Re: [ESPResSo-users] Problem MPI+GPU using LBM

[Axel Arnold]

Hi Markus,

I just noticed that you first set the lb walls and then configure lb to 
use the GPU. Unfortunately, that means that the walls are lost, since 
they still go to the CPU code. That should not cause a crash, but 
certainly will not do what you expect. So, can you initialize the fluid 
first, before setting the walls?

Axel

On 07/01/2013 11:57 AM, Markus Gusenbauer wrote:
> Ok, found it. I ran same simulation with the current Master. 
> Unfortunately no difference in the output.
>
> Markus
>
> On 01.07.2013 11:35, Axel Arnold wrote:
> > ... meaning, the current git master...
> >
> > Axel
> >
> > On 07/01/2013 11:34 AM, Axel Arnold wrote:
> > > Hi Markus,
> > >
> > > which version are you using? There are currently a number of 
> > > restructure attempts going on in the LBGPU code, so it is quite 
> > > important that you use the latest version.
> > >
> > > Axel
> > >
> > > On 07/01/2013 11:17 AM, Markus Gusenbauer wrote:
> > > > Hi all,
> > > >
> > > > I've tried to run a simple simulation using 2 CPU + GPU. I have a 
> > > > channel with lbfluid, from the left I put a certain velocity. 
> > > > Without lbboundary the simulation runs fine. As soon as I add a 
> > > > lbboundary it crashes:
> > > >
> > > > [mgusenbauerMint13:12827] *** An error occurred in MPI_Bcast
> > > > [mgusenbauerMint13:12827] *** on communicator MPI_COMMUNICATOR 3
> > > > [mgusenbauerMint13:12827] *** MPI_ERR_TRUNCATE: message truncated
> > > > [mgusenbauerMint13:12827] *** MPI_ERRORS_ARE_FATAL (your MPI job 
> > > > will now abort)
> > > > -------------------------------------------------------------------------- 
> > > >
> > > > mpirun has exited due to process rank 1 with PID 12827 on
> > > > node mgusenbauerMint13 exiting without calling "finalize". This may
> > > > have caused other processes in the application to be
> > > > terminated by signals sent by mpirun (as reported here).
> > > > -------------------------------------------------------------------------- 
> > > >
> > > >
> > > >
> > > > Here is the tcl-script:
> > > >
> > > >
> > > > setmd time_step 0.1
> > > > setmd skin 0.2
> > > > thermostat off
> > > >
> > > > setmd box_l 40 40 100
> > > >
> > > > lbboundary wall normal 1 0 0 dist 0.5 type 501
> > > > lbboundary wall normal -1 0 0 dist -39.5 type 501
> > > > lbboundary wall normal 0 1 0 dist 0.5 type 501
> > > > lbboundary wall normal 0 -1 0 dist -39.5 type 501
> > > >
> > > >
> > > > lbfluid gpu grid 1 dens 1.0 visc 1.5 tau 0.1 friction 0.5
> > > >
> > > > set i 0
> > > > while { $i < 100 } {
> > > >     puts "$i / 100 \r"
> > > >
> > > >     for { set iii 0 } { $iii < 40} { incr iii } {
> > > >         for { set jjj 0 } { $jjj < 40 } { incr jjj } {
> > > >             for { set kkk 0 } { $kkk < 1 } { incr kkk } {
> > > >                 lbnode $iii $jjj $kkk set u 0.0 0.0 0.1
> > > >             }
> > > >         }
> > > >     }
> > > >
> > > >     integrate 1
> > > >     incr i
> > > > }
> > > >
> > > >
> > > > Same simulation works fine using MPI+CPU. Any ideas?
> > > >
> > > > Markus


--
JP Dr. Axel Arnold
ICP, Universität Stuttgart
Pfaffenwaldring 27
70569 Stuttgart, Germany
Email: address@hidden
Tel: +49 711 685 67609