jianmu
/
lammps
forked from lijiext/lammps
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

git-svn-id: svn://svn.icms.temple.edu/lammps-ro/trunk@8812 f3b2605a-c512-4ea7-a41b-209d697bcdaa

This commit is contained in:
sjplimp 2012-09-21 16:04:48 +00:00
parent 77baebf99a
commit 110a7d3dcc
6 changed files with 169 additions and 30 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

40
doc/package.html
View File

@ -25,22 +25,24 @@
last = ID of last GPU to be used on each node
split = fraction of particles assigned to the GPU
zero or more keyword/value pairs may be appended
keywords = <I>threads_per_atom</I>
<I>threads_per_atom</I> value = Nthreads
Nthreads = # of GPU threads used per atom
keywords = <I>threads_per_atom</I> or <I>cellsize</I>
<I>threads_per_atom</I> value = Nthreads
Nthreads = # of GPU threads used per atom
<I>cellsize</I> value = dist
dist = length (distance units) in each dimension for neighbor bins
<I>cuda</I> args = keyword value ...
one or more keyword/value pairs may be appended
keywords = <I>gpu/node</I> or <I>gpu/node/special</I> or <I>timing</I> or <I>test</I> or <I>override/bpa</I>
<I>gpu/node</I> value = N
N = number of GPUs to be used per node
<I>gpu/node/special</I> values = N gpu1 .. gpuN
N = number of GPUs to be used per node
gpu1 .. gpuN = N IDs of the GPUs to use
<I>timing</I> values = none
<I>test</I> values = id
id = atom-ID of a test particle
<I>override/bpa</I> values = flag
flag = 0 for TpA algorithm, 1 for BpA algorithm
<I>gpu/node</I> value = N
N = number of GPUs to be used per node
<I>gpu/node/special</I> values = N gpu1 .. gpuN
N = number of GPUs to be used per node
gpu1 .. gpuN = N IDs of the GPUs to use
<I>timing</I> values = none
<I>test</I> values = id
id = atom-ID of a test particle
<I>override/bpa</I> values = flag
flag = 0 for TpA algorithm, 1 for BpA algorithm
<I>omp</I> args = Nthreads mode
Nthreads = # of OpenMP threads to associate with each MPI process
mode = force or force/neigh (optional)
@ -133,6 +135,18 @@ large cutoffs or with a small number of particles per GPU, increasing
the value can improve performance. The number of threads per atom must
be a power of 2 and currently cannot be greater than 32.
</P>
<P>The <I>cellsize</I> keyword can be used to control the size of the cells used
for binning atoms in neighbor list calculations. Setting this value is
normally not needed; the optimal value is close to the default
(equal to the cutoff distance for the short range interactions
plus the neighbor skin). GPUs can perform efficiently with much larger cutoffs
than CPUs and this can be used to reduce the time required for long-range
calculations or in some cases to eliminate them with models such as
<A HREF = "pair_coul.html">coul/wolf</A> or <A HREF = "pair_coul.html">coul/dsf</A>. For very large cutoffs,
it can be more efficient to use smaller values for cellsize in parallel
simulations. For example, with a cutoff of 20*sigma and a neighbor skin of
sigma, a cellsize of 5.25*sigma can be efficient for parallel simulations.
</P>
<HR>
<P>The <I>cuda</I> style invokes options associated with the use of the

40
doc/package.txt
View File

@ -20,22 +20,24 @@ args = arguments specific to the style :l
last = ID of last GPU to be used on each node
split = fraction of particles assigned to the GPU
zero or more keyword/value pairs may be appended
keywords = {threads_per_atom}
{threads_per_atom} value = Nthreads
Nthreads = # of GPU threads used per atom
keywords = {threads_per_atom} or {cellsize}
{threads_per_atom} value = Nthreads
Nthreads = # of GPU threads used per atom
{cellsize} value = dist
dist = length (distance units) in each dimension for neighbor bins
{cuda} args = keyword value ...
one or more keyword/value pairs may be appended
keywords = {gpu/node} or {gpu/node/special} or {timing} or {test} or {override/bpa}
{gpu/node} value = N
N = number of GPUs to be used per node
{gpu/node/special} values = N gpu1 .. gpuN
N = number of GPUs to be used per node
gpu1 .. gpuN = N IDs of the GPUs to use
{timing} values = none
{test} values = id
id = atom-ID of a test particle
{override/bpa} values = flag
flag = 0 for TpA algorithm, 1 for BpA algorithm
{gpu/node} value = N
N = number of GPUs to be used per node
{gpu/node/special} values = N gpu1 .. gpuN
N = number of GPUs to be used per node
gpu1 .. gpuN = N IDs of the GPUs to use
{timing} values = none
{test} values = id
id = atom-ID of a test particle
{override/bpa} values = flag
flag = 0 for TpA algorithm, 1 for BpA algorithm
{omp} args = Nthreads mode
Nthreads = # of OpenMP threads to associate with each MPI process
mode = force or force/neigh (optional) :pre
@ -127,6 +129,18 @@ large cutoffs or with a small number of particles per GPU, increasing
the value can improve performance. The number of threads per atom must
be a power of 2 and currently cannot be greater than 32.
The {cellsize} keyword can be used to control the size of the cells used
for binning atoms in neighbor list calculations. Setting this value is
normally not needed; the optimal value is close to the default
(equal to the cutoff distance for the short range interactions
plus the neighbor skin). GPUs can perform efficiently with much larger cutoffs
than CPUs and this can be used to reduce the time required for long-range
calculations or in some cases to eliminate them with models such as
"coul/wolf"_pair_coul.html or "coul/dsf"_pair_coul.html. For very large cutoffs,
it can be more efficient to use smaller values for cellsize in parallel
simulations. For example, with a cutoff of 20*sigma and a neighbor skin of
sigma, a cellsize of 5.25*sigma can be efficient for parallel simulations.
:line
The {cuda} style invokes options associated with the use of the

