides xGF(2543549 ,8,3)
9*2^2543551+1 Divides xGF(2543542 ,9,2)
9*2^2543551+1 Divides xGF(2543547 ,9,8)
9*2^2543551+1 Divides GF(2543550,11)
9*2^2543551+1 Divides xGF(2543550,11,2)
9*2^2543551+1 Divides xGF(2543550,11,3)
9*2^2543551+1 Divides xGF(2543550,11,4)
9*2^2543551+1 Divides xGF(2543550,11,6)
9*2^2543551+1 Divides xGF(2543550,11,8)
9*2^2543551+1 Divides xGF(2543550,11,9)
9*2^2543551+1 Divides GF(2543549 ,12)
9*2^2543551+1 Divides xGF(2543550,12,11)

This is not only a success for Scott but for everyone else participating as well. Congratulations to you all!!!
____________

Microcruncher*
Volunteer tester

Joined: 28 Jun 09
Posts: 391
ID: 42625
Credit: 45,226,534
RAC: 0

Message 37398 - Posted: 23 Jun 2011 | 6:49:15 UTC

Congratulations to the discoverer (Scott Brown) and to the PrimeGrid team (John, Lennart and Rytis) for the discovery.
____________

Jim Wilkins

Joined: 4 Sep 07
Posts: 188
ID: 11784
Credit: 78,667,983
RAC: 111,790

Message 37399 - Posted: 23 Jun 2011 | 7:01:57 UTC - in response to Message 37394.

Prime Fermat Divisor

Scott finds one of those "elusive" prime Fermat divisors. As hoped, searching the lower k's to higher n's has paid off. After further testing of Scott's k=9 prime, it was shown to divide F2543548. See news post for more info. Testing is not complete, but we do know the following:
9*2^2543551+1 Divides F2543548
9*2^2543551+1 Divides GF(2543549,3)
9*2^2543551+1 Divides xGF(2543549,3,2)
9*2^2543551+1 Divides xGF(2543550-? ,4,3)
9*2^2543551+1 Divides GF(2543549,6)
9*2^2543551+1 Divides xGF(2543550-? ,8,3)
9*2^2543551+1 Divides xGF(2543550-? ,9,2)
9*2^2543551+1 Divides xGF(2543550-? ,9,8)
9*2^2543551+1 Divides GF(2543550,11)
9*2^2543551+1 Divides xGF(2543550,11,2)
9*2^2543551+1 Divides xGF(2543550,11,3)
9*2^2543551+1 Divides xGF(2543550,11,4)
9*2^2543551+1 Divides xGF(2543550,11,6)
9*2^2543551+1 Divides xGF(2543550,11,8)
9*2^2543551+1 Divides xGF(2543550,11,9)
9*2^2543551+1 Divides GF(2543550-? ,12)
9*2^2543551+1 Divides xGF(2543550,12,11)

? will be updated once testing is complete.

This is not only a success for Scott but for everyone else participating as well. Congratulations to you all!!!

I am somewhat aware of what Fermat numbers are, but what do the notations GF(a,b) and xGF(a,b,c) mean?

____________
Thanks,
Jim

pschoefer
Volunteer developer
Volunteer tester

Joined: 20 Sep 05
Posts: 667
ID: 845
Credit: 2,377,371,150
RAC: 155,183

Message 37402 - Posted: 23 Jun 2011 | 8:49:44 UTC - in response to Message 37399.

I am somewhat aware of what Fermat numbers are, but what do the notations GF(a,b) and xGF(a,b,c) mean?

GF(a,b) is the Generalized Fermat number b^2^a + 1.
xGF(a,b,c) is the extended Generalized Fermat number b^2^a + c^2^a.
____________

Jim Wilkins

Joined: 4 Sep 07
Posts: 188
ID: 11784
Credit: 78,667,983
RAC: 111,790

Message 37413 - Posted: 23 Jun 2011 | 11:51:05 UTC - in response to Message 37402.

I am somewhat aware of what Fermat numbers are, but what do the notations GF(a,b) and xGF(a,b,c) mean?

GF(a,b) is the Generalized Fermat number b^2^a + 1.
xGF(a,b,c) is the extended Generalized Fermat number b^2^a + c^2^a.

Ahh...

____________
Thanks,
Jim

Hammeh

Joined: 11 Nov 08
Posts: 114
ID: 31582
Credit: 4,377,201
RAC: 0

Message 37417 - Posted: 23 Jun 2011 | 14:52:29 UTC

Congratulations guys =D
____________

My Top 5000 Primes:
110059!+1 is prime! (FPS)
14493618614235*2^666666-1 is prime! (SGS)

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 37422 - Posted: 23 Jun 2011 | 16:05:31 UTC

Progress Update

The leading edge for PPS (LLR) has reached n=2.75M (that's 150k in the last 24 hours). Testing times now are ~8.1x's longer than they were at n=1M. They will continue to increase. There are 22k candidates remaining to reach 3M.

A lot can happen in 24 hours. Two primes and a World Record Prime Fermat divisor. Congratulations to everyone for this success. While we didn't bump it back up to 200k/day, 150k/day is quite respectable considering testing times continue to increase.

Note: Next range will be 1200<k<10000 for 670k<n<672k.
____________

Honza
Volunteer moderator
Volunteer tester
Project scientist

Joined: 15 Aug 05
Posts: 1905
ID: 352
Credit: 4,082,937,699
RAC: 3,907,008

Message 37425 - Posted: 23 Jun 2011 | 16:43:34 UTC - in response to Message 37361.

[EDIT] WOW!!! The second success quickly followed: 9*2^2543551+1 (765687 digits - ranked 57th) found by Scott Brown (Scott Brown) of the Duke University team. A top 20 prime at PrimeGrid. :)

Congrats Scott.
I think I'm double-checker. It there a link to computer reported this WU, how long it took etc.?

____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186

Rick Reynolds

Joined: 13 Aug 10
Posts: 8159
ID: 65483
Credit: 6,709,342,304
RAC: 53,411

Message 37428 - Posted: 23 Jun 2011 | 18:23:24 UTC

Congrats to both of you!
____________
@AggieThePew

Christopher Siegert

Joined: 5 Jul 09
Posts: 232
ID: 42986
Credit: 77,011,345
RAC: 0

Message 37431 - Posted: 23 Jun 2011 | 19:30:15 UTC

At http://www.prothsearch.com/fermat.html, I noticed that a Fermat divisor was found on March 12th within PrimeGrid: 7333 Â· 2^138560 + 1 divides F138557.

What's the story behind this discovery? Was a mistake made back in March?

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 37434 - Posted: 23 Jun 2011 | 20:17:03 UTC - in response to Message 37431.

At http://www.prothsearch.com/fermat.html, I noticed that a Fermat divisor was found on March 12th within PrimeGrid: 7333 Â· 2^138560 + 1 divides F138557.

What's the story behind this discovery? Was a mistake made back in March?

In short, it was discovered on 12 March 2011 but was not realized until a double check effort of all Proth primes n<210k revealed it had been overlooked. Please see news post for more info.
____________

Honza
Volunteer moderator
Volunteer tester
Project scientist

Joined: 15 Aug 05
Posts: 1905
ID: 352
Credit: 4,082,937,699
RAC: 3,907,008

Message 37443 - Posted: 24 Jun 2011 | 9:28:21 UTC - in response to Message 37425.

Congrats Scott.
I think I'm double-checker. Is there a link to computer reported this WU, how long it took etc.?

Found it - http://www.primegrid.com/workunit.php?wuid=191438931
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 37480 - Posted: 25 Jun 2011 | 14:12:12 UTC

Progress Update

The leading edge for PPS (LLR) has reached n=2.99M (that's 240k in the last 48 hours). Testing times now are ~8.2x's longer than they were at n=1M. They will continue to increase. There are 1.7k candidates remaining to reach 3M.

We'll complete the last of the 3M work today and enter the next range of 1200<k<10000 for 670k<n<672k. There are 262k WU's in this range.

Thanks for all the help as we begin this new phase of the search. We'll return to low k work (10<k<100 for 1M<n<1.5M) probably in a few weeks.

[EDIT] As of this edit, new work is now going out.
____________

toms83

Joined: 17 Dec 09
Posts: 238
ID: 51975
Credit: 433,804,879
RAC: 0

Message 37492 - Posted: 25 Jun 2011 | 18:58:01 UTC

Testing times increasing pretty fast.

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 37493 - Posted: 25 Jun 2011 | 19:22:39 UTC - in response to Message 37492.

Testing times increasing pretty fast.

Not now

Honza
Volunteer moderator
Volunteer tester
Project scientist

Joined: 15 Aug 05
Posts: 1905
ID: 352
Credit: 4,082,937,699
RAC: 3,907,008

Message 37494 - Posted: 25 Jun 2011 | 19:36:24 UTC

I switched to PSP for the time being (ie. ruby badge), all tasks in high priority (duration correction factor above 20). Let's see how long until it settles down...
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186

Joined: 1 Oct 10
Posts: 256
ID: 69151
Credit: 111,892,661
RAC: 26,778

Message 37510 - Posted: 26 Jun 2011 | 17:06:26 UTC

I did the same thing Honza. I may set my grid back to PPS though if the PSPs aren't gonna run well on my older chips. If the PPS runtimes increase too, I suppose credits will as well? As I recall, PPS never game a ton of credits but it's got a great chance of finding (or in this case, rediscovering) primes.
____________

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 37512 - Posted: 26 Jun 2011 | 18:00:58 UTC

Progress Update

All work for k=5,7,9 up to n=3M is out. Current work is being drawn from the 1200<k<10000 for 670k<n<672k range so testing times have returned to "normal". We'll use n=1M as the base testing time. These WU's are almost 2x's faster than tests at n=1M. Depending on completion rates, n may be increased to n=675k.

Thanks for all the help as we begin this new phase of the search. We'll return to low k work (10<k<100 for 1M<n<1.5M) once the current range is out.

FYI, k<1200 for n<1M is complete. This is a major accomplishment! Congratulations everyone!!!
____________

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 37929 - Posted: 4 Jul 2011 | 10:50:59 UTC

Progress Update

All work is out for 1200<k<10000 for 670k<n<675k range. Yes, the range went quickly so it was increased to 675k.

We have returned to low k work (10<k<100 for 1M<n<1.5M). Testing times are back up to the 1M baseline. In fact, we are already at 1.14M so times are about 1.3x's that at 1M. These will continue to increase as n increases.

A lot of the work up to 1.2M will be double check work from previous efforts. A bit more first pass will open up from 1.2M-1.4M. Afterwards, over half the range will be first pass.

Thanks again for all your help and support.
____________

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 38222 - Posted: 12 Jul 2011 | 2:06:37 UTC

Progress Update

k=5, 7, 9 for 1M<n<3M is COMPLETE!
1200<k<10000 for 670k<n<675k is COMPLETE!

EXCELLENT work!!! Well done everyone.

10<k<100 for 1M<n<2M is the current range. Progress now is at n=1375289. Testing times are about 1.8x's that at 1M. These will continue to increase as n increases.

Our double check work is getting less and less as you can see here:

31% first pass at 1.2M
44% first pass at 1.3M
62% first pass at 1.4M
69% first pass at 1.5M
78% first pass at 1.6M
87% first pass at 1.8M
91% first pass at 2.0M
96% first pass at 2.5M

At n=2M, we'll pause again and complete a PPSE range of 1200<k<10000 for 675k<n<680k or 685k. Afterwards, we are looking at 10<k<32 to n=2.5M.

Thanks again for all your help and support.
____________

Omega

Joined: 20 Apr 08
Posts: 181
ID: 21694
Credit: 27,596,829
RAC: 1

Message 39127 - Posted: 12 Aug 2011 | 16:37:58 UTC

Looks like progress of 10<k<100 is near n=2M...
Is the plan still to work on 1200<k<10000 afterwards..?

