PrimeGrid Autumnal Equinox Challenge September 23 - October 6

[somethingsketchy]

Schweet! We did beat Team BOINC@Poland. I'll admit I was getting nervous for the last two days and whetever they would steal our 9th spot

 

[biodoc]

Great race everyone!! :thumbsup:

 

[rabrittain]

Thanks to Ken_g6 and petrusbroder for the stats!!!

Great race!!

 

[sangyup81]

Thank you keeping us up to date! I'm itching to upgrade!

 

[RobertE]

Good luck, bad luck or both, I'm not sure. About 3 hours after the race ended, my power supply in my main rig gave up the ghost. Just a click and everything went dark.

Thankfully, I've got a Microcenter close by, new PS in place and up and running again.

I think it simply overheated. It's in a case with the PS intake on the bottom that had a little dust filter. The filter did it's job, but I had totally forgotten about it. Oops.

 

[theAnimal]

Good luck, bad luck or both, I'm not sure. About 3 hours after the race ended, my power supply in my main rig gave up the ghost. Just a click and everything went dark.

Thankfully, I've got a Microcenter close by, new PS in place and up and running again.

I think it simply overheated. It's in a case with the PS intake on the bottom that had a little dust filter. The filter did it's job, but I had totally forgotten about it. Oops.

Hopefully you got a good quality unit, an overheated PSU should simply shut off.

 

[RobertE]

Hopefully you got a good quality unit, an overheated PSU should simply shut off.

It did. It smelled hot, not burnt. Computer is back up and running. It did not revive itself (power supply) after the trip to microcenter and back. So now to deal with OCZs RMA process. At least nothing appears to be dead.

 

[somethingsketchy]

Ouch that sucks. I noticed, too, my fan filters are covered with dust. Definitely cleaning those tonight after work.

 

[Ken g6]

Quite a few unstable overclocks out there it seems.

667____172161_____waffleironhead

Today:
735____144239_____waffleironhead

Yeah, apparently quite a few!

 

[petrusbroder]

I'll keep track of the development of the stats and how much the stats change during the clean-up phase.

Teams' stats as of October 7, 17:00:46
06 ______ PrimeSearch Team_______ 36 004 816.19 _____ (+/-0)
07 ______ Czech National Team ___ 31 422 841.19 _____ (-60 884.81)
08 ______ Dutch Power Cows ______ 23 336 031.62 _____ (+/-0)
09 ______ TeAm AnandTech ________ 17 780 900.19 _____ (-84 072.81) :\
10 ______ BOINC@Poland __________ 17 252 481.57 _____ (-58 128.43)
11 ______ BOINCstats ____________ 16 879 494.55 _____ (+/-0)
12 ______ Duke University _______ 16 155 748.93 _____ (-2 481.07)

Not too bad, though ... but I really hope that we will not loose 500K points!

 

[waffleironhead]

Today:
735____144239_____waffleironhead

Yeah, apparently quite a few!

My temps spiked one day. No overclock, just poor ventilation. :'( I saw the temp over 60 and I knew i was in trouble. ouch.

 

[VirtualLarry]

My temps spiked one day. No overclock, just poor ventilation. :'( I saw the temp over 60 and I knew i was in trouble. ouch.

Is that an AMD rig? An Intel rig should just shrug its shoulders at 60C.

 

[RobertE]

only 5 pending left for me, so far so good.

 

[petrusbroder]

198 results validated, 13 results pending, 0 with errors, 4 marked as invalid.

The 4 "invalid results" reached consensus before my result was sent in. Thus I crunched the WU, sent in successfully a correct result and got no points.
<Petrus shrugs>.
Those 4 results would have given some 105K points in the race, which would have given me one step up the ladder.
Thus, if they send out 4 - 5 replications and got two incorrect results and three correct results, then only the two, which arrive first, get the points.
Hmmm, I am not so sure I like that.

 

[RobertE]

198 results validated, 13 results pending, 0 with errors, 4 marked as invalid.

The 4 "invalid results" reached consensus before my result was sent in. Thus I crunched the WU, sent in successfully a correct result and got no points.
<Petrus shrugs>.
Those 4 results would have given some 105K points in the race, which would have given me one step up the ladder.
Thus, if they send out 4 - 5 replications and got two incorrect results and three correct results, then only the two, which arrive first, get the points.
Hmmm, I am not so sure I like that.

I'm sure Ken will correct me if I'm wrong, but as long as you return the result within the challenge window and it validates as "valid" against the other results, then you should get points for it.

