PrimeGrid Challenges 2020

[Ken g6]

Thanks, I've had too many plates to spin lately and I forgot about the ball.

 

[Ken g6]

OK, quick stats, 6 hours from the end:

Rank___Credits____Username
83_____378621_____VirtualLarry
88_____349174_____Orange Kid
96_____318411_____Icecold-Team Anandtech
136____176366_____Ken_g6
171____121887_____Lane42
203____87826______SlangNRox
240____57335______zzuupp
273____38447______Howdy
297____30150___10esseeTony
321____18953______waffleironhead
336____14234______biodoc
430____2308_______emoga
523____534________geecee

Rank__Credits____Team
13____2069233____UK BOINC Team
14____1984479____BOINC@AUSTRALIA
15____1865415____SETI.USA
16____1593835____TeAm AnandTech
17____1529301____The Scottish Boinc Team
18____1251539____Rechenkraft.net
19____1243574____BOINCstats

As usual, we're competing with the Scots.

 

[VirtualLarry]

Whoa. I'm at the head of the stats list for the TeAm in this challenge? That's a bit different. This time, I had 4x 6C/12T Ryzen 5 CPUs, rather than just three.

That's obviously because all of the "big guns" were away at Formula BOINC camp....

Edit: Looks like IceCold pulled away at the end, to tag the first spot for our team. Good thing, now my world-view isn't shattered anymore. That, and it kept the Scots from pulling ahead of us too.

 

[Ken g6]

Preliminary final stats:

Rank___Credits____Username
79_____433781_____Icecold-Team Anandtech
84_____413010_____VirtualLarry
89_____381742_____Orange Kid
133____196647_____Ken_g6
167____142563_____Lane42
207____91343___10esseeTony
209____90212______SlangNRox
240____62346______zzuupp
274____42622______Howdy
328____18953______waffleironhead
343____14234______biodoc
436____2308_______emoga
538____534________geecee

Rank__Credits____Team
13____2136337____UK BOINC Team
14____2104831____BOINC@AUSTRALIA
15____2054008____SETI.USA
16____1890301____TeAm AnandTech
17____1756200____The Scottish Boinc Team
18____1368557____Rechenkraft.net
19____1358197____BOINCstats

A nice turnout. But we could have done better without that giant Yoyo in the way.

 

[TennesseeTony]

Thanks, I've had too many plates to spin lately and I forgot about the ball.

We really do appreciate you! Don't break any dishes just to do a stats update. Especially on a Holiday weekend.

What is this one anyway, Labor Day? I entered a Labor Day photo contest some years ago, small group, but I won.


If the neighbors did not already know that I am nuttier than a squirrel turd, they discovered it on this day.

 

[Orange Kid]

Thanks for all the stats.

 

[Pokey]

a veritable army of Tonys. God help us all.

 

[Icecold]

Whoa. I'm at the head of the stats list for the TeAm in this challenge? That's a bit different. This time, I had 4x 6C/12T Ryzen 5 CPUs, rather than just three.

That's obviously because all of the "big guns" were away at Formula BOINC camp....

Edit: Looks like IceCold pulled away at the end, to tag the first spot for our team. Good thing, now my world-view isn't shattered anymore. That, and it kept the Scots from pulling ahead of us too.

For what it's worth, your 4 6 core Ryzen CPU's are way more efficient , easier to manage, etc. than what I had to run to pull ahead, so you were 1st place as far as that goes. I have a stack of random old Haswell Dell PC's, etc. using a ton of power(along with a small amount of newer stuff) that I ran to try to keep us ahead of TSBT. Those 4 PC's you ran put out a lot of points!

 

[VirtualLarry]

I have a stack of random old Haswell Dell PC's, etc. using a ton of power(along with a small amount of newer stuff) that I ran to try to keep us ahead of TSBT.

That's amazing! Thank you for your impressive contribution! And pulling ahead of my Ryzens, with older tech, even.

