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

Merge remote-tracking branch 'upstream/master' into python_refactoring

This commit is contained in:
Richard Berger 2017-05-06 16:00:22 -04:00
parent 2e1f8b4aef a73402ad93
commit f8af7edf92
477 changed files with 33719 additions and 21114 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
doc/src/Manual.txt
View File

@ -1,7 +1,7 @@
<!-- HTML_ONLY -->
<HEAD>
<TITLE>LAMMPS Users Manual</TITLE>
<META NAME="docnumber" CONTENT="11 Apr 2017 version">
<META NAME="docnumber" CONTENT="4 May 2017 version">
<META NAME="author" CONTENT="http://lammps.sandia.gov - Sandia National Laboratories">
<META NAME="copyright" CONTENT="Copyright (2003) Sandia Corporation. This software and manual is distributed under the GNU General Public License.">
</HEAD>
@ -21,7 +21,7 @@
<H1></H1>
LAMMPS Documentation :c,h3
11 Apr 2017 version :c,h4
4 May 2017 version :c,h4
Version info: :h4
@ -158,12 +158,11 @@ END_RST -->
2.1 "What's in the LAMMPS distribution"_start_1 :ulb,b
2.2 "Making LAMMPS"_start_2 :b
2.3 "Making LAMMPS with optional packages"_start_3 :b
2.4 "Building LAMMPS via the Make.py script"_start_4 :b
2.5 "Building LAMMPS as a library"_start_5 :b
2.6 "Running LAMMPS"_start_6 :b
2.7 "Command-line options"_start_7 :b
2.8 "Screen output"_start_8 :b
2.9 "Tips for users of previous versions"_start_9 :ule,b
2.4 "Building LAMMPS as a library"_start_4 :b
2.5 "Running LAMMPS"_start_5 :b
2.6 "Command-line options"_start_6 :b
2.7 "Screen output"_start_7 :b
2.8 "Tips for users of previous versions"_start_8 :ule,b
"Commands"_Section_commands.html :l
3.1 "LAMMPS input script"_cmd_1 :ulb,b
3.2 "Parsing rules"_cmd_2 :b

2
doc/src/Section_commands.txt
View File

@ -1052,7 +1052,7 @@ package"_Section_start.html#start_3.
"oxdna2/excv"_pair_oxdna2.html,
"oxdna2/stk"_pair_oxdna2.html,
"quip"_pair_quip.html,
"reax/c (k)"_pair_reax_c.html,
"reax/c (k)"_pair_reaxc.html,
"smd/hertz"_pair_smd_hertz.html,
"smd/tlsph"_pair_smd_tlsph.html,
"smd/triangulated/surface"_pair_smd_triangulated_surface.html,

6
doc/src/Section_errors.txt
View File

@ -11171,6 +11171,12 @@ Self-explanatory. :dd
If the fix changes the timestep, the dump dcd file will not
reflect the change. :dd
{Energy due to X extra global DOFs will be included in minimizer energies} :dt
When using fixes like box/relax, the potential energy used by the minimizer
is augmented by an additional energy provided by the fix. Thus the printed
converged energy may be different from the total potential energy. :dd
{Energy tally does not account for 'zero yes'} :dt
The energy removed by using the 'zero yes' flag is not accounted

10
doc/src/Section_intro.txt
View File

@ -249,8 +249,12 @@ Pizza.py WWW site"_pizza. :l
Specialized features :h5
These are LAMMPS capabilities which you may not think of as typical
molecular dynamics options:
LAMMPS can be built with optional packages which implement a variety
of additional capabilities. An overview of all the packages is "given
here"_Section_packages.html.
These are some LAMMPS capabilities which you may not think of as
typical classical molecular dynamics options:
"static"_balance.html and "dynamic load-balancing"_fix_balance.html
"generalized aspherical particles"_body.html
@ -515,7 +519,7 @@ the packages they have written are somewhat unique to LAMMPS and the
code would not be as general-purpose as it is without their expertise
and efforts.
Axel Kohlmeyer (Temple U), akohlmey at gmail.com, SVN and Git repositories, indefatigable mail list responder, USER-CG-CMM and USER-OMP packages
Axel Kohlmeyer (Temple U), akohlmey at gmail.com, SVN and Git repositories, indefatigable mail list responder, USER-CGSDK and USER-OMP packages
Roy Pollock (LLNL), Ewald and PPPM solvers
Mike Brown (ORNL), brownw at ornl.gov, GPU package
Greg Wagner (Sandia), gjwagne at sandia.gov, MEAM package for MEAM potential

