PrimeGrid Challenges 2022

Page 15 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Ken g6

Programming Moderator, Elite Member
Moderator
Dec 11, 1999
16,472
4,323
75
Day 1 stats:

Rank___Credits____Username
1______97032000___[TA]Skillz
4______61800000___crashtech
11_____27636000___Pokey
12_____21012000___markfw
24_____13740000___mmonnin
33_____8676000____biodoc
38_____7680000____xii5ku
42_____6780000____Orange Kid
48_____5820000____Fardringle
52_____5460000____w a h
66_____4584000____cellarnoise2
82_____3408000____parsnip soup in a clay bowl
91_____2952000____kiska
112____2028000____Skivelitis2
172____816000_____Lane42
220____348000_____waffleironhead
244____216000_____Ken_g6
284____48000______geecee

Rank__Credits____Team
1_____270036000___TeAm AnandTech
2_____138768000___Antarctic Crunchers
3_____100284000___Czech National Team
4_____91632000___BOINC@AUSTRALIA

Wow, quite a turnout! My slow card can't keep up, especially when it's too hot to run it.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
26,678
15,737
136
I had to turn off a bunch last week, as my brother from Hawaii was staying with me for my moms funeral. But I fired them up sat, just 4 of them a little late.
 
Reactions: crashtech

Ken g6

Programming Moderator, Elite Member
Moderator
Dec 11, 1999
16,472
4,323
75
Day 2 stats:

Rank___Credits____Username
1______205320000___[TA]Skillz
6______125532000___crashtech
10_____60396000___Pokey
12_____45540000___markfw
25_____27828000___mmonnin
31_____20868000___biodoc
40_____15348000___xii5ku
41_____14004000___Orange Kid
48_____11316000___Fardringle
57_____9696000____w a h
65_____9312000____cellarnoise2
84_____6888000____parsnip soup in a clay bowl
98_____5196000____kiska
151____2376000____Skivelitis2
184____1500000____Lane42
231____720000_____waffleironhead
255____396000_____Ken_g6
291____120000_____geecee

Rank__Credits____Team
1_____562356000___TeAm AnandTech
2_____308820000___Antarctic Crunchers
3_____222528000___AMD Users
4_____215388000___Czech National Team

We're first and I'm not quite last.
 

Ken g6

Programming Moderator, Elite Member
Moderator
Dec 11, 1999
16,472
4,323
75
More or less final stats:

Rank___Credits____Username
1______342264000___[TA]Skillz
6______179208000___crashtech
10_____94200000___Pokey
13_____70068000___markfw
25_____41976000___mmonnin
30_____34044000___biodoc
41_____22992000___xii5ku
42_____21228000___Orange Kid
51_____17160000___Fardringle
58_____14688000___cellarnoise2
85_____10368000___parsnip soup in a clay bowl
88_____9696000____w a h
105____7308000____kiska
189____2376000____Skivelitis2
191____2184000____Lane42
245____816000_____waffleironhead
265____516000_____Ken_g6
301____180000_____geecee
324____108000_____SlangNRox

Rank__Credits____Team
1_____871380000___TeAm AnandTech
2_____507456000___Antarctic Crunchers
3_____405084000___AMD Users
4_____359652000___BOINC@AUSTRALIA

This challenge may be done, but save some space for the World Space Week Challenge starting two weeks from today!
 

StefanR5R

Elite Member
Dec 10, 2016
6,275
9,591
136
41_____22992000___xii5ku
42_____21228000___Orange Kid
I used 9 (nine) GPUs for that, all Pascal — neglecting a few points which came from short-lived experiments with cpu_WW_mt with obvious outcome. (CPU's aren't worthwhile in WW.) I guess @Orange Kid got to the point to almost overtaking me with fewer but more sensibly chosen GPUs…
 

Orange Kid

Elite Member
Oct 9, 1999
4,403
2,191
146
I used everything I had. 3070ti, 3070, 2070, 1070, 970, 5950x (24 cores), 3950x (24 cores), 5900 (18 cores) and a 2700x (12 cores). I don't use my laptops anymore on DC.
I guess I should have use all the cores and maybe....😀😀😀
 
Reactions: StefanR5R

crashtech

Lifer
Jan 4, 2013
10,630
2,200
146
Pascal is seeming a bit long in the tooth at this point, though I still run a 1080ti and 2 1070tis. The 1070ti looks bad even compared to a lowly 2060...
 
Reactions: Ken g6

StefanR5R

Elite Member
Dec 10, 2016
6,275
9,591
136
Tue 10/4 - 10/11 PrimeGrid World Space Week Challenge: TRP-LLR (CPUs only)
starts and ends at 12:00 UTC (08:00 EDT, 05:00 PDT)
From the "edit PrimeGrid preferences" webpage:
Recent average CPU time: 31:08:37
This project has a 10% long job credit bonus and a 10% conjecture credit bonus.
Re bonuses: ESP-LLR, PSP-LLR, SoB-LLR, SR5-LLR, TRP-LLR, and the currently inactive TRP-Sieve are considered conjecture projects. Specifically, the goal of TRP-LLR is to find out whether or not 509,203 is the smallest Riesel number. And the means to this end is to look for primes within a certain search space: If a prime of the form of k*2^n-1 is found, then k is not a Riesel number. There is just a few of numbers smaller than 509,203 left, but I suppose the search space which is based off of these few numbers is pretty big in terms of computational effort. After all, PrimeGrid already awards a little "long job" credit bonus here. And in the comparably unlikely event that a prime is found via TRP-LLR, it will make it to rank 52 of the current Top 5000 Primes list, according to PrimeGrid's home page.

Like all of the LLR subprojects except SGS-LLR (AFAICS), TRP-LLR makes use of the LLR2 application which splits the work into a big "main task" for the actual testing of a prime candidate, and a small "verification task" as validation of the main task's result.