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 39170 - Posted: 14 Aug 2011 | 15:49:08 UTC - in response to Message 39127.

I have a lot of PPS work units that are being given to me without a wing man. Will those be double checked eventually?
And...

Looks like progress of 10<k<100 is near n=2M...
Is the plan still to work on 1200<k<10000 afterwards..?

I am also interested in the answer to this.
____________

Pooh Bear 27

Joined: 10 May 09
Posts: 732
ID: 39821
Credit: 609,349,634
RAC: 3,751,423

Message 39171 - Posted: 14 Aug 2011 | 15:59:32 UTC - in response to Message 39170.

I have a lot of PPS work units that are being given to me without a wing man. Will those be double checked eventually?

PPS and SGS (and I think one other) have adaptive handout. If your machine proves itself for a set amount of period it can be considered good enough to do work without a wingperson. It will get dual jobs and most of the time you will be the second receiver to a person who's not been deemed good enough to work on their own.

Enjoy the pleasures of getting work that is immediately validated.

Omega

Joined: 20 Apr 08
Posts: 181
ID: 21694
Credit: 27,596,829
RAC: 1

Message 39172 - Posted: 14 Aug 2011 | 16:00:45 UTC - in response to Message 39170.

I have a lot of PPS work units that are being given to me without a wing man. Will those be double checked eventually?

No, SGS (LLR), PPS (LLR), and TRP (LLR) use Adaptive replication. That's why sometimes you see a minimum quorum of 1.

Gary Craig
Volunteer tester

Joined: 30 Dec 09
Posts: 3213
ID: 52890
Credit: 1,005,618,748
RAC: 0

Message 39183 - Posted: 14 Aug 2011 | 23:37:10 UTC - in response to Message 39172.

I *think* that any WU where the result is a prime is double-checked, even if the initial replication is 1.

--Gary

Vato
Volunteer tester

Joined: 2 Feb 08
Posts: 796
ID: 18447
Credit: 385,651,415
RAC: 304,980

Message 39186 - Posted: 15 Aug 2011 | 8:39:49 UTC - in response to Message 39183.

I guess the question is more to do with if there is double-checking eventually for the candidates where adaptive replication only sent out one WU and it was reported as composite. There could be missing primes in there. If you're just looking for top 5000 primes, it doesn't matter too much per se, but it does mean that recording the limits of the search are more complex. It does matter a lot more for conjecture proving, and also as n increases the error rates get higher.
____________

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 39194 - Posted: 15 Aug 2011 | 20:07:24 UTC - in response to Message 39186.

I guess the question is more to do with if there is double-checking eventually for the candidates where adaptive replication only sent out one WU and it was reported as composite.

This was my question.
____________

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 39229 - Posted: 17 Aug 2011 | 15:41:24 UTC

Hi all crunchers :)

We only have ~13k numbers left on PPS (n=1M-2M) so all help is needed to finish them.

If you can ? please switch to PPS LLR to get all wu's out there done.

I will raise buffer on frontpage to 10k That way you can see progress.

When it get close to zero I will load PPS n= 675k-680k.

Why ? We have a challenge soon :) We need to try to get all 1M-2M done
before the challenge

Thanks

Lennart

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 39231 - Posted: 17 Aug 2011 | 17:45:28 UTC - in response to Message 39229.

I've got it chugging along on PPS LLR, I should hit silver on it by the end of the challenge.
____________

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 39236 - Posted: 17 Aug 2011 | 19:21:25 UTC - in response to Message 39231.

Thanks for all help.

We are now down below 10k

Lennart

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 39239 - Posted: 17 Aug 2011 | 23:30:53 UTC - in response to Message 39236.

What range is the next stage going to be?
____________

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 39240 - Posted: 18 Aug 2011 | 0:10:35 UTC - in response to Message 39239.

First we will do n=675k-680k (maybe higher)

After that we follow the plan k= 100-500 up to n=1.5M

Lennart

roberto4u2*

Joined: 11 Aug 11
Posts: 24
ID: 108167
Credit: 60,880,364
RAC: 0

Message 39241 - Posted: 18 Aug 2011 | 0:13:58 UTC - in response to Message 39240.

I'm pretty new to this whole thing. Will the crunching times go up with the increase in k, or is the amount of actual work done in a WU the same? (that should show my ignorance of the math behind it)...

Just trying to figure out if it is the size of "n" or the size of "k" which has the most impact on crunching times...

Thx

Edit: I said "size" when I really meant "k"

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 39244 - Posted: 18 Aug 2011 | 1:15:26 UTC - in response to Message 39241.

I'm pretty new to this whole thing. Will the crunching times go up with the increase in k, or is the amount of actual work done in a WU the same? (that should show my ignorance of the math behind it)...

Just trying to figure out if it is the size of "n" or the size of "k" which has the most impact on crunching times...

Thx

Edit: I said "size" when I really meant "k"

It is the size of n that increase time. There is a little increase when k's get bigger but nothing you will notice here on pps.

The range we will start now (675k-680k) is much faster then those we are running now ( n=1M-2M)

Lennart

mdettweiler

Joined: 25 Apr 07
Posts: 271
ID: 7489
Credit: 7,617,052
RAC: 2,537

Message 39246 - Posted: 18 Aug 2011 | 1:24:46 UTC - in response to Message 39241.

I'm pretty new to this whole thing. Will the crunching times go up with the increase in k, or is the amount of actual work done in a WU the same? (that should show my ignorance of the math behind it)...

Just trying to figure out if it is the size of "n" or the size of "k" which has the most impact on crunching times...

Thx

Edit: I said "size" when I really meant "k"

Short answer: just the n. The k's effect is essentially negligible.

It's mostly the n that makes the difference in crunching time. Think of it this way: if you have two sequences of numbers, like this:

1*2=2; 2*2=4; 3*2=6; 4*2=8; 5*2=10
versus
2^1=2; 2^2=4; 2^3=8; 2^4=16; 2^5=32

the latter increases in size much more quickly, being an exponential sequence instead of a linear one. The numbers being tested in the Proth Prime Search are defined by the sequence k*2^n+1. Since n is an exponent and k merely a multiplier, the n will have a much greater effect on the size of the number, particularly as we generalize this to large values of k and n. The k does still have an effect, but it is minuscule compared to that of the n. And since crunching time is proportional to the size of the number (read: number of digits), it is by extension affected primarily by the n.

Note that the k does have a secondary effect which can make a more significant difference in crunching time. For calculations of this type, the most efficient way to run them on a computer is to use what's called a Fast Fourier Transform (FFT). The full details of such an algorithm are, needless to say, very complicated, but the part relevant to this discussion is that computers are programmed to handle FFTs in such a way that their sizes increase in a stepwise manner. It's somewhat like gears in a car with a manual transmission, where (for instance) first gear is good for everything up to 15 MPH, second gear up to 20-25 MPH, third gear up to 35-40 MPH, and so on. You can drive at 40 MPH in first gear, but you'll waste a lot of gas and probably ruin your engine; the reverse, however, is not true, for you'll stall the car if you drive at 10 MPH in third gear. With FFTs, you can safely use too big an FFT (at the expense of additional crunching time), but using too small an FFT will produce an inaccurate result. The trick is to find the lowest FFT that you can use safely. The LLR code at the core of PrimeGrid's worker application does this automatically, and makes its decision by looking at the k and n and comparing them with a hardcoded table of what's safe to run with what FFT. So the k does have some effect on the crunching time, though still not so great an effect as the n; and the difference still evens out to an exponential progression as n increases with k held constant. For this reason, the lowest k's are considered the most desirable, because you can find primes of the same size with less work. That said, the difference isn't huge until the size of k increases greatly (for instance, k=500 will not take much longer to test than k=10, but k=100000 will take noticeably longer).

Hope this helps! :-)

roberto4u2*

Joined: 11 Aug 11
Posts: 24
ID: 108167
Credit: 60,880,364
RAC: 0

Message 39247 - Posted: 18 Aug 2011 | 1:54:51 UTC - in response to Message 39246.

Actually both answers were helpful. Thanks to you both! My trade is as a systems administrator and starter programmer, not as a scientist (or mathematician more accurately). Wikipedia is woefully full of too much information on primes in general to grasp in one go (and generally terrible at describing the different search algos from a computing perspective) :)

Any suggestions on what sub-project could benefit most from an additional 336 2.6Ghz Nehalem cores crunching on them (That's sort of tongue in cheek since I'm posting to the PPS LLR status thread)? I've got an (actual) small server farm to burn-in which comes online within the next couple of weeks if I do my job right :p . Most of the project status posts (and the preferences pages) seem to be out of date.

The good news, is that the administration of the server farm is under my control (including whether or not it runs primegrid after business hours :D).

Thanks

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 39359 - Posted: 22 Aug 2011 | 15:55:52 UTC - in response to Message 39240.

First we will do n=675k-680k (maybe higher)

After that we follow the plan k= 100-500 up to n=1.5M

Lennart

We're already at 680K, so John's suggested 685K looks
likely if we're going to have any of these left for the Challenge?

I've been running TRP LLR's on the 32-bit blade cluster
(looking for "gold"), but am planning on finishing those
before/during the Challenge, to get mostly PPS LLR's.

-Bruce*

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 39394 - Posted: 23 Aug 2011 | 22:57:17 UTC

Thank you all :)

All wu's pps 1M-2M is done

Lennart
____________

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 39398 - Posted: 24 Aug 2011 | 2:37:26 UTC - in response to Message 39240.

First we will do n=675k-680k (maybe higher)

After that we follow the plan k= 100-500 up to n=1.5M

Lennart

I though that you were referring to n=680k-685k (which
from 682k already. Any chance "maybe higher" might stretch
to 690k?

Nothing wrong with some portion of the contest running in
"k= 100-500 up to n=1.5M", but I can certainly confirm that
these n < 700k are giving a lot more hits (of the type appealing
to people dropping in for the contest) than the larger n's we
were running just before. -Bruce*

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 39418 - Posted: 24 Aug 2011 | 17:52:44 UTC - in response to Message 39398.

First we will do n=675k-680k (maybe higher)

After that we follow the plan k= 100-500 up to n=1.5M

Lennart

I though that you were referring to n=680k-685k (which
from 682k already. Any chance "maybe higher" might stretch
to 690k?

Nothing wrong with some portion of the contest running in
"k= 100-500 up to n=1.5M", but I can certainly confirm that
these n < 700k are giving a lot more hits (of the type appealing
to people dropping in for the contest) than the larger n's we
were running just before. -Bruce*

We will continue to at least 690K or 695k :)

Lennart

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 39435 - Posted: 25 Aug 2011 | 10:30:46 UTC - in response to Message 39418.

First we will do n=675k-680k (maybe higher)

After that we follow the plan k= 100-500 up to n=1.5M

Lennart

I though that you were referring to n=680k-685k (which
from 682k already. Any chance "maybe higher" might stretch
to 690k?

Nothing wrong with some portion of the contest running in
"k= 100-500 up to n=1.5M", but I can certainly confirm that
these n < 700k are giving a lot more hits (of the type appealing
to people dropping in for the contest) than the larger n's we
were running just before. -Bruce*

We will continue to at least 690K or 695k :)

Lennart

Hi all PPS freak :)

We did load all up to 700k.

After 700k ther are 2-3 options. We will decide after the challenge what will come after 700k.

Happy crunching...

Lennart

____________

rogue
Volunteer developer

Joined: 8 Sep 07
Posts: 1223
ID: 12001
Credit: 18,565,548
RAC: 0

Message 39838 - Posted: 31 Aug 2011 | 12:44:51 UTC - in response to Message 39435.