3173
doc/src/Section_packages.txt
View File

File diff suppressed because it is too large Load Diff

534
doc/src/Section_start.txt
View File

@ -14,12 +14,11 @@ experienced users.
2.1 "What's in the LAMMPS distribution"_#start_1
2.2 "Making LAMMPS"_#start_2
2.3 "Making LAMMPS with optional packages"_#start_3
2.4 "Building LAMMPS via the Make.py script"_#start_4
2.5 "Building LAMMPS as a library"_#start_5
2.6 "Running LAMMPS"_#start_6
2.7 "Command-line options"_#start_7
2.8 "Screen output"_#start_8
2.9 "Tips for users of previous versions"_#start_9 :all(b)
2.5 "Building LAMMPS as a library"_#start_4
2.6 "Running LAMMPS"_#start_5
2.7 "Command-line options"_#start_6
2.8 "Screen output"_#start_7
2.9 "Tips for users of previous versions"_#start_8 :all(b)
:line
@ -80,7 +79,7 @@ This section has the following sub-sections:
Read this first :h5,link(start_2_1)
If you want to avoid building LAMMPS yourself, read the preceding
If you want to avoid building LAMMPS yourself, read the preceeding
section about options available for downloading and installing
executables. Details are discussed on the "download"_download page.
@ -96,7 +95,7 @@ make serial :pre
Note that on a facility supercomputer, there are often "modules"
loaded in your environment that provide the compilers and MPI you
should use. In this case, the "mpicxx" compile/link command in
Makefile.mpi should just work by accessing those modules.
Makefile.mpi should simply work by accessing those modules.
It may be the case that one of the other Makefile.machine files in the
src/MAKE sub-directories is a better match to your system (type "make"
@ -107,33 +106,35 @@ make stampede :pre
If any of these builds (with an existing Makefile.machine) works on
your system, then you're done!
If you need to install an optional package with a LAMMPS command you
want to use, and the package does not depend on an extra library, you
can simply type
make name :pre
before invoking (or re-invoking) the above steps. "Name" is the
lower-case name of the package, e.g. replica or user-misc.
If you want to do one of the following:
use optional LAMMPS features that require additional libraries
use optional packages that require additional libraries
use optional accelerator packages that require special compiler/linker settings
run on a specialized platform that has its own compilers, settings, or other libs to use :ul
use a LAMMPS command that requires an extra library (e.g. "dump image"_dump_image.html)
build with a package that requires an extra library
build with an accelerator package that requires special compiler/linker settings
run on a machine that has its own compilers, settings, or libraries :ul
then building LAMMPS is more complicated. You may need to find where
auxiliary libraries exist on your machine or install them if they
don't. You may need to build additional libraries that are part of
the LAMMPS package, before building LAMMPS. You may need to edit a
extra libraries exist on your machine or install them if they don't.
You may need to build extra libraries that are included in the LAMMPS
distribution, before building LAMMPS itself. You may need to edit a
Makefile.machine file to make it compatible with your system.
Note that there is a Make.py tool in the src directory that automates
several of these steps, but you still have to know what you are doing.
"Section 2.4"_#start_4 below describes the tool. It is a convenient
way to work with installing/un-installing various packages, the
Makefile.machine changes required by some packages, and the auxiliary
libraries some of them use.
Please read the following sections carefully. If you are not
comfortable with makefiles, or building codes on a Unix platform, or
running an MPI job on your machine, please find a local expert to help
you. Many compilation, linking, and run problems that users have are
often not really LAMMPS issues - they are peculiar to the user's
system, compilers, libraries, etc. Such questions are better answered
by a local expert.
you. Many compilation, linking, and run problems users experience are
often not LAMMPS issues - they are peculiar to the user's system,
compilers, libraries, etc. Such questions are better answered by a
local expert.
If you have a build problem that you are convinced is a LAMMPS issue
(e.g. the compiler complains about a line of LAMMPS source code), then
@ -251,7 +252,7 @@ re-compile, after typing "make clean" (which will describe different
clean options).
The LMP_INC variable is used to include options that turn on ifdefs
within the LAMMPS code. The options that are currently recognized are:
within the LAMMPS code. The options that are currently recogized are:
-DLAMMPS_GZIP
-DLAMMPS_JPEG
@ -362,7 +363,7 @@ installed on your platform. If MPI is installed on your system in the
usual place (under /usr/local), you also may not need to specify these
3 variables, assuming /usr/local is in your path. On some large
parallel machines which use "modules" for their compile/link
environments, you may simply need to include the correct module in
environements, you may simply need to include the correct module in
your build environment, before building LAMMPS. Or the parallel
machine may have a vendor-provided MPI which the compiler has no
trouble finding.
@ -430,7 +431,7 @@ use the KISS library described above.
You may also need to set the FFT_INC, FFT_PATH, and FFT_LIB variables,
so the compiler and linker can find the needed FFT header and library
files. Note that on some large parallel machines which use "modules"
for their compile/link environments, you may simply need to include
for their compile/link environements, you may simply need to include
the correct module in your build environment. Or the parallel machine
may have a vendor-provided FFT library which the compiler has no
trouble finding.
@ -450,12 +451,13 @@ you must also manually specify the correct library, namely -lsfftw or
The FFT_INC variable also allows for a -DFFT_SINGLE setting that will
use single-precision FFTs with PPPM, which can speed-up long-range
calculations, particularly in parallel or on GPUs. Fourier transform
calulations, particularly in parallel or on GPUs. Fourier transform
and related PPPM operations are somewhat insensitive to floating point
truncation errors and thus do not always need to be performed in
double precision. Using the -DFFT_SINGLE setting trades off a little
accuracy for reduced memory use and parallel communication costs for
transposing 3d FFT data.
transposing 3d FFT data. Note that single precision FFTs have only
been tested with the FFTW3, FFTW2, MKL, and KISS FFT options.
Step 7 :h6
@ -507,13 +509,13 @@ You should get the executable lmp_foo when the build is complete.
Errors that can occur when making LAMMPS: h5 :link(start_2_3)
NOTE: If an error occurs when building LAMMPS, the compiler or linker
will state very explicitly what the problem is. The error message
should give you a hint as to which of the steps above has failed, and
what you need to do in order to fix it. Building a code with a
Makefile is a very logical process. The compiler and linker need to
find the appropriate files and those files need to be compatible with
LAMMPS source files. When a make fails, there is usually a very
If an error occurs when building LAMMPS, the compiler or linker will
state very explicitly what the problem is. The error message should
give you a hint as to which of the steps above has failed, and what
you need to do in order to fix it. Building a code with a Makefile is
a very logical process. The compiler and linker need to find the
appropriate files and those files need to be compatible with LAMMPS
settings and source files. When a make fails, there is usually a very
simple reason, which you or a local expert will need to fix.
Here are two non-obvious errors that can occur:
@ -556,7 +558,8 @@ Typing "make clean-all" or "make clean-machine" will delete *.o object
files created when LAMMPS is built, for either all builds or for a
particular machine.
Changing the LAMMPS size limits via -DLAMMPS_SMALLBIG or -DLAMMPS_BIGBIG or -DLAMMPS_SMALLSMALL :h6
Changing the LAMMPS size limits via -DLAMMPS_SMALLBIG or
-DLAMMPS_BIGBIG or -DLAMMPS_SMALLSMALL :h6
As explained above, any of these 3 settings can be specified on the
LMP_INC line in your low-level src/MAKE/Makefile.foo.
@ -655,11 +658,6 @@ This section has the following sub-sections:
2.3.3 "Packages that require extra libraries"_#start_3_3
2.3.4 "Packages that require Makefile.machine settings"_#start_3_4 :all(b)
Note that the following "Section 2.4"_#start_4 describes the Make.py
tool which can be used to install/un-install packages and build the
auxiliary libraries which some of them use. It can also auto-edit a
Makefile.machine to add settings needed by some packages.
:line
Package basics: :h5,link(start_3_1)
@ -669,235 +667,221 @@ are always included, plus optional packages. Packages are groups of
files that enable a specific set of features. For example, force
fields for molecular systems or granular systems are in packages.
"Section 4"_Section_packages.html in the manual has details
about all the packages, including specific instructions for building
LAMMPS with each package, which are covered in a more general manner
"Section 4"_Section_packages.html in the manual has details about all
the packages, which come in two flavors: [standard] and [user]
packages. It also has specific instructions for building LAMMPS with
any package which requires an extra library. General instructions are
below.
You can see the list of all packages by typing "make package" from
within the src directory of the LAMMPS distribution. This also lists
various make commands that can be used to manipulate packages.
within the src directory of the LAMMPS distribution. It will also
list various make commands that can be used to manage packages.
If you use a command in a LAMMPS input script that is part of a
package, you must have built LAMMPS with that package, else you will
get an error that the style is invalid or the command is unknown.
Every command's doc page specifies if it is part of a package. You can
also type
Every command's doc page specfies if it is part of a package. You can
type
lmp_machine -h :pre
to run your executable with the optional "-h command-line
switch"_#start_7 for "help", which will simply list the styles and
commands known to your executable, and immediately exit.
There are two kinds of packages in LAMMPS, standard and user packages.
More information about the contents of standard and user packages is
given in "Section 4"_Section_packages.html of the manual. The
difference between standard and user packages is as follows:
Standard packages, such as molecule or kspace, are supported by the
LAMMPS developers and are written in a syntax and style consistent
with the rest of LAMMPS. This means we will answer questions about
them, debug and fix them if necessary, and keep them compatible with
future changes to LAMMPS.
User packages, such as user-atc or user-omp, have been contributed by
users, and always begin with the user prefix. If they are a single
command (single file), they are typically in the user-misc package.
Otherwise, they are a set of files grouped together which add a
specific functionality to the code.
User packages don't necessarily meet the requirements of the standard
packages. If you have problems using a feature provided in a user
package, you may need to contact the contributor directly to get help.
Information on how to submit additions you make to LAMMPS as single
files or either a standard or user-contributed package are given in
"this section"_Section_modify.html#mod_15 of the documentation.
switch"_#start_7 for "help", which will list the styles and commands
known to your executable, and immediately exit.
:line
Including/excluding packages :h5,link(start_3_2)
To use (or not use) a package you must include it (or exclude it)
before building LAMMPS. From the src directory, this is typically as
simple as:
To use (or not use) a package you must install it (or un-install it)
before building LAMMPS. From the src directory, this is as simple as:
make yes-colloid
make mpi :pre
or
make no-manybody
make no-user-omp
make mpi :pre
NOTE: You should NOT include/exclude packages and build LAMMPS in a
NOTE: You should NOT install/un-install packages and build LAMMPS in a
single make command using multiple targets, e.g. make yes-colloid mpi.
This is because the make procedure creates a list of source files that
will be out-of-date for the build if the package configuration changes
within the same command.
Some packages have individual files that depend on other packages
being included. LAMMPS checks for this and does the right thing.
I.e. individual files are only included if their dependencies are
already included. Likewise, if a package is excluded, other files
Any package can be installed or not in a LAMMPS build, independent of
all other packages. However, some packages include files derived from
files in other packages. LAMMPS checks for this and does the right
thing. I.e. individual files are only included if their dependencies
are already included. Likewise, if a package is excluded, other files
dependent on that package are also excluded.
NOTE: The one exception is that we do not recommend building with both
the KOKKOS package installed and any of the other acceleration
packages (GPU, OPT, USER-INTEL, USER-OMP) also installed. This is
because of how Kokkos sometimes builds using a wrapper compiler which
can make it difficult to invoke all the compile/link flags correctly
for both Kokkos and non-Kokkos files.
If you will never run simulations that use the features in a
particular packages, there is no reason to include it in your build.
For some packages, this will keep you from having to build auxiliary
libraries (see below), and will also produce a smaller executable
which may run a bit faster.
For some packages, this will keep you from having to build extra
libraries, and will also produce a smaller executable which may run a
bit faster.
When you download a LAMMPS tarball, these packages are pre-installed
in the src directory: KSPACE, MANYBODY,MOLECULE, because they are so
commonly used. When you download LAMMPS source files from the SVN or
Git repositories, no packages are pre-installed.
When you download a LAMMPS tarball, three packages are pre-installed
in the src directory -- KSPACE, MANYBODY, MOLECULE -- because they are
so commonly used. When you download LAMMPS source files from the SVN
or Git repositories, no packages are pre-installed.
Packages are included or excluded by typing "make yes-name" or "make
no-name", where "name" is the name of the package in lower-case, e.g.
name = kspace for the KSPACE package or name = user-atc for the
USER-ATC package. You can also type "make yes-standard", "make
no-standard", "make yes-std", "make no-std", "make yes-user", "make
no-user", "make yes-lib", "make no-lib", "make yes-all", or "make
no-all" to include/exclude various sets of packages. Type "make
package" to see all of the package-related make options.
Packages are installed or un-installed by typing
NOTE: Inclusion/exclusion of a package works by simply moving files
back and forth between the main src directory and sub-directories with
the package name (e.g. src/KSPACE, src/USER-ATC), so that the files
are seen or not seen when LAMMPS is built. After you have included or
excluded a package, you must re-build LAMMPS.
make yes-name
make no-name :pre
Additional package-related make options exist to help manage LAMMPS
files that exist in both the src directory and in package
sub-directories. You do not normally need to use these commands
unless you are editing LAMMPS files or have downloaded a patch from
the LAMMPS WWW site.
where "name" is the name of the package in lower-case, e.g. name =
kspace for the KSPACE package or name = user-atc for the USER-ATC
package. You can also type any of these commands:
Typing "make package-update" or "make pu" will overwrite src files
with files from the package sub-directories if the package has been
included. It should be used after a patch is installed, since patches
only update the files in the package sub-directory, but not the src
files. Typing "make package-overwrite" will overwrite files in the
package sub-directories with src files.
make yes-all | install all packages
make no-all | un-install all packages
make yes-standard or make yes-std | install standard packages
make no-standard or make no-std| un-install standard packages
make yes-user | install user packages
make no-user | un-install user packages
make yes-lib | install packages that require extra libraries
make no-lib | un-install packages that require extra libraries
make yes-ext | install packages that require external libraries
make no-ext | un-install packages that require external libraries :tb(s=|)
which install/un-install various sets of packages. Typing "make
package" will list all the these commands.
NOTE: Installing or un-installing a package works by simply moving
files back and forth between the main src directory and
sub-directories with the package name (e.g. src/KSPACE, src/USER-ATC),
so that the files are included or excluded when LAMMPS is built.
After you have installed or un-installed a package, you must re-build
LAMMPS for the action to take effect.
The following make commands help manage files that exist in both the
src directory and in package sub-directories. You do not normally
need to use these commands unless you are editing LAMMPS files or have
downloaded a patch from the LAMMPS web site.
Typing "make package-status" or "make ps" will show which packages are
currently included. For those that are included, it will list any
currently installed. For those that are installed, it will list any
files that are different in the src directory and package
sub-directory. Typing "make package-diff" lists all differences
between these files. Again, type "make package" to see all of the
package-related make options.
sub-directory.
Typing "make package-update" or "make pu" will overwrite src files
