PrimeGrid Challenges 2022

[Markfw]

@Markfw you are on 5.4 now you need 5.10 or higher.

@crashtech is running 5.13

My brain was thinking of the first digit. 4 is higher than 1, but its 10.....

Duhhhhh Have another beer Mark !

 

[Ken g6]

Day 1.5 stats:

Rank___Credits____Username
1______14532908___Skillz
5______7803899____crashtech
18_____3253910____Icecold
23_____2638241____biodoc
26_____2412027____markfw
35_____1758185____xii5ku
41_____1449130____Orange Kid
48_____1249378____parsnip soup in a clay
71_____873336_____Fardringle
97_____564792_____Skivelitis2
100____512130_____Lane42
108____486509_____VirtualLarry
243____135999_____waffleironhead
267____112135_____Ken_g6
299____77608______emoga
519____1393_______geecee

Rank__Credits____Team
1_____37855553___TeAm AnandTech
2_____27616789___Antarctic Crunchers
3_____26883366___Czech National Team
4_____25663049___SETI.Germany

I think @Fardringle's stats are better than mine.

Oh, and his stats reports are too.

 

[Markfw]

Moving up... to 5th. I added a 3950x and 2080TI today

 

[StefanR5R]

my 5900X is dropping to 0.52Ghz every 10s. I think my CM AIO LC is shot. Or BOINC now temp-throttles the CPU when it hits 80C??? I couldn't find a setting.

BOINC does not monitor hardware sensors.

Your 5950x is faster than my lowly integrated GPU. lol
my vega 11(2400g) is running 48 mins per wu. heh
I really didnt think it was that comparable.

To be fair, this challenge is better for CPUs than GPUs, but running both just gets "bonus" credits.

There was an application update in December 2020 which extended the useful range of the CPU transforms, and at the same time improved their performance dramatically.

In other news, SETI.Germany finished 2nd last year on this same race with GFN-17 low only and their recent output has consistently been the 2nd highest behind us. Will they have enough to surpass the Artic Crunchers and the Czech Team?

One of their main producers is having an as yet unexplained outage since late last night though, which puts a big dent in SG's production now.

 

[StefanR5R]

Tasks for all remaining GFN-17-Low workunits have been sent to hosts by now, and the counter on the home page shows 0 available tasks. But as there are still 38k GFN-17-Low tasks in progress right now, there will certainly still be a few resends making their rounds before this subproject is completed for good.

 

[Skillz]

Yes I'm seeing this now. We were talking about boss' sudden increase in production, too. Scott Brown was worried Boss was gonna challenge him for 3rd place.

 

[lane42]

I seem to have a lot of Pendings, i think .
One has over 400 pendings, one has over 200,
Is that a lot ?
Nevermind. i see some guy's with over 600 pendings

 

[Skillz]

Your pendings count towards the challenge. So don't worry about having a lot of them.

The only draw back is if they end up being invalid then they'll deduct those points from the challenge score.

 

[Fardringle]

Day 2 stats:

Rank	Team Name	18-Jan	19-Jan	20-Jan
1	TeAm AnandTech	19,906,097	50,041,357
2	Antarctic Crunchers	15,597,906	36,332,052
3	Czech National Team	15,817,817	34,076,607
4	SETI.Germany	12,653,791	31,455,798
5	Aggie The Pew	14,142,939	31,050,134
6	AMD Users	7,404,489	27,494,898
7	[H]ard|OCP	5,838,842	13,139,706
8	Storm	5,778,266	11,639,359
9	Crunching@EVGA	4,120,441	9,080,557
10	Metal Archives	4,633,664	8,718,368

Rank	Name	18-Jan	19-Jan	20-Jan
1	Skillz	6,659,008	19,768,985
5	crashtech	4,653,904	9,802,688
16	Icecold	1,888,951	4,229,104
22	biodoc	1,347,922	3,459,026
24	markfw	1,267,498	3,344,948
37	xii5ku	1,148,986	2,146,999
42	Orange Kid	799,279	1,853,776
44	parsnip soup in a clay bowl	643,527	1,803,592
70	Fardringle	521,958	1,098,832
96	Lane42	245,740	725,542
98	VirtualLarry	146,521	704,654
106	Skivelitis2	382,891	620,539
198	waffleironhead	64,508	263,293
262	Ken_g6	65,610	146,675
320	emoga	77,608	77,608
478	geecee	1,393	4,204

 

[crashtech]

I have over 2,000 pendings

 

[Markfw]

I have over 2,000 pendings

I have 1420 pending, and I have a low power ppd than you !

 

[Skillz]

Mine says I have too many to display subtotals.

edit
Went to the last page, apparently I have 3535 pendings.

 

[Ken g6]

I collected Day 2.5 stats, but forgot to post them.

