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

git-svn-id: svn://svn.icms.temple.edu/lammps-ro/trunk@15178 f3b2605a-c512-4ea7-a41b-209d697bcdaa

Browse Source
This commit is contained in:
sjplimp
2016-06-14 22:43:51 +00:00
parent c2c73fd8a3
commit 94242eb591
4 changed files with 37 additions and 37 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
doc/html/_sources/pair_mgpt.txt
View File

@ -103,15 +103,15 @@ compiler and compiler switches (-msse3 on x86, and using the IBM XL
compiler on BG/Q), these optimized routines are used automatically.
For BG/Q machines, building with the default Makefile for that
architecture (e.g., "make bgq") should enable the optimized algebra
routines. For x-86 machines, the here provided Makefile.mpi_fastmgpt
(build with "make mpi_fastmgpt") enables the fast algebra routines.
The user will be informed in the output files of the matrix kernels in
use. To further improve speed, on x86 the option precision single can
be added to the :doc:`pair_coeff <pair_coeff>` command line, which
improves speed (up to a factor of two) at the cost of doing matrix
calculations with 7 digit precision instead of the default 16. For
consistency the default option can be specified explicitly by the
option precision double.
routines. For x-86 machines, there is a provided Makefile.mgptfast
which enables the fast algebra routines, i.e. build LAMMPS with "make
mgptfast". The user will be informed in the output files of the
matrix kernels in use. To further improve speed, on x86 the option
precision single can be added to the :doc:`pair_coeff <pair_coeff>`
command line, which improves speed (up to a factor of two) at the cost
of doing matrix calculations with 7 digit precision instead of the
default 16. For consistency the default option can be specified
explicitly by the option precision double.
All remaining potential and control data are contained with the parmin
and potin files, including cutoffs, atomic mass, and other basic MGPT
@ -138,9 +138,9 @@ before proceeding to more complex simulations.
For good performance, LAMMPS should be built with the compiler
flags "-O3 -msse3 -funroll-loops" when including this pair style. The
src/MAKE/OPTIONS/Makefile.mpi_fastmgpt is an example machine Makefile
with these options included as part of a standard MPI build. Note
that as-is it will build with whatever low-level compiler (g++, icc,
src/MAKE/OPTIONS/Makefile.mgptfast is an example machine Makefile with
these options included as part of a standard MPI build. Note that it
as provided, it will build with whatever low-level compiler (g++, icc,
etc) is the default for your MPI installation.

24
doc/html/pair_mgpt.html
View File

@ -216,15 +216,15 @@ compiler and compiler switches (-msse3 on x86, and using the IBM XL
compiler on BG/Q), these optimized routines are used automatically.
For BG/Q machines, building with the default Makefile for that
architecture (e.g., &#8220;make bgq&#8221;) should enable the optimized algebra
routines. For x-86 machines, the here provided Makefile.mpi_fastmgpt
(build with &#8220;make mpi_fastmgpt&#8221;) enables the fast algebra routines.
The user will be informed in the output files of the matrix kernels in
use. To further improve speed, on x86 the option precision single can
be added to the <a class="reference internal" href="pair_coeff.html"><span class="doc">pair_coeff</span></a> command line, which
improves speed (up to a factor of two) at the cost of doing matrix
calculations with 7 digit precision instead of the default 16. For
consistency the default option can be specified explicitly by the
option precision double.</p>
routines. For x-86 machines, there is a provided Makefile.mgptfast
which enables the fast algebra routines, i.e. build LAMMPS with &#8220;make
mgptfast&#8221;. The user will be informed in the output files of the
matrix kernels in use. To further improve speed, on x86 the option
precision single can be added to the <a class="reference internal" href="pair_coeff.html"><span class="doc">pair_coeff</span></a>
command line, which improves speed (up to a factor of two) at the cost
of doing matrix calculations with 7 digit precision instead of the
default 16. For consistency the default option can be specified
explicitly by the option precision double.</p>
<p>All remaining potential and control data are contained with the parmin
and potin files, including cutoffs, atomic mass, and other basic MGPT
variables. Specific MGPT potential data for the transition metals
@ -248,9 +248,9 @@ before proceeding to more complex simulations.</p>
<p class="first admonition-title">Note</p>
<p class="last">For good performance, LAMMPS should be built with the compiler
flags &#8220;-O3 -msse3 -funroll-loops&#8221; when including this pair style. The
src/MAKE/OPTIONS/Makefile.mpi_fastmgpt is an example machine Makefile
with these options included as part of a standard MPI build. Note
that as-is it will build with whatever low-level compiler (g++, icc,
src/MAKE/OPTIONS/Makefile.mgptfast is an example machine Makefile with
these options included as part of a standard MPI build. Note that it
as provided, it will build with whatever low-level compiler (g++, icc,
etc) is the default for your MPI installation.</p>
</div>
<hr class="docutils" />

2
doc/html/searchindex.js
View File

File diff suppressed because one or more lines are too long