Error

zyx2017

Recently, I have recompiled the src file, but when I used the GPU to run the example, there was an error as follows. The reason I did not find, so could you tell me why did it occur? I opened the Error_BoundaryOut.vtk file, but I could not see anything.

Error information:

Particle summary:
Fixed....: 1887 id:(0-1886) MKs:1 (11)
Moving...: 0
Floating.: 29506 id:(1887-31392) MKs:171 (14-184)
Fluid....: 49897 id:(31393-81289) MKs:1 (1)

Total particles: 81290 (bound=31393 (fx=1887 mv=0 ft=29506) fluid=49897)
Total MK blocks: 173 (bound=172 (fx=1 mv=0 ft=171) fluid=1)

Allocated memory in CPU: 7514484 (7.17 MB)
Allocated memory in GPU: 21119644 (20.14 MB)
Part_0000 81290 particles successfully stored

[Initialising simulation (6z0tkmuy) 31-07-2018 20:51:37]
PART PartTime TotalSteps Steps Time/Sec Finish time
========= ============ ============ ======= ========= ===================

*** ERROR: Some boundary particle was excluded. ***
TimeStep: 0.000816 (Nstep: 3)
Total boundary: 4053 (fixed=0 moving=0 floating=4053)
Excluded for: position=4053 rhop=0 velocity=0


*** Exception (JSphGpuSingle::AbortBoundOut)
Text: Fixed, moving or floating particles were excluded. Checks VTK file Error_BoundaryOut.vtk with excluded particles.

TPouzol

Hi,

It seems that some of yours floatings particules are excluded thus causing the error:
"Total boundary: 4053 (fixed=0 moving=0 floating=4053)"
In my experience that is caused by either 1) geometry not generated correctly, or 2) floatings becoming "crazy"...
You should post process "part000" only (usin the "-last:0" option) and check the results in Paraview.

bye

TPouzol

Also check the "Error_BoundaryOut.vtk" against your geometry to known where lies the problem

zyx2017

Thanks, TPouzol. I could not see anything at Error_BoundaryOut.vtk. What is going on here?

zyx2017

As you said, what cause the floatings becoming "crazy"? The example could run well when I used the CPU, so the geometry was generated correctly.

TPouzol

If you could send the data, maybe I can have a look, otherwise I can't try to help you more.

zyx2017

Thanks. The files you could download in https://pan.baidu.com/s/1ZljKJZkQ0KuI79c1b5Osug

TPouzol

OK, I'd have the same kind of problem. I'm not sure I understand it right but its a problem regarding the domain of computation. As I understand it, after initialisation Dual redefine the computation domain depending on the location of fluid particule (at least for the GPU).

As you have fluid only on one side, all your floatings are excluded... Its weird but... try to instantiate some fluid particule on the other side of your floatings (top left corner ?). They will just fall and be excluded but that has done the trick for me so far

Maybe there is a better, elegant solution ? (Although : http://dual.sphysics.org/index.php/faq/#Q_12)

Bye