git-svn-id: svn://svn.icms.temple.edu/lammps-ro/trunk@9726 f3b2605a-c512-4ea7-a41b-209d697bcdaa

This commit is contained in:
sjplimp 2013-04-02 15:15:35 +00:00
parent 9b753bd19f
commit 45d187463d
4 changed files with 94 additions and 18 deletions

View File

@ -333,15 +333,21 @@ command can be used to insure all timestep numbers are the same length
(e.g. 00010), which can make it easier to read a series of dump files
in order by some post-processing tools.
</P>
<P>If a "%" character appears in the filename, then one file is written
for each processor and the "%" character is replaced with the
processor ID from 0 to P-1. For example, tmp.dump.% becomes
<P>If a "%" character appears in the filename, then each of P processors
writes a portion of the dump file, and the "%" character is replaced
with the processor ID from 0 to P-1. For example, tmp.dump.% becomes
tmp.dump.0, tmp.dump.1, ... tmp.dump.P-1, etc. This creates smaller
files and can be a fast mode of output on parallel machines that
support parallel I/O for output. This option is not available for the
<I>dcd</I>, <I>xtc</I>, and <I>xyz</I> styles.
</P>
<P>Note that the "*" and "%" characters can be used together to produce a
<P>By default, P = the number of processors meaning one file per
processor, but P can be set to a smaller value via the <I>nfile</I> or
<I>fileper</I> keywords of the <A HREF = "dump_modify.html">dump_modify</A> command.
These options can be the most efficient way of writing out dump files
when running on large numbers of processors.
</P>
<P>Note that using the "*" and "%" characters together can produce a
large number of small dump files!
</P>
<P>If the filename ends with ".bin", the dump file (or files, if "*" or

View File

@ -321,15 +321,21 @@ command can be used to insure all timestep numbers are the same length
(e.g. 00010), which can make it easier to read a series of dump files
in order by some post-processing tools.
If a "%" character appears in the filename, then one file is written
for each processor and the "%" character is replaced with the
processor ID from 0 to P-1. For example, tmp.dump.% becomes
If a "%" character appears in the filename, then each of P processors
writes a portion of the dump file, and the "%" character is replaced
with the processor ID from 0 to P-1. For example, tmp.dump.% becomes
tmp.dump.0, tmp.dump.1, ... tmp.dump.P-1, etc. This creates smaller
files and can be a fast mode of output on parallel machines that
support parallel I/O for output. This option is not available for the
{dcd}, {xtc}, and {xyz} styles.
Note that the "*" and "%" characters can be used together to produce a
By default, P = the number of processors meaning one file per
processor, but P can be set to a smaller value via the {nfile} or
{fileper} keywords of the "dump_modify"_dump_modify.html command.
These options can be the most efficient way of writing out dump files
when running on large numbers of processors.
Note that using the "*" and "%" characters together can produce a
large number of small dump files!
If the filename ends with ".bin", the dump file (or files, if "*" or

View File

@ -66,12 +66,16 @@
<I>every</I> arg = N
N = dump every this many timesteps
N can be a variable (see below)
<I>fileper</I> arg = Np
Np = write one file for every this many processors
<I>first</I> arg = <I>yes</I> or <I>no</I>
<I>format</I> arg = C-style format string for one line of output
<I>flush</I> arg = <I>yes</I> or <I>no</I>
<I>image</I> arg = <I>yes</I> or <I>no</I>
<I>label</I> arg = string
string = character string (e.g. BONDS) to use in header of dump local file
<I>nfile</I> arg = Nf
Nf = write this many files, one from each of Nf processors
<I>pad</I> arg = Nchar = # of characters to convert timestep to
<I>precision</I> arg = power-of-10 value from 10 to 1000000
<I>region</I> arg = region-ID or "none"
@ -96,7 +100,7 @@
dump_modify myDump image yes scale no flush yes
dump_modify 1 region mySphere thresh x < 0.0 thresh epair >= 3.2
dump_modify xtcdump precision 10000
dump_modify 1 every 1000
dump_modify 1 every 1000 nfile 20
dump_modify 1 every v_myVar
dump_modify 1 amap min max cf 0.0 3 min green 0.5 yellow max blue boxcolor red
</PRE>
@ -436,6 +440,8 @@ the first two fields (atom id and type) are not actually written into
the CFG file, though you must include formats for them in the format
string.
</P>
<P>The <I>fileper</I> keyword is documented below with the <I>nfile</I> keyword.
</P>
<HR>
<P>The <I>image</I> keyword applies only to the dump <I>atom</I> style. If the
@ -463,6 +469,28 @@ e.g. BONDS or ANGLES.
</P>
<HR>
<P>The <I>nfile</I> or <I>fileper</I> keywords can be used in conjunction with the
"%" wildcard character in the specified dump file name, for all dump
styles except the <I>dcd</I>, <I>xtc</I>, and <I>xyz</I> styles (for which "%" is not
allowed). As explained on the <A HREF = "dump.html">dump</A> command doc page, the
"%" character causes the dump file to be written in pieces, one piece
for each of P processors. By default P = the number of processors the
simulation is running on. The <I>nfile</I> or <I>fileper</I> keyword can be
used to set P to a smaller value, which can be more efficient when
running on a large number of processors.
</P>
<P>The <I>nfile</I> keyword sets P to the specified Nf value. For example, if
Nf = 4, and the simulation is running on 100 processors, 4 files will
be written, by processors 0,25,50,75. Each will collect information
from itself and the next 24 processors and write it to a dump file.
</P>
<P>For the <I>fileper</I> keyword, the specified value of Np means write one
file for every Np processors. For example, if Np = 4, every 4th
processor (0,4,8,12,etc) will collect information from itself and the
next 3 processors and write it to a dump file.
</P>
<HR>
<P>The <I>pad</I> keyword only applies when the dump filename is specified
with a wildcard "*" character which becomes the timestep. If <I>pad</I> is
0, which is the default, the timestep is converted into a string of
@ -509,10 +537,14 @@ sort</A> option is turned on, which it is by default, to
improve performance. A sort value of <I>id</I> means sort the output by
atom ID. A sort value of N or -N means sort the output by the value
in the Nth column of per-atom info in either ascending or descending
order. The dump <I>local</I> style cannot be sorted by atom ID, since
there are typically multiple lines of output per atom. Some dump
styles, such as <I>dcd</I> and <I>xtc</I>, require sorting by atom ID to format
the output file correctly.
order.
</P>
<P>The dump <I>local</I> style cannot be sorted by atom ID, since there are
typically multiple lines of output per atom. Some dump styles, such
as <I>dcd</I> and <I>xtc</I>, require sorting by atom ID to format the output
file correctly. If multiple processors are writing the dump file, via
the "%" wildcard in the dump filename, then sorting cannot be
performed.
</P>
<P>IMPORTANT NOTE: Unless it is required by the dump style, sorting dump
file output requires extra overhead in terms of CPU and communication