27
doc/pair_coul.html
View File

@ -17,6 +17,10 @@
</H3>
<H3>pair_style coul/debye/omp command
</H3>
<H3>pair_style coul/dsf command
</H3>
<H3>pair_style coul/dsf/gpu command
</H3>
<H3>pair_style coul/long command
</H3>
<H3>pair_style coul/long/omp command
@ -31,6 +35,7 @@
</P>
<PRE>pair_style coul/cut cutoff
pair_style coul/debye kappa cutoff
pair_style coul/dsf alpha cutoff
pair_style coul/long cutoff
pair_style coul/long/gpu cutoff
pair_style coul/wolf alpha cutoff
@ -49,6 +54,9 @@ pair_coeff 2 2 3.5
pair_coeff * *
pair_coeff 2 2 3.5
</PRE>
<PRE>pair_style coul/dsf 0.05 10.0
pair_coeff * *
</PRE>
<PRE>pair_style coul/long 10.0
pair_coeff * *
</PRE>
@ -75,6 +83,17 @@ Coulombic term, given by
<P>where kappa is the Debye length. This potential is another way to
mimic the screening effect of a polar solvent.
</P>
<P>Style <I>coul/dsf</I> computes Coulombic interactions via the damped
shifted force model described in <A HREF = "#Fennell">Fennell</A>, given by:
</P>
<CENTER><IMG SRC = "Eqs/pair_coul_dsf.jpg">
</CENTER>
<P>where <I>alpha</I> is the damping parameter and erfc() is the
complementary error-function. The potential corrects issues in the
Wolf model (described below) to provide consistent forces and energies
(the Wolf potential is not differentiable at the cutoff) and smooth
decay to zero.
</P>
<P>Style <I>coul/wolf</I> computes Coulombic interactions via the Wolf
summation method, described in <A HREF = "#Wolf">Wolf</A>, given by:
</P>
@ -193,5 +212,11 @@ hybrid/overlay</A>
<A NAME = "Wolf"></A>
<P><B>(Wolf)</B> D. Wolf, P. Keblinski, S. R. Phillpot, J. Eggebrecht, J Chem
Phys, 110, 8254 (1999).</P>
Phys, 110, 8254 (1999).
</P>
<A NAME = "Fennell"></A>
<P><B>(Fennell)</B> C. J. Fennell, J. D. Gezelter, J Chem Phys, 124,
234104 (2006).
</P>
</HTML>

