Cannot (yet) use single precision with MSM

LAMMPS LAMMPS Mailing List Mirror
1

lammps version 24 Dec 2015

I am using the following input file:

boundary f f f
units real
atom_style full
bond_style class2
angle_style class2
dihedral_style class2
improper_style class2
pair_style soft 12
pair_style lj/class2/coul/long 12
pair_modify tail yes table 12
read_data C.data

kspace_style msm 1e-4
neighbor 1 bin

neigh_modify delay 0 every 1 check yes page 1000000
thermo 100
thermo_style custom step temp press vol density etotal ke pe time
minimize 1.0e-06 1.0e-06 100000 100000

The run stops and this error appears:

Cannot (yet) use single precision with MSM (remove -DFFT_SINGLE from Makefile and recompile) (…/msm.cpp:169)

The above solution is impossible for me and the machine I am using. Is there any other solution I can deal with this issue?

Thank you for your help
-X

2

Besides, in another simulation, I applied OPLS, and according to this thread:

http://lammps.sandia.gov/threads/msg40084.html

I used:

lj/cut/coul/msm 12 instead of lj/cut/coul/long 12

However, It does not work, and the same error appears.
I look forward to your solutions and suggestions

Thanks

3

lammps version 24 Dec 2015

I am using the following input file:

boundary f f f
units real
atom_style full
bond_style class2
angle_style class2
dihedral_style class2
improper_style class2
pair_style soft 12
pair_style lj/class2/coul/long 12
pair_modify tail yes table 12
read_data C.data
kspace_style msm 1e-4
neighbor 1 bin
neigh_modify delay 0 every 1 check yes page 1000000
thermo 100
thermo_style custom step temp press vol density etotal ke pe time
minimize 1.0e-06 1.0e-06 100000 100000

The run stops and this error appears:

Cannot (yet) use single precision with MSM (remove -DFFT_SINGLE from
Makefile and recompile) (../msm.cpp:169)

The above solution is impossible for me and the machine I am using. Is there
any other solution I can deal with this issue?

no.

4

I don’t understand this:
The above solution is impossible for me and the machine I am using.

You simply build LAMMPS w/out the -DFFT_SINGLE flag in Makefile.machine and

then you can use MSM, either for kspace or pair styles

Since you must be building LAMMPS with a special Makefile.machine that

has the flag, why can’t you also build an executable w/out the flag?

Steve

5

Also, it is required to use an MSM pair_style with kspace_style msm. Using pair_style lj/class2/coul/long won’t work with MSM.

Stan

6

Hi Steve,
I mean, the operators of the supercomputers won’t build a separate version of lammps for an specific application of a user.
This is why I was looking for any other solution :slight_smile:
Thanks anyway

7

If you mean your sysadmin provides one version of pre-built LAMMPS
and decided to do that with enabling the -DFFT_SINGLE flag (which
is not the default), then someone should fine him for abuse of
sysadmin powers. I think that’s a felony in most labs.

Also, nothing is stopping you from logging into the machine
and building a copy of LAMMPS yourself. Just cover your
tracks (rename the exe), and don’t tell the sysadmins
we told you that though …

Steve