View File

@ -60,12 +60,16 @@ keyword = {acolor} or {adiam} or {amap} or {append} or {bcolor} or {bdiam} or {b
{every} arg = N
N = dump every this many timesteps
N can be a variable (see below)
{fileper} arg = Np
Np = write one file for every this many processors
{first} arg = {yes} or {no}
{format} arg = C-style format string for one line of output
{flush} arg = {yes} or {no}
{image} arg = {yes} or {no}
{label} arg = string
string = character string (e.g. BONDS) to use in header of dump local file
{nfile} arg = Nf
Nf = write this many files, one from each of Nf processors
{pad} arg = Nchar = # of characters to convert timestep to
{precision} arg = power-of-10 value from 10 to 1000000
{region} arg = region-ID or "none"
@ -89,7 +93,7 @@ dump_modify 1 format "%d %d %20.15g %g %g" scale yes
dump_modify myDump image yes scale no flush yes
dump_modify 1 region mySphere thresh x < 0.0 thresh epair >= 3.2
dump_modify xtcdump precision 10000
dump_modify 1 every 1000
dump_modify 1 every 1000 nfile 20
dump_modify 1 every v_myVar
dump_modify 1 amap min max cf 0.0 3 min green 0.5 yellow max blue boxcolor red :pre
@ -429,6 +433,8 @@ the first two fields (atom id and type) are not actually written into
the CFG file, though you must include formats for them in the format
string.
The {fileper} keyword is documented below with the {nfile} keyword.
:line
The {image} keyword applies only to the dump {atom} style. If the
@ -456,6 +462,28 @@ e.g. BONDS or ANGLES.
:line
The {nfile} or {fileper} keywords can be used in conjunction with the
"%" wildcard character in the specified dump file name, for all dump
styles except the {dcd}, {xtc}, and {xyz} styles (for which "%" is not
allowed). As explained on the "dump"_dump.html command doc page, the
"%" character causes the dump file to be written in pieces, one piece
for each of P processors. By default P = the number of processors the
simulation is running on. The {nfile} or {fileper} keyword can be
used to set P to a smaller value, which can be more efficient when
running on a large number of processors.
The {nfile} keyword sets P to the specified Nf value. For example, if
Nf = 4, and the simulation is running on 100 processors, 4 files will
be written, by processors 0,25,50,75. Each will collect information
from itself and the next 24 processors and write it to a dump file.
For the {fileper} keyword, the specified value of Np means write one
file for every Np processors. For example, if Np = 4, every 4th
processor (0,4,8,12,etc) will collect information from itself and the
next 3 processors and write it to a dump file.
:line
The {pad} keyword only applies when the dump filename is specified
with a wildcard "*" character which becomes the timestep. If {pad} is
0, which is the default, the timestep is converted into a string of
@ -502,10 +530,14 @@ sort"_atom_modify.html option is turned on, which it is by default, to
improve performance. A sort value of {id} means sort the output by
atom ID. A sort value of N or -N means sort the output by the value
in the Nth column of per-atom info in either ascending or descending
order. The dump {local} style cannot be sorted by atom ID, since
there are typically multiple lines of output per atom. Some dump
styles, such as {dcd} and {xtc}, require sorting by atom ID to format
the output file correctly.
order.
The dump {local} style cannot be sorted by atom ID, since there are
typically multiple lines of output per atom. Some dump styles, such
as {dcd} and {xtc}, require sorting by atom ID to format the output
file correctly. If multiple processors are writing the dump file, via
the "%" wildcard in the dump filename, then sorting cannot be
performed.
IMPORTANT NOTE: Unless it is required by the dump style, sorting dump
file output requires extra overhead in terms of CPU and communication