TPS Record Rally: September 16-21
 

Hi everyone!

We've reached the optimal sieve depth for the variable-n range, so we're having a rally to celebrate its completion. It will start on September 16, 00:01 CDT and end on September 21, 23:59 CDT. The rally will be on both PRPNet (port 12000) and LLRNet (port 12050). As usual, the person and team that tests the most k/n pairs and/or finds the most primes will win the competition.

The records to break for this rally are:
- number of k/n pairs tested in a day (current record is 29961 pairs on January 16, 2011)
- largest results file generated in a day (current record is 2.6 MB for the PRPNet server on January 16)
- most primes found in a 24 hour period (current record is 8 primes found between May 3, 2:00 CDT and May 4, 2:00 CDT)
- largest known twin prime (current record is 100355 digits found on August 2009)

We'll see how many of those records fall in the upcoming days. In the meantime, happy prime and twin hunting :smile:
--------------------------------------------------------
(original message)
Hi everyone!

We haven't had a rally for quite some time, so we'll soon have a big and long one to make up for that. :smile: It will start on September 14th and end a week later on the 21st.

Like last time, we'll be having a sieve rally again. With the variable-n range sieve entering it's third year of operation, we figured that it's about time the optimal sieve depth was reached. We'll be working on n=495K-500K, which has a bit more than 1P to go. To participate, head over to the sieve reservation thread here: [URL]http://www.mersenneforum.org/showthread.php?t=12260[/URL]

The rules will be similar to last time: you can reserve a range at any time on the 14th, but you'll need to finish it within 168 hours if you want full credit. That is, a range reserved on Sept. 14, 10:32 GMT will have to be completed on or before Sept. 21, 10:32 GMT.

During the previous rally, our average rate was 1797M p/sec, or 155T/day. If we average a little above that during next week's rally, it will be all that's needed to put this sieve effort behind us once and for all. So let's dust off those computers, finish off the range, and git-r-done!

The overall stats for the rally are at the bottom of this post.

Rally stats for September 16:
[code]
User k/n pairs tested Primes found
odicin 6817 1
Oddball 1483 1
mdettweiler 399 0
Robert47 3 0

Team k/n pairs tested Primes found
BOINC_Confederation 6817 1
TwinFinders 1483 1
RaidersOfTheLostPrimes 399 0
Everyone else 3 0

Total: 8702 pairs tested, 2 primes found
Rate: 1 pair every 9.93 seconds
[/code]

Rally stats for September 17:
[code]
User k/n pairs tested Primes found
odicin 6984 2
sm5ymt 4380 3
Lumiukko 2872 0
Rick_Reynolds 1769 1
Oddball 1536 0
mdettweiler 528 0
Tim 505 0
Robert47 51 0
Tarmo_Ilves 48 0

Team k/n pairs tested Primes found
PrimeSearchTeam 7252 3
BOINC_Confederation 6984 2
Aggie_The_Pew 2322 1
TwinFinders 1536 0
RaidersOfTheLostPrimes 528 0
Everyone else 51 0

Total: 18673 pairs tested, 6 primes found
Rate: 1 pair every 4.63 seconds
[/code]

Rally stats for September 18:
[code]
User k/n pairs tested Primes found
sm5ymt 27195 5
Lumiukko 7063 3
odicin 6788 2
Rick_Reynolds 2689 1
Oddball 1545 1
Caravaggio 759 0
mdettweiler 722 0
Tim 545 0
Ian_Smith 21 0

Team k/n pairs tested Primes found
PrimeSearchTeam 34258 8
BOINC_Confederation 6788 2
Aggie_The_Pew 4014 1
TwinFinders 1545 1
RaidersOfTheLostPrimes 722 0

Total: 47327 pairs tested, 12 primes found
Rate: 1 pair every 1.83 seconds
[/code]

Rally stats for September 19:
[code]
User k/n pairs tested Primes found
sm5ymt 27789 6
Lumiukko 6872 1
odicin 5656 3
Rick_Reynolds 2666 0
mdettweiler 1527 0
Oddball 1398 0
Caravaggio 792 0
Tim 425 0
Ian_Smith 116 0
Robert47 14 0

Team k/n pairs tested Primes found
PrimeSearchTeam 34661 7
BOINC_Confederation 5656 3
Aggie_The_Pew 3999 0
RaidersOfTheLostPrimes 1527 0
TwinFinders 1398 0
Everyone else 14 0

Total: 47255 pairs tested, 10 primes found
Rate: 1 pair every 1.83 seconds
[/code]

Rally stats for September 20:
[code]
User k/n pairs tested Primes found
sm5ymt 33408 7
Lumiukko 8146 2
odicin 5403 1
Rick_Reynolds 2806 0
Mark_Doom 2031 0
mdettweiler 1728 0
Oddball 1527 0
Caravaggio 867 0
Puzzle-Peter 820 0
Tim 520 0
flat 352 0
rebirther 294 0
Ian_Smith 225 0
Gary_Craig 130 0
Tarmo_Ilves 6 0

Team k/n pairs tested Primes found
PrimeSearchTeam 41554 9
Aggie_The_Pew 6585 0
BOINC_Confederation 5697 1
RaidersOfTheLostPrimes 1728 0
TwinFinders 1527 0
SETI.Germany 1172 0

Total: 58263 pairs tested, 10 primes found
Rate: 1 pair every 1.48 seconds
[/code]

