huge RAM requirements in coupled photon-gamma simulation

Report all good and bad behavior here
Post Reply
nico
Posts: 23
Joined: Wed Feb 28, 2018 1:57 pm
Security question 1: No
Security question 2: 93

huge RAM requirements in coupled photon-gamma simulation

Post by nico » Mon Oct 14, 2019 6:57 pm

Dear Serpent developers,

I have encountered an issue regarding coupled neutron-photon transport.
When I run the coupled simulation, an error about memory allocation comes out.
Looking at the RAM requirements:
1) 100k neutrons with "set ngamma 1" and requires 42% of the RAM (i.e. 42 GB)
2) 100k neutrons with "set ngamma 0" and requires 42% of the RAM (i.e. 42 GB), photon data are loaded but only neutron transport is performed
3) 100k neutrons without photons (commenting the line "set ngamma 1") only 2.8% (almost 3 GB), no photons data and no photon transport

Why is there such a memory requirements? I'd need to run simulations with 1M neutrons, so I wonder if there is a way to fix this issue.

I attach the code here,

Code: Select all

%% -- EXTERNAL FILE INCLUSION
include "/home/pclab/serpent2models/alfred/materials/ALFRED_materials_673_673"  % materials
include "/home/pclab/serpent2models/alfred/geometry/ALFRED_geometry"            % geometry

%% -- LIBRARY FOR NUCLEAR DATA
set acelib "/opt/serpent/xsdata/endfb7/sss_endfb7.xsdata"
set pdatadir "/opt/serpent/xsdata/photon_data"

%% --------------------------------------------- Core lattice loading ---------------------------------------------
% Core
surf s1 sqc 0.0 0.0 25	      % wrapping inner perimeter

lat core 3 0.0 0.0 5 5 17.1      % Core with CR and SR extracted
ifa ifa ifa ifa ifa
 ifa ifa ifa ifa ifa
  ifa ifa ifa ifa ifa
   ifa ifa ifa ifa ifa
    ifa ifa ifa ifa ifa

%% --------------------------------------------------------------------------------------------------------------------------------

cell c100 0 fill core -s1  % Filling the cylinder with the core
cell c101 0 outside s1     % Placing the vacuum outside the reactor


%% -- SIMULATION SETUP
set bc 2
set pop 100000 500 50

set power 1              % fix power to 1 W
set ures 1               % sample fast unresolved resonance region (urr)
set his yes              % Check on fission source convergence
set ngamma 1             % set coupled neutron-photon calculation


det pprod n dm fueli de mygrid dr -5   void % photons source
det psrc  p dm fueli de mygrid dr -11  void 
det ptot  p du ifa             dr -25  void % total macroscopic XS for photons
Thank you for your kind attention,
Nicolo'

PS Sorry, I already posted it but maybe you did not see it since it was under another topic!

User avatar
Jaakko Leppänen
Site Admin
Posts: 2446
Joined: Thu Mar 18, 2010 10:43 pm
Security question 2: 0
Location: Espoo, Finland
Contact:

Re: huge RAM requirements in coupled photon-gamma simulation

Post by Jaakko Leppänen » Tue Oct 15, 2019 5:55 pm

Could you also include the additional input files with geometry and material definitions+
- Jaakko

nico
Posts: 23
Joined: Wed Feb 28, 2018 1:57 pm
Security question 1: No
Security question 2: 93

Re: huge RAM requirements in coupled photon-gamma simulation

Post by nico » Tue Oct 15, 2019 6:39 pm

Sorry, you can find the files attached to this post!
Attachments
missing_files.7z
(6.89 KiB) Downloaded 121 times

User avatar
Jaakko Leppänen
Site Admin
Posts: 2446
Joined: Thu Mar 18, 2010 10:43 pm
Security question 2: 0
Location: Espoo, Finland
Contact:

Re: huge RAM requirements in coupled photon-gamma simulation

Post by Jaakko Leppänen » Wed Oct 16, 2019 5:53 pm

Looks like the memory allocation is done wrong for coupled neutron photon transport problems. Try commenting out the if-else stucture on line 189-191 of inithistories.c, so that the value of np is calculated as:

Code: Select all

    np = (long)(RDB[DATA_PART_GBUF_FACTOR]*RDB[DATA_OMP_MAX_THREADS]);
- Jaakko

nico
Posts: 23
Joined: Wed Feb 28, 2018 1:57 pm
Security question 1: No
Security question 2: 93

Re: huge RAM requirements in coupled photon-gamma simulation

Post by nico » Tue Nov 05, 2019 12:00 pm

Dear Jaakko,

thank you so much for your support. We extensively tested the coupled neutron-photon simulations and now it works quite well.

Nicolo'

Post Reply