with files from the package sub-directories if the package is
installed. It should be used after a patch has been applied, since
patches only update the files in the package sub-directory, but not
the src files.
Typing "make package-overwrite" will overwrite files in the package
sub-directories with src files.
Typing "make package-diff" lists all differences between these files.
Again, just type "make package" to see all of the package-related make
options.
:line
Packages that require extra libraries :h5,link(start_3_3)
A few of the standard and user packages require additional auxiliary
libraries. Many of them are provided with LAMMPS, in which case they
must be compiled first, before LAMMPS is built, if you wish to include
that package. If you get a LAMMPS build error about a missing
library, this is likely the reason. See the
"Section 4"_Section_packages.html doc page for a list of
packages that have these kinds of auxiliary libraries.
A few of the standard and user packages require extra libraries. See
"Section 4"_Section_packages.html for two tables of packages which
indicate which ones require libraries. For each such package, the
Section 4 doc page gives details on how to build the extra library,
including how to download it if necessary. The basic ideas are
summarized here.
The lib directory in the distribution has sub-directories with package
names that correspond to the needed auxiliary libs, e.g. lib/gpu.
Each sub-directory has a README file that gives more details. Code
for most of the auxiliary libraries is included in that directory.
Examples are the USER-ATC and MEAM packages.
[System libraries:]
A few of the lib sub-directories do not include code, but do include
instructions (and sometimes scripts) that automate the process of
downloading the auxiliary library and installing it so LAMMPS can link
to it. Examples are the KIM, VORONOI, USER-MOLFILE, and USER-SMD
packages.
Packages in the tables "Section 4"_Section_packages.html with a "sys"
in the last column link to system libraries that typically already
exist on your machine. E.g. the python package links to a system
Python library. If your machine does not have the required library,
you will have to download and install it on your machine, in either
the system or user space.
The lib/python directory (for the PYTHON package) contains only a
choice of Makefile.lammps.* files. This is because no auxiliary code
or libraries are needed, only the Python library and other system libs
that should already available on your system. However, the
Makefile.lammps file is needed to tell LAMMPS which libs to use and
where to find them.
[Internal libraries:]
For libraries with provided code, the sub-directory README file
(e.g. lib/atc/README) has instructions on how to build that library.
This information is also summarized in "Section
4"_Section_packages.html. Typically this is done by typing
something like:
Packages in the tables "Section 4"_Section_packages.html with an "int"
in the last column link to internal libraries whose source code is
included with LAMMPS, in the lib/name directory where name is the
package name. You must first build the library in that directory
before building LAMMPS with that package installed. E.g. the gpu
package links to a library you build in the lib/gpu dir. You can
often do the build in one step by typing "make lib-name args=..."
from the src dir, with appropriate arguments. You can leave off the
args to see a help message. See "Section 4"_Section_packages.html for
details for each package.
make -f Makefile.g++ :pre
[External libraries:]
If one of the provided Makefiles is not appropriate for your system
you will need to edit or add one. Note that all the Makefiles have a
setting for EXTRAMAKE at the top that specifies a Makefile.lammps.*
file.
Packages in the tables "Section 4"_Section_packages.html with an "ext"
in the last column link to exernal libraries whose source code is not
included with LAMMPS. You must first download and install the library
before building LAMMPS with that package installed. E.g. the voronoi
package links to the freely available "Voro++ library"_voronoi. You
can often do the download/build in one step by typing "make lib-name
args=..." from the src dir, with appropriate arguments. You can leave
off the args to see a help message. See "Section
4"_Section_packages.html for details for each package.
If the library build is successful, it will produce 2 files in the lib
directory:
:link(voronoi,http://math.lbl.gov/voro++)
libpackage.a
Makefile.lammps :pre
[Possible errors:]
The Makefile.lammps file will typically be a copy of one of the
Makefile.lammps.* files in the library directory.
There are various common errors which can occur when building extra
libraries or when building LAMMPS with packages that require the extra
libraries.
Note that you must insure that the settings in Makefile.lammps are
appropriate for your system. If they are not, the LAMMPS build may
fail. To fix this, you can edit or create a new Makefile.lammps.*
file for your system, and copy it to Makefile.lammps.
If you cannot build the extra library itself successfully, you may
need to edit or create an appropriate Makefile for your machine, e.g.
with appropriate compiler or system settings. Provided makefiles are
typically in the lib/name directory. E.g. see the Makefile.* files in
lib/gpu.
As explained in the lib/package/README files, the settings in
Makefile.lammps are used to specify additional system libraries and
their locations so that LAMMPS can build with the auxiliary library.
For example, if the MEAM package is used, the auxiliary library
consists of F90 code, built with a Fortran complier. To link that
library with LAMMPS (a C++ code) via whatever C++ compiler LAMMPS is
built with, typically requires additional Fortran-to-C libraries be
included in the link. Another example are the BLAS and LAPACK
libraries needed to use the USER-ATC or USER-AWPMD packages.
The LAMMPS build often uses settings in a lib/name/Makefile.lammps
file which either exists in the LAMMPS distribution or is created or
copied from a lib/name/Makefile.lammps.* file when the library is
built. If those settings are not correct for your machine you will
need to edit or create an appropriate Makefile.lammps file.
For libraries without provided code, the sub-directory README file has
information on where to download the library and how to build it,
e.g. lib/voronoi/README and lib/smd/README. The README files also
describe how you must either (a) create soft links, via the "ln"
command, in those directories to point to where you built or installed
the packages, or (b) check or edit the Makefile.lammps file in the
same directory to provide that information.
Package-specific details for these steps are given in "Section
4"_Section_packages.html an in README files in the lib/name
directories.
Some of the sub-directories, e.g. lib/voronoi, also have an install.py
script which can be used to automate the process of
downloading/building/installing the auxiliary library, and setting the
needed soft links. Type "python install.py" for further instructions.
[Compiler options needed for accelerator packages:]
As with the sub-directories containing library code, if the soft links
or settings in the lib/package/Makefile.lammps files are not correct,
the LAMMPS build will typically fail.
Several packages contain code that is optimized for specific hardware,
e.g. CPU, KNL, or GPU. These are the OPT, GPU, KOKKOS, USER-INTEL,
and USER-OMP packages. Compiling and linking the source files in
these accelerator packages for optimal performance requires specific
settings in the Makefile.machine file you use.
:line
Packages that require Makefile.machine settings :h5,link(start_3_4)
A few packages require specific settings in Makefile.machine, to
either build or use the package effectively. These are the
USER-INTEL, KOKKOS, USER-OMP, and OPT packages, used for accelerating
code performance on CPUs or other hardware, as discussed in "Section
5.3"_Section_accelerate.html#acc_3.
A summary of what Makefile.machine changes are needed for each of
these packages is given in "Section 4"_Section_packages.html.
The details are given on the doc pages that describe each of these
accelerator packages in detail:
A summary of the Makefile.machine settings needed for each of these
packages is given in "Section 4"_Section_packages.html. More info is
given on the doc pages that describe each package in detail:
5.3.1 "USER-INTEL package"_accelerate_intel.html
5.3.2 "GPU package"_accelerate_intel.html
5.3.3 "KOKKOS package"_accelerate_kokkos.html
5.3.4 "USER-OMP package"_accelerate_omp.html
5.3.5 "OPT package"_accelerate_opt.html :all(b)
You can also look at the following machine Makefiles in
src/MAKE/OPTIONS, which include the changes. Note that the USER-INTEL
and KOKKOS packages allow for settings that build LAMMPS for different
hardware. The USER-INTEL package builds for CPU and the Xeon Phi, the
KOKKOS package builds for OpenMP, GPUs (Cuda), and the Xeon Phi.
You can also use or examine the following machine Makefiles in
src/MAKE/OPTIONS, which include the settings. Note that the
USER-INTEL and KOKKOS packages can use settings that build LAMMPS for
different hardware. The USER-INTEL package can be compiled for Intel
CPUs and KNLs; the KOKKOS package builds for CPUs (OpenMP), GPUs
(Cuda), and Intel KNLs.
Makefile.intel_cpu
Makefile.intel_phi
@ -907,127 +891,9 @@ Makefile.kokkos_phi
Makefile.omp
Makefile.opt :ul
Also note that the Make.py tool, described in the next "Section
2.4"_#start_4 can automatically add the needed info to an existing
machine Makefile, using simple command-line arguments.
:line
2.4 Building LAMMPS via the Make.py tool :h4,link(start_4)
The src directory includes a Make.py script, written in Python, which
can be used to automate various steps of the build process. It is
particularly useful for working with the accelerator packages, as well
as other packages which require auxiliary libraries to be built.
The goal of the Make.py tool is to allow any complex multi-step LAMMPS
build to be performed as a single Make.py command. And you can
archive the commands, so they can be re-invoked later via the -r
(redo) switch. If you find some LAMMPS build procedure that can't be
done in a single Make.py command, let the developers know, and we'll
see if we can augment the tool.
You can run Make.py from the src directory by typing either:
Make.py -h
python Make.py -h :pre
which will give you help info about the tool. For the former to work,
you may need to edit the first line of Make.py to point to your local
Python. And you may need to insure the script is executable:
chmod +x Make.py :pre
Here are examples of build tasks you can perform with Make.py:
Install/uninstall packages: Make.py -p no-lib kokkos omp intel
Build specific auxiliary libs: Make.py -a lib-atc lib-meam
Build libs for all installed packages: Make.py -p cuda gpu -gpu mode=double arch=31 -a lib-all
Create a Makefile from scratch with compiler and MPI settings: Make.py -m none -cc g++ -mpi mpich -a file
Augment Makefile.serial with settings for installed packages: Make.py -p intel -intel cpu -m serial -a file
Add JPG and FFTW support to Makefile.mpi: Make.py -m mpi -jpg -fft fftw -a file
Build LAMMPS with a parallel make using Makefile.mpi: Make.py -j 16 -m mpi -a exe
Build LAMMPS and libs it needs using Makefile.serial with accelerator settings: Make.py -p gpu intel -intel cpu -a lib-all file serial :tb(s=:)
The bench and examples directories give Make.py commands that can be
used to build LAMMPS with the various packages and options needed to
run all the benchmark and example input scripts. See these files for
more details:
bench/README
bench/FERMI/README
bench/KEPLER/README
bench/PHI/README
examples/README
examples/accelerate/README
examples/accelerate/make.list :ul
All of the Make.py options and syntax help can be accessed by using
the "-h" switch.
E.g. typing "Make.py -h" gives
Syntax: Make.py switch args ...
switches can be listed in any order
help switch:
-h prints help and syntax for all other specified switches
switch for actions:
-a lib-all, lib-dir, clean, file, exe or machine
list one or more actions, in any order
machine is a Makefile.machine suffix, must be last if used
one-letter switches:
-d (dir), -j (jmake), -m (makefile), -o (output),
-p (packages), -r (redo), -s (settings), -v (verbose)
switches for libs:
-atc, -awpmd, -colvars, -cuda
-gpu, -meam, -poems, -qmmm, -reax
switches for build and makefile options:
-intel, -kokkos, -cc, -mpi, -fft, -jpg, -png :pre
Using the "-h" switch with other switches and actions gives additional
info on all the other specified switches or actions. The "-h" can be
anywhere in the command-line and the other switches do not need their
arguments. E.g. type "Make.py -h -d -atc -intel" will print:
-d dir
dir = LAMMPS home dir
if -d not specified, working dir must be lammps/src :pre
-atc make=suffix lammps=suffix2
all args are optional and can be in any order
make = use Makefile.suffix (def = g++)
lammps = use Makefile.lammps.suffix2 (def = EXTRAMAKE in makefile) :pre
-intel mode
mode = cpu or phi (def = cpu)
build Intel package for CPU or Xeon Phi :pre
Note that Make.py never overwrites an existing Makefile.machine.
Instead, it creates src/MAKE/MINE/Makefile.auto, which you can save or
rename if desired. Likewise it creates an executable named
src/lmp_auto, which you can rename using the -o switch if desired.
The most recently executed Make.py command is saved in
src/Make.py.last. You can use the "-r" switch (for redo) to re-invoke
the last command, or you can save a sequence of one or more Make.py
commands to a file and invoke the file of commands using "-r". You
can also label the commands in the file and invoke one or more of them
by name.
A typical use of Make.py is to start with a valid Makefile.machine for
your system, that works for a vanilla LAMMPS build, i.e. when optional
packages are not installed. You can then use Make.py to add various
settings (FFT, JPG, PNG) to the Makefile.machine as well as change its
compiler and MPI options. You can also add additional packages to the
build, as well as build the needed supporting libraries.
You can also use Make.py to create a new Makefile.machine from
scratch, using the "-m none" switch, if you also specify what compiler
and MPI options to use, via the "-cc" and "-mpi" switches.
:line
2.5 Building LAMMPS as a library :h4,link(start_5)
2.4 Building LAMMPS as a library :h4,link(start_4)
LAMMPS can be built as either a static or shared library, which can
then be called from another application or a scripting language. See
@ -1063,7 +929,7 @@ src/MAKE/Makefile.foo and perform the build in the directory
Obj_shared_foo. This is so that each file can be compiled with the
-fPIC flag which is required for inclusion in a shared library. The
build will create the file liblammps_foo.so which another application
can link to dynamically. It will also create a soft link liblammps.so,
can link to dyamically. It will also create a soft link liblammps.so,
which will point to the most recently built shared library. This is
the file the Python wrapper loads by default.
@ -1149,7 +1015,7 @@ interface and how to extend it for your needs.
:line
2.6 Running LAMMPS :h4,link(start_6)
2.5 Running LAMMPS :h4,link(start_5)
By default, LAMMPS runs by reading commands from standard input. Thus
if you run the LAMMPS executable by itself, e.g.
@ -1281,7 +1147,7 @@ more processors or setup a smaller problem.
:line
2.7 Command-line options :h4,link(start_7)
2.6 Command-line options :h4,link(start_6)
At run time, LAMMPS recognizes several optional command-line switches
which may be used in any order. Either the full word or a one-or-two
@ -1415,8 +1281,8 @@ LAMMPS is compiled with CUDA=yes.
numa Nm :pre
This option is only relevant when using pthreads with hwloc support.
In this case Nm defines the number of NUMA regions (typically sockets)
on a node which will be utilized by a single MPI rank. By default Nm
In this case Nm defines the number of NUMA regions (typicaly sockets)
on a node which will be utilizied by a single MPI rank. By default Nm
= 1. If this option is used the total number of worker-threads per
MPI rank is threads*numa. Currently it is always almost better to
assign at least one MPI rank per NUMA region, and leave numa set to
@ -1480,7 +1346,7 @@ replica runs on on one or a few processors. Note that with MPI
installed on a machine (e.g. your desktop), you can run on more
(virtual) processors than you have physical processors.
To run multiple independent simulations from one input script, using
To run multiple independent simulatoins from one input script, using
multiple partitions, see "Section 6.4"_Section_howto.html#howto_4
of the manual. World- and universe-style "variables"_variable.html
are useful in this context.
@ -1711,7 +1577,7 @@ negative numeric value. It is OK if the first value1 starts with a
:line
2.8 LAMMPS screen output :h4,link(start_8)
2.7 LAMMPS screen output :h4,link(start_7)
As LAMMPS reads an input script, it prints information to both the
screen and a log file about significant actions it takes to setup a
@ -1759,7 +1625,7 @@ The first section provides a global loop timing summary. The {loop time}
is the total wall time for the section. The {Performance} line is
provided for convenience to help predicting the number of loop
continuations required and for comparing performance with other,
similar MD codes. The {CPU use} line provides the CPU utilization per
similar MD codes. The {CPU use} line provides the CPU utilzation per
MPI task; it should be close to 100% times the number of OpenMP
threads (or 1 of no OpenMP). Lower numbers correspond to delays due
to file I/O or insufficient thread utilization.
@ -1867,7 +1733,7 @@ communication, roughly 75% in the example above.
:line
2.9 Tips for users of previous LAMMPS versions :h4,link(start_9)
2.8 Tips for users of previous LAMMPS versions :h4,link(start_8)
The current C++ began with a complete rewrite of LAMMPS 2001, which
was written in F90. Features of earlier versions of LAMMPS are listed

