Info PrimeGrid Challenges 2025

[Skillz]

 

[Markfw]

What is this page ? Should I do something where is says "I am a member of this code " ?

 

[emoga]

What is this page ? Should I do something where is says "I am a member of this code " ?

View attachment 122360

No, there’s nothing you need to do right now; just have a bit of patience while the verification runs its course.

 

[Ken g6]

More-or-less final stats:

Rank___Credits____Username
2______84588651___Icecold
3______78986466___markfw (+1 prime!)
6______38336606___crashtech
9______26980413___ChelseaOilman
10_____25888251___cellarnoise2
20_____11187410___10esseeTony
24_____9156655____[TA]Skillz
31_____6237245____Orange Kid
38_____5102794____w a h
82_____2428061____waffleironhead
88_____2178044____mmonnin
94_____2064798____johnnevermind
107____1707914____Ken_g6
290____104261_____IEC

Rank__Credits____Team
1_____294947576___TeAm AnandTech
2_____112350011___Czech National Team
3_____94378765___Antarctic Crunchers
4_____71583258___[H]ard|OCP

 

[Ken g6]

@Markfw here's your official announcement. Congrats!

 

[Markfw]

@Markfw here's your official announcement. Congrats!

Thanks !!!! The only problem was the computer has 192 gig ram, not 196. Other than that totally on track !!!! And obviously the biggest prime by TeamAnandtech ???

 

[emoga]

Congrats on getting the sacred badge 😏

 

[TennesseeTony]

World Record Generalized Cullen Prime!
On 16 April 2025, 11:37:45 UTC, PrimeGrid's Generalized Cullen/Woodall PrimeSearch found the largest known Generalized Cullen Prime:

4052186 * 69 4052186 +1

Generalized Cullen numbers are of the form: n*bn+1. Generalized Cullen numbers that are prime are called Generalized Cullen primes. For more information, please see “Cullen prime” in The Prime Glossary.

The prime is 7,451,366 digits long and enters “The Largest Known Primes Database” ranked 1st for Generalized Cullen Primes and 16th overall. This is the second largest prime ever found by PrimeGrid.

Base 69 was one of 9 primeless Generalized Cullen bases for b ≤121 that PrimeGrid is searching. The remaining bases are 13, 29, 47, 49, 55, 101, 109 & 121.

The discovery was made by Mark Williams of the United States using 8 cores of an AMD EPYC 9554 64-Core Processor with 196GB RAM, running Microsoft Windows 10 Professional x64 Edition. This computer took about 10 hours, 15 minutes to complete the probable prime (PRP) test using PRST. Mark is a member of TeAm AnandTech.

The PRP was confirmed prime on 17 April 2025 by an AMD Ryzen 9 7950X3D @ 4.20GHz with 128GB RAM, running Debian 12.5. This computer, using 8 cores, took 12 hours and 32 minutes to complete the primality test using PRST.

For more details, please see the official announcement.

 

[TennesseeTony]

And obviously the biggest prime by TeamAnandtech ???

Your dedication has paid off!

 

[Markfw]

Congrats on getting the sacred View attachment 122747 badge 😏

 

[igor_kavinski]

Congratulations, Markfw!

Proof that the Universe cares

 

[Markfw]

Its now also on the BOINC manager as the first message !

 

[biodoc]

Congrats on the massive prime @Markfw ! I was catching up on this thread and saw the news.

 

[igor_kavinski]

Markfw is VERY lucky. Right after his prime discovery, I saw news of this: https://www.scirp.org/pdf/apm2024145_65302433.pdf

I find it VERY interesting that the root primes for the prime periodic table are 2,3,5 and 7. Add them up and it's 17 which adds up to 8. I consider 8 to be the number of absolute chaos and it is out of absolute chaos that the Universe was born (anyone who thinks that the Big Bang wasn't chaotic, has a lot of explaining to do ).

If this study gets validated, Markfw's prime may be the last ever large prime "discovered" through DC. This new periodic table opens the possibilities of generating new primes with almost 100 percent certainty so it will no longer be like looking for a needle in the haystack of infinity.