Rank___Credits____Username
1______36839369___Skillz
5______13072657___crashtech
16_____5800178____Icecold
20_____5034785____markfw
21_____4889810____biodoc
37_____2845822____xii5ku
39_____2664027____parsnip soup in a clay bowl
41_____2583625____Orange Kid
71_____1501714____Fardringle
93_____1090014____Lane42
103____922873_____VirtualLarry
138____620539_____Skivelitis2
185____414867_____waffleironhead
257____204234_____Ken_g6
286____162163_____kiska
340____77608______emoga
484____6602_______geecee

Rank__Credits____Team
1_____78719862___TeAm AnandTech
2_____51660465___Antarctic Crunchers
3_____46966947___Czech National Team
4_____43704261___Aggie The Pew

 

[Fardringle]

I'll wait to post the "Day 3" stats when the challenge timer ends in about 2.5 hours so that they will be the final stats as well as the Day 3 stats.

 

[Kiska]

257____204234_____Ken_g6
286____162163_____kiska

I am coming up on you ken!

 

[StefanR5R]

I noticed something weird:

About 100 results of mine (or more) were marked "validation inconclusive". A spot check on several of these revealed that both tasks of the respective workunit were sent to hosts of mine --- i.e. not to one host of mine and one host of another user.

After the inconclusive validation, the server created a third task of course, and sent that to a third host. I spotted one case in which the third host already returned its result, and the outcome then was that all three results became valid. Furthermore, I don't have any invalid results so far.

My interpretation:
PrimeGrid naturally wants the quorum of 2 be fulfilled by different users, not just different hosts. However, the scheduler may still assign tasks of one and the same workunit to hosts of one and the same user. In this case, when results are returned, the validator marks them inconclusive, and re-validates them as soon as another result from a different user comes in.

 

[StefanR5R]

Also: Hello @crashtech! Good to see you. :-)

 

[crashtech]

@StefanR5R , been expecting you, I'll just step aside now!

 

[StefanR5R]

No worries. I throttled my GPUs too much this time. :-D And didn't look for optimum CPU settings.

 

[Fardringle]

Day 3 (final) stats:

Rank	Team Name	18-Jan	19-Jan	20-Jan
1	TeAm AnandTech	19,906,097	50,041,357	106,645,769
2	Antarctic Crunchers	15,597,906	36,332,052	63,016,032
3	Czech National Team	15,817,817	34,076,607	56,182,448
4	SETI.Germany	12,653,791	31,455,798	54,621,092
5	Aggie The Pew	14,142,939	31,050,134	53,097,771
6	AMD Users	7,404,489	27,494,898	40,149,277
7	[H]ard|OCP	5,838,842	13,139,706	21,845,012
8	Storm	5,778,266	11,639,359	16,343,961
9	Crunching@EVGA	4,120,441	9,080,557	15,533,804
10	BOINC@MIXI	--	--	11,874,934

(BOINC@MIXI wasn't anywhere near the top 10 until today, so I don't have numbers for the other days for them.)

Rank	Name	18-Jan	19-Jan	20-Jan
1	Skillz	6,659,008	19,768,985	44,750,168
5	crashtech	4,653,904	9,802,688	15,230,900
6	xii5ku	1,148,986	2,146,999	14,915,459
17	Icecold	1,888,951	4,229,104	6,982,291
21	markfw	1,267,498	3,344,948	6,191,043
22	biodoc	1,347,922	3,459,026	5,845,255
38	parsnip soup in a clay bowl	643,527	1,803,592	3,334,397
41	Orange Kid	799,279	1,853,776	3,055,037
76	Fardringle	521,958	1,098,832	1,654,262
93	Lane42	245,740	725,542	1,336,618
102	VirtualLarry	146,521	704,654	1,080,256
130	kiska	--	--	788,627
160	Skivelitis2	382,891	620,539	620,539
175	waffleironhead	64,508	263,293	538,880
259	Ken_g6	65,610	146,675	245,024
362	emoga	77,608	77,608	77,609
484	geecee	1,393	4,204	7,562

 

[Orange Kid]

Thanks all for the stats.

 

[Skillz]

You are correct @StefanR5R PG has it setup so a different user and different computer need to validate a task. I guess some slip by when you download a few thousand tasks at once though.

 

[StefanR5R]

*Maybe* they even want to have the results from computers of users of different teams? -- If yes, that could be one of the possible explanations why they have implemented it in the validator, but not in the scheduler.

(A host cannot change ownership during the lifetime of a task. But a user can change team affiliation during the lifetime of a task.)

(Another potential explanation could be that it may be 'cheaper' to implement in the validator than in the scheduler. Cheaper for them of course, but a little more expensive for the user community.)

 

[Skillz]

I was talking to them about that in discord yesterday. They only block the same user and same host from validating the same result. Team affiliation doesn't matter.

 

[Skillz]

@StefanR5R can you PM me a few of those tasks? The admins at PG wanna look at it.