23
doc/pair_coul.txt
View File

@ -10,6 +10,8 @@ pair_style coul/cut command :h3
pair_style coul/cut/omp command :h3
pair_style coul/debye command :h3
pair_style coul/debye/omp command :h3
pair_style coul/dsf command :h3
pair_style coul/dsf/gpu command :h3
pair_style coul/long command :h3
pair_style coul/long/omp command :h3
pair_style coul/long/gpu command :h3
@ -20,6 +22,7 @@ pair_style coul/wolf/omp command :h3
pair_style coul/cut cutoff
pair_style coul/debye kappa cutoff
pair_style coul/dsf alpha cutoff
pair_style coul/long cutoff
pair_style coul/long/gpu cutoff
pair_style coul/wolf alpha cutoff :pre
@ -38,6 +41,9 @@ pair_style coul/debye 1.4 3.0
pair_coeff * *
pair_coeff 2 2 3.5 :pre
pair_style coul/dsf 0.05 10.0
pair_coeff * * :pre
pair_style coul/long 10.0
pair_coeff * * :pre
@ -64,6 +70,17 @@ Coulombic term, given by
where kappa is the Debye length. This potential is another way to
mimic the screening effect of a polar solvent.
Style {coul/dsf} computes Coulombic interactions via the damped
shifted force model described in "Fennell"_#Fennell, given by:
:c,image(Eqs/pair_coul_dsf.jpg)
where {alpha} is the damping parameter and erfc() is the
complementary error-function. The potential corrects issues in the
Wolf model (described below) to provide consistent forces and energies
(the Wolf potential is not differentiable at the cutoff) and smooth
decay to zero.
Style {coul/wolf} computes Coulombic interactions via the Wolf
summation method, described in "Wolf"_#Wolf, given by:
@ -181,4 +198,8 @@ hybrid/overlay"_pair_hybrid.html
:link(Wolf)
[(Wolf)] D. Wolf, P. Keblinski, S. R. Phillpot, J. Eggebrecht, J Chem
Phys, 110, 8254 (1999).
Phys, 110, 8254 (1999).
:link(Fennell)
[(Fennell)] C. J. Fennell, J. D. Gezelter, J Chem Phys, 124,
234104 (2006).

36
doc/pair_lj.html
View File

@ -37,6 +37,10 @@
</H3>
<H3>pair_style lj/cut/coul/debye/omp command
</H3>
<H3>pair_style lj/cut/coul/dsf command
</H3>
<H3>pair_style lj/cut/coul/dsf/gpu command
</H3>
<H3>pair_style lj/cut/coul/long command
</H3>
<H3>pair_style lj/cut/coul/long/cuda command
@ -57,7 +61,7 @@
</P>
<PRE>pair_style style args
</PRE>
<UL><LI>style = <I>lj/cut</I> or <I>lj/cut/coul/cut</I> or <I>lj/cut/coul/debye</I> or <I>lj/cut/coul/long</I> or <I>lj/cut/coul/long/tip4p</I>
<UL><LI>style = <I>lj/cut</I> or <I>lj/cut/coul/cut</I> or <I>lj/cut/coul/debye</I> or <I>lj/cut/coul/dsf</I> or <I>lj/cut/coul/long</I> or <I>lj/cut/coul/long/tip4p</I>
<LI>args = list of arguments for a particular style
</UL>
<PRE> <I>lj/cut</I> args = cutoff
@ -69,6 +73,10 @@
kappa = inverse of the Debye length (inverse distance units)
cutoff = global cutoff for LJ (and Coulombic if only 1 arg) (distance units)
cutoff2 = global cutoff for Coulombic (optional) (distance units)
<I>lj/cut/coul/dsf</I> args = alpha cutoff (cutoff2)
alpha = damping parameter (inverse distance units)
cutoff = global cutoff for LJ (and Coulombic if only 1 arg) (distance units)
cutoff2 = global cutoff for Coulombic (distance units)
<I>lj/cut/coul/long</I> args = cutoff (cutoff2)
cutoff = global cutoff for LJ (and Coulombic if only 1 arg) (distance units)
cutoff2 = global cutoff for Coulombic (optional) (distance units)
@ -97,6 +105,10 @@ pair_coeff * * 1.0 1.0
pair_coeff 1 1 1.0 1.5 2.5
pair_coeff 1 1 1.0 1.5 2.5 5.0
</PRE>
<PRE>pair_style lj/cut/coul/dsf 0.05 2.5 10.0
pair_coeff * * 1.0 1.0
pair_coeff 1 1 1.0 1.0 2.5
</PRE>
<PRE>pair_style lj/cut/coul/long 10.0
pair_style lj/cut/coul/long 10.0 8.0
pair_coeff * * 100.0 3.0
@ -135,6 +147,23 @@ to the Coulombic term, given by
<P>where kappa is the inverse of the Debye length. This potential is
another way to mimic the screening effect of a polar solvent.
</P>
<P>Style <I>lj/cut/coul/dsf</I> computes the Coulombic term via the damped
shifted force model described in <A HREF = "#Fennell">Fennell</A>, given by:
</P>
<CENTER><IMG SRC = "Eqs/pair_coul_dsf.jpg">
</CENTER>
<P>where <I>alpha</I> is the damping parameter and erfc() is the complementary
error-function. This potential is essentially a short-range,
spherically-truncated, charge-neutralized, shifted, pairwise <I>1/r</I>
summation. The potential is based on Wolf summation, proposed as an
alternative to Ewald summation for condensed phase systems where
charge screening causes electrostatic interactions to become
effectively short-ranged. In order for the electrostatic sum to be
absolutely convergent, charge neutralization within the cutoff radius
is enforced by shifting the potential through placement of image
charges on the cutoff sphere. Convergence can often be improved by
setting <I>alpha</I> to a small non-zero value.
</P>
<P>Style <I>lj/cut/coul/long</I> computes the same Coulombic interactions as
style <I>lj/cut/coul/cut</I> except that an additional damping factor is
applied to the Coulombic term so it can be used in conjunction with
@ -283,4 +312,9 @@ default.
<P><B>(Jorgensen)</B> Jorgensen, Chandrasekhar, Madura, Impey, Klein, J Chem
Phys, 79, 926 (1983).
</P>
<A NAME = "Fennell"></A>
<P><B>(Fennell)</B> C. J. Fennell, J. D. Gezelter, J Chem Phys, 124,
234104 (2006).
</P>
</HTML>