9
doc/src/Section_tools.txt
View File

@ -369,15 +369,18 @@ supports it. It has its own WWW page at
msi2lmp tool :h4,link(msi)
The msi2lmp sub-directory contains a tool for creating LAMMPS input
data files from BIOVIA's Materias Studio files (formerly Accelrys'
The msi2lmp sub-directory contains a tool for creating LAMMPS template
input and data files from BIOVIA's Materias Studio files (formerly Accelrys'
Insight MD code, formerly MSI/Biosym and its Discover MD code).
This tool was written by John Carpenter (Cray), Michael Peachey
(Cray), and Steve Lustig (Dupont). Several people contributed changes
to remove bugs and adapt its output to changes in LAMMPS.
See the README file for more information.
This tool has several known limitations and is no longer under active
development, so there are no changes except for the occasional bugfix.
See the README file in the tools/msi2lmp folder for more information.
:line

2
doc/src/angle_sdk.txt
View File

@ -46,7 +46,7 @@ from the pair_style.
[Restrictions:]
This angle style can only be used if LAMMPS was built with the
USER-CG-CMM package. See the "Making
USER-CGSDK package. See the "Making
LAMMPS"_Section_start.html#start_3 section for more info on packages.
[Related commands:]

1
doc/src/bonds.txt
View File

@ -16,7 +16,6 @@ Bond Styles :h1
bond_none
bond_nonlinear
bond_oxdna
bond_oxdna2
bond_quartic
bond_table
bond_zero

34
doc/src/compute_sna_atom.txt
View File

@ -24,7 +24,7 @@ twojmax = band limit for bispectrum components (non-negative integer) :l
R_1, R_2,... = list of cutoff radii, one for each type (distance units) :l
w_1, w_2,... = list of neighbor weights, one for each type :l
zero or more keyword/value pairs may be appended :l
keyword = {diagonal} or {rmin0} or {switchflag} or {bzeroflag} :l
keyword = {diagonal} or {rmin0} or {switchflag} or {bzeroflag} or {quadraticflag}:l
{diagonal} value = {0} or {1} or {2} or {3}
{0} = all j1, j2, j <= twojmax, j2 <= j1
{1} = subset satisfying j1 == j2
@ -36,7 +36,10 @@ keyword = {diagonal} or {rmin0} or {switchflag} or {bzeroflag} :l
{1} = use switching function
{bzeroflag} value = {0} or {1}
{0} = do not subtract B0
{1} = subtract B0 :pre
{1} = subtract B0
{quadraticflag} value = {0} or {1}
{0} = do not generate quadratic terms
{1} = generate quadratic terms :pre
:ule
[Examples:]
@ -151,7 +154,7 @@ linear mapping from radial distance to polar angle {theta0} on the
The argument {twojmax} and the keyword {diagonal} define which
bispectrum components are generated. See section below on output for a
detailed explanation of the number of bispectrum components and the
ordered in which they are listed
ordered in which they are listed.
The keyword {switchflag} can be used to turn off the switching
function.
@ -162,6 +165,14 @@ the calculated bispectrum components. This optional keyword is only
available for compute {sna/atom}, as {snad/atom} and {snav/atom}
are unaffected by the removal of constant terms.
The keyword {quadraticflag} determines whether or not the
quadratic analogs to the bispectrum quantities are generated.
These are formed by taking the outer product of the vector
of bispectrum components with itself.
See section below on output for a
detailed explanation of the number of quadratic terms and the
ordered in which they are listed.
NOTE: If you have a bonded system, then the settings of
"special_bonds"_special_bonds.html command can remove pairwise
interactions between atoms in the same bond, angle, or dihedral. This
@ -180,7 +191,7 @@ command that includes all pairs in the neighbor list.
Compute {sna/atom} calculates a per-atom array, each column
corresponding to a particular bispectrum component. The total number
of columns and the identities of the bispectrum component contained in
of columns and the identity of the bispectrum component contained in
each column depend on the values of {twojmax} and {diagonal}, as
described by the following piece of python code:
@ -213,6 +224,19 @@ block contains six sub-blocks corresponding to the {xx}, {yy}, {zz},
notation. Each of these sub-blocks contains one column for each
bispectrum component, the same as for compute {sna/atom}
For example, if {K}=30 and ntypes=1, the number of columns in the per-atom
arrays generated by {sna/atom}, {snad/atom}, and {snav/atom}
are 30, 90, and 180, respectively. With {quadratic} value=1,
the numbers of columns are 930, 2790, and 5580, respectively.
If the {quadratic} keyword value is set to 1, then additional
columns are appended to each per-atom array, corresponding to
a matrix of quantities that are products of two bispectrum components. If the
number of bispectrum components is {K}, then the number of matrix elements
is {K}^2. These are output in subblocks of {K}^2 columns, using the same
ordering of columns and sub-blocks as was used for the bispectrum
components.
These values can be accessed by any command that uses per-atom values
from a compute as input. See "Section
6.15"_Section_howto.html#howto_15 for an overview of LAMMPS output
@ -231,7 +255,7 @@ LAMMPS"_Section_start.html#start_3 section for more info.
[Default:]
The optional keyword defaults are {diagonal} = 0, {rmin0} = 0,
{switchflag} = 1, {bzeroflag} = 0.
{switchflag} = 1, {bzeroflag} = 1, {quadraticflag} = 0,
:line

