Please note that our servers will be unavailable for a while on Tuesday 5th of May until Wednesday 6th due to a change of storage backend.

Hence uploads may fail and other BOINC client requests time out. Many thanks for your understanding and happy crunching!

Thanks Nils,
Always nice to have time to prepare and with CMS I can have enough to run and last until wednesday ( but I wish "no limit" actually worked that way)

You do realise Boinc auto gets another project, right? Never need to prepare.

, You must have been really bored to come up with that goofy comment.

yawn...what a show...

Greetings,

Noticed that a forthcoming shutdown of the Collider is happening soon, any idea on how it will affect the BOINC side of things.

Kind Regards

There will be plenty of simulations for upgrades of the LHC and the experiments for HL-LHC also when the accelerator is stopped.

(BOINC was never used to distribute analysis data from the LHC, the data sets are too big to be suitable for volunteer computing.)

Now sadly our downtime will probably drag on until Thursday, as the copy of the data is slower than expected. Sorry for that.

I have stuff ready to upload, but I got two massive tasks from you guys. 1 pythia at nearly 48hrs non stop and only 75% done with integrating and a Herwig integrate at 36hrs and counting with no init yet. So I'm not hurting.

(BOINC was never used to distribute analysis data from the LHC, the data sets are too big to be suitable for volunteer computing.)

How big? I can buy more disk. I love big downloads!

In reply to Mr P Hucker's message of 7 May 2026:

(BOINC was never used to distribute analysis data from the LHC, the data sets are too big to be suitable for volunteer computing.)

How big? I can buy more disk. I love big downloads!

Maybe you'd have to buy more RAM too which is much more expensive these days.

---

In reply to Nils's message of 6 May 2026:

There will be plenty of simulations for upgrades of the LHC and the experiments for HL-LHC also when the accelerator is stopped.

(BOINC was never used to distribute analysis data from the LHC, the data sets are too big to be suitable for volunteer computing.)

Now sadly our downtime will probably drag on until Thursday, as the copy of the data is slower than expected. Sorry for that.

Hi Nils

Thankyou for your answer.

Kind Regards

In reply to Garrulus glandarius's message of 7 May 2026:

Maybe you'd have to buy more RAM too which is much more expensive these days.

I noticed! AI tells me AI stole it all. But I have a 128GB computer, plus two 64GB and one 48GB.

It's not about physical size, its about the data.

This is how gemini interprets the parameters of the task: Since you are running in Minimum Bias (mb-inelastic) mode, Pythia is tasked with simulating 7,000 individual proton-proton collisions. Once those 7,000 events are generated and processed through the internal physics hooks, the task will reach true completion.

So one of the tasks is this: (Pythia): Is aiming for 7,000 events and has just finished its 2-day "warm-up" to start Event 1. It took 2 days on a Ryzen 3700x to work that. Now its starting the next phase of creating events. This has 420 events but its a lot of math.

The other is a Herwig type task and that also took 2 days to build the database and it is now working on that data. It needs to work through 42,000 events in this task.

Further analysis says its about a 3-3.5 day run. 2 days to build up the data and 1.5 to analyze it. Thats the Herwig. The Pythia, that's about a week. 2 days data building and about 4 days running it.

So if you can commit a core to a week of 24/7 running maybe you will get one of these.
For me that is in the 2922 group. But not everything from that group is large. There were plenty of normal ones.

Memory is not an issue. I have 48 gigs which I needed to run Rosetta Python tasks back in the old days along with all the other projects I run.

In reply to greg_be's message of 7 May 2026:

So if you can commit a core to a week of 24/7 running maybe you will get one of these.

Maybe I'm not the only one who also runs CPDN when available. Those tasks take 2-3 days on top of the line rigs and about a week on "regular" machines. Many years ago (well before I joined) they had tasks that would run several weeks, up to 2 months. So yeah, running a task for a week shouldn't be an issue, especially if we'd have a decent runtime estimate (unlike for current Theory tasks).

---

In reply to Garrulus glandarius's message of 7 May 2026:

Maybe I'm not the only one who also runs CPDN when available.

Has there been a blue moon? I thought that project had closed, I haven't seen a task in years.

So one of the tasks is this: (Pythia): Is aiming for 7,000 events and has just finished its 2-day "warm-up" to start Event 1. It took 2 days on a Ryzen 3700x to work that. Now its starting the next phase of creating events. This has 420 events but its a lot of math.

The other is a Herwig type task and that also took 2 days to build the database and it is now working on that data. It needs to work through 42,000 events in this task.

Further analysis says its about a 3-3.5 day run. 2 days to build up the data and 1.5 to analyze it. Thats the Herwig. The Pythia, that's about a week. 2 days data building and about 4 days running it.

Great analysis and summary! I have also seen powheg-box Theory tasks can take a long time to run.

If the BOINC agent can detect it, it would be helpful eventually to see something appear once in the Notice tab indicating that the "LHC@home task nnnnnn may take longer than others to complete".

Not all users will have the knowledge and the time to open the Work unit and find the meta data field nor use a tail command.

In reply to Mr P Hucker's message of 7 May 2026:

In reply to Garrulus glandarius's message of 7 May 2026:

Maybe I'm not the only one who also runs CPDN when available.

Has there been a blue moon? I thought that project had closed, I haven't seen a task in years.

There have been some software improvements and a fresh batch is just about to be finished (well, except for all the tasks that are already sent out to slow PCs). More work is expected so I recommend reading the latest discussions and checking in from time to time.

---

If any of you get this:
Theory_2922-4865899-922 I wish you good luck.
It's a 7,000 event Diffractive physics (according to bot that I posted a lot of data from runrivet to) and its stuck in a infinite calculation on event 6 (2 days processing) in what it calls: "Spin-Lock" or a "Numerical Divergence." Maybe that changes on a new system? I don't know. But i'm tossing it.

Back to the original topic..does anyone know when the servers will be back?

In reply to Nils's message of 6 May 2026:

There will be plenty of simulations for upgrades of the LHC and the experiments for HL-LHC also when the accelerator is stopped.

(BOINC was never used to distribute analysis data from the LHC, the data sets are too big to be suitable for volunteer computing.)

Now sadly our downtime will probably drag on until Thursday, as the copy of the data is slower than expected. Sorry for that.

^ Just wait for this to be finished ;)

---

---------- 24/7 Crunching since 2011 -----------
DSKAG Austria: http://www.dskag.at