Project Update

[Dark Angel]

PrimeGrid: SGS may end sooner than expected
With the server upgrade accomplished, we were able to convert SGS to fast double checking. This, by itself, essentially doubles the speed of SGS, and when combined with increased participation this has SGS on a pace to exhaust its sieve file (i.e., the nominal end of the project) before the planned shutdown date of December 15th. The new plan is therefore to let SGS run to the end of the sieve file, regardless of when that occurs.

The forum discussion thread about the SGS shutdown is here .

[Dingo]

[PrimeGrid] PPS Sieve ends Feb-7(ish); AP27 to continue
PPS Sieve will be suspended around February 7th. The exact date may vary a little. AP27, however, will be continuing. We've decided to continue searching for an AP28. More information can be found here.

View article · Wed, 10 Jan 2024 16:18:24 +0000

[Dingo]

[PrimeGrid] Hilbert's Birthday Challenge starts 23 January
The first challenge of the 2024 Series will be a 3-day challenge celebrating David Hilbert's birthday! The challenge will be offered on the PPSE-LLR application, running from January 23 01:00 UTC to January 26 01:00 UTC. To participate in the Challenge, please select only the Proth Prime Search Extended LLR (PPSE) project in your PrimeGrid preferences section. Join the discussion at https://www.primegrid.com/forum_thread.php?id=10425.

View article · Mon, 15 Jan 2024 12:25:35 +0000

[Dingo]

[PrimeGrid] Unplanned outage

This afternoon PrimeGrid experienced an unplanned outage. For reasons unknown, the database locked up and started consuming 100% CPU on all 32 cores, while doing nothing at all. We proceeded cautiously, not wanting to risk damaging the database, and slowly shut down the system, checked the database, and brought everything back up. Everything is back to normal, and the servers are operating once again.

View article · Sat, 27 Jan 2024 00:49:20 +0000

[Dingo]

[PrimeGrid] Tour de Primes 2024
February is almost upon us, and with it, PrimeGrid's annual Tour de Primes. This is a month long event where all can earn special Tour de Primes badges and a lucky few will win the coveted Tour de Primes jerseys. More information can be found here.

View article · Wed, 31 Jan 2024 17:11:00 +0000

[ryzenmulti]

I have set up 4x Titan V GPU's on 2 different computers. The sole GPU task is genefer_extreme (do I feel lucky) and each card completes 1 in almost exactly 24 hrs. So 4x 960K credits every day (3.8M creds) just from these GPU's plus whatever the CPU's get done.

If credits won badges, what do we get? Apart from that, sweet f/a?  According to the developer of genefer22, it is not yet optimised for FP64 cards so RTX 4080's outperform the Titan V's - but they only cost me $700 and are only dual slot, not a triple slot hog. So I am winning regardless.

[Dingo]

[PrimeGrid] Policy regarding banning computers
Effective immediately, PrimeGrid reserves the right to ban computers that exhibit problematic behavior. Please see this forum thread for more information.

View article · Thu, 22 Feb 2024 17:04:48 +0000

[Dark Angel]

They're getting miffed at people downloading 30-40 thousand tasks at a time, particularly when they don't get completed in time or are aborted after the FB sprints are finished.

[tazzduke]

It was a rogue user, who amassed over 60k tasks, who did not know what they were doing, it also highlighted the shortcomings of the BOINC system both on user side and server side.

They notified with a general message and somehow people got there knickers in a twist.

This here quote from another user on the PG forums sums it up.

Ok, so I'm a simple man (unless you ask my wife, then all bets of off lol) but to understand you in general terms here: this was a single incident which you handled. Then gave us all the heads up if/when this happens again you reserve the right to take appropriate action. That for systems/users than can process the much larger number of tasks that's fine...so long as you actually DO the work you take. I mean I get whey questions are being asked. Nobody wanna break the rules and/or cause problems. Me, I just prefer the feeling you have conveyed in every response in this thread. Unless, of course I've gotten it wrong.

Regards

[Dark Angel]

That's pretty much it.  They're not worried about people that grab 2/3/4/5,000 units and are actually returning them, it's the extreme case that's the problem.

[ryzenmulti]

Some stats going forwards for those looking to set records -
GFN 16 returned a prime on average every 14801 wu's; ave time per wu roughly 6 mins (GPU and CPU)
GFN 17mega returned a prime on average every 28032 wu's; ave time per wu roughly 8 mins (GPU and CPU)
GFN18 returned a prime on average every 57815 wu's; ave time per wu roughly 10 mins (GPU and CPU)

In the last week I have done 19600 GFN18's with 100% focus and nada/nil/zero/zilch. Not only do they take longer, but are returned fewer so it is a non-linear double wammy. When compared to GFN16'6, GFN18's will take forever.

[Dingo]

[PrimeGrid] JUICE's Birthday Challenge
From April 14th 12:14:00 UTC to April 21th 12:14:00 PrimeGrid will be running a 7 day challenge on the Riesel Problem project (TRP). Note the unusual start and end times! For more information, please see this forum thread.

View article · Wed, 10 Apr 2024 15:22:22 +0000