Re: Respin Fluka2008.3b.2 released (fwd)

From: Joseph Comfort <Joseph.Comfort_at_asu.edu>
Date: Sun, 18 Oct 2009 23:23:47 -0700 (MST)

Dear Alfredo,

I reran the cases and you are right -- no crashes. Before running, I
checked the version, and it was correct. The only thing I did was to
delete the old ran... files that had been left over from earlier
attempts to run the cases that failed, just in case there was some
confusion between such files and the specified random number seed.

I'm thinking that a more likely explanation is with the specific cpu
cores or board(s). Of the 40 jobs (run in parallel), those in directory
ID numbers 25-32 failed. It is curious that they are all together. The
cluster has many quad-core boards, and I think even a bunch of 8-core
boards.

I will let the IT people know about this failure. Perhaps they can
identify the board(s) and check them out.

Thank you,
Joe

On Sat, 17 Oct 2009, Alfredo Ferrari wrote:

> Dear Joseph
>
> I ran your input with your random file with no problem/crash. I launched a
> series of runs with various random numbers, and now I have 39 of them
> successfully completed and no crash at all. Are you sure you are using an
> executable with the latest respin, and not by mistake with the previous one?
>
> Ciao
> Alfredo
>
>
> +----------------------------------------------------------------------+
> | Alfredo Ferrari || Tel.: +41.22.76.76119 |
> | CERN-EN/STI || Fax.: +41.22.76.69474 |
> | 1211 Geneva 23 || e-mail: Alfredo.Ferrari_at_cern.ch |
> | Switzerland || |
> +----------------------------------------------------------------------+
>
> On Fri, 16 Oct 2009, Joseph Comfort wrote:
>
> > Hi Alfredo,
> >
> > Thank you for the respin. I have tried it and, while better, the bug is
> > not fully resolved yet. I submitted a run for 10 cycles on 40 cpu cores.
> > 34 of them finished fine. Of the other six, 5 of them bombed in the first
> > cycle, and 1 in the second cycle. The only differences in the 40 runs is
> > the random number seed.
> >
> > A tar.gz file is attached of the relevant files, including a user mgdraw.f
> > file.
> >
> > Thank you,
> > Joe Comfort
> >
> > On Thu, 15 Oct 2009, Alfredo Ferrari wrote:
> >
> > > (smtp1.mi.infn.it [192.84.138.69]); Thu, 15 Oct 2009 16:42:58 +0200 (CEST)
> > > Sender: owner-fluka-discuss_at_mi.infn.it
> > >
> > > Dear Fluka users
> > >
> > > a new respin, Fluka2008.3b.2, has been pushed to the web site.
> > >
> > > The respin corrects for the following bugs/crashes:
> > >
> > > a) two "array out of bounds" problems with Peanut at medium/high energies
> > > (pointed out by Joseph Comfort)
> > > b) a possible (rare) infinite loop for interactions at intermediate
> > > energies (found on 4-He targets internally at CERN)
> > > c) a possible (rare) crash for interactions on 4-He (found when solving
> > > the previous problem)
> > > d) a crash for interactions on Nitrogen by low energy neutrons when the
> > > NEUTRONS default is selected, and transport thresholds for protons
> > > are kept very high (pointed out by somebody on the Fluka list)
> > > e) a (rare) crash in rQMD (pointed out by somebody on the Fluka list)
> > > f) a problem when using an isomer as radioactive isotope in the BEAM
> > > card (pointed out by S.Roesler)
> > > g) a qui-pro-quo in the provided example source routine in the part
> > > dealing with radioactive isotope beams
> > > h) a bug in R-Phi-Z USRBIN scoring of activities: the decay weights were
> > > improperly dealt with (they were never applied). Please note this
> > > can affect all R-Phi-Z USRBIN activity (only activity!) scoring.
> > > No impact on all other USRBIn scorings, or on RESNUCLEi scorings
> > > i) a crash when using the new Compton scattering (with account for
> > > electron binding) with a material containing Plutonium
> > > j) a crash when asking for nucleon decays (of interest only for
> > > members of the Icarus experiment)
> > >
> > > Please note that the corrections for b) and c) *WILL CHANGE* sooner ot
> > > later
> > > the random number sequence of runs. Because of this, the development
> > > team will be unable to reproduce possible new crashes showing up in
> > > Fluka2008.3b.1: only problems found when using Fluka2008.3b.2 will
> > > be taken into consideration. Therefore, all users, including those that
> > > are
> > > not
> > > affected by any of the (rare) bugs listed above, are kindly requested to
> > > move
> > > to Fluka2008.3b.2.
> > >
> > > I take the chance to thanks all those reporting problems. If you meet
> > > a problem/crash, after checking it is not obviously due to your input
> > > cards,
> > > please report it to the list, including the input file, the latest random
> > > number left in fluka_xxx, possible messages in the .log file, and all
> > > extra informations/user routines which could be required in order to
> > > reproduce the problem.
> > >
> > > I would like also to draw your attention to the content of the message of
> > > Francesco Cerutti of 25 Sep 2009 (Subject Re: 131-Iodine): in short,
> > > whenever
> > > non analogue radioactive decays are asked for, the energy balance
> > > as well as the energy deposition in the individual regions reported in the
> > > output are meaningless for what concern the decay radiation, since
> > > the calculation is intrinsically biased. Obviously the estimators
> > > associated
> > > with cooling times are meaningful.
> > > If one wants to get the infos reported in the output correct,
> > > simply change the decay procedure into an analogue one (changing
> > > WHAT(1) -> 2. in the RADDECAY card).
> > >
> > >
> > > Ciao
> > > Alfredo
> > > x the Fluka development team
> > >
> > >
> > > +----------------------------------------------------------------------+
> > > | Alfredo Ferrari || Tel.: +41.22.76.76119 |
> > > | CERN-EN/STI || Fax.: +41.22.76.69474 |
> > > | 1211 Geneva 23 || e-mail: Alfredo.Ferrari_at_cern.ch |
> > > | Switzerland || |
> > > +----------------------------------------------------------------------+
> > >
> > >
>
>
Received on Mon Oct 19 2009 - 09:32:55 CEST

This archive was generated by hypermail 2.2.0 : Mon Oct 19 2009 - 09:32:59 CEST