PrimeGrid Leonids Challenge Nov 18-23

[somethingsketchy]

64 moar cores crunching as of right now

Yah!

 

[Ken g6]

Day 3 stats +1 hour:

Rank___Credits____Username
15_____2525809____petrusbroder
48_____1191898____biodoc
93_____702949_____filibusterman
98_____681678_____Uppsala9496
100____669026_____kimmyl
108____599888_____ZipSpeed
168____378684_____SlangNRox
193____330486_____Robert Everly
194____329682_____GLeeM
210____300706_____Larry
260____231077_____Ken_g6
261____230079_____Chokko
266____212779_____Ryan
309____181268_____zzuupp
363____132989_____waffleironhead
382____129618_____sangyup81
473____83371______blckgrffn
489____69722______Chris
562____65440______geecee

Rank__Credits____Team
6_____11020325___Czech National Team
7_____10343024___PrimeSearchTeam
8_____9435079____Dutch Power Cows
9_____9047158____TeAm AnandTech
10____9001624____Aggie The Pew
11____6533281____The Knights Who Say Ni!
12____5863897____US Navy

We're back between the Cows and the rats again. Seems like we've been here before.

Welcome to Chris, and anyone else I missed. :$ We have 19 participants in this race - I think that's a record!

Now is a good time to start thinking about juggling WUs. I already juggled one more onto my dual-core, and I'm planning to juggle 3 more onto my quad-core (a.k.a. juggle one out so three more will complete on time) tomorrow morning.

 

[petrusbroder]

Thanks for the stats.
It seems to me that you do the eveningstast (UTC) and I do the morning stats (UTC).
Sounds very good to me.

 

[biodoc]

64 moar cores crunching as of right now

Awesome!

 

[somethingsketchy]

I've already shut down any receipt of new tasks and plan on burning through the queue that I have (roughly 10 tasks). I've also up'd the number of threads on the main rig to hopefully burn through those tasks.

 

[filibusterman]

I've already shut down any receipt of new tasks and plan on burning through the queue that I have (roughly 10 tasks). I've also up'd the number of threads on the main rig to hopefully burn through those tasks.

If you do not allow new tasks you may have to manually flush before the end of the race or is there a way to avoid this?

 

[Ken g6]

If you do not allow new tasks you may have to manually flush before the end of the race or is there a way to avoid this?

True. There is a configuration setting to flush immediately when a WU finishes, but I can't find it right now.

Another option is to judge when a machine has downloaded all the work it's going to complete for the race, and then switch it to a "location" (home, work, school, or (none)), running something fast like PPS LLR. It should then download that work frequently, and upload your results at the same time.

Edit: My shell script, in the same thread with my juggling guide, will flush for you if you're on Linux.

 

[blckgrffn]

So, they need to be reported in before the race concludes, is that it?

 

[somethingsketchy]

True. There is a configuration setting to flush immediately when a WU finishes, but I can't find it right now.

Another option is to judge when a machine has downloaded all the work it's going to complete for the race, and then switch it to a "location" (home, work, school, or (none)), running something fast like PPS LLR. It should then download that work frequently, and upload your results at the same time.

Edit: My shell script, in the same thread with my juggling guide, will flush for you if you're on Linux.

This is pretty much what I am doing. I am betting that I already have enough WUs to complete the competition on time.

So, they need to be reported in before the race concludes, is that it?

That's correct, all WUs need to be done on time, in order to be counted for the competition.

 

[blckgrffn]

This is pretty much what I am doing. I am betting that I already have enough WUs to complete the competition on time.



That's correct, all WUs need to be done on time, in order to be counted for the competition.

I guess I assumed they had some sort of completion time stamp on them - but I suppose there would be those that cheat at even something like this.

 

[somethingsketchy]

I guess I assumed they had some sort of completion time stamp on them - but I suppose there would be those that cheat at even something like this.

Well after doing some pre-computing of some WUs before the competition, I figured it would be roughly 20-22 hrs/WU for my main rig.

Given that we're less than 40 hours from the end of the competition, I'm making a bet that I can complete all 11 tasks (one will be done within the next two hours) on time (even while running 24/7).

 

[petrusbroder]

Individual stats as of November 22, 05:45:58

Rank _____ Name ______________________ Points
017 ______ petrusbroder ___________ 2 985 408.61
051 ______ biodoc _________________ 1 317 145.02
091 ______ filibusterman ____________ 865 406.79
096 ______ Kymmyl ___________________ 828 713.83
097 ______ Uppsala9496 ______________ 823 106.20
121 ______ ZipSpeed _________________ 599 888.81
148 ______ SlangNRox ________________ 516 981.71
174 ______ Robert Everly ____________ 433 970.65
207 ______ Larry ____________________ 368 443.35
215 ______ GLeeM ____________________ 347 416.13
236 ______ Ken_g6 ___________________ 317 096.26
292 ______ Chokko ___________________ 230 079.49
294 ______ ryan _____________________ 229 082.37
325 ______ zzuupp ___________________ 198 645.21
401 ______ sangyup81 ________________ 146 023.90
430 ______ waffleironhead ___________ 132 989.78
497 ______ blckgrffn _________________ 99 972.95
556 ______ Chris _____________________ 69 722.20
623 ______ geecee ____________________ 65 440.70

1206 crunchers are priming.