I checked the current active search range of TRP on a Haswell CPU, using the current Linux application binary sllr2_1.3.0_linux64_220821, and got FMA3 FFT lengths between 1120K and 1344K. This means 8.75…10.5 MBytes FFT data sizes. Hence, typical older 4-core desktop CPUs don't have enough processor cache to fit all the "hot data", but e.g. a Zen 2 CCX with 16 MBytes level 3 cache does.

Corresponding to the varied computational payload in TRP-LLR's active search space, credit for main tasks currently varies between ≈7,800 and ≈9,300 according to what I am seeing in random computer's results tables.
 

Ken g6

Programming Moderator, Elite Member
Moderator
Dec 11, 1999
16,472
4,323
75
So, the challenge started about an hour ago. I slept late, but my computers didn't.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
26,678
15,737
136
I have a 12700F, a 5950x and a 7950x. Surprisingly, the 12700F with only on 8c task runs 2:15 a task, while the 5950x is 3:15 and the slowest is the 7950x at 3:40 (memory problems). The the 5950x and 7950x are running 2 tasks at once. Also, the 7950x is running windows, thats a disadvantage.
 
Last edited:

crashtech

Lifer
Jan 4, 2013
10,630
2,200
146
Hmm, I got a late start. Also was unable to test TRP in Windows, so that's just a crapshoot. The Linux boxes should fare better, hopefully.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
26,678
15,737
136
I added another 5950x, this one runs linux. So I have a total of 9 8c tasks running, but one is a screamer (the 12700F)
 
Last edited:

StefanR5R

Elite Member
Dec 10, 2016
6,275
9,591
136
Surprisingly, the 12700F with only on 8c task runs 2:15 a task, while the 5950x is 3:15 and the slowest is the 7950x at 3:40 (memory problems).
The 12700F has got a single unified last-level cache. In contrast, the 5950x and 7950x have got two CCXs = two last-level caches. Program threads which are not running on the same CCX can synchronize only through main memory.

If nothing is done to tie all program threads of an LLR2 instance to cores of the same CCX, these dual-CCX Ryzen models take a large performance hit.

(Not very well comparable to these Ryzens, but: On a Zen2 32-core (8-CCX) EPYC, task durations are increased by almost 50 % and task energy is increased by ≥50 % if cross-CCX scheduling is not prevented.)

So, what to do about it? — With EPYC BIOSes, that's pretty easy. I suppose Ryzen BIOSes don't have the necessary BIOS option. In that case, there are two alternatives:
  • Start as many BOINC client instances as there are CCXs, and tie each client instance to cores of a dedicated CCX. (The LLR tasks which a client instance launches subsequently will automatically be tied to these cores as well.)
  • Monitor LLR tasks as they are starting and ending. Tie a newly started task to available cores of a suitable CCX.
Implementations of either alternative have been posted in some more or less arcane corners of the internet. On dual-socket EPYCs, I either use the BIOS option or the "multiple client" alternative. On dual-socket computers, the "monitoring" alternative would have the drawback that a task may have started on one NUMA node and already allocated memory there, then be tied to cores of the other NUMA node by the monitor, after which it would access far memory.

Edit:
Shoehorning llrTRP tasks into specific cores such that they don't cross CCX boundaries has another potential benefit on Zen3 and Zen4: Instead of running 2 concurrent tasks on a dual-CCX Zen3/Zen4 Ryzen, one can run 4 concurrent tasks ( = 2 per CCX). Zen3 and Zen4 have enough cache per CCX to support that.¹ Task durations will be longer then, but less than double. I.e. total throughput will increase.
 
Last edited:

crashtech

Lifer
Jan 4, 2013
10,630
2,200
146
I have briefly looked around for a "taskset for dummies" tutorial while at work today. I think it's been talked about a lot over on TAAT but it doesn't have it's own thread afaik.
 

mmonnin03

Senior member
Nov 7, 2006
264
249
116
Dang I adjusted to EST the wrong direction thinking it would have been 16:00 instead of 8:00, otherwise I would have set PG to 0% with them allowing work and changed PG status from work. But just starting now with all CPUs.
 
Reactions: biodoc

StefanR5R

Elite Member
Dec 10, 2016
6,275
9,591
136
Main power went down at home for a minute, just when I went off to work. I came back briefly to bring everything up again. Let's see if the interrupted tasks finish valid. They probably will; LLR2 seems to be robust against such incidents.
 

Ken g6

Programming Moderator, Elite Member
Moderator
Dec 11, 1999
16,472
4,323
75
Day 1 stats:

Rank___Credits____Username
3______2005484____[TA]Skillz
6______1786641____xii5ku
7______1659692____crashtech
10_____1231133____parsnip soup in a clay bowl
18_____625116_____markfw
19_____621545_____biodoc
23_____590142_____emoga
43_____330714_____Orange Kid
50_____242091_____mmonnin
61_____189824_____cellarnoise2
80_____143957_____waffleironhead
98_____109810_____Ken_g6
144____54021______mnelsonx
164____37523______Lane42
262____373___10esseeTony

Rank__Credits____Team
1_____9628073____TeAm AnandTech
2_____7107653____Antarctic Crunchers
3_____6194506____Czech National Team
4_____5092349____SETI.Germany

@TennesseeTony looks innocent with those 373 points, but that usually indicates someone who's finished several doublechecks. Which means he's probably running several full WUs that haven't completed yet. xii5ku followed the same pattern, and look at him now!
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
26,678
15,737
136
Damn windows up[date ! Can someone give me a link to top teams and top users for this contest ? I lost everything. Damn windows restored tabs from years ago.

Edit: I found them Nevermind.
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |