Join PrimeGrid
Returning Participants
Community
Leader Boards
Results
Other
drummerslowrise

Message boards :
Aggie The Pew message board
: Our "own" primes
Author 
Message 
CrunchiVolunteer tester
Send message
Joined: 25 Nov 09 Posts: 3008 ID: 50683 Credit: 59,175,698 RAC: 1,834

As I write in CRUS project thread: and if someone is interested: we, as team, can do our own sieve and search our "own" primes in any base we like.
Of course you will not be credited here, but since goal is same: find prime: you will get your name near prime.
Since we have very good computer power, we, as team can choose one sequence to search and if we all participate will be over in one month.
So if someone , will try can do that in this thread....
Everything can be and will be discussed open, so we all know the rules :)
____________
92*10^^{1439761}1 REPDIGIT PRIME :) :) :)
314187728^^{131072}+1 GENERALIZED FERMAT
31*332^^{367560}+1 CRUS PRIME
Proud member of team Aggie The Pew. Go Aggie!  

NeoVolunteer tester
Send message
Joined: 28 Oct 10 Posts: 710 ID: 71509 Credit: 91,178,992 RAC: 0

CrunChi,
I love the way you think. I've been waiting to start an AtP only prime search.
Two weeks ago, I reserved base R138 over at mersenne in the CRUS program.
There we 8K's left to solve for that base.
I started my own sieve using srsieve, then used Sr2sieve to take it to 5000e9, sieving for 100,000 < n < 250,000
I was very fortunate and found my first prime early
372*138^1031601 (220,753 digits)
I then removed K=372 from my sieve file and sieved it super deep. LLR's are running at 220 seconds per candidate, and the removal rate when sieving was 400+ seconds per candidate. So, my LLR candidate file is super ripe for primes.
7K's left to go. In about 10 hours I will have LLR'd all the way to N=120,000.
If anyone wants my LLR candidate file to test for for a certain range of N, I am more than happy to break it up into sections.

Alternatively,
I am doing manual sieving for proth/riesel over at Primsearchteam on my GPU's for Primegrid's next proth search run (9Million < N < 12Million). tpsievecuda is extremely fast on my 770 gtx. I get 17M/p sec. I have about 6 more days to go before I finish my reserved range.
Due to the huge amount of computing power our team has, we could actually use tpsieve to sieve for a small range of k and a relatively small range of N values, but at a high N...
So, maybe a range like
11 < K <113 and 35,000,000 < N < 36,000,000
Tpsievecuda would find factors in this range for both +1 and 1's...
Those would be some big candidates to test, and that range of N above was merely an example. But the lower K values would mean smaller FFT values which means faster crunching. We just have to make sure we are not duplicating work already completed in the 321, 27, and 121 search projects...
I'm game.
Neo
AtP  

CrunchiVolunteer tester
Send message
Joined: 25 Nov 09 Posts: 3008 ID: 50683 Credit: 59,175,698 RAC: 1,834

Neo, first of all thanks for answer.
CRUS , as project is ok: but it is just resolving one problem. I was refer to another approach.
If you look at Top 5000 you can see there is few repdigit primes.
They are interesting form since only one digit is different from all other ( and all other is same)
I found 95*10^4660021 ( 949999....999)
There is only ONE repdigit prime above 1M. I am currently deeply in those range for 3K but still no results.
Or look another: there is no prime above 380000 digits with base 7. I do also test in this range, without success ( for now) but will find one.
For me, important is that I found some prime that prime was some kind of special ( first in the world). Resolving CRUS base will not do that , primes are to small. And since the use bigger base, primes got very rare.
As you know primes on base 10 are 3.13 x rare then primes on base 2.
So lets take only one K: with ten CPU working only 48 hours ( 40 cores) we can make any sieve deep enough. And then we evenly distribute data from sieve and process that.
Whatever we find will be greatest in world of primes. And then if you wont come back to your usual activity on Primegrid or take some other K and process again.
Do not hunt loans and patches , find bigger, special primes ( and few days of work less on Primegrid noone will notice) . When you find primes ( and most of us here search for primes as first goal) our name will be entered in Top 5000. And on that site there is no place where is written how many points you have on Primegrid and on what project you have biggest badge. It is totally irrelevant. But fact you find prime is fact that is recorded!
In any case your computers will work...
____________
92*10^^{1439761}1 REPDIGIT PRIME :) :) :)
314187728^^{131072}+1 GENERALIZED FERMAT
31*332^^{367560}+1 CRUS PRIME
Proud member of team Aggie The Pew. Go Aggie!  

NeoVolunteer tester
Send message
Joined: 28 Oct 10 Posts: 710 ID: 71509 Credit: 91,178,992 RAC: 0

CrunChi,
What sieving program do you use to start your search?
Is the sieve GPU based?
The format of 95*10^4660021 looks as if it's a CRUS project...
See here:
http://www.noprimeleftbehind.net/crus/Rieselconjecturereserves.htm
If this is the case, I know how to start the sieve and then take it deep.
Neo  

CrunchiVolunteer tester
Send message
Joined: 25 Nov 09 Posts: 3008 ID: 50683 Credit: 59,175,698 RAC: 1,834

CrunChi,
What sieving program do you use to start your search?
Is the sieve GPU based?
The format of 95*10^4660021 looks as if it's a CRUS project...
See here:
http://www.noprimeleftbehind.net/crus/Rieselconjecturereserves.htm
If this is the case, I know how to start the sieve and then take it deep.
Neo
I start sieve with Newpgen and since I only make sieve for one K use sr1sieve ( much faster then sr2sieve) . So it cannot be made with GPU.
95*10^n1 is not CRUS project. Only CRUS for base 10 is 4421 (n1) and 7666 for (n+1) Only about 100 factor (K) (2,3,5,6,8,9,92,93,95,96,98,99....) can produce repdigit prime....
____________
92*10^^{1439761}1 REPDIGIT PRIME :) :) :)
314187728^^{131072}+1 GENERALIZED FERMAT
31*332^^{367560}+1 CRUS PRIME
Proud member of team Aggie The Pew. Go Aggie!  

Message boards :
Aggie The Pew message board
: Our "own" primes 