|Message: Irreproducible results||Not Logged In (login)|
Click on the Forum title, e.g. on the "Forums by Category" page, to read a sequence of postings to the Forum and its threads all in one page. If you are only interested in one thread or the thread following a specific posting, click the thread or the posting, which takes you to a smaller page, which contains only the part you are interested in and may be easier to navigate.
Messages are "chained" if there are only replies at the first level, i.e. 1/1.html, 1/1/1.html etc. In case of "chained" messages the message number is replaced by the icon and there is no indentation.
Inline: Display the subject line only or also the text of the posting(s); for the choice "All" the "Outline" choices are switched off.
|1||0||1||no text / full text of posting|
|2||1||All||text for level 1 only / text for All postings|
Outline: Choose the depth of the posting thread, successive toggle controls provide increasing detail.
|1||2||1||2 levels / 1 level (original posting)|
|2||3||2||3 levels / 2 levels|
|3||3||All||3 levels / all levels (all postings)|
When I compare results from a non-MPI run to an MPI run with one worker node, I get different results.
I am comparing position and momentum for each step of each track. Not every track is different, but many of them are. I'm tracking 200 MeV/c mu+ through a 500 mm absorber containing LH2, using the QGSP_BERT physics list. If I replace the LH2 with Vacuum, the results are identical.
When using Geant4 9.6.p02 I got identical results, but with Geant4 10.1 they are different.
What has changed between 9.6.p02 and 10.1?
NOTE: I am using my own MPI code, which pre-dates the Geant4 multi-threading effort by several years; this is a single executable file which can be run either with or without MPI. The code manages random numbers such that the results should be identical, independent of event order or # ranks; it prints the result of RandGauss::shoot() in BeginOfEventAction() and in EndOfEventAction() -- these prints are identical between the MPI and non-MPI runs. So the same sequence of random numbers was used, but the tracking is not identical.
I'm using OpenMPI on a Mac Pro with 16 cores, running Mavericks (OS 10.9.5).
This is worrisome -- I see no reason why these runs should not give identical results. And in the past they did.
|Inline Depth:||Outline Depth:||Add message:|