Development for LHC@home

In reply to [AF>Le_Pommier] Jerome_C2005's message of 15 Sep 2025:

I have 4 tasks running on AMD EPYC Milan for more than 26 hours...

So, seems not a "problem" of cpu..
I have some wus that have exceeded 40 hours of calculation.

I think there is something to do on the app

Longest workunits so far ran for 62 hours.

Ended after 60 hours or are still running ?

My longest one was 60h, got 110 tasks finished successfully so far, no errors. I guess the clue is really to just let them run. They will finish eventually. Checkpoints seem to work fine. But is it just me or do they keep getting longer and longer? My first ones went a lot quicker. Maybe they don't like SMT because the change to longer units happened right around when I started to fully load my hosts.

In reply to Drago75's message of 16 Sep 2025:

My longest one was 60h, got 110 tasks finished successfully so far, no errors. I guess the clue is really to just let them run. They will finish eventually. Checkpoints seem to work fine. But is it just me or do they keep getting longer and longer? My first ones went a lot quicker. Maybe they don't like SMT because the change to longer units happened right around when I started to fully load my hosts.

Yes they do seem to be getting longer. I seem to have a higher % tasks going past 1 day run time.

In reply to Drago75's message of 16 Sep 2025:

My longest one was 60h, got 110 tasks finished successfully so far, no errors.

Have you got one of these wus validated?
I've over 15 wus pending validation, some of these since 3 days ago....waiting for wingman :-)

But is it just me or do they keep getting longer and longer? My first ones went a lot quicker. Maybe they don't like SMT because the change to longer units happened right around when I started to fully load my hosts.

+1

Meanwhile I got 30 tasks validated and 90+ pending. You can expect around 100 points per 10,000 seconds calculation time. Many tasks were propably aborted in the process because people got impatienend. So that lengthenes confirmation for others.

In reply to Drago75's message of 16 Sep 2025:

You can expect around 100 points per 10,000 seconds calculation time

So, for a 140k seconds wu can be 1400 points...

You don't crunch here for credit :D

Oh I just saw my first 4 tasks ended during the night : between 149k and 189k seconds. All pending so far.

In reply to [AF>Le_Pommier] Jerome_C2005's message of 16 Sep 2025:

You don't crunch here for credit :D

You know i'm not crunching for credits :-)
But, if wus are validated, we are doing science and we are helping developers to debug/optimize/etc the code...

I have some wus that have exeeded 3 days (72 hrs) of calculation... :-O

From LHC site:

LHC@home: New application Xtrack
As mentioned in this thread , we finally have the new Xtrack BOINC application available for beta testing.

Please refer to the Sixtrack/Xtrack forum for further details and feedback discussion. Tasks should hopefully be available by tomorrow.

The LHC@home team
16/09/2025

Maybe to stop the queue here and collect the results to see if it's all ok before passing in production...

(74 hours reached)

They are validating. I have hundreds validated. Some tasks are like 30 min while others are several days.

Have many Xtrack Tasks with deadline 25/09/19.
Xtrack varying from a few minutes to many days in runtime.

So, no calculation in Boinc, how to find a good runtime.

After three hours of runtime, Boinc show no endtime in Boincmanager anymore.
Help in Slotfolder with stderr.txt is also not possible.

In reply to mmonnin's message of 17 Sep 2025:

They are validating. I have hundreds validated.

Good for you. Still no validated tasks (now are 17 my wus pending validation).
And server status shows only 2 wus pending :-P

In reply to maeax's message of 17 Sep 2025:

Xtrack varying from a few minutes to many days in runtime.

Up to now, no xtrack "few minutes", only "many days".
But today i download a different kind of simulation: before are "camotran_true_scan_end_of_collapse_round_pos", now are "camotran_true_scan_end_of_levelling_B2".
Maybe these simulation are shorter...

I am still running 8 of
camontan__tune_scan_end_of_collapse_round_scan_*__fine_scan_*

I haven't had any finish for several days now and many running over 80 hours and many pending of the 75 hour tasks too
And most production members won't run tasks that take that long when they have others that only run for a couple hours each.

Good thing I'm not in a hurry
only about 160 running and I know some have 1000+

---

Mad Scientist For Life

In reply to boboviz's message of 17 Sep 2025:

But today i download a different kind of simulation: before are "camotran_true_scan_end_of_collapse_round_pos", now are "camotran_true_scan_end_of_levelling_B2".
Maybe these simulation are shorter...

No way, 11hrs for "levelling_B2"...