Hello everybody.
I would be really thankful if you help me with this error encountered during the LAMMPS run for calculating and saving the data of RDF between a pair of atoms.
I’ve made sure that the “compute rdf” command works fine but the problem stands for the fix ave/time which records the RDF data. I used 100 and 100 as Nevery and Nfreq respectively but the run stopped and MPI_abort encountered exactly before the 100th timestep (for instance timestep 99).
compute 8 all rdf 100 349 210
fix c8 all ave/time 100 1 100 c_8[1] c_8[2] c_8[3] file RDF.rdf mode vector
The error you show has nothing to do with compute rdf or fix ave/time. It is a failure of using the GPU. It should go away when you run without requesting GPU acceleration.
This statement makes no sense. A compute is not executed unless it is “consumed”, i.e. when you add a compute command, it will be only checked for correct syntax but will never be executed unless there is some other command that references it, e.g. fix ave/time.
Thanks for your response.
I had tried what you said but neither the GPU nor the CPU works properly. MPI_abort error encounters even without using the GPU.
By changing Nevery and Nfreq to 10000, the system is run with no problem until it reaches the 10000th timestep. That’s why I believe that there must be something with fix ave/time.
It is impossible to debug such a thing from remote and based on screenshots providing only partial data. Since you input seems rather complex and convoluted, my suggestion is that you take the “rhodo” benchmark input deck from the “bench” folder and change it:
- Adjust the thermo and run settings to be similar to your input
- Add similar compute rdf and fix ave/time commands like your input. It would be sufficient to only add computations for atom types 33 and 4 (which are the water oxygen and hydrogen of the system). That would have the advantage, that those should give results similar to bulk water and thus allow to validate the compute rdf settings.
Then you can see, if you can reproduce the same kind of error with this test input, and if yes, you can post it here and it will be easier to debug what is going on.
Please also state what version of LAMMPS exactly you are using and what platform you are running on.
1 Like
