[Halld-offline] Fine-mesh field and EVIO

[Richard Jones]

Dave,

That sounds believable.  However, Jake reports that when he runs the same analysis as I am doing, but on his cluster, he does not see anything about a fine-mesh map in his logs.  That is the mystery I referred to in my comment.  Some builds seem to invoke it, and others not.

To have a local cache of the fine-mesh map, this relies on the availability of the evio library.  As soon as there is a public release of this library, I will build with it, but I don't want to work with private copies.

-Richard J.


On 2/22/2011 8:58 AM, David Lawrence wrote:
> Hi Richard,
>
>       I noticed in your log message below the comment about the mystery
> of the fine-mesh map sometimes being computed and sometimes not. This is
> likely due to a feature built-in to the code where if the Hall-D
> software was built with EVIO support (identified with EVIOROOT
> environment variable being set at compile time) it will write out the
> fine-mesh map to the local directory. This can then be read in on
> subsequent invocations. If you change directories, then it won't find it
> and will write out another one. This will cost some additional disk
> space as these 125MB "finemesh.evio" files will be left around.
>
>       For those running the reconstruction code a lot to look at only a
> few events (<100), they may find benefit from using this feature as it
> saves startup time. If you're doing a lot of events however, it will be
> of limited benefit due to the time saved being so small compared to the
> total processing time spent.
>
>
> r7462 | jonesrt | 2011-02-18 19:02:51 -0500 (Fri, 18 Feb 2011) | 10 lines
>
> DApplication.cc [rtj]
>       * fix the problem that THREAD_TIMEOUT is hard-wired to 30 seconds,
>         disabling the -PTHREAD_TIMEOUT= command line parameter.  Now it
>         respects this parameter on the command line, and defaults to 30
>         seconds if not given.  This was a disaster before, because it
>         typically takes 2 minutes or so to read the magnetic field map
>         from the calibdb and compute the fine-mesh map from what it pulls
>         in from the database.  Mysteriously, sometimes it wants to compute
>         the fine-mesh map and sometimes not -- don't ask.
>
>
>
> Regards,
> -Dave


-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 4092 bytes
Desc: S/MIME Cryptographic Signature
URL: <https://mailman.jlab.org/pipermail/halld-offline/attachments/20110222/19de10d4/attachment.p7s>