Hi all PPS freak :)

We did load all up to 700k.

After 700k ther are 2-3 options. We will decide after the challenge what will come after 700k.

Apparently the decision was already made as I see tasks for n>700000 going out for k>1200. What was that decision?

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 39861 - Posted: 31 Aug 2011 | 16:57:57 UTC - in response to Message 39838.

After 700k ther are 2-3 options. We will decide after the challenge what will come after 700k.

Apparently the decision was already made as I see tasks for n>700000 going out for k>1200. What was that decision?

This doesn't sound like the previously mentioned

"After that we follow the plan k= 100-500 up to n=1.5M"

looks more like

"1200<k<4000 to 800K". -Bruce*

rogue
Volunteer developer

Joined: 8 Sep 07
Posts: 1223
ID: 12001
Credit: 18,565,548
RAC: 0

Message 39862 - Posted: 31 Aug 2011 | 17:49:05 UTC - in response to Message 39861.

Apparently the decision was already made as I see tasks for n>700000 going out for k>1200. What was that decision?

This doesn't sound like the previously mentioned

"After that we follow the plan k= 100-500 up to n=1.5M"

looks more like

"1200<k<4000 to 800K". -Bruce*[/quote]

Lennart let me know via PM that he had to load up to n=710000 to ensure that the server wouldn't run out of work during the challenge.

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 39879 - Posted: 31 Aug 2011 | 22:11:56 UTC - in response to Message 39862.

...
Lennart let me know via PM that he had to load up to n=710000 to ensure that the server wouldn't run out of work during the challenge.

Thanks; so I can stop counting 9's to find when the next range
starts arriving. That's n=690xxx, n=6990xx, n=69990x that I've
been seeing in my valid task list, PPS LLR. -Bruce

Arjant2

Joined: 14 Aug 10
Posts: 198
ID: 65556
Credit: 215,419,354
RAC: 2

Message 40063 - Posted: 7 Sep 2011 | 10:37:10 UTC

I was looking for an explanation why my PPS LLR units suddenly increase from a 9 minute runtime to 80 minutes runtime.
I guess this has something to do with it?

____________

[DPC]Charley

Joined: 15 Apr 11
Posts: 803
ID: 95137
Credit: 94,993,523
RAC: 0

Message 40064 - Posted: 7 Sep 2011 | 10:53:04 UTC - in response to Message 40063.

I was looking for an explanation why my PPS LLR units suddenly increase from a 9 minute runtime to 80 minutes runtime.
I guess this has something to do with it?

Yes and no. Yes it is because of the larger n's, but not the ones mentioned here if I'm not mistaken. I guess you're getting pps_llr instead of pps_llr_extended, like me and some of other power cows on our own forum? If you look at the results you'll see that _extended is with n's in the range of 700k - 7??k, but the non extended are for n's over 2M. This apparently costs a lot more time.

Crun-chi
Volunteer tester

Joined: 25 Nov 09
Posts: 3059
ID: 50683
Credit: 63,361,547
RAC: 0

Message 40068 - Posted: 7 Sep 2011 | 12:08:52 UTC - in response to Message 40064.

More time, more credits: so it looks fair enough :)

____________
92*10^1439761-1 NEAR-REPDIGIT PRIME :) :) :)
4â€Š*â€Š650^498101-1 CRUS PRIME
314187728^131072+1 GENERALIZED FERMAT
Proud member of team Aggie The Pew. Go Aggie!

Arjant2

Joined: 14 Aug 10
Posts: 198
ID: 65556
Credit: 215,419,354
RAC: 2

Message 40069 - Posted: 7 Sep 2011 | 12:25:21 UTC - in response to Message 40068.

Credit is not the most important for me on LLR, I am more interested in finding primes, so just wanted to know what the impact is of these longer WU's for finding a prime.
If the chance prime/not prime stays the same, then the chance of finding a prime has decreased (ie. 9 times smaller).
____________

[DPC]Camulos

Joined: 23 Mar 11
Posts: 86
ID: 91898
Credit: 104,976,575
RAC: 0

Message 40070 - Posted: 7 Sep 2011 | 12:42:10 UTC

I am with Arjant2 :D just wanna find primes :)

TheDawgz

Joined: 7 Jul 08
Posts: 2235
ID: 25193
Credit: 600,923,360
RAC: 155,910

Message 40078 - Posted: 7 Sep 2011 | 16:10:21 UTC

TheDawgz are frequently fuzzy on these type of things so please correct us if we are wrong ... but
It looks like we have moved into a range that John had mentioned back in July consisting of low k's and n's > 2M.
If so, then this should be mostly 1st pass work and the primes should make the Top 100 with digit counts >600k.
____________
There's someone in our head but it's not us.

Rick Reynolds

Joined: 13 Aug 10
Posts: 8159
ID: 65483
Credit: 6,709,342,304
RAC: 53,411

Message 40079 - Posted: 7 Sep 2011 | 16:13:03 UTC - in response to Message 40078.

TheDawgz are frequently fuzzy on these type of things so please correct us if we are wrong ... but
It looks like we have moved into a range that John had mentioned back in July consisting of low k's and n's > 2M.
If so, then this should be mostly 1st pass work and the primes should make the Top 100 with digit counts >600k.

Looks like a mixture of ranges on my end. I've got wu's running between 15 minutes and 1:30 hours on the same machine. So, here's hoping to find one of the "big" primes.

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 40085 - Posted: 7 Sep 2011 | 17:32:24 UTC - in response to Message 40063.

I was looking for an explanation why my PPS LLR units suddenly increase from a 9 minute runtime to 80 minutes runtime.
I guess this has something to do with it?

Our apologies. The tasks were from our next range: 10<k<32 from 2M<n<2.5M The work was uploaded yesterday and the WU generator started sending them out prematurely. There's still ~100k tasks remaining in the 709k-710k range. The issue has been resolved and the work has returned to that range.

Here's an update to the remaining PPS ranges on the todo list:
10<k<32 from 2M<n<2.5M
100<k<500 from 1M<n<1.5M
500<k<1200 from 1M<n<1.3M
1200<k<4000 from 710k<n<800k

Additionally, we are looking at splitting out the single digit k's to achieve the following levels:

3 @ 7M (321 Prime Search - ongoing effort)
5 @ 5M
7 @ 4M
9 @ 3M (already at 3M)

So basically, it's only 5 & 7 that would increase. Our effort remains to search lower k's to higher n's intermixed with work from the 1200<k<10000 lower n ranges.
____________

Arjant2

Joined: 14 Aug 10
Posts: 198
ID: 65556
Credit: 215,419,354
RAC: 2

Message 40098 - Posted: 7 Sep 2011 | 20:28:07 UTC

Thanx John for clarifying.
____________

rogue
Volunteer developer

Joined: 8 Sep 07
Posts: 1223
ID: 12001
Credit: 18,565,548
RAC: 0

Message 40117 - Posted: 7 Sep 2011 | 21:49:06 UTC - in response to Message 40085.

Additionally, we are looking at splitting out the single digit k's to achieve the following levels:
3 @ 7M (321 Prime Search - ongoing effort)
5 @ 5M
7 @ 4M
9 @ 3M (already at 3M)

As I maintain prothsearch.com, I recommend taking 5, 7, and 9 to 5M.

Rick Reynolds

Joined: 13 Aug 10
Posts: 8159
ID: 65483
Credit: 6,709,342,304
RAC: 53,411

Message 40157 - Posted: 8 Sep 2011 | 17:49:52 UTC

Have we moved into the newer ranges now? I notice most of my wu's are now taking much longer to run.

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 40169 - Posted: 9 Sep 2011 | 1:45:18 UTC - in response to Message 40166.

Just because you add or subtract 1 from a given number, you do not necessarily end up with a prime number.

But in my guess it could be either even or odd.

If they are even, they should not be prime numbers.

Put that number into a formula and try deduce whether it is prime or not.

Anyway, I just put it together a little bit and wrote 37*49*67*73 into my calculator.

Only 49 (7*7) is not a prime, the other three numbers are.

So, if a given number happens to possibly return a factor in it, that number is not prime, but what about any number in it (like 73)? They could be prime numbers on their own.

The fundamental theorem of arithmetic says that every integer greater than one can be written as a unique product (up to ordering) of primes. So the prime factorization of the number you gave is 7*7*37*67*73.
____________

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 40179 - Posted: 9 Sep 2011 | 15:24:20 UTC - in response to Message 40157.

Have we moved into the newer ranges now? I notice most of my wu's are now taking much longer to run.

Yes. 10<k<32 from 2M<n<2.5M is now in the queue. It will be followed by 500<k<1200 from 1M<n<1.3M and then an n range in 1200<k<4000 from 710k<n<800k.
____________

[DPC]Charley

Joined: 15 Apr 11
Posts: 803
ID: 95137
Credit: 94,993,523
RAC: 0

Message 40181 - Posted: 9 Sep 2011 | 15:55:43 UTC - in response to Message 40166.

Just because you add or subtract 1 from a given number, you do not necessarily end up with a prime number.

But in my guess it could be either even or odd.

[...].

The number they add or subtract 1 to/from is always even, because it is a multiple of 2. Therefor every number tested must be odd, exactly because every even number other than 2 can not be prime.

____________
PrimeGrid Challenge Overall standings --- Last update: From Pi to Paddy (2016)

Rick Reynolds

Joined: 13 Aug 10
Posts: 8159
ID: 65483
Credit: 6,709,342,304
RAC: 53,411

Message 40183 - Posted: 9 Sep 2011 | 16:08:58 UTC - in response to Message 40179.

Have we moved into the newer ranges now? I notice most of my wu's are now taking much longer to run.

Yes. 10<k<32 from 2M<n<2.5M is now in the queue. It will be followed by 500<k<1200 from 1M<n<1.3M and then an n range in 1200<k<4000 from 710k<n<800k.

Great, thanks for the update.

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 40216 - Posted: 10 Sep 2011 | 20:36:11 UTC

Progress Update

Progress through the 10<k<32 for 2M<n<2.5M range has been surprisingly quick. We are currently at 2.375M after only a couple of days. At this rate, we should reach 2.5M in 3-4 days. That is quite IMPRESSIVE!!!

Next up is the 500<k<1200 for 1M<n<1.3M range. We are skipping over the 100<k<500 for 1M<n<1.5M range (for now) to give participants at the Proth Search Site time to complete their reservations. Afterwards, we'll complete a section of the 1200<k<4000 to 800K range.

Proth Prime Search Update
k=3 to 25M is ongoing in the 321 Prime Search (currently at n=7457987)
k=5 to 5M (currently at n=3M)
k=7 to 4M (currently at n=3M)
k=9 to 3M (currently at n=3M)
10<k<32 to 2.5M is ongoing (currently at n=2004074)
32<k<100 to 2M (currently at n=2M)
100<k<500 to 1.5M (currently at n=1M)
500<k<1200 to 1.3M (currently at n=1M)
1200<k<4000 to 800K is ongoing (currently at n=691891)
4000<k<10000 to 700K is ongoing (currently at n=691891)
____________

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 40241 - Posted: 12 Sep 2011 | 17:12:43 UTC

Progress Update

Faster than expected, work for the 10<k<32 for 2M<n<2.5M range has been emptied. We are now in the 500<k<1200 for 1M<n<1.3M range. Next up we'll complete a section of the 1200<k<4000 to 800K range...most likely 710k-725k, more if necessary.
____________

JackTheSkipper

Joined: 12 Aug 05
Posts: 18
ID: 251
Credit: 42,661,011
RAC: 0

Message 40264 - Posted: 13 Sep 2011 | 9:39:49 UTC

Hi all,

I have some questions concerning PPS:

When i look at the range statistics page, I wonder why there are only 1186 tasks completed for k=4801 when 21210 are done for k=4785 ?

