What is the section "Other" in MPI task timing breakdown

LAMMPS LAMMPS General Discussion
1

Hello,

I am trying to understand the time in MPI task timing breakdown, can you please help me know what is the other in this case?

Probably this is not enough information, what information do you need?

Is it correct to assume it’s fix brownian or propel/self that consumes the time, since everything else is optimized?

I am trying to simulate a 2D active Brownian particles.

# WCA potential (purely repulsive)
pair_style lj/cut ${rcut}
pair_coeff * * ${e_epsilon} 1.0 ${rcut}
pair_modify shift yes

set atom * dipole/random ${seed} 1 #0 0

fix step all brownian/sphere ${temp} ${seed} gamma_t ${gamma_t} gamma_r ${gamma_r} rng ${rng}

fix align_field all efield  0 ${ef} 0
fix active all propel/self dipole ${fp}

Performance: 150280.362 tau/day, 347871.207 timesteps/s, 360.742 Matom-step/s
99.4% CPU use with 1 MPI tasks x no OpenMP threads

MPI task timing breakdown:
Section |  min time  |  avg time  |  max time  |%varavg| %total
---------------------------------------------------------------
Pair    | 0          | 0          | 0          |   0.0 |  0.00
Neigh   | 0          | 0          | 0          |   0.0 |  0.00
Comm    | 0.011076   | 0.011076   | 0.011076   |   0.0 |  3.85
Output  | 7.76e-06   | 7.76e-06   | 7.76e-06   |   0.0 |  0.00
Modify  | 0.008371   | 0.008371   | 0.008371   |   0.0 |  2.91
Other   |            | 0.268      |            |       | 93.23

Thank you in advance

2

Very obviously, everything that is not in any of the other categories.

No.

Your simulation is not running for a sufficient amount of time to make any meaningful assessment. Since your quoted input and output is truncated, it is impossible to say more.

3

Thank you, I looked more and figured that I had “post no” which turned off the stats after the run. Now I have a more meaningful result.

I think I have to figure out how to speed up the output.

Loop time of 2.92343 on 128 procs for 1000 steps with 671089 atoms

Performance: 29.554 tau/day, 342.064 timesteps/s, 229.555 Matom-step/s
73.0% CPU use with 128 MPI tasks x 1 OpenMP threads

MPI task timing breakdown:
Section |  min time  |  avg time  |  max time  |%varavg| %total
---------------------------------------------------------------
Pair    | 0.14962    | 0.15759    | 0.16574    |   0.7 |  5.39
Neigh   | 0.0034633  | 0.0035771  | 0.003699   |   0.1 |  0.12
Comm    | 0.023559   | 0.026872   | 0.030202   |   1.0 |  0.92
Output  | 1.5713     | 1.6684     | 1.7637     |   4.5 | 57.07
Modify  | 0.83133    | 0.84841    | 0.93337    |   1.1 | 29.02
Other   |            | 0.2186     |            |       |  7.48

Nlocal:        5242.88 ave        5310 max        5171 min
Histogram: 2 1 9 18 30 32 18 10 4 4
Nghost:        379.836 ave         396 max         352 min
Histogram: 1 1 5 6 14 30 21 24 14 12
Neighs:        11007.5 ave       11293 max       10608 min
Histogram: 1 2 2 8 27 33 24 18 8 5
FullNghs:      21643.6 ave       22236 max       20854 min
Histogram: 1 2 2 11 28 32 25 18 5 4
4

There is not much that can be done to speed up output except for writing output less frequently.