Hi Bruno,
just to clarify, Test 2 is done by increasing the number of iterations (1x, 3x
and 12x the number of iterations specified in checkPerf.py). This means the
number of interactions should increase as well and, hence, particle velocities
should decrease because of more interactions.
I ad
revno: 3899
committer: Anton Gladky
timestamp: Wed 2014-04-09 16:03:16 +0200
message:
Add Cs parameters to particles, visualize them in VTK.
modified:
core/Body.hpp
pkg/common/SPHEngine.cpp
pkg/common/SPHEngine.hpp
pkg/dem/VTKR
revno: 3904
committer: Anton Gladky
timestamp: Wed 2014-04-09 16:03:21 +0200
message:
Add example script to test different kernel functions.
added:
examples/sph/testKernelFunc.py
--
lp:yade
https://code.launchpad.net/~yade-pkg/yade
revno: 3900
committer: Anton Gladky
timestamp: Wed 2014-04-09 16:03:16 +0200
message:
Export in VTK some more SPH-parameters.
modified:
pkg/dem/VTKRecorder.cpp
--
lp:yade
https://code.launchpad.net/~yade-pkg/yade/git-trunk
Your te
revno: 3898
committer: Anton Gladky
timestamp: Wed 2014-04-09 16:03:16 +0200
message:
Produce meaningful error message.
modified:
pkg/common/SPHEngine.cpp
pkg/common/SPHEngine.hpp
pkg/dem/ViscoelasticPM.hpp
--
lp:yade
https://c
revno: 3901
committer: Anton Gladky
timestamp: Wed 2014-04-09 16:03:16 +0200
message:
Get a coordination number from body.
modified:
core/Body.cpp
core/Body.hpp
pkg/dem/VTKRecorder.cpp
--
lp:yade
https://code.launchpad.net/~yad
revno: 3905
committer: Anton Gladky
timestamp: Wed 2014-04-09 16:03:21 +0200
message:
Add README to examples SPH.
added:
examples/sph/README
--
lp:yade
https://code.launchpad.net/~yade-pkg/yade/git-trunk
Your team Yade developers
revno: 3902
committer: Anton Gladky
timestamp: Wed 2014-04-09 16:03:16 +0200
message:
Use bool return type for force calculation.
Would be good to implement it for all Law_*, like
discussed recently.
modified:
pkg/common/SPHEn
revno: 3897
committer: Anton Gladky
timestamp: Wed 2014-04-09 16:03:16 +0200
message:
Normalize parameters of kernel functions.
modified:
pkg/common/SPHEngine.cpp
--
lp:yade
https://code.launchpad.net/~yade-pkg/yade/git-trunk
Your
revno: 3903
committer: Anton Gladky
timestamp: Wed 2014-04-09 16:03:16 +0200
message:
Add watercolumn example for SPH.
added:
examples/sph/box.geo
examples/sph/box.mesh
examples/sph/cpt/
examples/sph/cpt/a
examples/sph/waterc
revno: 3894
committer: Anton Gladky
timestamp: Wed 2014-04-09 11:11:49 +0200
message:
Add gradients and laplacian of the kernel functions.
modified:
pkg/common/SPHEngine.cpp
pkg/common/SPHEngine.hpp
--
lp:yade
https://code.launch
revno: 3896
committer: Anton Gladky
timestamp: Wed 2014-04-09 16:03:16 +0200
message:
Remove duplicated code.
modified:
pkg/common/SPHEngine.cpp
pkg/common/SPHEngine.hpp
pkg/dem/ViscoelasticPM.cpp
pkg/dem/ViscoelasticPM.hpp
-
revno: 3895
committer: Anton Gladky
timestamp: Wed 2014-04-09 16:03:15 +0200
message:
Implement some more kernel functions.
modified:
pkg/common/SPHEngine.cpp
pkg/common/SPHEngine.hpp
pkg/dem/ViscoelasticPM.cpp
pkg/dem/Viscoela
revno: 3893
committer: Anton Gladky
timestamp: Wed 2014-04-09 11:11:49 +0200
message:
Implement experimental SPH-model
Basic implementation of smoothed particles hydrodynamics
Squashed commit of the following:
commit 2e4
Hi guys,
just to let you know. I updated the results on the wiki [1]. Still performance
test but with more iterations and up to 1 million particles.
Cheers,
Klaus
[1] https://yade-dem.org/wiki/Performance_Test#Test_2
___
Mailing list: https://launch
Thanks!
If I understand correctly, particles velocities are decreasing with
iterations. So, more iterations means less weight for the collider
overall (hence less effect of parallelizing it).
>From you results with 1million, I see for the collider T(j8)=T(j1)/5.8.
Could you tell if the collider tim
So a) is a lot better.
I can update the "compile" part of the doc to explain how to compile with
openblas+openmp on 12.04.
--
You received this bug notification because you are a member of Yade
developers, which is subscribed to Yade.
https://bugs.launchpad.net/bugs/1304878
Title:
Openblas br
2014-04-09 12:48 GMT+02:00 Bruno Chareyre <1304...@bugs.launchpad.net>:
> >Sure, we can switch to atlas, if you think it will fix (at least
> partially)
>
> It should fix completely.
> Could we have alternatives in the dependencies? I think I've seen that in
> deb packages sometimes. E.g. installi
>I do not see any problems with parallelisation on Debian Jessie
Good to know that it stabilizes after 6 years of struggling with libs,
just for solving a linear system!
--
You received this bug notification because you are a member of Yade
developers, which is subscribed to Yade.
https://bugs.l
>it seems there are good reasons not to use Openblas+OpenMP.
Well, we do use openblas+openmp for years. I was always compiling
openblas (with, namely, USE_OPENMP = 1), that is why I never noticed the
problem we see now. When running yade, PFV uses openblas while python
libs use atlas (see [1] #4)
Hi Anton,
I checked the Pournin's implementation you made, and I may have found a
little error. According to the formula (10) in the paper, the four formulas
(22) are the resulting stiffness/viscosity_coefs of a contact, and you
rightly transcript them. But in your algorithm, you set
kn1 = kn2 = *
Actually, I do not see any problems with parallelisation on Debian Jessie.
We can probably switch to atlas only on 12.04, which is pretty old already.
Anton
--
You received this bug notification because you are a member of Yade
developers, which is subscribed to Yade.
https://bugs.launchpad.net
** Bug watch added: Debian Bug tracker #737675
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=737675
--
You received this bug notification because you are a member of Yade
developers, which is subscribed to Yade.
https://bugs.launchpad.net/bugs/1304878
Title:
Openblas breaks openmp
Statu
> I'm not sure how to best fix yadedaily. I noticed openblas-omp packages
> available for Fedora and others, but not for debian.
>
See [1], it seems there are good reasons not to use Openblas+OpenMP.
[1] https://bugs.debian.org/737675
> One option is to replace libopenblas dependency by libatla
Public bug reported:
Linking libopenblas as it is in ubuntu 12.04 disable multithread tasks (known
issue with other applications to).
Yade -jN run the parallel tasks in a sequential way, and therefore it actually
slows down for increasing N due to overheads, as noticed by [1].
The bug affects ya
25 matches
Mail list logo