On the 5, 000 largest known prime list, I read for instance that 27*2^2218064+1 is prime and it was found in 2009, but it is in the n=2-2.5M range we just finished to check. So, my first question is what project found this prime in 2009 ? The second is how do you take this former discovery into account ?

Fab

____________

Honza
Volunteer moderator
Volunteer tester
Project scientist

Joined: 15 Aug 05
Posts: 1905
ID: 352
Credit: 4,082,937,699
RAC: 3,907,008

Message 40268 - Posted: 13 Sep 2011 | 14:24:22 UTC

I found mentioned prime on PRPNET (27121 Prime Search subproject: k=27 & 121 for k*2^n+-1 for n<10M)

This particular range of k=27 (now n~3,77M) was already covered, same apply for k=121 (now n~4.39M)
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 40269 - Posted: 13 Sep 2011 | 16:44:10 UTC - in response to Message 40264.

When i look at the range statistics page, I wonder why there are only 1186 tasks completed for k=4801 when 21210 are done for k=4785 ?

The k's have different densities. After sieving, k's have a different number of candidates remaining. In this case, k=4785 has a great deal more candidates remaining than k=4801.

Nash weights are typically used to compare remaining candidate densities between k's. Chris Nash created the Nash Sieve. From this, in collaboration with Paul Jobling, came psieve. psieve offered an option that would calculate the Nash weight of k's. This is simply the count of those n remaining after applying the Nash sieve. The standard Nash weight is calculated over the range 100000 < n < 110000.

Another comparison tool is Jack Brennen's Proth Weight applet. If you input 4785 to 4801 with step 2, you'll see for that range that k=4785 is the highest weight and 4801 is the lowest weight.

Note that these are general comparison tools.

On the 5, 000 largest known prime list, I read for instance that 27*2^2218064+1 is prime and it was found in 2009, but it is in the n=2-2.5M range we just finished to check. So, my first question is what project found this prime in 2009 ? The second is how do you take this former discovery into account ?

As Honza points out, 27*2^2218064+1 was found in a previous PrimeGrid effort in PRPNet. When new work is loaded, we mark all known primes so they are not tested again.

Additionally, there are individual searchers who are prime finding on their own. We invite all individual searchers to upload their residues to save PrimeGrid time in (double) checking their ranges.
____________

Pooh Bear 27

Joined: 10 May 09
Posts: 732
ID: 39821
Credit: 609,349,634
RAC: 3,751,423

Message 40270 - Posted: 13 Sep 2011 | 17:03:14 UTC - in response to Message 40269.

When new work is loaded, we mark all known primes so they are not tested again.

Most of the time. There are those few that sneak through every so often. Take it from someone who had one :-)

JackTheSkipper

Joined: 12 Aug 05
Posts: 18
ID: 251
Credit: 42,661,011
RAC: 0

Message 40271 - Posted: 13 Sep 2011 | 17:30:56 UTC - in response to Message 40269.

Merci :)
____________

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 40731 - Posted: 24 Sep 2011 | 19:01:11 UTC

Progress Update

Progress through the 500<k<1200 for 1M<n<1.3M range has been trudging along. Today we have reached n=1.1M. At that rate, we have AT LEAST 24 more days to reach 1.3M.

Next up is the 1200<k<4000 to 800K range (710k-725k). It is probably while we're in this range that we'll see the end to SGS single primes making it into the Top 5000 list. :(

Proth Prime Search Update
k=3 to 25M is ongoing in the 321 Prime Search (currently at n=7505609)
k=5 to 5M (currently at n=3M)
k=7 to 4M (currently at n=3M)
k=9 to 3M (currently at n=3M)
10<k<32 to 2.5M (currently at n=2.5M)
32<k<100 to 2M (currently at n=2M)
100<k<500 to 1.5M (currently at n=1M)
500<k<1200 to 1.3M is ongoing (currently at n=1.1M)
1200<k<4000 to 800K (currently at n=710k)
4000<k<10000 to 700K (currently at n=710k)

FYI, as of this post, n entry levels in the Top 5000 list:

350th - n=1338620
400th - n=1269198
500th - n=1168975
600th - n=1086798
____________

[AF>Libristes] nico8313

Joined: 19 Jan 09
Posts: 2
ID: 34414
Credit: 183,207,042
RAC: 0

Message 41523 - Posted: 8 Oct 2011 | 9:47:06 UTC

Hi,

Proth Prime Search (LLR)

The beta cuda is ready ???
OpenCL is used ??

thanks

rroonnaalldd
Volunteer developer
Volunteer tester

Joined: 3 Jul 09
Posts: 1213
ID: 42893
Credit: 34,634,263
RAC: 0

Message 41524 - Posted: 8 Oct 2011 | 10:01:33 UTC - in response to Message 41523.

____________
Best wishes. Knowledge is power. by jjwhalen

Omega

Joined: 20 Apr 08
Posts: 181
ID: 21694
Credit: 27,596,829
RAC: 1

Message 42347 - Posted: 30 Oct 2011 | 14:51:16 UTC

The leading edge is now at n=1280000 close to the 1,3M edge.
If the suspension of PPS prime reporting continues, may I suggest PPS continues at k=5 (and/or 7) instead of 1200<k<4000 since there are fewer primes in the higher n range (lower k)...

Omega

Joined: 20 Apr 08
Posts: 181
ID: 21694
Credit: 27,596,829
RAC: 1

Message 42561 - Posted: 3 Nov 2011 | 16:53:33 UTC - in response to Message 42347.

The leading edge is now at n=1296000, hours away from the 1,3M cut.
Since the PPS prime reporting halt is still ongoing, i'd like to re-visit my suggestion below to continue with lower k's (5 and/or 7) instead of 1200<k<4000....

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 42563 - Posted: 3 Nov 2011 | 17:42:58 UTC - in response to Message 40085.

...
Here's an update to the remaining PPS ranges on the todo list:
10<k<32 from 2M<n<2.5M
100<k<500 from 1M<n<1.5M
500<k<1200 from 1M<n<1.3M
1200<k<4000 from 710k<n<800k

Additionally, we are looking at splitting out the single digit k's to achieve the following levels:

3 @ 7M (321 Prime Search - ongoing effort)
5 @ 5M
7 @ 4M
9 @ 3M (already at 3M)

So basically, it's only 5 & 7 that would increase. Our effort remains to search lower k's to higher n's intermixed with work from the 1200<k<10000 lower n ranges.

Alternatively, if 5 & 7 were split out, people could switch from PPS LLR to the
5-split or 7-split projects. I might run some of each; keeping an eye on the top20
list (by number, not by score! tied at twelve, but with enough backlog for 11th).

Meanwhile, the return of electric here has produced a SGS hit; my first since April 14.

-Bruce*

Omega

Joined: 20 Apr 08
Posts: 181
ID: 21694
Credit: 27,596,829
RAC: 1

Message 42628 - Posted: 4 Nov 2011 | 15:31:18 UTC

well now the cue is empty...
are we now to believe that PPS is on hold completely until the end of SGS primes to top5000 ?!

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 42644 - Posted: 4 Nov 2011 | 17:10:14 UTC - in response to Message 42628.

well now the cue is empty...
are we now to believe that PPS is on hold completely until the end of SGS primes to top5000 ?!

The queue is filled again --- not sure what with, but something. One
of my managers was reporting no tasks for c. 5 hours. -Bruce*

(Two SGS hits ... These don't seem to be getting reported either;
nothing from PrimeGrid in the past 48 hours --- only one,
"11/2/2011 21:20:06 CDT", West Tennessee time.)

Omega

Joined: 20 Apr 08
Posts: 181
ID: 21694
Credit: 27,596,829
RAC: 1

Message 42645 - Posted: 4 Nov 2011 | 17:12:10 UTC

It was filled with 1200<k<4000...
I guess our suggestions were ignored and not even honored with a reply...

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 42646 - Posted: 4 Nov 2011 | 17:29:58 UTC - in response to Message 42645.

It was filled with 1200<k<4000...
I guess our suggestions were ignored and not even honored with a reply...

Yes, looks like
3061*2^710016+1 is not prime.

Given that not even SGS primes are being reported, looks like
our Admins are busy elsewhere. If one compares SGS and the
present PPS LLR range, n = 710000 is going to be well above
the lower edge of the Top5000, for a while, at least.

Regards, Bruce*

TheDawgz

Joined: 7 Jul 08
Posts: 2235
ID: 25193
Credit: 600,923,360
RAC: 155,910

Message 42649 - Posted: 4 Nov 2011 | 18:17:39 UTC - in response to Message 42645.

well now the cue is empty...
are we now to believe that PPS is on hold completely until the end of SGS primes to top5000 ?!

It was filled with 1200<k<4000...
I guess our suggestions were ignored and not even honored with a reply...

With my apologies to the Admins and the other 45,568 other registered users here at PrimeGrid ..

____________
There's someone in our head but it's not us.

Omega

Joined: 20 Apr 08
Posts: 181
ID: 21694
Credit: 27,596,829
RAC: 1

Message 42650 - Posted: 4 Nov 2011 | 18:23:35 UTC - in response to Message 42649.

excuse me?!
I mean no disrespect to the admins.
I think they do a fantastical job running PG.
But the PPS-reporting-halt is a mess (in my opinion, which -my dear TheDawgz- I hope I still may have) of their own making...

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 42652 - Posted: 4 Nov 2011 | 18:57:35 UTC - in response to Message 42563.

Alternatively, if 5 & 7 were split out, people could switch from PPS LLR to the
5-split or 7-split projects.

I apologize for the miscommunication. I should have said "singled out within the PPS (LLR) project" instead of splitting out. k=5 & 7 will remain within the PPS (LLR) project. There are currently no plans on separating them out like k=3 which btw was a collaboration with the 321 Search project. :)
____________

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 42653 - Posted: 4 Nov 2011 | 19:04:01 UTC - in response to Message 42650.

But the PPS-reporting-halt is a mess

Yes, the lack of prime submissions external to PrimeGrid has caught us by surprise...meaning that PPS primes will have to wait a little bit longer before submission is resumed. However we view this as a positive as it allows more users searching SGS to submit their primes into the Top 5000 list.

PPS prime submission will continue to be suspended until SGS falls out of the Top 5000 list. The bottom 50 can be viewed here. Current PPS primes awaiting to be submitted will enter at position ~330th to 690th. Their expected lifespan in the Top 5000 list is upwards of a decade. Therefore, there's no concern of them falling out any time soon.

The 710k-725k primes will enter at position ~3260th. Again, no concern of falling out any time soon.

Thank you for your patience as we navigate through SGS's end to the Top 5000 list.
____________

Michael Goetz
Volunteer moderator

Joined: 21 Jan 10
Posts: 13634
ID: 53948
Credit: 281,106,278
RAC: 30,422

Message 42659 - Posted: 4 Nov 2011 | 19:40:39 UTC - in response to Message 42653.

...meaning that PPS primes will have to wait a little bit longer before submission is resumed. However we view this as a positive as it allows more users searching SGS to submit their primes into the Top 5000 list.

Might I suggest that at this point, all that is being accomplished is delaying the inevitable, at the cost of artificially withholding submissions that by all rights aught to have already been submitted.

Although I consider it unlikely, it is conceivable that someone other than PrimeGrid might also submit the prime to the top 5k list. I don't think you want to be in the position of defending the non-submission to the prime finder if someone jumps ahead of him/her and submits it manually while the primes are being withheld.

My opinion, for what it's worth, is that more harm than good is being done right now.

____________
My lucky number is 75898524288+1

DaveB

Joined: 20 Jun 09
Posts: 351
ID: 42198
Credit: 11,898,570
RAC: 0

Message 42689 - Posted: 4 Nov 2011 | 21:57:42 UTC

I think the idea is that no one will do SGS if they do not get into the top 5000.