52
doc/src/dump.txt
View File

@ -7,12 +7,12 @@
:line
dump command :h3
"dump custom/vtk"_dump_custom_vtk.html command :h3
"dump vtk"_dump_vtk.html command :h3
"dump h5md"_dump_h5md.html command :h3
"dump molfile"_dump_molfile.html command :h3
"dump netcdf"_dump_netcdf.html command :h3
"dump image"_dump_image.html command :h3
"dump movie"_dump_image.html command :h3
"dump molfile"_dump_molfile.html command :h3
"dump nc"_dump_nc.html command :h3
[Syntax:]
@ -20,7 +20,7 @@ dump ID group-ID style N file args :pre
ID = user-assigned name for the dump :ulb,l
group-ID = ID of the group of atoms to be dumped :l
style = {atom} or {atom/gz} or {atom/mpiio} or {cfg} or {cfg/gz} or {cfg/mpiio} or {dcd} or {xtc} or {xyz} or {xyz/gz} or {xyz/mpiio} or {h5md} or {image} or {movie} or {molfile} or {local} or {custom} or {custom/gz} or {custom/mpiio} :l
style = {atom} or {atom/gz} or {atom/mpiio} or {cfg} or {cfg/gz} or {cfg/mpiio} or {custom} or {custom/gz} or {custom/mpiio} or {dcd} or {h5md} or {image} or or {local} or {molfile} or {movie} or {netcdf} or {netcdf/mpiio} or {vtk} or {xtc} or {xyz} or {xyz/gz} or {xyz/mpiio} :l
N = dump every this many timesteps :l
file = name of file to write dump info to :l
args = list of arguments for a particular style :l
@ -30,33 +30,22 @@ args = list of arguments for a particular style :l
{cfg} args = same as {custom} args, see below
{cfg/gz} args = same as {custom} args, see below
{cfg/mpiio} args = same as {custom} args, see below
{custom}, {custom/gz}, {custom/mpiio} args = see below
{dcd} args = none
{h5md} args = discussed on "dump h5md"_dump_h5md.html doc page
{image} args = discussed on "dump image"_dump_image.html doc page
{local} args = see below
{molfile} args = discussed on "dump molfile"_dump_molfile.html doc page
{movie} args = discussed on "dump image"_dump_image.html doc page
{netcdf} args = discussed on "dump netcdf"_dump_netcdf.html doc page
{netcdf/mpiio} args = discussed on "dump netcdf"_dump_netcdf.html doc page
{vtk} args = same as {custom} args, see below, also "dump vtk"_dump_vtk.html doc page
{xtc} args = none
{xyz} args = none :pre
{xyz/gz} args = none :pre
{xyz} args = none
{xyz/gz} args = none
{xyz/mpiio} args = none :pre
{custom/vtk} args = similar to custom args below, discussed on "dump custom/vtk"_dump_custom_vtk.html doc page :pre
{h5md} args = discussed on "dump h5md"_dump_h5md.html doc page :pre
{image} args = discussed on "dump image"_dump_image.html doc page :pre
{movie} args = discussed on "dump image"_dump_image.html doc page :pre
{molfile} args = discussed on "dump molfile"_dump_molfile.html doc page
{nc} args = discussed on "dump nc"_dump_nc.html doc page :pre
{local} args = list of local attributes
possible attributes = index, c_ID, c_ID\[I\], f_ID, f_ID\[I\]
index = enumeration of local values
c_ID = local vector calculated by a compute with ID
c_ID\[I\] = Ith column of local array calculated by a compute with ID, I can include wildcard (see below)
f_ID = local vector calculated by a fix with ID
f_ID\[I\] = Ith column of local array calculated by a fix with ID, I can include wildcard (see below) :pre
{custom} or {custom/gz} or {custom/mpiio} args = list of atom attributes
{custom} or {custom/gz} or {custom/mpiio} args = list of atom attributes :l
possible attributes = id, mol, proc, procp1, type, element, mass,
x, y, z, xs, ys, zs, xu, yu, zu,
xsu, ysu, zsu, ix, iy, iz,
@ -94,6 +83,15 @@ args = list of arguments for a particular style :l
v_name = per-atom vector calculated by an atom-style variable with name
d_name = per-atom floating point vector with name, managed by fix property/atom
i_name = per-atom integer vector with name, managed by fix property/atom :pre
{local} args = list of local attributes :l
possible attributes = index, c_ID, c_ID\[I\], f_ID, f_ID\[I\]
index = enumeration of local values
c_ID = local vector calculated by a compute with ID
c_ID\[I\] = Ith column of local array calculated by a compute with ID, I can include wildcard (see below)
f_ID = local vector calculated by a fix with ID
f_ID\[I\] = Ith column of local array calculated by a fix with ID, I can include wildcard (see below) :pre
:ule
[Examples:]