Edit: I am still running 2x Ryzen R5 1600 "AE" (old-school) CPUs, which I intended to upgrade at least one of them to a Zen2 6C/12T 3600 CPU, but AMD (or their retailers???) increased the price from a fairly affordable $155 to $199, which was currently a no-go for me this month, after I dropped $100 on a GTX 1060 6GB from someone here for a friend, and then splurged just today, I think, on that 32" 1440P 144Hz Westinghouse FreeSync monitor. (See Hot Deals thread on it.) It was $240 + $5 ship. Seemed like a decent deal.

 

[Ken g6]

Bump for the next challenge.

In addition, this challenge is using LLR2. LLR2 returns large result files to be quickly checked by another user. For this reason, PrimeGrid asks that we return results as soon as the work is done. A.k.a. "No bunkering!"

LLR2 may also use more RAM than you're used to; you may need large swap files.

 

[zzuupp]

bump for getting even closer

 

[Ken g6]

In case anyone's up this late, the challenge has started.

Anyone may or may not include me. 😴

 

[VirtualLarry]

I just woke up from the worst darn nightmare that I've had in a long time. Did this challenge start at midnight or something? Le sigh.

Will have to do some work to switch hosts over, I don't have BOINC on my GPU miner. (Does this challenge include GPUs?)

Edit: OK, doesn't appear to utilize GPUs, so that makes it easier. Got it going on a Ryzen R5 3600 and 1600. (I'm down one 3600, and the other 1600 PC got sold.) Sorry TeAm about my lack of relative firepower for this one.

 

[Ken g6]

Day 0.5 stats:

Rank___Credits____Username
60_____85657______Orange Kid
80_____55102______emoga
98_____44728______Ken_g6
125____31383______waffleironhead
177____13422______VirtualLarry
186____12820______zzuupp

Rank__Credits____Team
15____396200_____Richard Dawkins Foundation for Reason and
16____275144_____BOINC@AUSTRALIA
17____243114_____TeAm AnandTech
18____240912_____BOINC@Poland
19____210995_____Rechenkraft.net
20____189976_____Team China

I hope some other people wake up to the fact the challenge has started.

 

[Orange Kid]

I used the "allow transfer only between times" to do an automagical start.
First time I tried that and it seems to have worked.
Who says an old dog can't learn new tricks.
I'll add others as they finish up stuff they are running.

 

[motqalden]

I used the "allow transfer only between times" to do an automagical start.
First time I tried that and it seems to have worked.
Who says an old dog can't learn new tricks.
I'll add others as they finish up stuff they are running.

what so you set it up to run PG and then set allow transfer so it doesn't try to download new tasks till after the start? what a great idea!

 

[motqalden]

looking for input on how you guys are running this on zen2. HT on? SMT? how many?

 

[zzuupp]

63 Orange Kid.... 162 175.38 244
75 emoga......... 136 780.27 227
110 Ken_g6........ 084 273.02 170
165 VirtualLarry.. 040 974.62 165
170 waffleironhead. 037 640.28 101
212 zzuupp......... 024 764.00 112
264 Icecold........ 015 021.13 14

16 Rechenkraft.net 730 852.90 1 251
17 BOINC@AUSTRALIA 544 732.63 970
18 BOINC@Poland 505 810.85 1 076
19 TeAm AnandTech 501 628.69 1 033
20 UK BOINC Team 395 514.11 640
21 Team China 389 566.44 700
22 LinusTechTips_Team 318 697.68 1 554

LAZY 23 hour stats

Nighty Nite!

 

[Ken g6]

Day 2-ish stats:

Rank___Credits____Username
69_____366830_____Orange Kid
73_____326201_____emoga
118____182319_____Ken_g6
152____125692_____waffleironhead
154____124277_____VirtualLarry
210____66052______biodoc
228____54043______zzuupp
278____38242______SlangNRox
345____15021______Icecold

Rank__Credits____Team
15____1983520____Richard Dawkins Foundation for Reason and
16____1743232____SETI.USA
17____1298646____TeAm AnandTech
18____1192424____BOINC@AUSTRALIA
19____1186348____BOINC@Poland
20____1122939____LinusTechTips_Team

We're still missing a few people.