I think it is rubbish, consider that many happily do Sieves and SOB with no chance of finding a prime (SOB = very little chance with only 5 left to find and 17 for the whole project).

I agree that admins should let nature take its course. I have no personal gain in ths as I have not got any primes (not even waiting ones).
____________
Member team AUSTRALIA
My lucky number is 9291*2^1085585+1

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 42696 - Posted: 4 Nov 2011 | 22:28:34 UTC - in response to Message 42689.

I think the idea is that no one will do SGS if they do not get into the top 5000.

No, that is not the case. It's simply to optimize the end of SGS on the Top 5000 list. BTW, TPS continued quite well after it fell from the Top 5000 list and eventually, the twin was found.
____________

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 42897 - Posted: 7 Nov 2011 | 18:07:20 UTC - in response to Message 42696.

I think the idea is that no one will do SGS if they do not get into the top 5000.

No, that is not the case. It's simply to optimize the end of SGS on the Top 5000 list. BTW, TPS continued quite well after it fell from the Top 5000 list and eventually, the twin was found.

That would be the pair with n = 333333, 8/13/2009. Sounds like
n = 333333 fell off of the top5000 in 2009, with n = 666666 falling
off in 2011. Would a plausible extrapolation be that n = 999999
will fall off some time in 2013? If so, n < 1M lasts only two years.

That's a longer time-frame than the next couple of weeks, to one
month, maybe for PPS LLR reporting to be unfrozen. My first result
in distributed computing was c. 1/3 of the sieving for breaking RSA-120
in June 1993. I'm starting to wonder whether the next two years
of PPS LLR might merit a higher priority than weeks 3 and 4 of keeping
SGS in the top5000; which would suggest a hard deadline of
Nov 21. My current backlog of 20 hits (18 PPS, 2 SGS) is a moderate
career's-worth. My most recent was 10/14; next week will be a month's
worth of computing with no feedback on results. Just wondering ...

Bruce*

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 43360 - Posted: 17 Nov 2011 | 18:31:06 UTC

Progress Update

We are well into 1200<k<4000 to 800K range (710k-725k). SGS has officially dropped out of the Top 5000 list and the backlog of PPS primes have been submitted.

Next up is the 100<k<500 to 1.5M range. We expect to enter that range within the next two weeks.

Proth Prime Search Update
k=3 to 25M is ongoing in the 321 Prime Search (currently at n=7505609)
k=5 to 5M (currently at n=3M)
k=7 to 4M (currently at n=3M)
k=9 to 3M (currently at n=3M)
10<k<32 to 2.5M (currently at n=2.5M)
32<k<100 to 2M (currently at n=2M)
100<k<500 to 1.5M (currently at n=1M)
500<k<1200 to 1.3M (currently at n=1.3M)
1200<k<4000 to 800K ongoing (currently at n=710k)
4000<k<10000 to 725K ongoing (currently at n=710k)
____________

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 43533 - Posted: 19 Nov 2011 | 2:36:46 UTC - in response to Message 43360.

Progress Update

We are well into 1200<k<4000 to 800K range (710k-725k). SGS has officially dropped out of the Top 5000 list and the backlog of PPS primes have been submitted.

Next up is the 100<k<500 to 1.5M range. We expect to enter that range within the next two weeks.
...

Looks like these are starting early, n=1M, before the finish to 725k --- likely
to keep traffic down for the duration of the Challenge?

I'm still seeing a mixture of the two, but the earlier range is still in
n = 722k, so perhaps the two weeks worth is being deferred.

Looking to get my gold on 321 during or soon after the Challenge,
then maybe work on TRP. -Bruce*

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 43534 - Posted: 19 Nov 2011 | 3:27:15 UTC - in response to Message 43533.

Looks like these are starting early, n=1M, before the finish to 725k --- likely
to keep traffic down for the duration of the Challenge?

It's the nature of the WU generator. When both PPS (4<k<1200) and PPSE (1200<k<10000) work is in the queue, both will go out together. Otherwise, it's typically by n.

At the rate PPSE (1200<k<10000 for 710k<n<725k) was going out, we were going to be empty in a day or two.
____________

TroubledBunny
Volunteer tester

Joined: 4 Jun 07
Posts: 607
ID: 8931
Credit: 470,669,737
RAC: 0

Message 43746 - Posted: 22 Nov 2011 | 11:04:22 UTC

No PPS LLR WUs available?
____________

35 x 2^3587843+1 is prime!

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 44060 - Posted: 28 Nov 2011 | 16:01:51 UTC

Progress Update

Currently, the active range is 100<k<500 to 1.5M.

Proth Prime Search Update
k=3 to 25M is ongoing in the 321 Prime Search (trailing edge n=7620569; leading edge n=8180124)
k=5 to 5M (currently at n=3M)
k=7 to 4M (currently at n=3M)
k=9 to 3M (currently at n=3M)
10<k<32 to 2.5M (currently at n=2.5M)
32<k<100 to 2M (currently at n=2M)
100<k<500 to 1.5M ongoing (trailing edge n=1005938; leading edge n=1103385)
500<k<1200 to 1.3M (currently at n=1.3M)
1200<k<4000 to 800K ongoing (trailing edge n=724010; leading edge n=725k)
4000<k<10000 to 725K ongoing (trailing edge n=723984; leading edge n=725k)

Green is completed goal. :)
____________

DoES
Volunteer tester

Joined: 11 Oct 08
Posts: 784
ID: 30382
Credit: 74,895,590
RAC: 0

Message 44660 - Posted: 10 Dec 2011 | 6:20:05 UTC - in response to Message 44060.

Here is a milestone of sorts -- My first prime I got since coming back to PG in July 2011 is now ranked 5035 -- It entered the top 5000 ranked 3513 in August 2011

5035 --- 2337 Â· 2^678383 + 1 -- 204218 -- L2522 -- Aug 2011

We are moving at a fast pace it seems

____________
Member of AtP

Shown here is an Australian native rat (Ratus Kickarsus)

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 45232 - Posted: 20 Dec 2011 | 3:32:15 UTC

Progress Update

The March to the Solstice has made good progress towards accomplishing these goals.

NOTE: A snafu in loading the next range allowed k=7 to also be included. Therefore, we'll be taking both k=5 & 7 to 5M. This will weaken our chances of finishing the goals by the end of the year.

Currently, the active range is k=5 & 7 from 3M to 5M.

Proth Prime Search Update
k=3 to 25M is ongoing in the 321 Prime Search (trailing edge n=7673589; leading edge n=8192966)
k=5 to 5M ongoing (trailing edge n=3000949; leading edge n=4246999)
k=7 to 5M ongoing (trailing edge n=3000104; leading edge n=4246816)
k=9 to 3M (currently at n=3M)
10<k<32 to 2.5M (currently at n=2.5M)
32<k<100 to 2M (currently at n=2M)
100<k<500 to 1.5M ongoing (trailing edge n=1187833; leading edge n=1.2M)
500<k<1200 to 1.3M (currently at n=1.3M)
1200<k<4000 to 800K ongoing (trailing edge n=739038; leading edge n=775k)
4000<k<10000 to 725K (currently at n=725k)

Green is completed goal. :)
____________

Honza
Volunteer moderator
Volunteer tester
Project scientist

Joined: 15 Aug 05
Posts: 1905
ID: 352
Credit: 4,082,937,699
RAC: 3,907,008

Message 45240 - Posted: 20 Dec 2011 | 7:46:37 UTC

John, how many candidates are there for active range is k=5 & 7 from 3M to 5M?
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 45264 - Posted: 20 Dec 2011 | 15:05:33 UTC - in response to Message 45240.

John, how many candidates are there for active range is k=5 & 7 from 3M to 5M?

k=5: 29,519 total
k=7: 71,585 total
____________

Cannibal Corpse

Joined: 7 Jan 11
Posts: 9
ID: 80318
Credit: 45,087,490
RAC: 0

Message 45315 - Posted: 21 Dec 2011 | 9:51:09 UTC

My up-most appolgy to PG and the wing man..For some reason my PPS.llr than ran for 15 minutes turned into 3:30 to 4:30 hours. Even tho my cach is a day on all..well I had to abort quit a bit of wu's because of time dead lines. Sorry..

____________

My first Prime
20208428036625*2^666666-1 (SGS)
PROUD MEMBER OF TEAM CARL SAGAN

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 46020 - Posted: 2 Jan 2012 | 3:56:24 UTC

The End of 2011

Here are the numbers for the end of 2011. Looks like a good starting point for 2012! :)

May 2012 be a healthy and prosperous year for all!

Proth Prime Search Update
k=3 to 25M is ongoing in the 321 Prime Search (trailing edge n=7743792; leading edge n=8192966) excluded from year end goals ;)
k=5 to 5M ongoing (trailing edge n=3377139; leading edge n=5M)
k=7 to 5M ongoing (trailing edge n=3307796; leading edge n=5M)
k=9 to 3M (currently at n=3M)
10<k<32 to 2.5M (currently at n=2.5M)
32<k<100 to 2M (currently at n=2M)
100<k<500 to 1.5M ongoing (trailing edge n=1.2M; leading edge n=1223663)
500<k<1200 to 1.3M (currently at n=1.3M)
1200<k<4000 to 800K (currently at n=775k)
4000<k<10000 to 725K (currently at n=725k)

Green is completed goal. :)
____________

Genn
Volunteer tester

Joined: 16 Jul 09
Posts: 50
ID: 43504
Credit: 91,204,089
RAC: 0

Message 47071 - Posted: 16 Jan 2012 | 14:44:10 UTC

I see that 500<k<1200 has been loaded with workunits over 1.3M. When can we expect goals for 2012?

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 47134 - Posted: 17 Jan 2012 | 4:11:46 UTC

2012 Goals

Here are the new goals for 2012. Yes, they are much more aggressive than last year. However, with the promise of the new AVX builds available for Sandy Bridges, we have confidence that these goals are attainable.

Proth Prime Search Update
k=3 to 9M - ongoing in the 321 Prime Search (trailing edge n=7841640; leading edge n=8192966)
k=5 to 7M - (currently at n=5M)
k=7 to 6M - (currently at n=5M)
k=9 to 4M - (currently at n=3M)
10<k<32 to 3M - (currently at n=2.5M)
32<k<100 to 2.5M - (currently at n=2M)
100<k<500 to 2M - ongoing (trailing edge n=1245850; leading edge n=1373125)
500<k<1200 to 1.5M - ongoing (trailing edge n=1.3M; leading edge n=1303923)
1200<k<4000 to 1.25M - (currently at n=775k)
4000<k<10000 to 1M - (currently at n=725k)

Green is completed goal. :)
____________

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 48691 - Posted: 4 Feb 2012 | 1:54:58 UTC

2012 Goals - Progress update

The Year of the Dragon completed a huge chunk of work. We are very close to achieving the first goal of the year. One month down, 11 to go. :)

Thank you to everyone for participating. Once n=1.5M has been reach for both 100<k<500 and 500<k<1200, we'll move to lower n's at 4000<k<10000.

Proth Prime Search Update
k=3 to 9M - ongoing in the 321 Prime Search (trailing edge n=7981182; leading edge n=8192966)
k=5 to 7M - (currently at n=5M)
k=7 to 6M - (currently at n=5M)
k=9 to 4M - (currently at n=3M)
10<k<32 to 3M - (currently at n=2.5M)
32<k<100 to 2.5M - (currently at n=2M)
100<k<500 to 2M - ongoing (trailing edge n=1406274; leading edge n=1484139)
500<k<1200 to 1.5M - ongoing (trailing edge n=1320383; leading edge n=1484139)
1200<k<4000 to 1.25M - (currently at n=775k)
4000<k<10000 to 1M - (currently at n=725k)

Green is completed goal. :)
____________

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 49679 - Posted: 15 Feb 2012 | 18:58:48 UTC

2012 Goals - Progress update

The 4000<k<10000 range has made some GREAT strides over the past week going from n=725k to n~747k. That's pretty AMAZING!!! And just as expected, producing a lot of primes for the Tour. BTW, the effort to get to 1.5M resulted in completing the first goal of 2012. Congratulations to all!

