ICODE-ADC/lammps
4
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix spelling error and reformat paragraph

Browse Source
This commit is contained in:
Axel Kohlmeyer
2021-08-29 22:08:49 -04:00
parent 5d03028028
commit 284ed98fb8
1 changed files with 13 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
doc/src/pair_snap.rst
View File

@ -191,22 +191,19 @@ pair_coeff command, to avoid ambiguity in the number of coefficients.
The keywords *chunksize* and *parallelthresh* are only applicable when The keywords *chunksize* and *parallelthresh* are only applicable when
using the pair style *snap* with the KOKKOS package on GPUs and are using the pair style *snap* with the KOKKOS package on GPUs and are
ignored otherwise. ignored otherwise. The *chunksize* keyword controls the number of atoms
The *chunksize* keyword controls in each pass used to compute the bispectrum components and is used to
the number of atoms in each pass used to compute the bispectrum avoid running out of memory. For example if there are 8192 atoms in the
components and is used to avoid running out of memory. For example simulation and the *chunksize* is set to 4096, the bispectrum
if there are 8192 atoms in the simulation and the *chunksize* calculation will be broken up into two passes (running on a single GPU).
is set to 4096, the bispectrum calculation will be broken up The *parallelthresh* keyword controls a crossover threshold for
into two passes (running on a single GPU). performing extra parallelism. For small systems, exposing additional
The *parallelthresh* keyword controls parallelism can be beneficial when there is not enough work to fully
a crossover threshold for performing extra parallelism. For saturate the GPU threads otherwise. However, the extra parallelism also
small systems, exposing additional parallism can be beneficial when leads to more divergence and can hurt performance when the system is
there is not enough work to fully saturate the GPU threads otherwise. already large enough to saturate the GPU threads. Extra parallelism
However, the extra parallelism also leads to more divergence will be performed if the *chunksize* (or total number of atoms per GPU)
and can hurt performance when the system is already large enough to is smaller than *parallelthresh*.
saturate the GPU threads. Extra parallelism will be performed if the
*chunksize* (or total number of atoms per GPU) is smaller than
*parallelthresh*.
Detailed definitions for all the other keywords Detailed definitions for all the other keywords
are given on the :doc:`compute sna/atom <compute_sna_atom>` doc page. are given on the :doc:`compute sna/atom <compute_sna_atom>` doc page.