33
doc/pair_lj.txt
View File

@ -20,6 +20,8 @@ pair_style lj/cut/coul/debye command :h3
pair_style lj/cut/coul/debye/cuda command :h3
pair_style lj/cut/coul/debye/gpu command :h3
pair_style lj/cut/coul/debye/omp command :h3
pair_style lj/cut/coul/dsf command :h3
pair_style lj/cut/coul/dsf/gpu command :h3
pair_style lj/cut/coul/long command :h3
pair_style lj/cut/coul/long/cuda command :h3
pair_style lj/cut/coul/long/gpu command :h3
@ -33,7 +35,7 @@ pair_style lj/cut/coul/long/tip4p/opt command :h3
pair_style style args :pre
style = {lj/cut} or {lj/cut/coul/cut} or {lj/cut/coul/debye} or {lj/cut/coul/long} or {lj/cut/coul/long/tip4p}
style = {lj/cut} or {lj/cut/coul/cut} or {lj/cut/coul/debye} or {lj/cut/coul/dsf} or {lj/cut/coul/long} or {lj/cut/coul/long/tip4p}
args = list of arguments for a particular style :ul
{lj/cut} args = cutoff
cutoff = global cutoff for Lennard Jones interactions (distance units)
@ -44,6 +46,10 @@ args = list of arguments for a particular style :ul
kappa = inverse of the Debye length (inverse distance units)
cutoff = global cutoff for LJ (and Coulombic if only 1 arg) (distance units)
cutoff2 = global cutoff for Coulombic (optional) (distance units)
{lj/cut/coul/dsf} args = alpha cutoff (cutoff2)
alpha = damping parameter (inverse distance units)
cutoff = global cutoff for LJ (and Coulombic if only 1 arg) (distance units)
cutoff2 = global cutoff for Coulombic (distance units)
{lj/cut/coul/long} args = cutoff (cutoff2)
cutoff = global cutoff for LJ (and Coulombic if only 1 arg) (distance units)
cutoff2 = global cutoff for Coulombic (optional) (distance units)
@ -72,6 +78,10 @@ pair_coeff * * 1.0 1.0
pair_coeff 1 1 1.0 1.5 2.5
pair_coeff 1 1 1.0 1.5 2.5 5.0 :pre
pair_style lj/cut/coul/dsf 0.05 2.5 10.0
pair_coeff * * 1.0 1.0
pair_coeff 1 1 1.0 1.0 2.5 :pre
pair_style lj/cut/coul/long 10.0
pair_style lj/cut/coul/long 10.0 8.0
pair_coeff * * 100.0 3.0
@ -110,6 +120,23 @@ to the Coulombic term, given by
where kappa is the inverse of the Debye length. This potential is
another way to mimic the screening effect of a polar solvent.
Style {lj/cut/coul/dsf} computes the Coulombic term via the damped
shifted force model described in "Fennell"_#Fennell, given by:
:c,image(Eqs/pair_coul_dsf.jpg)
where {alpha} is the damping parameter and erfc() is the complementary
error-function. This potential is essentially a short-range,
spherically-truncated, charge-neutralized, shifted, pairwise {1/r}
summation. The potential is based on Wolf summation, proposed as an
alternative to Ewald summation for condensed phase systems where
charge screening causes electrostatic interactions to become
effectively short-ranged. In order for the electrostatic sum to be
absolutely convergent, charge neutralization within the cutoff radius
is enforced by shifting the potential through placement of image
charges on the cutoff sphere. Convergence can often be improved by
setting {alpha} to a small non-zero value.
Style {lj/cut/coul/long} computes the same Coulombic interactions as
style {lj/cut/coul/cut} except that an additional damping factor is
applied to the Coulombic term so it can be used in conjunction with
@ -256,3 +283,7 @@ default.
:link(Jorgensen)
[(Jorgensen)] Jorgensen, Chandrasekhar, Madura, Impey, Klein, J Chem
Phys, 79, 926 (1983).
:link(Fennell)
[(Fennell)] C. J. Fennell, J. D. Gezelter, J Chem Phys, 124,
234104 (2006).