Teams' stats as of November 22, 05:45:32
Rank ____ Name ____________________ Points
05 ______ SETI.Germany ____________ 15 361 988.49
06 ______ Czech National Team _____ 13 325 182.99
07 ______ PrimeSearch Team ________ 11 593 175.30
08 ______ Dutch Power Cows ________ 11 266 973.94
09 ______ TeAm AnandTech __________ 10 575 533.97
10 ______ Aggie The Pew ___________ 10 459 199.89
11 ______ The Knights Who Say Ni! __ 7 556 344.99
12 ______ US Navy __________________ 7 385 249.27
13 ______ BOINC@Poland _____________ 7 069 285.26


214 teams are in the race.

 

[somethingsketchy]

So close...I wonder if I can catch up to Chokko before the end of the competition

292 ______ Chokko ___________________ 230 079.49
294 ______ ryan _____________________ 229 082.37

 

[blckgrffn]

Ha... looks like these cores are going to need ~36+ hours to get through a WU. Good thing I kicked them all off with ~48 hours to go in the race!

 

[ZipSpeed]

The 12 tasks currently crunching should make the deadline before the end of the race with a couple hours to spare.

 

[geecee]

Sadly, I think I'm almost done with the race already. The last WU that I downloaded looks like it's going to finish about an hour after the race ends. I have one more WU that will finish in the next two hours, and then any WU that gets pulled down after that will finish about 4-5 hours after the race is over. It's a G6950 box that's taking around 30-31 hours to finish each unit.

I was originally going to dedicate 6 cores to this race, but the 4 old Nocona Xeon cores I had available were going to take more than the whole race time allotted to finish a single unit. I was hoping I might be able to complete one unit on each core during the race period, but it was not going to happen.

 

[Ken g6]

FYI, older machines like Core 2 Quads are bottlenecked in their RAM bandwidth. If 4 WUs aren't going to finish in time, abort 2 or 3.

 

[Uppsala9496]

Just took my fastest machine offline for a bit. Went to best buy and bought a Galaxy GTX 460 for $111.99 on clearance. Has a $30 rebate too. Need to get all of the drivers installed in linux and make sure the sound works (pain in the rear). Should be able to get another 8 units completed on it before the end of the race (finishes them in just over 15 hours and have 4 currently going).

 

[blckgrffn]

FYI, older machines like Core 2 Quads are bottlenecked in their RAM bandwidth. If 4 WUs aren't going to finish in time, abort 2 or 3.

Awesome, the bandwidth starved 4S Core 2 based systems are perfect for this then!

 

[filibusterman]

Figures looks like I have a power outage at my house...maybe I will have to go home at lunch to start the rigs back up.

 

[Ken g6]

Day 4 stats:
Rank___Credits____Username
16_____3615645____petrusbroder
52_____1614886____biodoc
72_____1246581____kimmyl
88_____1065755____filibusterman
95_____993031_____Uppsala9496
112____809785_____ZipSpeed
157____586409_____SlangNRox
174____523563_____Larry
211____433970_____Robert Everly
249____367129_____Ken_g6
250____365149_____GLeeM
253____353173_____Chokko
323____247156_____Ryan
351____215906_____zzuupp
485____132989_____waffleironhead
488____132277_____blckgrffn
497____129579_____sangyup81
585____85816______Chris
614____83082______geecee

Rank__Credits____Team
6_____16536931___Czech National Team
7_____13757394___Dutch Power Cows
8_____13715610___PrimeSearchTeam
9_____13001890___TeAm AnandTech
10____12708097___Aggie The Pew
11____9118242____US Navy
12____9017144____The Knights Who Say Ni!

We're getting a bit better lead on the rats, but not very much. The cows and PrimeSearchTeam are also awfully close targets.

24 hours to go. Juggle like your score depended on it!

 

[somethingsketchy]

Now juggling all 10 WUs on my main rig. Hopefully I'll make it on time (though I may lose two of them I predict).

 

[SlangNRox]

I'll only have 8 more wu that will finish before the race is over

 

[petrusbroder]

Juggling fouriously ... I hope to deliver at least some 40 of not 42 results ...

 

[Uppsala9496]

92 filibusterman TeAm AnandTech 1065755.70
93 Uppsala9496 TeAm AnandTech 1065537.65

Wonder if I can catch up...
One of my machines is done with the race (about 24 hours to finish a WU).
Only have about 10 remaining WU's.