347
doc/src/dump_custom_vtk.txt
View File

@ -1,347 +0,0 @@
"LAMMPS WWW Site"_lws - "LAMMPS Documentation"_ld - "LAMMPS Commands"_lc :c
:link(lws,http://lammps.sandia.gov)
:link(ld,Manual.html)
:link(lc,Section_commands.html#comm)
:line
dump custom/vtk command :h3
[Syntax:]
dump ID group-ID style N file args :pre
ID = user-assigned name for the dump :ulb,l
group-ID = ID of the group of atoms to be dumped :l
style = {custom/vtk} :l
N = dump every this many timesteps :l
file = name of file to write dump info to :l
args = list of arguments for a particular style :l
{custom/vtk} args = list of atom attributes
possible attributes = id, mol, proc, procp1, type, element, mass,
x, y, z, xs, ys, zs, xu, yu, zu,
xsu, ysu, zsu, ix, iy, iz,
vx, vy, vz, fx, fy, fz,
q, mux, muy, muz, mu,
radius, diameter, omegax, omegay, omegaz,
angmomx, angmomy, angmomz, tqx, tqy, tqz,
c_ID, c_ID\[N\], f_ID, f_ID\[N\], v_name :pre
id = atom ID
mol = molecule ID
proc = ID of processor that owns atom
procp1 = ID+1 of processor that owns atom
type = atom type
element = name of atom element, as defined by "dump_modify"_dump_modify.html command
mass = atom mass
x,y,z = unscaled atom coordinates
xs,ys,zs = scaled atom coordinates
xu,yu,zu = unwrapped atom coordinates
xsu,ysu,zsu = scaled unwrapped atom coordinates
ix,iy,iz = box image that the atom is in
vx,vy,vz = atom velocities
fx,fy,fz = forces on atoms
q = atom charge
mux,muy,muz = orientation of dipole moment of atom
mu = magnitude of dipole moment of atom
radius,diameter = radius,diameter of spherical particle
omegax,omegay,omegaz = angular velocity of spherical particle
angmomx,angmomy,angmomz = angular momentum of aspherical particle
tqx,tqy,tqz = torque on finite-size particles
c_ID = per-atom vector calculated by a compute with ID
c_ID\[I\] = Ith column of per-atom array calculated by a compute with ID, I can include wildcard (see below)
f_ID = per-atom vector calculated by a fix with ID
f_ID\[I\] = Ith column of per-atom array calculated by a fix with ID, I can include wildcard (see below)
v_name = per-atom vector calculated by an atom-style variable with name
d_name = per-atom floating point vector with name, managed by fix property/atom
i_name = per-atom integer vector with name, managed by fix property/atom :pre
:ule
[Examples:]
dump dmpvtk all custom/vtk 100 dump*.myforce.vtk id type vx fx
dump dmpvtp flow custom/vtk 100 dump*.%.displace.vtp id type c_myD\[1\] c_myD\[2\] c_myD\[3\] v_ke :pre
The style {custom/vtk} is similar to the "custom"_dump.html style but
uses the VTK library to write data to VTK simple legacy or XML format
depending on the filename extension specified. This can be either
{*.vtk} for the legacy format or {*.vtp} and {*.vtu}, respectively,
for the XML format; see the "VTK
homepage"_http://www.vtk.org/VTK/img/file-formats.pdf for a detailed
description of these formats. Since this naming convention conflicts
with the way binary output is usually specified (see below),
"dump_modify binary"_dump_modify.html allows to set the binary
flag for this dump style explicitly.
[Description:]
Dump a snapshot of atom quantities to one or more files every N
timesteps in a format readable by the "VTK visualization
toolkit"_http://www.vtk.org or other visualization tools that use it,
e.g. "ParaView"_http://www.paraview.org. The timesteps on which dump
output is written can also be controlled by a variable; see the
"dump_modify every"_dump_modify.html command for details.
Only information for atoms in the specified group is dumped. The
"dump_modify thresh and region"_dump_modify.html commands can also
alter what atoms are included; see details below.
As described below, special characters ("*", "%") in the filename
determine the kind of output.
IMPORTANT NOTE: Because periodic boundary conditions are enforced only
on timesteps when neighbor lists are rebuilt, the coordinates of an
atom written to a dump file may be slightly outside the simulation
box.
IMPORTANT NOTE: Unless the "dump_modify sort"_dump_modify.html
option is invoked, the lines of atom information written to dump files
will be in an indeterminate order for each snapshot. This is even
true when running on a single processor, if the "atom_modify
sort"_atom_modify.html option is on, which it is by default. In this
case atoms are re-ordered periodically during a simulation, due to
spatial sorting. It is also true when running in parallel, because
data for a single snapshot is collected from multiple processors, each
of which owns a subset of the atoms.
For the {custom/vtk} style, sorting is off by default. See the
"dump_modify"_dump_modify.html doc page for details.
:line
The dimensions of the simulation box are written to a separate file
for each snapshot (either in legacy VTK or XML format depending on
the format of the main dump file) with the suffix {_boundingBox}
appended to the given dump filename.
For an orthogonal simulation box this information is saved as a
rectilinear grid (legacy .vtk or .vtr XML format).
Triclinic simulation boxes (non-orthogonal) are saved as
hexahedrons in either legacy .vtk or .vtu XML format.
Style {custom/vtk} allows you to specify a list of atom attributes
to be written to the dump file for each atom. Possible attributes
are listed above. In contrast to the {custom} style, the attributes
are rearranged to ensure correct ordering of vector components
(except for computes and fixes - these have to be given in the right
order) and duplicate entries are removed.
You cannot specify a quantity that is not defined for a particular
simulation - such as {q} for atom style {bond}, since that atom style
doesn't assign charges. Dumps occur at the very end of a timestep,
so atom attributes will include effects due to fixes that are applied
during the timestep. An explanation of the possible dump custom/vtk attributes
is given below. Since position data is required to write VTK files "x y z"
do not have to be specified explicitly.
The VTK format uses a single snapshot of the system per file, thus
a wildcard "*" must be included in the filename, as discussed below.
Otherwise the dump files will get overwritten with the new snapshot
each time.
:line
Dumps are performed on timesteps that are a multiple of N (including
timestep 0) and on the last timestep of a minimization if the
minimization converges. Note that this means a dump will not be
performed on the initial timestep after the dump command is invoked,
if the current timestep is not a multiple of N. This behavior can be
changed via the "dump_modify first"_dump_modify.html command, which
can also be useful if the dump command is invoked after a minimization
ended on an arbitrary timestep. N can be changed between runs by
using the "dump_modify every"_dump_modify.html command.
The "dump_modify every"_dump_modify.html command
also allows a variable to be used to determine the sequence of
timesteps on which dump files are written. In this mode a dump on the
first timestep of a run will also not be written unless the
"dump_modify first"_dump_modify.html command is used.
Dump filenames can contain two wildcard characters. If a "*"
character appears in the filename, then one file per snapshot is
written and the "*" character is replaced with the timestep value.
For example, tmp.dump*.vtk becomes tmp.dump0.vtk, tmp.dump10000.vtk,
tmp.dump20000.vtk, etc. Note that the "dump_modify pad"_dump_modify.html
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 with some post-processing tools.
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 preceded by an underscore character.
For example, tmp.dump%.vtp becomes tmp.dump_0.vtp, tmp.dump_1.vtp, ...
tmp.dump_P-1.vtp, etc. This creates smaller files and can be a fast
mode of output on parallel machines that support parallel I/O for output.
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.
For the legacy VTK format "%" is ignored and P = 1, i.e., only
processor 0 does write files.
Note that using the "*" and "%" characters together can produce a
large number of small dump files!
If {dump_modify binary} is used, the dump file (or files, if "*" or
"%" is also used) is written in binary format. A binary dump file
will be about the same size as a text version, but will typically
write out much faster.
:line
This section explains the atom attributes that can be specified as
part of the {custom/vtk} style.
The {id}, {mol}, {proc}, {procp1}, {type}, {element}, {mass}, {vx},
{vy}, {vz}, {fx}, {fy}, {fz}, {q} attributes are self-explanatory.
{Id} is the atom ID. {Mol} is the molecule ID, included in the data
file for molecular systems. {Proc} is the ID of the processor (0 to
Nprocs-1) that currently owns the atom. {Procp1} is the proc ID+1,
which can be convenient in place of a {type} attribute (1 to Ntypes)
for coloring atoms in a visualization program. {Type} is the atom
type (1 to Ntypes). {Element} is typically the chemical name of an
element, which you must assign to each type via the "dump_modify
element"_dump_modify.html command. More generally, it can be any
string you wish to associated with an atom type. {Mass} is the atom
mass. {Vx}, {vy}, {vz}, {fx}, {fy}, {fz}, and {q} are components of
atom velocity and force and atomic charge.
There are several options for outputting atom coordinates. The {x},
{y}, {z} attributes write atom coordinates "unscaled", in the
appropriate distance "units"_units.html (Angstroms, sigma, etc). Use
{xs}, {ys}, {zs} if you want the coordinates "scaled" to the box size,
so that each value is 0.0 to 1.0. If the simulation box is triclinic
(tilted), then all atom coords will still be between 0.0 and 1.0.
I.e. actual unscaled (x,y,z) = xs*A + ys*B + zs*C, where (A,B,C) are
the non-orthogonal vectors of the simulation box edges, as discussed
in "Section 6.12"_Section_howto.html#howto_12.
Use {xu}, {yu}, {zu} if you want the coordinates "unwrapped" by the
image flags for each atom. Unwrapped means that if the atom has
passed thru a periodic boundary one or more times, the value is
printed for what the coordinate would be if it had not been wrapped
back into the periodic box. Note that using {xu}, {yu}, {zu} means
that the coordinate values may be far outside the box bounds printed
with the snapshot. Using {xsu}, {ysu}, {zsu} is similar to using
{xu}, {yu}, {zu}, except that the unwrapped coordinates are scaled by
the box size. Atoms that have passed through a periodic boundary will
have the corresponding coordinate increased or decreased by 1.0.
The image flags can be printed directly using the {ix}, {iy}, {iz}
attributes. For periodic dimensions, they specify which image of the
simulation box the atom is considered to be in. An image of 0 means
it is inside the box as defined. A value of 2 means add 2 box lengths
to get the true value. A value of -1 means subtract 1 box length to
get the true value. LAMMPS updates these flags as atoms cross
periodic boundaries during the simulation.
The {mux}, {muy}, {muz} attributes are specific to dipolar systems
defined with an atom style of {dipole}. They give the orientation of
the atom's point dipole moment. The {mu} attribute gives the
magnitude of the atom's dipole moment.
The {radius} and {diameter} attributes are specific to spherical
particles that have a finite size, such as those defined with an atom
style of {sphere}.
The {omegax}, {omegay}, and {omegaz} attributes are specific to
finite-size spherical particles that have an angular velocity. Only
certain atom styles, such as {sphere} define this quantity.
The {angmomx}, {angmomy}, and {angmomz} attributes are specific to
finite-size aspherical particles that have an angular momentum. Only
the {ellipsoid} atom style defines this quantity.
The {tqx}, {tqy}, {tqz} attributes are for finite-size particles that
can sustain a rotational torque due to interactions with other
particles.
The {c_ID} and {c_ID\[I\]} attributes allow per-atom vectors or arrays
calculated by a "compute"_compute.html to be output. The ID in the
attribute should be replaced by the actual ID of the compute that has
been defined previously in the input script. See the
"compute"_compute.html command for details. There are computes for
calculating the per-atom energy, stress, centro-symmetry parameter,
and coordination number of individual atoms.
Note that computes which calculate global or local quantities, as
opposed to per-atom quantities, cannot be output in a dump custom/vtk
command. Instead, global quantities can be output by the
"thermo_style custom"_thermo_style.html command, and local quantities
can be output by the dump local command.
If {c_ID} is used as a attribute, then the per-atom vector calculated
by the compute is printed. If {c_ID\[I\]} is used, then I must be in
the range from 1-M, which will print the Ith column of the per-atom
array with M columns calculated by the compute. See the discussion
above for how I can be specified with a wildcard asterisk to
effectively specify multiple values.
The {f_ID} and {f_ID\[I\]} attributes allow vector or array per-atom
quantities calculated by a "fix"_fix.html to be output. The ID in the
attribute should be replaced by the actual ID of the fix that has been
defined previously in the input script. The "fix
ave/atom"_fix_ave_atom.html command is one that calculates per-atom
quantities. Since it can time-average per-atom quantities produced by
any "compute"_compute.html, "fix"_fix.html, or atom-style
"variable"_variable.html, this allows those time-averaged results to
be written to a dump file.
If {f_ID} is used as a attribute, then the per-atom vector calculated
by the fix is printed. If {f_ID\[I\]} is used, then I must be in the
range from 1-M, which will print the Ith column of the per-atom array
with M columns calculated by the fix. See the discussion above for
how I can be specified with a wildcard asterisk to effectively specify
multiple values.
The {v_name} attribute allows per-atom vectors calculated by a
"variable"_variable.html to be output. The name in the attribute
should be replaced by the actual name of the variable that has been
defined previously in the input script. Only an atom-style variable
can be referenced, since it is the only style that generates per-atom
values. Variables of style {atom} can reference individual atom
attributes, per-atom atom attributes, thermodynamic keywords, or
invoke other computes, fixes, or variables when they are evaluated, so
this is a very general means of creating quantities to output to a
dump file.
The {d_name} and {i_name} attributes allow to output custom per atom
floating point or integer properties that are managed by
"fix property/atom"_fix_property_atom.html.
See "Section 10"_Section_modify.html of the manual for information
on how to add new compute and fix styles to LAMMPS to calculate
per-atom quantities which could then be output into dump files.
:line
[Restrictions:]
The {custom/vtk} style does not support writing of gzipped dump files.
The {custom/vtk} dump style is part of the USER-VTK package. It is
only enabled if LAMMPS was built with that package. See the "Making
LAMMPS"_Section_start.html#start_3 section for more info.
To use this dump style, you also must link to the VTK library. See
the info in lib/vtk/README and insure the Makefile.lammps file in that
directory is appropriate for your machine.
The {custom/vtk} dump style neither supports buffering nor custom
format strings.
[Related commands:]
"dump"_dump.html, "dump image"_dump_image.html,
"dump_modify"_dump_modify.html, "undump"_undump.html
[Default:]
By default, files are written in ASCII format. If the file extension
is not one of .vtk, .vtp or .vtu, the legacy VTK file format is used.

18
doc/src/dump_h5md.txt
View File

@ -17,9 +17,7 @@ group-ID = ID of the group of atoms to be imaged :l
h5md = style of dump command (other styles {atom} or {cfg} or {dcd} or {xtc} or {xyz} or {local} or {custom} are discussed on the "dump"_dump.html doc page) :l
N = dump every this many timesteps :l
file.h5 = name of file to write to :l
args = list of data elements to dump, with their dump "subintervals".
At least one element must be given and image may only be present if
position is specified first. :l
args = list of data elements to dump, with their dump "subintervals"
position options
image
velocity options
@ -29,15 +27,17 @@ position is specified first. :l
box value = {yes} or {no}
create_group value = {yes} or {no}
author value = quoted string :pre
:ule
For the elements {position}, {velocity}, {force} and {species}, one
may specify a sub-interval to write the data only every N_element
Note that at least one element must be specified and image may only be
present if position is specified first.
For the elements {position}, {velocity}, {force} and {species}, a
sub-interval may be specified to write the data only every N_element
iterations of the dump (i.e. every N*N_element time steps). This is
specified by the option
specified by this option directly following the element declaration:
every N_element :pre
that follows directly the element declaration.
every N_element :pre
:ule

66
doc/src/dump_nc.txt
View File

@ -1,66 +0,0 @@
"LAMMPS WWW Site"_lws - "LAMMPS Documentation"_ld - "LAMMPS Commands"_lc :c
:link(lws,http://lammps.sandia.gov)
:link(ld,Manual.html)
:link(lc,Section_commands.html#comm)
:line
dump nc command :h3
dump nc/mpiio command :h3
[Syntax:]
dump ID group-ID nc N file.nc args
dump ID group-ID nc/mpiio N file.nc args :pre
ID = user-assigned name for the dump :ulb,l
group-ID = ID of the group of atoms to be imaged :l
{nc} or {nc/mpiio} = style of dump command (other styles {atom} or {cfg} or {dcd} or {xtc} or {xyz} or {local} or {custom} are discussed on the "dump"_dump.html doc page) :l
N = dump every this many timesteps :l
file.nc = name of file to write to :l
args = list of per atom data elements to dump, same as for the 'custom' dump style. :l,ule
[Examples:]
dump 1 all nc 100 traj.nc type x y z vx vy vz
dump_modify 1 append yes at -1 global c_thermo_pe c_thermo_temp c_thermo_press :pre
dump 1 all nc/mpiio 1000 traj.nc id type x y z :pre
[Description:]
Dump a snapshot of atom coordinates every N timesteps in Amber-style
NetCDF file format. NetCDF files are binary, portable and
self-describing. This dump style will write only one file on the root
node. The dump style {nc} uses the "standard NetCDF
library"_netcdf-home all data is collected on one processor and then
written to the dump file. Dump style {nc/mpiio} used the "parallel
NetCDF library"_pnetcdf-home and MPI-IO; it has better performance on
a larger number of processors. Note that 'nc' outputs all atoms sorted
by atom tag while 'nc/mpiio' outputs in order of the MPI rank.
In addition to per-atom data, also global (i.e. not per atom, but per
frame) quantities can be included in the dump file. This can be
variables, output from computes or fixes data prefixed with v_, c_ and
f_, respectively. These properties are included via
"dump_modify"_dump_modify.html {global}.
:link(netcdf-home,http://www.unidata.ucar.edu/software/netcdf/)
:link(pnetcdf-home,http://trac.mcs.anl.gov/projects/parallel-netcdf/)
:line
[Restrictions:]
The {nc} and {nc/mpiio} dump styles are part of the USER-NC-DUMP
package. It is only enabled if LAMMPS was built with that
package. See the "Making LAMMPS"_Section_start.html#start_3 section
for more info.
:line
[Related commands:]
"dump"_dump.html, "dump_modify"_dump_modify.html, "undump"_undump.html

82
doc/src/dump_netcdf.txt Normal file
View File

@ -0,0 +1,82 @@
"LAMMPS WWW Site"_lws - "LAMMPS Documentation"_ld - "LAMMPS Commands"_lc :c
:link(lws,http://lammps.sandia.gov)
:link(ld,Manual.html)
:link(lc,Section_commands.html#comm)
:line
dump netcdf command :h3
dump netcdf/mpiio command :h3
[Syntax:]
dump ID group-ID netcdf N file args
dump ID group-ID netcdf/mpiio N file args :pre
ID = user-assigned name for the dump :ulb,l
group-ID = ID of the group of atoms to be imaged :l
{netcdf} or {netcdf/mpiio} = style of dump command (other styles {atom} or {cfg} or {dcd} or {xtc} or {xyz} or {local} or {custom} are discussed on the "dump"_dump.html doc page) :l
N = dump every this many timesteps :l
file = name of file to write dump info to :l
args = list of atom attributes, same as for "dump_style custom"_dump.html :l,ule
[Examples:]
dump 1 all netcdf 100 traj.nc type x y z vx vy vz
dump_modify 1 append yes at -1 global c_thermo_pe c_thermo_temp c_thermo_press
dump 1 all netcdf/mpiio 1000 traj.nc id type x y z :pre
[Description:]
Dump a snapshot of atom coordinates every N timesteps in Amber-style
NetCDF file format. NetCDF files are binary, portable and
self-describing. This dump style will write only one file on the root
node. The dump style {netcdf} uses the "standard NetCDF
library"_netcdf-home. All data is collected on one processor and then
written to the dump file. Dump style {netcdf/mpiio} uses the
"parallel NetCDF library"_pnetcdf-home and MPI-IO to write to the dump
file in parallel; it has better performance on a larger number of
processors. Note that style {netcdf} outputs all atoms sorted by atom
tag while style {netcdf/mpiio} outputs atoms in order of their MPI
rank.
NetCDF files can be directly visualized via the following tools:
Ovito (http://www.ovito.org/). Ovito supports the AMBER convention and
all of the above extensions. :ule,b
VMD (http://www.ks.uiuc.edu/Research/vmd/). :l
AtomEye (http://www.libatoms.org/). The libAtoms version of AtomEye
contains a NetCDF reader that is not present in the standard
distribution of AtomEye. :l,ule
In addition to per-atom data, global data can be included in the dump
file, which are the kinds of values output by the
"thermo_style"_thermo_style.html command . See "Section howto
6.15"_Section_howto.html#howto_15 for an explanation of per-atom
versus global data. The global output written into the dump file can
be from computes, fixes, or variables, by prefixing the compute/fix ID
or variable name with "c_" or "f_" or "v_" respectively, as in the
example above. These global values are specified via the "dump_modify
global"_dump_modify.html command.
:link(netcdf-home,http://www.unidata.ucar.edu/software/netcdf/)
:link(pnetcdf-home,http://trac.mcs.anl.gov/projects/parallel-netcdf/)
:line
[Restrictions:]
The {netcdf} and {netcdf/mpiio} dump styles are part of the
USER-NETCDF package. They are only enabled if LAMMPS was built with
that package. See the "Making LAMMPS"_Section_start.html#start_3
section for more info.
:line
[Related commands:]
"dump"_dump.html, "dump_modify"_dump_modify.html, "undump"_undump.html

179
doc/src/dump_vtk.txt Normal file
View File

@ -0,0 +1,179 @@
"LAMMPS WWW Site"_lws - "LAMMPS Documentation"_ld - "LAMMPS Commands"_lc :c
:link(lws,http://lammps.sandia.gov)
:link(ld,Manual.html)
:link(lc,Section_commands.html#comm)
:line
dump vtk command :h3
[Syntax:]
dump ID group-ID vtk N file args :pre
ID = user-assigned name for the dump
group-ID = ID of the group of atoms to be dumped
vtk = style of dump command (other styles {atom} or {cfg} or {dcd} or {xtc} or {xyz} or {local} or {custom} are discussed on the "dump"_dump.html doc page)
N = dump every this many timesteps
file = name of file to write dump info to
args = same as arguments for "dump_style custom"_dump.html :ul
[Examples:]
dump dmpvtk all vtk 100 dump*.myforce.vtk id type vx fx
dump dmpvtp flow vtk 100 dump*.%.displace.vtp id type c_myD\[1\] c_myD\[2\] c_myD\[3\] v_ke :pre
[Description:]
Dump a snapshot of atom quantities to one or more files every N
timesteps in a format readable by the "VTK visualization
toolkit"_http://www.vtk.org or other visualization tools that use it,
e.g. "ParaView"_http://www.paraview.org. The timesteps on which dump
output is written can also be controlled by a variable; see the
"dump_modify every"_dump_modify.html command for details.
This dump style is similar to "dump_style custom"_dump.html but uses
the VTK library to write data to VTK simple legacy or XML format
depending on the filename extension specified for the dump file. This
can be either {*.vtk} for the legacy format or {*.vtp} and {*.vtu},
respectively, for XML format; see the "VTK
homepage"_http://www.vtk.org/VTK/img/file-formats.pdf for a detailed
description of these formats. Since this naming convention conflicts
with the way binary output is usually specified (see below), the
"dump_modify binary"_dump_modify.html command allows setting of a
binary option for this dump style explicitly.
Only information for atoms in the specified group is dumped. The
"dump_modify thresh and region"_dump_modify.html commands can also
alter what atoms are included; see details below.
As described below, special characters ("*", "%") in the filename
determine the kind of output.
IMPORTANT NOTE: Because periodic boundary conditions are enforced only
on timesteps when neighbor lists are rebuilt, the coordinates of an
atom written to a dump file may be slightly outside the simulation
box.
IMPORTANT NOTE: Unless the "dump_modify sort"_dump_modify.html option
is invoked, the lines of atom information written to dump files will
be in an indeterminate order for each snapshot. This is even true
when running on a single processor, if the "atom_modify
sort"_atom_modify.html option is on, which it is by default. In this
case atoms are re-ordered periodically during a simulation, due to
spatial sorting. It is also true when running in parallel, because
data for a single snapshot is collected from multiple processors, each
of which owns a subset of the atoms.
For the {vtk} style, sorting is off by default. See the
"dump_modify"_dump_modify.html doc page for details.
:line
The dimensions of the simulation box are written to a separate file
for each snapshot (either in legacy VTK or XML format depending on the
format of the main dump file) with the suffix {_boundingBox} appended
to the given dump filename.
For an orthogonal simulation box this information is saved as a
rectilinear grid (legacy .vtk or .vtr XML format).
Triclinic simulation boxes (non-orthogonal) are saved as
hexahedrons in either legacy .vtk or .vtu XML format.
Style {vtk} allows you to specify a list of atom attributes to be
written to the dump file for each atom. The list of possible attributes
is the same as for the "dump_style custom"_dump.html command; see
its doc page for a listing and an explanation of each attribute.
NOTE: Since position data is required to write VTK files the atom
attributes "x y z" do not have to be specified explicitly; they will
be included in the dump file regardless. Also, in contrast to the
{custom} style, the specified {vtk} attributes are rearranged to
ensure correct ordering of vector components (except for computes and
fixes - these have to be given in the right order) and duplicate
entries are removed.
The VTK format uses a single snapshot of the system per file, thus
a wildcard "*" must be included in the filename, as discussed below.
Otherwise the dump files will get overwritten with the new snapshot
each time.
:line
Dumps are performed on timesteps that are a multiple of N (including
timestep 0) and on the last timestep of a minimization if the
minimization converges. Note that this means a dump will not be
performed on the initial timestep after the dump command is invoked,
if the current timestep is not a multiple of N. This behavior can be
changed via the "dump_modify first"_dump_modify.html command, which
can also be useful if the dump command is invoked after a minimization
ended on an arbitrary timestep. N can be changed between runs by
using the "dump_modify every"_dump_modify.html command.
The "dump_modify every"_dump_modify.html command
also allows a variable to be used to determine the sequence of
timesteps on which dump files are written. In this mode a dump on the
first timestep of a run will also not be written unless the
"dump_modify first"_dump_modify.html command is used.
Dump filenames can contain two wildcard characters. If a "*"
character appears in the filename, then one file per snapshot is
written and the "*" character is replaced with the timestep value.
For example, tmp.dump*.vtk becomes tmp.dump0.vtk, tmp.dump10000.vtk,
tmp.dump20000.vtk, etc. Note that the "dump_modify pad"_dump_modify.html
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 with some post-processing tools.
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 preceded by an underscore character.
For example, tmp.dump%.vtp becomes tmp.dump_0.vtp, tmp.dump_1.vtp, ...
tmp.dump_P-1.vtp, etc. This creates smaller files and can be a fast
mode of output on parallel machines that support parallel I/O for output.
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.
For the legacy VTK format "%" is ignored and P = 1, i.e., only
processor 0 does write files.
Note that using the "*" and "%" characters together can produce a
large number of small dump files!
If {dump_modify binary} is used, the dump file (or files, if "*" or
"%" is also used) is written in binary format. A binary dump file
will be about the same size as a text version, but will typically
write out much faster.
:line
[Restrictions:]
The {vtk} style does not support writing of gzipped dump files.
The {vtk} dump style is part of the USER-VTK package. It is
only enabled if LAMMPS was built with that package. See the "Making
LAMMPS"_Section_start.html#start_3 section for more info.
To use this dump style, you also must link to the VTK library. See
the info in lib/vtk/README and insure the Makefile.lammps file in that
directory is appropriate for your machine.
The {vtk} dump style supports neither buffering or custom format
strings.
[Related commands:]
"dump"_dump.html, "dump image"_dump_image.html,
"dump_modify"_dump_modify.html, "undump"_undump.html
[Default:]
By default, files are written in ASCII format. If the file extension
is not one of .vtk, .vtp or .vtu, the legacy VTK file format is used.

