LHC@home

Hello,

I completed yesterday a very short workunit, please take a look here:

wb0055_v65b0055__1__64.33_59.34__8_10__6__60_1_sixvf_boinc288030

I first thought it was invalid and has been stopped early due to errors, but according to the results, they all seem valid. The other returned results are also very short ranging from 5 to 277 seconds computation time only (mine is the 277 seconds result btw :-) ).

I'm not crunching much for LHC, mostly for other projects, so my question: Are there more such workunits send out or is this something irregular?

Best regards

Alex

---

Hi Alex,

yes these are fairly common on LHC - you will get the relevant credit for the work done providing the tasks also ran short on the other machines that were sent the same work. In fact, since your post the validator has awarded you 0.08 cobblestones credit for this work, valuing it as 1/1200th of a days work for the BOINC standard box - ie about a minute's work at 1GHz

What happens is that the app is simulating the particles whizzing round the LHC. Sometimes if the sim is of an unstable configuration of fields in the LHC, then the particles hit the wall of the tube on the first time round the loop, instead of going round 100,000 or 1000,000 times. This is, of course, exactly what the engineers want to know, so it is not an error.

It can be irritating to participants, as these short jobs tend to come in clumps - at present for example I have one machine still running long jobs from yesterday's downloads, but another identical machine that downloaed the same number of WUs got completed in a few hours as they were mostly short running jobs. I would be much happier if the long- and short-runnign jobs were evenly spread over the machines.

This is a human reaction based on feelings, not logic. Of course, if the engineers knew which would be the short jobs in advance they would not need to run them at all.

It is part of supporting this project to accept that this happens. Even when we win the lottery to get some WU at all, sometimes it happens that they only run for a short time, but then the work has run out again and other people got WU that ran longer.

Some people, quite fairly, have chosen to support other projects instead simply because they do not like this feature. That is their choice; personally I prefer to support this project anyway as I have contributed as much to the science when I send back a short job as when I send back a long one. My boxes will automatically go back to crunching my other projects as soon as the LHC job finishes, whether it is soon or late.

River~~

ps - I liked the sig :)

---

Hi River~~,

Thanks a lot for your detailed reply. And I appreciate the 0.08 cobblestones very much as it brings me one step closer to 150k on BOINC combined. ;-)

then the particles hit the wall of the tube on the first time round the loop, instead of going round 100,000 or 1000,000 times. This is, of course, exactly what the engineers want to know, so it is not an error.

Ok, makes perfect sense to me. There are similar workunits in other projects. For example CPDN cancels calculation for a climate model, when values (i.e. temperature) become too extreme.

My boxes will automatically go back to crunching my other projects as soon as the LHC job finishes, whether it is soon or late.

Yep, would be the same here, so I have no problem with short WUs. I keep monitoring LHC and I'm ready to donate more resources as soon as available work will increase.

Regards

Alex