 

[Ken g6]

That prime periodic table thing is from a year ago. I guess nothing came of it. I think they just rediscovered wheel factorization, but I'm not sure.

 

[TennesseeTony]

Just taking some notes for the next 2 Challenges.

GFN 20 GPU 4090@450watts- 26m20s
GFN 20 CPU AMD 7945HX, 16 threads- 298minutes

AP27 GPU 4090@450watts- 1m27s
AP27 CPU AMD 7945HX, 16 threads- 51m
AP27 CPU AMD 7945HX, 8 threads- 99m
AP27 CPU AMD 7950X, 4 threads- 184m @165 watts
AP27 CPU AMD 7945HX, 4 threads- 200m
AP27 CPU AMD 7945HX, 2 threads- 405m



Neither GFN 20 or AP27 use much system RAM.

GFN 20 uploads are fairly large, 32MB.

 

[StefanR5R]

On CPUs, Genefer-20's data allocation is 20.3 MBytes. AMD Zen 3...5 CPUs (except for some APUs) should give best throughput with one task running on each CCX, such that datasharing across CCXs is avoided.

Here is one of several sources for simple scripts which pin Genefer tasks to those CPUs which belong to the same CCX:
xii5ku_AT in SETI.Germany's forum on December 17, 2023
Save the Windows script with a name like "gfn_affinity.ps1" (*.ps1 is for PowerShell scripts), edit the $cores=... line according to your CPU and according to whether or not the SMT threads shall be used, and then let it execute in parallel with BOINC.

 

[Ken g6]

Nobody bumped this yet? Challenge #4 for 2025 is underway!

 

[Markfw]

Nobody bumped this yet? Challenge #4 for 2025 is underway!

I would, but until my first load of CPU tasks dump (~4 hours) I have nothing to really say. As a team we are killing it though.

 

[StefanR5R]

Single RTX 4090 @ 360W, air cooled, driven by a i7-7700K:
..... avg. 1,530 s/task ...... 55.5 tasks/d ...... 2,000 kPPD ..... 420 W ..... 4.7 kPPD/W

Dual RTX 4090 @ 320 W, water cooled, driven by a i7-7700K:
..... avg. 1,570 s/task ... 2×55.0 tasks/d ... 2×1,950 kPPD ... 760 W ..... 5.1 kPPD/W

Epyc 9554P @ 400 W:
... avg. 20,300 s/task ... 8×4.26 tasks/d ...... 1,200 kPPD .... 500 W ..... 3.0 kPPD/W

Edit: I haven't checked yet what happens to processor clock speed if the air cooled GPU is limited to 320 W also. I would expect genefer22g to easily use all the Watts it can get, this far below the default board power limit. Those 9% difference in performance-per-Watt are probably coming from a combination of the lower GPU temperature (< 55 °C versus < 65 °C), better chip selection of the watercooled GPUs, and from having two GPUs hosted by a single computer rather than just one GPU per computer.

 

[TennesseeTony]

deleted. My refresh didn't refresh.

 

[Markfw]

Well, this is not the official day one, but its close. Top 5 teams. And seti germany came from I don't know where. I did not see them anywhere close to the top one hour ago.They could even overtake us. (I could be wrong on that)

 

[Ken g6]

Well, this is not the official day one, but its close.

So's this. Yours was early (I think?) but mine is late.

Day 1 stats:

Rank___Credits____Username
3______23859558___markfw
5______17824734___crashtech
19_____5978131____Icecold
20_____5934781____ChelseaOilman
21_____5773545____[TA]Skillz
24_____5066586___10esseeTony
28_____4743336____Orange Kid
30_____4539229____mmonnin
34_____4000348____biodoc
53_____2336924____cellarnoise2
74_____1537012____w a h
91_____1068572____Ken_g6
129____542704_____johnnevermind

Rank__Credits____Team
1_____83205465___TeAm AnandTech
2_____47436308___SETI.Germany
3_____46615719___AMD Users
4_____44386097___Antarctic Crunchers

 

[Markfw]

So's this. Yours was early (I think?) but mine is late.

The day ended at 1:40 PDT and I posted at 2:08 PDT, so I was late.

as a team we are absolutely kicking hiney.

I am 3rd, so that's just honorable mention.