Thank you to everyone for participating. Should we reach n=775k next week, the 1200<k<4000 range will be added and both will be loaded up to n=800k. That should definitely last through the end of the Tour.

Proth Prime Search Update
k=3 to 9M - ongoing in the 321 Prime Search (trailing edge n=8037764; leading edge n=8192966)
k=5 to 7M - (currently at n=5M)
k=7 to 6M - (currently at n=5M)
k=9 to 4M - (currently at n=3M)
10<k<32 to 3M - (currently at n=2.5M)
32<k<100 to 2.5M - (currently at n=2M)
100<k<500 to 2M - (currently at n=1.5M)
500<k<1200 to 1.5M - (currently at n=1.5M)
1200<k<4000 to 1.25M - (currently at n=775k)
4000<k<10000 to 1M - ongoing (trailing edge n=728077; leading edge n=746811)

Green represents a completed goal. :)
____________

rogue
Volunteer developer

Joined: 8 Sep 07
Posts: 1223
ID: 12001
Credit: 18,565,548
RAC: 0

Message 49680 - Posted: 15 Feb 2012 | 19:11:58 UTC - in response to Message 49679.

k=5 to 7M - (currently at n=5M)
k=7 to 6M - (currently at n=5M)
k=9 to 4M - (currently at n=3M)

Would you consider testing all of these (or just 5 and 7) to the same n?

Christopher Siegert

Joined: 5 Jul 09
Posts: 232
ID: 42986
Credit: 77,011,345
RAC: 0

Message 49684 - Posted: 15 Feb 2012 | 21:15:05 UTC

Will there ever be an expansion above k=10,000?

I like seeing discoveries of Fermat divisors, and there are probably some low n primes in the 10000<k<50000 range.

Christopher Siegert

Joined: 5 Jul 09
Posts: 232
ID: 42986
Credit: 77,011,345
RAC: 0

Message 49686 - Posted: 15 Feb 2012 | 21:26:44 UTC

Top 5 Fermat factors with k>10000:

71007 * 2^49490 + 1
177795 * 2^38969 + 1
121531 * 2^30260 + 1
13323 * 2^19220 + 1
17217 * 2^14530 + 1

Those are some pretty low n values!

Christopher Siegert

Joined: 5 Jul 09
Posts: 232
ID: 42986
Credit: 77,011,345
RAC: 0

Message 49687 - Posted: 15 Feb 2012 | 21:28:41 UTC

Perhaps a separate project dedicated to high k value Fermat divisors would be more appropriate though. I doubt we'd find many top 5000 primes, if any.

rogue
Volunteer developer

Joined: 8 Sep 07
Posts: 1223
ID: 12001
Credit: 18,565,548
RAC: 0

Message 49689 - Posted: 15 Feb 2012 | 22:43:50 UTC - in response to Message 49684.

Will there ever be an expansion above k=10,000?

I like seeing discoveries of Fermat divisors, and there are probably some low n primes in the 10000<k<50000 range.

There is already a project that has a search, although it isn't coordinated with a client/server app. Go over to fermatsearch.org.

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 49694 - Posted: 15 Feb 2012 | 23:58:25 UTC - in response to Message 49687.

Perhaps a separate project dedicated to high k value Fermat divisors would be more appropriate though. I doubt we'd find many top 5000 primes, if any.

As rogue mentions, F E R M A T S E A R C H . O R G offers what you're looking for. Additionally, in their news section, you'll see PG's last two Fermat primes. :)

Also, if you take a look at Prime factors k Â· 2^n + 1 of Fermat numbers F(m) and complete factoring status compiled by Wilfrid Keller, you see at the bottom of the page Known search limits. In the 10000<k<50000 range you mention, here are the following limits:

up to k=10000 has been searched to n=210000 (actually, latest PRPNet progress shows n=313065)
up to k=20000 has been searched to n=77000
up to k=50000 has been searched to n=61000
____________

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 49695 - Posted: 16 Feb 2012 | 0:04:20 UTC - in response to Message 49680.

k=5 to 7M - (currently at n=5M)
k=7 to 6M - (currently at n=5M)
k=9 to 4M - (currently at n=3M)

Would you consider testing all of these (or just 5 and 7) to the same n?

Once all the goals have been met for the year, this would certainly be considered. Note that k=5 is such a low weight that getting it to 7M would only take a week or so. However, k=7 & 9 are heavy weights and would take quite a bit longer. Even just getting k=9 to 5M will take some time.
____________

rogue
Volunteer developer

Joined: 8 Sep 07
Posts: 1223
ID: 12001
Credit: 18,565,548
RAC: 0

Message 49698 - Posted: 16 Feb 2012 | 0:16:34 UTC - in response to Message 49695.

k=5 to 7M - (currently at n=5M)
k=7 to 6M - (currently at n=5M)
k=9 to 4M - (currently at n=3M)

Would you consider testing all of these (or just 5 and 7) to the same n?

Once all the goals have been met for the year, this would certainly be considered. Note that k=5 is such a low weight that getting it to 7M would only take a week or so. However, k=7 & 9 are heavy weights and would take quite a bit longer. Even just getting k=9 to 5M will take some time.

That just means more opportunities for mega primes!

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 49699 - Posted: 16 Feb 2012 | 0:17:54 UTC - in response to Message 49698.

k=5 to 7M - (currently at n=5M)
k=7 to 6M - (currently at n=5M)
k=9 to 4M - (currently at n=3M)

Would you consider testing all of these (or just 5 and 7) to the same n?

Once all the goals have been met for the year, this would certainly be considered. Note that k=5 is such a low weight that getting it to 7M would only take a week or so. However, k=7 & 9 are heavy weights and would take quite a bit longer. Even just getting k=9 to 5M will take some time.

That just means more opportunities for mega primes!

An optimist! I like it!!! :D
____________

[DPC]Charley

Joined: 15 Apr 11
Posts: 803
ID: 95137
Credit: 94,993,523
RAC: 0

Message 50802 - Posted: 28 Feb 2012 | 23:23:57 UTC

What's the plan after n=800k has been reached for the current range? Will the next range of n's be loaded for 1200<k<10000 or will we be switching back to lower k's/higher n's?
____________
PrimeGrid Challenge Overall standings --- Last update: From Pi to Paddy (2016)

Honza
Volunteer moderator
Volunteer tester
Project scientist

Joined: 15 Aug 05
Posts: 1905
ID: 352
Credit: 4,082,937,699
RAC: 3,907,008

Message 50823 - Posted: 29 Feb 2012 | 6:56:40 UTC - in response to Message 49679.

I think n>800k can be loaded for whole 1200<k<10000.

500<k<1200 to 1.5M - (currently at n=1.5M)
1200<k<4000 to 1.25M - (currently at n=775k)
4000<k<10000 to 1M - ongoing (trailing edge n=728077; leading edge n=746811)

____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186

John
Honorary cruncher

Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0

Message 50846 - Posted: 29 Feb 2012 | 15:10:05 UTC

2012 Goals - Progress update

The Tour de Primes has really done a nice job advancing PPS (LLR) towards its 2012 goals. Well done everyone!!!

Work will continue for the 1200-10000 ranges up to n=800k. Afterwards, a decision will be made where to go next. Most likely the search will stay at this level and push towards n=850k or even 900k. It all depends on progress. The past 3 weeks have just been incredible moving through a 55k n range. If the pace continued, 850k could be reached by the end of March.

As for other ranges, there's definitely plenty of work to do. The path forward will be to "cherry pick" some of the easier goals to achieve. It's nice to see bold green in the list. ;) Expect a pattern of venturing off to another range and then back to 1200-10000. This will be repeated over and over for the rest of the year.

Thank you again to everyone helping out with this effort. :)

Proth Prime Search Update
k=3 to 9M - ongoing in the 321 Prime Search (trailing edge n=8077245; leading edge n=8202120)
k=5 to 7M - (currently at n=5M)
k=7 to 6M - (currently at n=5M)
k=9 to 4M - (currently at n=3M)
10<k<32 to 3M - (currently at n=2.5M)
32<k<100 to 2.5M - (currently at n=2M)
100<k<500 to 2M - (currently at n=1.5M)
500<k<1200 to 1.5M - (currently at n=1.5M)
1200<k<4000 to 1.25M - ongoing (trailing edge n=775000; leading edge n=780000)
4000<k<10000 to 1M - ongoing (trailing edge n=763032; leading edge n=780733)

Green represents a completed goal. :)
____________

Honza
Volunteer moderator
Volunteer tester
Project scientist

Joined: 15 Aug 05
Posts: 1905
ID: 352
Credit: 4,082,937,699
RAC: 3,907,008

Message 51594 - Posted: 16 Mar 2012 | 8:34:13 UTC - in response to Message 50846.

Work will continue for the 1200-10000 ranges up to n=800k. Afterwards, a decision will be made where to go next. Most likely the search will stay at this level and push towards n=850k or even 900k.

We are now at 799k territory.
Since now further annoucement was made, I assume we will simply continue in most likely scenario.

FFT length: 64K
I wonder when it gets increased?
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 51732 - Posted: 19 Mar 2012 | 12:17:01 UTC - in response to Message 51594.

Work will continue for the 1200-10000 ranges up to n=800k. Afterwards, a decision will be made where to go next. Most likely the search will stay at this level and push towards n=850k or even 900k.

We are now at 799k territory.
Since now further annoucement was made, I assume we will simply continue in most likely scenario.

FFT length: 64K
I wonder when it gets increased?

Hi

800k-810k is loaded and I think we will go to at least 900k first and then we can discuss further directions.

Lennart
____________

Christopher Siegert

Joined: 5 Jul 09
Posts: 232
ID: 42986
Credit: 77,011,345
RAC: 0

Message 51756 - Posted: 20 Mar 2012 | 6:06:58 UTC

All I ask is that Primegrid hold off on k=63 & k=87. I have been independently crunching those k values for n>2,000,000. Once I'm finished, I'll gladly share the results.

I have been crunching these two k values, even at lower n, for quite some time.

(edit): Please hold off on "32<k<100 to 2.5M - (currently at n=2M)"
____________
15547296^32768 + 1 is prime!
63 Â· 2^1356980 + 1 is prime!
63 Â· 2^1356980 + 1 divides xGF(1356973,11,4)!

Death[Kiev]
Volunteer tester

Joined: 12 Jun 08
Posts: 126
ID: 24043
Credit: 3,622,648
RAC: 0

Message 54049 - Posted: 7 May 2012 | 6:41:50 UTC - in response to Message 51756.

any update?
____________

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 54079 - Posted: 7 May 2012 | 14:41:26 UTC - in response to Message 54049.

You can check the progress here also: http://www.primegrid.com/stats_pps_llr.php

It looks like we are sitting at 856K complete for the 1200<k<4000 and 4000<k<10000 ranges. So we completed about 100,000 n for each k in the last two months. I think we can definitely accomplish these goals by the end of the year if we keep making that kind of progress.
____________

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 54121 - Posted: 8 May 2012 | 1:35:00 UTC - in response to Message 54079.

PPSE 1200<k<4000 and 4000<k<10000 is loaded to 870k now.

Leading edge is at 859k now.

Lennart

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 54475 - Posted: 14 May 2012 | 2:10:22 UTC - in response to Message 54121.

PPSE 1200<k<4000 and 4000<k<10000 is loaded to 870k now.

Leading edge is at 859k now.

I was mentioning passing Buster for 7th over on the credit milestone
thread; but perhaps it's worth noting that credit here on PPS (LLR) is
pretty good. I blew past both Buster and Mumps (who hasn't been
moving) into 6th --- my RAC took a dip while running some TRPs during
the Challenge, but it's gone back up after switching back.

With a clear field up to 870k in view (and perhaps somewhat further
after that, towards 900k), I starting to contemplate the nearly unthinkable.
Another 1000 points to catch Lennart!

(This is on the Primegrid "Top Prime Finders" (by score) listing.)

Lennart

Honza
Volunteer moderator
Volunteer tester
Project scientist

Joined: 15 Aug 05
Posts: 1905
ID: 352
Credit: 4,082,937,699
RAC: 3,907,008

Message 54485 - Posted: 14 May 2012 | 11:30:25 UTC - in response to Message 54475.

I was mentioning passing Buster for 7th over on the credit milestone
thread; but perhaps it's worth noting that credit here on PPS (LLR) is
pretty good. I blew past both Buster and Mumps (who hasn't been
moving) into 6th --- my RAC took a dip while running some TRPs during
the Challenge, but it's gone back up after switching back.

...
Another 1000 points to catch Lennart!

Actually, I like this stats. It is a long way run and doesn't involve GPUs...yet.

I hope to get 10th (by score) soon. 5th by total prime count...it will not last long, will it?
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 54654 - Posted: 18 May 2012 | 2:40:42 UTC - in response to Message 54121.

PPSE 1200<k<4000 and 4000<k<10000 is loaded to 870k now.

Leading edge is at 859k now.

Lennart

This evening's email says
the following prime was submitted on your behalf: 107045 : 4341*2^870079+1 (261924 digits)

so perhaps the next range has been loaded ... if so,
how far? John's leap-day post mentions 900k as possible;
as well as 2012 year-end targets of 1.25M and 1M split
at k = 4000.

Regards, bdodson*

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 54658 - Posted: 18 May 2012 | 3:44:46 UTC - in response to Message 54654.

PPSE 1200<k<4000 and 4000<k<10000 is loaded to 870k now.

Leading edge is at 859k now.

Lennart

This evening's email says
the following prime was submitted on your behalf: 107045 : 4341*2^870079+1 (261924 digits)

so perhaps the next range has been loaded ... if so,
how far? John's leap-day post mentions 900k as possible;
as well as 2012 year-end targets of 1.25M and 1M split
at k = 4000.

Regards, bdodson*

We are definitely zipping through quite quickly.
____________

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 55091 - Posted: 31 May 2012 | 7:03:26 UTC - in response to Message 54658.

At what n are we going to transition to a different k range?
____________

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 55098 - Posted: 31 May 2012 | 11:11:12 UTC - in response to Message 55091.

At what n are we going to transition to a different k range?

We're getting real close to n=900K
7821*2^889737+1 is not prime. 2447*2^889707+1 is not prime. 6041*2^889705+1 is not prime.

The options appear to be to staying with these k = 1200,4000;
k = 4000,10000 perhaps heading towards n=1M or switching
to one of the smaller k ranges. Judging by the above we may
know soon what Lennart et.al. have decided.

(Well, fairly close.)

-bdodson*

Currently 3rd on the PrimePages Top20 by number of primes;
two behind Broadhurst for 2nd (actually 1+ epsilon).

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 55122 - Posted: 1 Jun 2012 | 6:07:26 UTC

It appears that we have transitioned to a different range.
____________

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 55125 - Posted: 1 Jun 2012 | 7:13:26 UTC - in response to Message 55122.

Edit to my previous post, I thought we had transitioned ranges because my computer was now taking 7+ hours to do a PPS LLR, but it turns out some WU I had aborted didn't end their processes and were hogging up all my CPU resources. D'oh.
____________

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 55126 - Posted: 1 Jun 2012 | 7:27:44 UTC

I am working on a new SGS range now so We will go up to 1M before we start on lower pps k's.

Lennart

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 55142 - Posted: 1 Jun 2012 | 14:33:51 UTC - in response to Message 55126.

Thanks for telling us the plan. Hopefully we make it there soon, and complete these goals by the end of the year.
____________

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 55203 - Posted: 3 Jun 2012 | 4:17:04 UTC

2012 Goals - Progress update

Good progress is being made in advancing PPS (LLR) towards its 2012 goals. Let's keep up the good work!!!

Work will continue for the 1200-10000 ranges up to n=1M. SGS is actively being sieved and we want to make sure PPS is high enough to remain on the Top 5000 when SGS starts.

After n=1M has been reached for 1200-10000, the path forward will be to "cherry pick" some of the easier goals to achieve. It's nice to see bold green in the list. ;) We will then return to 1200-4000 to push it to n=1.25M.

Thank you again to everyone helping out with this effort. :)

Proth Prime Search Update
k=3 to 9M - ongoing in the 321 Prime Search (trailing edge n=8374043; leading edge n=8399706)
k=5 to 7M - (currently at n=5M)
k=7 to 6M - (currently at n=5M)
k=9 to 4M - (currently at n=3M)
10<k<32 to 3M - (currently at n=2.5M)
32<k<100 to 2.5M - (currently at n=2M)
100<k<500 to 2M - (currently at n=1.5M)
500<k<1200 to 1.5M - (currently at n=1.5M)
1200<k<4000 to 1.25M - ongoing (trailing edge n=881184; leading edge n=891369)
4000<k<10000 to 1M - ongoing (trailing edge n=881974; leading edge n=891369)

Green represents a completed goal.

Lennart

cassetti

Joined: 19 Mar 12
Posts: 41
ID: 135491
Credit: 16,492,128
RAC: 91,065

Message 55216 - Posted: 3 Jun 2012 | 11:20:36 UTC - in response to Message 55203.

Awesome! Looking good there.
____________

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 55253 - Posted: 5 Jun 2012 | 16:25:59 UTC - in response to Message 55203.

2012 Goals - Progress update

Good progress is being made in advancing PPS (LLR) towards its 2012 goals. Let's keep up the good work!!!

Work will continue for the 1200-10000 ranges up to n=1M. SGS is actively being sieved and we want to make sure PPS is high enough to remain on the Top 5000 when SGS starts.
...

Thank you again to everyone helping out with this effort. :)

Proth Prime Search Update
...
100<k<500 to 2M - (currently at n=1.5M)
500<k<1200 to 1.5M - (currently at n=1.5M)
1200<k<4000 to 1.25M - ongoing (trailing edge n=881184; leading edge n=891369)
4000<k<10000 to 1M - ongoing (trailing edge n=881974; leading edge n=891369)

Green represents a completed goal.

Lennart

Some of the SGS timings appeared focused on n=1.29M with large k.
Is that the range of the new SGS/twin llr search? In a few years that
would bump all of 4000 < k < 10000 for sure, looks like all of
1200 < k < 4000 as well. Of course, all primes below 1000000-digits
will be bumped from the top5K eventually, presuming that we keep at this,
and a SGS search that far above 2^1M will find primes at a slower
rate than we're enjoying here below 2^1M.

To re-phrase the question, with the previous SGS search at
200700-digits, what's the new threshold? Looks like n=1.29M
suggests somewhere above 370000-digits, perhaps a round
400000-digits?

Best Regards, Bruce*

(currently 2nd on the PrimePages list by number of primes
on the top5K list)

j.sheridan
Volunteer tester

Joined: 21 Mar 11
Posts: 816
ID: 91622
Credit: 2,127,281,471
RAC: 2,545,876

Message 55254 - Posted: 5 Jun 2012 | 17:16:48 UTC - in response to Message 55203.

SGS is actively being sieved

where? can we join in?

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 55298 - Posted: 8 Jun 2012 | 2:45:33 UTC - in response to Message 55254.

SGS is actively being sieved

where? can we join in?

You don't have to ! The first range will be done in ~15 days.

Lennart

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 55300 - Posted: 8 Jun 2012 | 5:09:22 UTC - in response to Message 55298.

Does the first range being done mean that SGS will start again, or will you go further into the sieve before the project is started again? And what n level is it this time?
____________

Honza
Volunteer moderator
Volunteer tester
Project scientist

Joined: 15 Aug 05
Posts: 1905
ID: 352
Credit: 4,082,937,699
RAC: 3,907,008

Message 55307 - Posted: 8 Jun 2012 | 18:04:02 UTC

So, leading edge (and primes reported) are in >900k territory, well done.
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186

Honza
Volunteer moderator
Volunteer tester
Project scientist

Joined: 15 Aug 05
Posts: 1905
ID: 352
Credit: 4,082,937,699
RAC: 3,907,008

Message 55632 - Posted: 19 Jun 2012 | 9:30:45 UTC

Leading edge is almost ~915k and FFT size of 72K is stating to kick in with higher k's.
Expect some WUs with longer processing time of ~20%.
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 55647 - Posted: 19 Jun 2012 | 14:38:31 UTC - in response to Message 55632.

That will slow down our progress a bit, but hopefully it doesn't make too big of a dent in our progress to 1mil.
____________

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 56010 - Posted: 29 Jun 2012 | 14:48:59 UTC - in response to Message 55647.

Progress is quite good, we have increased the leading edge by about 35K since Lennart's update at the beginning of the month.
____________

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 56014 - Posted: 29 Jun 2012 | 19:02:21 UTC - in response to Message 56010.

Progress is quite good, we have increased the leading edge by about 35K since Lennart's update at the beginning of the month.

As nice round number today,
5385*2^925000+1, 278457 (decimal)

75K n's to go.

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 56136 - Posted: 5 Jul 2012 | 0:57:16 UTC - in response to Message 56014.

We are going at the pace of about 1000 n per day, so we should get to 1mil before mid September.
____________

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 57354 - Posted: 29 Aug 2012 | 6:32:41 UTC - in response to Message 56136.

The pace slowed, it looks like we won't hit 1mil until late September/possibly early October, regardless, we should finish off <1mil before the challenge which will mean we should be able to destroy the next range during the challenge (I'm guessing that will be the 1200<k<4000 range to 1.25mil goal that we will work on during the challenge since they would have the shortest times of the remaining goals. Although Lennart did say:

After n=1M has been reached for 1200-10000, the path forward will be to "cherry pick" some of the easier goals to achieve. It's nice to see bold green in the list. ;) We will then return to 1200-4000 to push it to n=1.25M.
Perhaps we will finally have a PPS challenge that won't challenge the servers?
____________

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 57498 - Posted: 6 Sep 2012 | 2:48:21 UTC - in response to Message 57354.

The pace slowed, it looks like we won't hit 1mil until late September/possibly early October, regardless, we should finish off <1mil before the challenge which will
mean ...

Here's another week's progress (of a sort)
Added 109143 : 9071*2^983279+1 (296001 digits)

296K-digits, with another c. 17K bits left to go to k=1M.

-bdodson*

(rapidly dropping from 200 primes on the top5000,
down to 180 primes; and down from 2nd to 4th
place by number of primes)

(255K-digit Proth primes from April/May 2011 - April/May 2012
dropping out of the top5000 more quickly than I'm
finding new 296K-digit primes ...)

Honza
Volunteer moderator
Volunteer tester
Project scientist

Joined: 15 Aug 05
Posts: 1905
ID: 352
Credit: 4,082,937,699
RAC: 3,907,008

Message 57526 - Posted: 7 Sep 2012 | 11:01:02 UTC

We will enter TOP2000 with PPS primes very soon...
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 57743 - Posted: 23 Sep 2012 | 6:51:52 UTC - in response to Message 57526.