Rally stats for September 21:
[code]
User k/n pairs tested Primes found
sm5ymt 22875 3
Lumiukko 7909 2
odicin 5775 2
Rick_Reynolds 2062 1
mdettweiler 1863 0
Oddball 1564 0
Puzzle-Peter 1095 0
flat 960 0
Caravaggio 947 0
Mark_Doom 857 0
Tim 255 0

Team k/n pairs tested Primes found
PrimeSearchTeam 30784 5
BOINC_Confederation 5775 2
Aggie_The_Pew 4121 1
SETI.Germany 2055 0
RaidersOfTheLostPrimes 1863 0
TwinFinders 1564 0

Total: 46162 pairs tested, 8 primes found
Rate: 1 pair every 1.87 seconds
[/code]

Total rally stats for September 16-21:
[code]
Rank User k/n pairs tested Primes found
1 sm5ymt 115647 24
2 odicin 37423 11
3 Lumiukko 32862 8
4 Rick_Reynolds 11992 3
5 Oddball 9053 2
6 mdettweiler 6767 0
7 Caravaggio 3365 0
8 Mark_Doom 2888 0
9 Tim 2250 0
10 Puzzle-Peter 1915 0
11 flat 1312 0
12 Ian_Smith 362 0
13 rebirther 294 0
14 Gary_Craig 130 0
15 Robert47 68 0
16 Tarmo_Ilves 54 0


Rank Team k/n pairs tested Primes found
1 PrimeSearchTeam 148509 32
2 BOINC_Confederation 37717 11
3 Aggie_The_Pew 21041 3
4 TwinFinders 9053 2
5 RaidersOfTheLostPrimes 6767 0
6 SETI.Germany 3227 0
7 Everyone else 68 0

Total: 226382 pairs tested, 48 primes found
Overall rate: 1 k/n pair every 2.29 seconds
[/code]

All ranges below the optimal sieve depth have been reserved, so the rally has been changed from a sieve rally to an LLR one. The starting time is postponed to September 16, 00:01 CDT in order to let sievers complete their ranges.

Is it possible to work together with Primegrid for PSA points?

[QUOTE=rebirther;271626]Is it possible to work together with Primegrid for PSA points?[/QUOTE]
There won't be any direct collaboration with Primegrid for this rally, but I'm considering that option for future ones. Of course, Rytis and/or Lennart would also have to agree on putting part of the variable-n range search on Primegrid's PRPNet.

Anyway, the rally is about to start in less than 3 hours :smile: Anyone in?

[QUOTE=Oddball;271803]Anyone in?[/QUOTE]

Yes... I'm switching cores to TPS-Server...

Regards Odi

[QUOTE=Oddball;271803]There won't be any direct collaboration with Primegrid for this rally, but I'm considering that option for future ones. Of course, Rytis and/or Lennart would also have to agree on putting part of the variable-n range search on Primegrid's PRPNet.

Anyway, the rally is about to start in less than 3 hours :smile: Anyone in?[/QUOTE]
Ah, whoops! I'd been planning to participate but the date came upon me rather quickly. (Things have been a bit hectic in the offline world lately.)

I'll see about switching some cores over tomorrow. :smile:

Say, I just noticed something rather interesting...at the rate we're currently processing, we are actually on track to finish the k<100K range up to n=500K in under 5 days! :shock: The rally will, of course, be over somewhat before then, but if a critical mass of participants stuck around for at least a few days after the rally, we should be able to finish it off.

I'll leave my 9 cores on for another week or so; anyone with me? :grin:

[QUOTE] anyone with me? :grin:[/QUOTE]

Not all cores, but I think i will left some more as usual at least to pg equinox challenge.

Regards Odi

[QUOTE=mdettweiler;272274]Say, I just noticed something rather interesting...at the rate we're currently processing, we are actually on track to finish the k<100K range up to n=500K in under 5 days! :shock: The rally will, of course, be over somewhat before then, but if a critical mass of participants stuck around for at least a few days after the rally, we should be able to finish it off.[/QUOTE]
I think a better goal would be to finish k<100K up to n=495K, not 500K. First of all, the 495K-500K sieve file is only sieved to p=10G, compared with 950T for the 490K-495K file. Secondly, we can reach n=495K before Primegrid's equinox rally starts if a lot of participants stick around, but it won't be possible to get to n=500K by that time even if we get an updated sieve file.

[quote]
I'll leave my 9 cores on for another week or so; anyone with me? :grin:[/quote]
I'm leaving most of my cores for another ~5 days (the rest will be used to doublecheck the primes found during the rally).

BTW, it appears that the PRPNet server will dry out soon. Max, could you load n=492.5K-495K into port 12000?

[QUOTE=Oddball;272293]I think a better goal would be to finish k<100K up to n=495K, not 500K. First of all, the 495K-500K sieve file is only sieved to p=10G, compared with 950T for the 490K-495K file. Secondly, we can reach n=495K before Primegrid's equinox rally starts if a lot of participants stick around, but it won't be possible to get to n=500K by that time even if we get an updated sieve file.[/quote]
Ah, right...I'd forgotten that the 495K-500K sieve file is currently only sieved to 10T. The sieving, of course, is nearly done, but the factors will need to be removed before that will be of any use for LLRing. Hopefully gribozavr will reappear soon bearing sieve files... :smile:
[quote]I'm leaving most of my cores for another ~5 days (the rest will be used to doublecheck the primes found during the rally).

BTW, it appears that the PRPNet server will dry out soon. Max, could you load n=492.5K-495K into port 12000?[/QUOTE]
Okay, will do.