[Gate-users] running multiple instances of Gate on the same node

[Giacomo]

Hi, 

we are trying to run multiple instances of Gate on the same computational
node to exploit the presence of multiple cores, but we are getting this kind
of output from the log: 

 

Processing ./MLEM_27sources1x106Bq_test_mod_140_250keV1.mac file...

0: paradent-9.rennes.grid5000.fr

MLEM_27sources1x106Bq_test_mod_140_250keV1.mac
100%   22KB  21.9KB/s   00:00

 

                [The first process seems OK, but for the other 3 we get..]

 

Processing ./MLEM_27sources1x106Bq_test_mod_140_250keV2.mac file...

1: paradent-9.rennes.grid5000.fr

MLEM_27sources1x106Bq_test_mod_140_250keV2.mac
100%   22KB  21.9KB/s   00:00

        [G4-cerr] [systems/SPECThead/base::IsValidAttachmentRequest]:

        [G4-cerr]       A volume creator ('SPECThead') is already attached
to this system component

        [G4-cerr]

        [G4-cerr] [systems/SPECThead/base::SetInserter]:

        [G4-cerr]       Ignoring attachment request

        [G4-cerr]

 

Processing ./MLEM_27sources1x106Bq_test_mod_140_250keV3.mac file...

2: paradent-9.rennes.grid5000.fr

MLEM_27sources1x106Bq_test_mod_140_250keV3.mac
100%   22KB  21.9KB/s   00:00

        [G4-cerr] [systems/SPECThead/base::IsValidAttachmentRequest]:

        [G4-cerr]       A volume creator ('SPECThead') is already attached
to this system component

        [G4-cerr]

        [G4-cerr] [systems/SPECThead/base::SetInserter]:

        [G4-cerr]       Ignoring attachment request

        [G4-cerr]

 

Processing ./MLEM_27sources1x106Bq_test_mod_140_250keV4.mac file...

3: paradent-9.rennes.grid5000.fr

MLEM_27sources1x106Bq_test_mod_140_250keV4.mac
100%   22KB  21.9KB/s   00:00

        [G4-cerr] [systems/SPECThead/base::IsValidAttachmentRequest]:

        [G4-cerr]       A volume creator ('SPECThead') is already attached
to this system component

        [G4-cerr]

        [G4-cerr] [systems/SPECThead/base::SetInserter]:

        [G4-cerr]       Ignoring attachment request

        [G4-cerr]

 

So essentially the other 3 Gate processes are using the same object
'SPECThead'.

Is this "sharing" of objects instantiated by the first computation
concurrency-OK? I mean.. can this fact lead to wrong results?

 

Thanks,

Giacomo Righetti & Sebnem Erturk