Hi Shy Shi,
The reason why it takes longer when you change the iron core by a lighter version (C or H2O) of the same dimensions, is that those other cores let more particles (primaries and showers) leak out, and when the escaping particles penetrate through the next shells of the target they split into more (due to your geometry biasing), and then more, and so on. Tracking all those additional particles is consuming your CPU time. But you should also obtain better statistics per starting electron, i.e. You could run much less than 1 million electrons and still obtain equally statistically significant results in the periphery of the dump if that is what you are looking for.
You should consider the following:
- If you are using geometry biasing you should tune the importances of the cells for each problem, typically leaving a constant population.
- What is the purpose of your geometry biasing? Is your interest outside the dump or at its core? If it is outside then geometry biasing can make sense (but it may not be worth the time to tune it), while if your interest was at the core, you would be wasting CPU.
- At the end of the day you should not JUST be concerned whether a simulation takes longer than another one. Your figure of merit for simulation efficiency should be (statistical_fluctuation*CPU_time), not just CPU_time
Finally, it may not have been necessary to attach a 9Mb error file...
-Mario
From: owner-fluka-discuss_at_mi.infn.it<mailto:owner-fluka-discuss_at_mi.infn.it> [mailto:owner-fluka-discuss_at_mi.infn.it] On Behalf Of shihy_at_ihep.ac.cn<mailto:shihy_at_ihep.ac.cn>
Sent: 15 August 2016 03:17
To: fluka-discuss <fluka-discuss_at_fluka.org<mailto:fluka-discuss_at_fluka.org>>
Subject: [fluka-discuss]: [fluka-discuss]Calculate too slow
Dear FLUKA Experts,
I'm designing a beam dump used for 6 GeV Electrons. The geometry is very simple with core and concrete surrounding the core. I want to use Graphite(Carbon), Iron and Water as the material of the core. When I set the core material to iron, it runs well and takes 1 day to calculate for one cycle with 1 Million particles. But when I change the material of core to carbon and water(the only thing I change is the "ASSIGNMAT" card) it takes more than 3 days to run for one cycle and has not finished. I killed the process and attached the input file(carbon version), out file and error file(more the 6 MB). What can I do to fix this?
Thank You in Advance.
Shy Shi
__________________________________________________________________________
You can manage unsubscription from this mailing list at
https://www.fluka.org/fluka.php?id=acc_info
Received on Sat Aug 20 2016 - 04:53:16 CEST