Leading edge just passed 999000, we should hit the goal by the end of Monday. Does anyone know what the next range will be and if it will loaded immediately upon completing this range?
____________

[DPC]Charley

Joined: 15 Apr 11
Posts: 803
ID: 95137
Credit: 94,993,523
RAC: 0

Message 57758 - Posted: 23 Sep 2012 | 12:14:15 UTC

We have to finish all up to 1M first so we might load 1M+ if we need more in the challenge and change to lower k's after the challenge. If we finish 1M before the challenge we might start on a lower k.

Seeing how we have a challenge that will only last 1 day I would think it's most interesting to load 1200<k<4000 to 1.25M is most interesting, as this will allow most people to participate.
____________
PrimeGrid Challenge Overall standings --- Last update: From Pi to Paddy (2016)

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 57772 - Posted: 24 Sep 2012 | 1:06:16 UTC - in response to Message 57758.

I agree with you. I also don't think it will take too terribly long to finish off that range.
____________

Genn
Volunteer tester

Joined: 16 Jul 09
Posts: 50
ID: 43504
Credit: 91,204,089
RAC: 0

Message 57776 - Posted: 24 Sep 2012 | 7:05:01 UTC

Good News Everyone! Leading edge is over 1M. But why 4000<k<10000 is also loaded over 1M? I thought we'll focus on 1200<k<4000 after we reach 1M.

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 57777 - Posted: 24 Sep 2012 | 7:19:00 UTC - in response to Message 57776.

I'm guessing code on the server needs to be updated in order for a new range to be sent out.
____________

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 57822 - Posted: 26 Sep 2012 | 6:51:25 UTC

So, any idea on when we are going to move onto a new range rather than keep plugging away past the previous goal?
____________

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 57854 - Posted: 27 Sep 2012 | 13:47:02 UTC - in response to Message 57776.

Good News Everyone! Leading edge is over 1M. But why 4000<k<10000 is also loaded over 1M? I thought we'll focus on 1200<k<4000 after we reach 1M.

Lennart's June 2012 Update gives 1M as the target for the year
on 4000 < k < 10000. Of the recent Proth's above 1M, three have
k < 4000 (mine's 3791*2^1001609+1 (301518 digits)), while there
are fourteen with 4000 < k < 10000. (That's relative to the Prime
Pages window of 72 hours.)

Seems like a majority of our current effort is focused off of the annual
targets, which includes 1200 < k < 4000 to 1.25M. There's also the
usual odds for a Proth prime to be a Fermat divisor of 1/k, which favors
k < 4000; although searching k > 4000 gets credit (on the 'Pages)
for larger difficulty.

Bruce*

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 57871 - Posted: 28 Sep 2012 | 7:02:58 UTC - in response to Message 57854.

Good News Everyone! Leading edge is over 1M. But why 4000<k<10000 is also loaded over 1M? I thought we'll focus on 1200<k<4000 after we reach 1M.

Lennart's June 2012 Update gives 1M as the target for the year
on 4000 < k < 10000. Of the recent Proth's above 1M, three have
k < 4000 (mine's 3791*2^1001609+1 (301518 digits)), while there
are fourteen with 4000 < k < 10000. (That's relative to the Prime
Pages window of 72 hours.)

Seems like a majority of our current effort is focused off of the annual
targets, which includes 1200 < k < 4000 to 1.25M. There's also the
usual odds for a Proth prime to be a Fermat divisor of 1/k, which favors
k < 4000; although searching k > 4000 gets credit (on the 'Pages)
for larger difficulty.

Bruce*

The leading edge is the same for 1200<k<4000 and 4000<k<10000, they aren't favoring either.
____________

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 57890 - Posted: 28 Sep 2012 | 17:54:09 UTC

2012 Goals - Progress update

Good progress is being made in advancing PPS (LLR) towards its 2012 goals. Let's keep up the good work!!!

Work will continue for the 1200-10000 ranges up to n=1.1M. SGS is online and we want to make sure PPS is high enough to remain on the Top 5000.

Thank you again to everyone helping out with this effort. :)

Proth Prime Search Update
k=3 to 9M - ongoing in the 321 Prime Search (trailing edge n=8.5M; leading edge n=8.62M)
k=5 to 6M - (currently at n=5M) (*)
k=7 to 6M - (currently at n=5M)
k=9 to 4M - (currently at n=3M)
10<k<32 to 2.5M - (currently at n=2.5M)
32<k<100 to 2.2M - (currently at n=2M)
100<k<500 to 2.2M - (currently at n=2M)
500<k<1200 to 2.2M - (currently at n=1.5M)
1200<k<10000 to 1.03M - ongoing (trailing edge n=998493; leading edge n=1.005M)

Green represents a completed goal.

Lennart

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 57893 - Posted: 28 Sep 2012 | 20:04:17 UTC - in response to Message 57890.

Good to know, but this makes me unsure of whether or not we will be able to hit all the goals for this year if we are pushing 1200<k<10000 that far, oh well. It amounts to less low range work for next year.
____________

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 57897 - Posted: 29 Sep 2012 | 4:53:08 UTC - in response to Message 57871.

... Of the recent Proth's above 1M, three have
k < 4000 ..., while there are fourteen with 4000 < k < 10000. ...

The leading edge is the same for 1200<k<4000 and 4000<k<10000, they aren't favoring either.

They? I'm suggesting two counting questions. three -vs- fourteen?
is one. The other is 4000 - 1200 = 2800 k's -vs- 10000 - 4000 =
6000 k's. There are way-more k's in the 2nd range being sent out,
and that "favors" the larger k range by 28-to-60. More candidates
submitted to LLR and more primes found. Substantially.

But, of course, Lennart's right. For a short term 1.0M - 1.1M we
can tolerate a few extra early 2013 primes; I think we were wondering
about 1.0M - 1.25M, but that isn't happening. OK?

-bdodson* (@Lennart: thanks for the update!)

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 57902 - Posted: 29 Sep 2012 | 11:16:41 UTC - in response to Message 57897.

We should have changed to a smaller k but I will keep the low n in the challenge.

It is only 24hr.

Lennart

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 57995 - Posted: 4 Oct 2012 | 14:52:01 UTC - in response to Message 57902.

We should have changed to a smaller k but I will keep the low n in the challenge.

It is only 24hr.

Lennart

Uhm; that's "low n" meaning large k, and continuing with
1200<k<10000 to 1.03M - ongoing (trailing edge n=998493; leading edge n=1.005M)

if I'm reading correctly. The smaller k's have n at/above 2M.

I'm seeing
7555*2^1010048+1 is not prime. 4869*2^1010097+1 is not prime.

for tasks that have just been sent, as in
4 Oct 2012 | 13:30:46 UTC 4 Oct 2012 | 13:47:48 UTC

With c. 27 hours to the start of the challenge, looks like
we'll be competing in n=1.01. -Bruce*

bdodson*

Joined: 13 Nov 09
Posts: 104
ID: 50079
Credit: 170,522,809
RAC: 0

Message 58139 - Posted: 7 Oct 2012 | 9:59:13 UTC - in response to Message 57995.

...
I'm seeing
7555*2^1010048+1 is not prime. 4869*2^1010097+1 is not prime.

for tasks that have just been sent, as in
4 Oct 2012 | 13:30:46 UTC 4 Oct 2012 | 13:47:48 UTC

With c. 27 hours to the start of the challenge, looks like
we'll be competing in n=1.01. -Bruce*

Looks like the leading edge at/near the end of the challenge
was in 1.017M
Is prime? 9125*2^1017239+1 is not prime. ... 413603656 290601 6 Oct 2012 | 15:17:36 UTC 6 Oct 2012 | 17:53:22 UTC Aborted by user 0.00 0.00 --- PPS (LLR) v1.01 413848949 187389 6 Oct 2012 | 17:56:38 UTC 7 Oct 2012 | 2:53:40 UTC Completed and validated 8,374.76 8,323.37 11.41 PPS (LLR) v6.13

shows 1.017239M sent at 15:17, then resent at 17:56, just short of
the 18:00 UTC close. And for one just past the close
Is prime? 6699*2^1017827+1 is not prime. ... 413686351 291102 6 Oct 2012 | 18:27:50 UTC 6 Oct 2012 | 22:23:40 UTC Aborted by user 0.00 0.00 --- PPS (LLR) v6.13 413853911 187389 6 Oct 2012 | 18:28:12 UTC 7 Oct 2012 | 5:11:13 UTC Completed and validated 8,568.37 8,507.76 9.89 PPS (LLR) v6.13 413872377 239220 6 Oct 2012 | 22:23:56 UTC 7 Oct 2012 | 0:37:04 UTC Completed and validated 551.37 456.55 9.89 PPS (LLR) v1.01

with quorum 2, sent at 18:27 UTC, for 1.0178M

Server seems to have gotten past the load from the contest; I'm
seeing 180 tasks to be validated now, under 100 a few minutes ago;
and "primes found" back up (rather than "down for contest").

-Bruce* (@peppert*: nice cores!)

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 58587 - Posted: 22 Oct 2012 | 0:06:00 UTC

2012 Goals - Progress update

Good progress is being made in advancing PPS (LLR) towards its 2012 goals. Let's keep up the good work!!!

Work will continue for the 1200-10000 ranges up to n=1.1M. SGS is online and we want to make sure PPS is high enough to remain on the Top 5000.

Thank you again to everyone helping out with this effort. :)

Proth Prime Search Update
k=3 to 9M - ongoing in the 321 Prime Search (trailing edge n=8.62M; leading edge n=8.68M)
k=5 to 6M - (currently at n=5.55M)
k=7 to 6M - (currently at n=5.55M)
k=9 to 4M - (currently at n=4M)
10<k<32 to 2.5M - (currently at n=2.5M)
32<k<100 to 2.2M - (currently at n=2M)
100<k<500 to 2.2M - (currently at n=2M)
500<k<1200 to 2.2M - (currently at n=1.5M)
1200<k<10000 to 1.03M - ongoing leading edge n=1.03M)

Lennart

Updated 2012-10-23 00:30 UTC

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 58616 - Posted: 23 Oct 2012 | 5:38:03 UTC

FYI, the PPS goals in the LLR priorities thread are outdated.
____________

Michael Millerick
Volunteer tester

Joined: 4 Feb 09
Posts: 790
ID: 35074
Credit: 302,410,770
RAC: 327,110

Message 59094 - Posted: 5 Nov 2012 | 7:35:37 UTC

k=3,5,7 have all met or exceeded their goals.
____________

Lennart SM5YMT
Honorary cruncher

Joined: 7 May 07
Posts: 1125
ID: 7989
Credit: 694,692,344
RAC: 0

Message 59868 - Posted: 28 Nov 2012 | 3:37:35 UTC

2012 Goals - Progress update

Good progress is being made in advancing PPS (LLR) towards its 2012 goals.
Only about one month left. Let's keep up the good work!!!

Work will continue for the 1200-10000 ranges up to n=1.1M. SGS is online and we want to make sure PPS is high enough to remain on the Top 5000.

Thank you again to everyone helping out with this effort. :)

Proth Prime Search Update
k=3 to 9M - ongoing in the 321 Prime Search (trailing edge n=8.99M; leading edge n=8.68M)
k=5 to 6M - (currently at n=6M) DONE
k=7 to 6M - (currently at n=6M) DONE
k=9 to 4M - (currently at n=4M) DONE
10<k<32 to 2.5M - (currently at n=2.5M) DONE
32<k<100 to 2.2M - (currently at n=2M)
100<k<500 to 2.2M - (currently at n=1.5M)
500<k<1200 to 2.2M - (currently at n~1.6M)
1200<k<10000 to 1.03M - ongoing leading edge n=1.03M) DONE

Lennart

Updated 2012-11-28 02:30 UTC

dancer42

Joined: 15 Oct 12
Posts: 99
ID: 175439
Credit: 19,009,646
RAC: 0