 

[somethingsketchy]

I wonder if there's some sort of threshold for invalid results. Like if the WU gets two invalids within a certain amount of time, then the WU is "dumped" and no one past the first couple of results gets the points...

 

[Ken g6]

I'm sure Ken will correct me if I'm wrong, but as long as you return the result within the challenge window and it validates as "valid" against the other results, then you should get points for it.

I actually asked about this at the start of the race, as I had one computer that was too slow to complete one WU in 7 days! (It took 9.5 or so.) I verified that the WU was sent in, late, and I got credit for it. I'm sure you did too, Peter.

I wonder if there's some sort of threshold for invalid results. Like if the WU gets two invalids within a certain amount of time, then the WU is "dumped" and no one past the first couple of results gets the points...

Note this line on every work unit page:

max # of error/total/success tasks 15, 15, 5

That means if there are 15 total attempts, or 5 successful but different results, no more work is sent out for that task until the admins take a look at it. This is rare enough that they usually send it out for at least one more round of 15/15/5.

 

[Ken g6]

Peter hasn't posted our progress regress today, so:

6_____36004816___PrimeSearchTeam (+/-0)
7_____31422841___Czech National Team (-60 885, ~2 bad WUs)
8_____23336031___Dutch Power Cows (+/-0)
9_____17665505___TeAm AnandTech (-199 468, ~6 bad WUs :\)
10____17132460___BOINC@Poland (-178 150, ~5 bad WUs)
11____16850512___BOINCstats (-28 983, ~1 bad WU)
12____16155748___Duke University (-2 482)

 

[sangyup81]

Interesting. I have never had an invalid result with any of my machines until this challenge =T

 

[VirtualLarry]

Is there any way to find out if the invalid result came from your username/account? I am curious if my overclock isn't truely stable. It passes one hour of OCCT Linpack 64-bit with 90&#37; RAM (out of 8GB), so I figured it was stable.

Edit: On my primary rig, I didn't have to boost the vcore. Since my Q9300 is essentially two E5200 chips sandwiched together, and E5200 chips often do 3.0 Ghz on stock voltage, I didn't think it was too much of a stretch for my chip to do 3.0 Ghz at stock voltage. My secondary rig, which has identical components (except for the case), needed a couple of minor clicks upward on the vcore to be OCCT one-hour stable.

Edit: I found a way on the PrimeGrid web site to drill down to the active computers, and then to the WUs. None of mine show Invalid, but there is one Woodall LLR that is still Pending credit.

 

[biodoc]

So far I have 105 WUs validated, 13 pending and no "invalid" WUs.

 

[Ken g6]

Regression testing and reporting. It's what I do for a living, and I'm doing it here now too. (Sort of.) :whiste:

6_____36004816___PrimeSearchTeam (+/-0)
7_____31392432___Czech National Team (-91 294, ~3 bad WUs)
8_____23336031___Dutch Power Cows (+/-0)
9_____17606690___TeAm AnandTech (-258 283, ~8 bad WUs :\)
10____16957322___BOINC@Poland (-353 288, ~12 bad WUs )
11____16820580___BOINCstats (-58 915, ~2 bad WUs)
12____16155748___Duke University (-2 482)

Well, doesn't look like we're going to fall back behind BOINC@Poland. But they might fall behind BOINCstats!

 

[somethingsketchy]

I'm wondering how the PrimeSearchTeam and the Cows had no invalid WUs? *demands recount*

 

[Ken g6]

PrimeSearchTeam consists of the PrimeGrid admins, plus others very experienced in prime searching. They know better than to overclock their systems too high. (And/or they're always testing for primes, so if something was producing errors they'd know about it long before the race.)

And, FYI, as of this morning, the Cows have lost one.

 

[Ken g6]

Faulty WU reports should be slowing down, so I thought I'd wait a bit until my next report. (Or Peter can do some reporting.)

6_____36004816___PrimeSearchTeam (+/-0)
7_____31266849___Czech National Team (-216 877, ~7 bad WUs)
8_____23278384___Dutch Power Cows (-57 647, ~2 bad WUs)
9_____17491217___TeAm AnandTech (-373 756, ~13 bad WUs :\)
10____16806669___BOINC@Poland (-503 941, ~15 bad WUs )
11____16763543___BOINCstats (-115 952, ~4 bad WUs)
12____16123998___Duke University (-34 232 ~1 bad WU)

Nobody's perfect, except PrimeSearchTeam. BOINC@Poland may still lose to BOINCstats - 2 more bad WUs would do it!

Edit: As of 10/12, BOINCstats is now 10th and BOINC@Poland is now 11th!