LHC@home

Seeing a limit of 20 tasks for my PC's, they have more CPU's (64).
Two months ago, no problem to fill the pipeline with more tasks.
Nothing changed from my side.

My computers seem to have a limit of 80 Theory tasks. They have 8 and 16 CPUs.

---

I stopped requesting new tasks, when I already got 148.
8-threads PC.

#date_d ngood nbad total
2023-05-02 380 14 394
2023-05-03 482 3 485

Max. Aufgaben pro Tag 2425
Anzahl der Aufgaben heute 54
20 Tasks is the limit from Theory, but why 485 finished in mcplot?

If you set Max # jobs &/or Max # CPUs to anything other than No limt then it could be there is a cap.

Given how long the tasks take I'm not sure why you need such a buffer of work, I have 0.25 day of work and have 70 on my computer, 36 running and 34 buffer

Have changed back to Atlas.
Only 20 Theory for 64 Cpu's is to low for this Threadripper.

2023-05-06 19:38:21 (11556): Guest Log: Probing /cvmfs/sft.cern.ch... Failed!
2023-05-06 19:38:21 (11556): Guest Log: 19:38:27 CEST +02:00 2023-05-06: cranky: [ERROR] 'cvmfs_config probe sft.cern.ch' failed.

Theory in Win11pro have Traffic in Download Yesterday 485 GByte. Is this because of CVMFS?

===> [runRivet] Mon May 8 22:20:29 UTC 2023 [boinc ppbar zinclusive 1960 -,-,50,120 - sherpa 2.1.0 default 18000 476]

full optimization: ( 58m 19s (32m 26s) elapsed / 23m 52s (13m 16s) left ) [23:26:09]
2.41927 pb +- ( 0.00710686 pb = 0.293761 % ) 240000 ( 893573 -> 28.3 % )
integration time: ( 1h 3m 40s (35m 23s) elapsed / 18m 34s (10m 20s) left ) [23:31:29]
My_File<FileType>::Close(): '0x3632ea8' returns 'out of memory'.
Exception_Handler::GenerateStackTrace(..): Generating stack trace
{
}

Exception_Handler::SignalHandler: Signal (6) caught.
Cannot continue.

Have changed back to Atlas.
Only 20 Theory for 64 Cpu's is to low for this Threadripper.

Now ok:
2023-05-07 1075 20 1095
2023-05-08 1200 22 1222
2023-05-09 1226 26 1252

working well atm:
2023-05-17 1278 13 1291
2023-05-18 1156 17 1173
80 MBit/s download, more than 400 GByte.

After changed CPU's to unlimited, now so many Tasks never seen before.
Only 20 Theory for 64 Cpu's is to low for this Threadripper.
22.07.2023 20:48:45 | LHC@home | choose_project: scanning
22.07.2023 20:49:21 | LHC@home | can't fetch CPU: project is backed off

12 hours later and hundreds of Theory-Tasks later, all is well so long.

#date_d ngood nbad total
2023-07-23 1013 15 1028
2023-07-24 1002 34 1036

Total CPU time: 448775770 s (456890 s by failed jobs)
Total jobs: 104785 (1692 failed, 2%)
Total events: 10295869000 (1G is reached on 2019-11-14, 10G is reached on 2023-07-25)

Is it possible to eliminate this lines?
ZAlign::ZAlign(): Q = 240.413 vs. 246.276, rel. diff. -nan
3900 events processed
ZAlign::ZAlign(): p_a*p_b = 172128 vs. 181493, rel. diff. -0.0515989
ZAlign::ZAlign(): Q = 612.276 vs. 609.686, rel. diff. 0.00424836

It's a Sherpa Problem. Stopped at the same Event (49100). Canceling now!!
This Theory task have stopped working.
Made a restart(boinc-VM in Virtualboxmanager).
First restart stopped with sft-Error during booting.
second restart working now:

Event 49100 ( 32m 11s elapsed / 33m 22s left ) -> ETA: Fri Jul 28 13:06
Rivet.Analysis.CMS_2012_I1090423: WARN Skipping histo with null area /CMS_2012_I1090423/d01-x01-y01
Rivet.Analysis.CMS_2012_I1090423: WARN Skipping histo with null area /CMS_2012_I1090423/d02-x01-y01
49100 events processed
===> [runRivet] Sun Jul 30 05:08:30 UTC 2023 [boinc pp jets 7000 80,-,960 - sherpa 1.4.2 default 100000 514]

Setting environment...
grep: /etc/redhat-release: No such file or directory
MCGENERATORS=/cvmfs/sft.cern.ch/lcg/releases/LCG_96/MCGenerators
g++ = /cvmfs/sft.cern.ch/lcg/releases/gcc/8.2.0-3fa06/x86_64-slc6/bin/g++
g++ version = 8.2.0