Mailing List picongpu-users@hzdr.de Message #67
From: René Widera <r.widera@hzdr.de>
Subject: Re: [PIConGPU-Users] [PIConGPU-Users] [PIConGPU-Users] Grid specification.
Date: Mon, 23 May 2016 15:20:16 +0200
To: <picongpu-users@hzdr.de>
Dear Danila,

The RingBuffer overflow say that your simulation is to large for the current used number of gpus.
It looks like you are using a older version of PIConGPU. I advice to switch to the current stable version release-0.1.2 [1].
This can help to avoid memory overflows but is no guarantee.

Your posted CFL is `number = 0.999981852222` but need to be grater equal to one.


René


[1] https://github.com/ComputationalRadiationPhysics/picongpu


On 23.05.2016 15:06, Khikhlukha Danila wrote:
Dear René
thank you so much for your prompt replay and the link.
I was able to find a supercell definition in the memory.param file. Proposed configuration worked like a charm,
although during  the initialization stage I observed messages like "[1,0]<stdout>:Ringbuffer: memory overflow", which  is a different story.
So am I correct that the precision for CFL condition checking is 3 digits after the point?

Thanks a lot,
Danila.
________________________________________
From: picongpu-users@hzdr.de [picongpu-users@hzdr.de] on behalf of René Widera [r.widera@hzdr.de]
Sent: Monday, May 23, 2016 11:11 AM
To: picongpu-users@hzdr.de
Subject: Re:  [PIConGPU-Users] [PIConGPU-Users] Grid specification.

Dear Danila,

thank you for using PIConGPU.
The grid sizes needs to be a multiple of the supercell size
(memory.param) per gpu. The default supercell size is 8x8x4 (direction X
x Y x Z).
If you run your example with `-d 2 8 2` than a working configuration is
`-g 256 3712 256`.
Please keep in mind if moving window `-m` is enabled the real simulation
volume is reduced by one gpu [1] in Y (laser propagation) direction.

best,

René

[1]
https://github.com/ComputationalRadiationPhysics/picongpu/wiki/PIConGPU-domain-definitions#moving-window

On 23.05.2016 10:58, Khikhlukha Danila wrote:
Hi all,
I have some problems trying to specify my computational grid in a
current realieaze version.
I'm not sure if I'm doing everything right. Could you please take a look?

1. In my simulation I was trying to setup a moving simulation box with
dimensions 100x100x100 um. I also was trying to
set the resolution in propagation direction much higher, so I specified
the number of cells like: "TBG_gridSize="-g 250 3750 250""
and TBG_movingWindow="-m"

2. In a gridConfig.param file I specified computational cell digestions
to fit the simulation box size  CELL_WIDTH_SI = 4.0e-7,
CELL_HEIGHT_SI = 2.6666e-8 and CELL_DEPTH_SI = CELL_WIDTH_SI. I also
specifed the time step to match CFL condition
DELTA_T_SI = 8.8558e-17

3. While compiling I got a static assert error about CFL condition being
violeted, whihc is a bit strange, since the numbers above
give Courant number = 0.999981852222.

4. Considering some possible round-off errors I decided to increase the
time step to make Courant number lower.
So I changed it to DELTA_T_SI = 8.855e-17 and it helped to compile.

5. Trying run it I recieved another assert error:
  > [1,0]<stderr>:picongpu:
/home/danila/tools/picongpu/src/picongpu/include/simulationControl/MySimulation.hpp:484:
  >void
picongpu::MySimulation::checkGridConfiguration(PMacc::DataSpace<DIM>,
PMacc::GridLayout<DIM>) [with unsigned int DIM = 3u]: Assertion
`globalGridSize[i] % MappingDesc::SuperCellSize::toRT()[i] == 0' failed.
I'm not sure I'm clear about it. What is a SuperCellSize?... I've
checked the source code and found a message that this check
is redundant. However further manipulations in the code unfortunately
are not very clear to me...

Could you please take a look if I'm doing something wrong?

Best regards,
Danila.

--
René Widera
Abteilung Laser-Teilchenbeschleunigung (FWKT)
Helmholtz-Zentrum Dresden-Rossendorf
Tel: +49 (0351) 260 3543
r.widera@hzdr.de
http://www.hzdr.de

Vorstand: Prof. Dr. Dr. h. c. Roland Sauerbrey,
            Prof. Dr. Dr. h. c. Peter Joehnk
Vereinsregister: VR 1693 beim Amtsgericht Dresden

#############################################################
This message is sent to you because you are subscribed to
   the mailing list <picongpu-users@hzdr.de>.
To unsubscribe, E-mail to: <picongpu-users-off@hzdr.de>
To switch to the DIGEST mode, E-mail to <picongpu-users-digest@hzdr.de>
To switch to the INDEX mode, E-mail to <picongpu-users-index@hzdr.de>
Send administrative queries to  <picongpu-users-request@hzdr.de>


#############################################################
This message is sent to you because you are subscribed to
   the mailing list <picongpu-users@hzdr.de>.
To unsubscribe, E-mail to: <picongpu-users-off@hzdr.de>
To switch to the DIGEST mode, E-mail to <picongpu-users-digest@hzdr.de>
To switch to the INDEX mode, E-mail to <picongpu-users-index@hzdr.de>
Send administrative queries to  <picongpu-users-request@hzdr.de>


--
René Widera
Abteilung Laser-Teilchenbeschleunigung (FWKT)
Helmholtz-Zentrum Dresden-Rossendorf
Tel: +49 (0351) 260 3543
r.widera@hzdr.de
http://www.hzdr.de

Vorstand: Prof. Dr. Dr. h. c. Roland Sauerbrey,
          Prof. Dr. Dr. h. c. Peter Joehnk
Vereinsregister: VR 1693 beim Amtsgericht Dresden
Subscribe (FEED) Subscribe (DIGEST) Subscribe (INDEX) Unsubscribe Mail to Listmaster