\|/ I really hope nobody is bunkering! \|/

 

[StefanR5R]

looking for input on how you guys are running this on zen2. HT on? SMT? how many?

I don't have measurements, but I suspect that

• single-threaded tasks, number of concurrent tasks = number of physical cores, give best throughput-per-Watt on recent AMD CPUs and recent Intel CPUs,
• best throughput, or at least almost best throughput, is still at those configurations which were identified for PPS-DIV a year ago in "PrimeGrid: CPU benchmarks" (posts #71–#87).

Still, due to somewhat larger workunits now, or/and due to the application update, the precise optimum may have shifted.

This guess of mine is based on a quick read elsewhere and on a quick look at some powermeters today in the morning.

Edit,
meanwhile I looked at some reported results of other users, and FMA3 FFT lengths are sometime 384K now (which is the same as the top end of WUs one year ago), sometimes 400K or 480K ( = more cache hungry). The range of active WUs looks more uniform now than it was a year ago.

 

[Ken g6]

Day 3 stats:

Rank___Credits____Username
20_____2166789____xii5ku
69_____569183_____Orange Kid
79_____475235_____emoga
104____360394_____biodoc
123____285080_____Ken_g6
141____236598_____VirtualLarry
159____198229_____waffleironhead
184____154438_____Icecold
216____107204_____SlangNRox
240____82561______zzuupp

Rank__Credits____Team
7_____7852875____AMD Users
8_____7280264____Storm
9_____7046801____[H]ard|OCP
10____4635659____TeAm AnandTech
11____4391748____BOINC@MIXI
12____4065953____Sicituradastra.
13____4037333____Rechenkraft.net


Thanks for the boost, @StefanR5R!

 

[Ken g6]

Day 4 stats:

Rank___Credits____Username
16_____3721323____xii5ku
65_____843446_____biodoc
70_____774001_____Orange Kid
88_____596954_____emoga
105____495137_____Icecold
125____385578_____Ken_g6
141____336025_____VirtualLarry
166____270531_____waffleironhead
202____178810_____SlangNRox
255____109414_____zzuupp

Rank__Credits____Team
7_____11452596___AMD Users
8_____9645050____Storm
9_____9562229____[H]ard|OCP
10____7711191____TeAm AnandTech
11____5930364____BOINC@MIXI
12____5798768____Rechenkraft.net
13____5501986____Sicituradastra.

Top 10 here we...stay. I hope.

 

[zzuupp]

thanks for the stats, kg6

Otherwise, more lazy stats were being posted

 

[Ken g6]

Fairly final stats for this challenge:

Rank___Credits____Username
23_____3721323____xii5ku
64_____1063804____biodoc
65_____1001584____Orange Kid
82_____819676_____Icecold
104____633511_____emoga
127____492466_____Ken_g6
145____422135_____VirtualLarry
167____344187_____waffleironhead
193____245847_____SlangNRox
258____139774_____zzuupp

Rank__Credits____Team
7_____14993653___AMD Users
8_____12257817___[H]ard|OCP
9_____11965960___Storm
10____8884278____TeAm AnandTech
11____7809356____Rechenkraft.net
12____7162814____BOINC@MIXI
13____6948162____Sicituradastra.

Aside from the fact we're solidly in 10th, invalid WUs aren't much of a thing anymore with Gerbicz checking. There have only been 13 in the first 4 days.

 

[VirtualLarry]

Aside from the fact we're solidly in 10th, invalid WUs aren't much of a thing anymore with Gerbicz checking. There have only been 13 in the first 4 days.

Oh how I wish that were true. It was point out to me by @biodoc , that I have generated 51 "error" WUs during this challenge, the vast majority of them being calculated by my Ryzen R5 1600 CPU. Including several AFTER I disabled "OpCache", which is supposed to be a way to work around the Zen1 early-production CPUs that had "flaws" (of which, mine is certainly one of them).

Either that, or my RAM or PSU is failing. This challenge with LLR2 sure pushes Ryzen CPUs hard. That's even AFTER I disabled CPB (core performance boost), so my CPU was sitting at "base clocks", and not even turboing. Even so, it was getting above 80C.