7
doc/src/fix_cmap.txt
View File

@ -87,8 +87,11 @@ the note below about how to include the CMAP energy when performing an
[Restart, fix_modify, output, run start/stop, minimize info:]
No information about this fix is written to "binary restart
files"_restart.html.
This fix writes the list of CMAP crossterms to "binary restart
files"_restart.html. See the "read_restart"_read_restart.html command
for info on how to re-specify a fix in an input script that reads a
restart file, so that the operation of the fix continues in an
uninterrupted fashion.
The "fix_modify"_fix_modify.html {energy} option is supported by this
fix to add the potential "energy" of the CMAP interactions system's

2
doc/src/fix_gcmc.txt
View File

@ -317,7 +317,7 @@ solution is to start a new simulation after the equilibrium density
has been reached.
With some pair_styles, such as "Buckingham"_pair_buck.html,
"Born-Mayer-Huggins"_pair_born.html and "ReaxFF"_pair_reax_c.html, two
"Born-Mayer-Huggins"_pair_born.html and "ReaxFF"_pair_reaxc.html, two
atoms placed close to each other may have an arbitrary large, negative
potential energy due to the functional form of the potential. While
these unphysical configurations are inaccessible to typical dynamical

9
doc/src/fix_gle.txt
View File

@ -67,9 +67,10 @@ target value as the {Tstart} and {Tstop} arguments, so that the diffusion
matrix that gives canonical sampling for a given A is computed automatically.
However, the GLE framework also allow for non-equilibrium sampling, that
can be used for instance to model inexpensively zero-point energy
effects "(Ceriotti2)"_#Ceriotti2. This is achieved specifying the
{noneq} keyword followed by the name of the file that contains the
static covariance matrix for the non-equilibrium dynamics.
effects "(Ceriotti2)"_#Ceriotti2. This is achieved specifying the {noneq}
keyword followed by the name of the file that contains the static covariance
matrix for the non-equilibrium dynamics. Please note, that the covariance
matrix is expected to be given in [temperature units].
Since integrating GLE dynamics can be costly when used together with
simple potentials, one can use the {every} optional keyword to
@ -148,7 +149,7 @@ dpd/tstat"_pair_dpd.html, "fix gld"_fix_gld.html
1170-80 (2010)
:link(GLE4MD)
[(GLE4MD)] "http://epfl-cosmo.github.io/gle4md/"_http://epfl-cosmo.github.io/gle4md/
[(GLE4MD)] "http://gle4md.org/"_http://gle4md.org/
:link(Ceriotti2)
[(Ceriotti2)] Ceriotti, Bussi and Parrinello, Phys Rev Lett 103,

4
doc/src/fix_qeq.txt
View File

@ -74,7 +74,7 @@ NOTE: The "fix qeq/comb"_fix_qeq_comb.html command must still be used
to perform charge equilibration with the "COMB
potential"_pair_comb.html. The "fix qeq/reax"_fix_qeq_reax.html
command can be used to perform charge equilibration with the "ReaxFF
force field"_pair_reax_c.html, although fix qeq/shielded yields the
force field"_pair_reaxc.html, although fix qeq/shielded yields the
same results as fix qeq/reax if {Nevery}, {cutoff}, and {tolerance}
are the same. Eventually the fix qeq/reax command will be deprecated.
@ -116,7 +116,7 @@ the shielded Coulomb is given by equation (13) of the "ReaxFF force
field"_#vanDuin paper. The shielding accounts for charge overlap
between charged particles at small separation. This style is the same
as "fix qeq/reax"_fix_qeq_reax.html, and can be used with "pair_style
reax/c"_pair_reax_c.html. Only the {chi}, {eta}, and {gamma}
reax/c"_pair_reaxc.html. Only the {chi}, {eta}, and {gamma}
parameters from the {qfile} file are used. This style solves partial
charges on atoms via the matrix inversion method. A tolerance of
1.0e-6 is usually a good number.

6
doc/src/fix_qeq_reax.txt
View File

@ -30,7 +30,7 @@ fix 1 all qeq/reax 1 0.0 10.0 1.0e-6 param.qeq :pre
Perform the charge equilibration (QEq) method as described in "(Rappe
and Goddard)"_#Rappe2 and formulated in "(Nakano)"_#Nakano2. It is
typically used in conjunction with the ReaxFF force field model as
implemented in the "pair_style reax/c"_pair_reax_c.html command, but
implemented in the "pair_style reax/c"_pair_reaxc.html command, but
it can be used with any potential in LAMMPS, so long as it defines and
uses charges on each atom. The "fix qeq/comb"_fix_qeq_comb.html
command should be used to perform charge equilibration with the "COMB
@ -42,7 +42,7 @@ The QEq method minimizes the electrostatic energy of the system by
adjusting the partial charge on individual atoms based on interactions
with their neighbors. It requires some parameters for each atom type.
If the {params} setting above is the word "reax/c", then these are
extracted from the "pair_style reax/c"_pair_reax_c.html command and
extracted from the "pair_style reax/c"_pair_reaxc.html command and
the ReaxFF force field file it reads in. If a file name is specified
for {params}, then the parameters are taken from the specified file
and the file must contain one line for each atom type. The latter
@ -106,7 +106,7 @@ be used for periodic cell dimensions less than 10 angstroms.
[Related commands:]
"pair_style reax/c"_pair_reax_c.html
"pair_style reax/c"_pair_reaxc.html
[Default:] none

6
doc/src/fix_reax_bonds.txt
View File

@ -28,7 +28,7 @@ fix 1 all reax/c/bonds 100 bonds.reaxc :pre
Write out the bond information computed by the ReaxFF potential
specified by "pair_style reax"_pair_reax.html or "pair_style
reax/c"_pair_reax_c.html in the exact same format as the original
reax/c"_pair_reaxc.html in the exact same format as the original
stand-alone ReaxFF code of Adri van Duin. The bond information is
written to {filename} on timesteps that are multiples of {Nevery},
including timestep 0. For time-averaged chemical species analysis,
@ -80,7 +80,7 @@ reax"_pair_reax.html be invoked. This fix is part of the REAX
package. It is only enabled if LAMMPS was built with that package,
which also requires the REAX library be built and linked with LAMMPS.
The fix reax/c/bonds command requires that the "pair_style
reax/c"_pair_reax_c.html be invoked. This fix is part of the
reax/c"_pair_reaxc.html be invoked. This fix is part of the
USER-REAXC package. It is only enabled if LAMMPS was built with that
package. See the "Making LAMMPS"_Section_start.html#start_3 section
for more info.
@ -88,6 +88,6 @@ for more info.
[Related commands:]
"pair_style reax"_pair_reax.html, "pair_style
reax/c"_pair_reax_c.html, "fix reax/c/species"_fix_reaxc_species.html
reax/c"_pair_reaxc.html, "fix reax/c/species"_fix_reaxc_species.html
[Default:] none

10
doc/src/fix_reaxc_species.txt
View File

@ -41,7 +41,7 @@ fix 1 all reax/c/species 1 100 100 species.out element Au O H position 1000 AuOH
[Description:]
Write out the chemical species information computed by the ReaxFF
potential specified by "pair_style reax/c"_pair_reax_c.html.
potential specified by "pair_style reax/c"_pair_reaxc.html.
Bond-order values (either averaged or instantaneous, depending on
value of {Nrepeat}) are used to determine chemical bonds. Every
{Nfreq} timesteps, chemical species information is written to
@ -65,7 +65,7 @@ symbol printed for each LAMMPS atom type. The number of symbols must
match the number of LAMMPS atom types and each symbol must consist of
1 or 2 alphanumeric characters. Normally, these symbols should be
chosen to match the chemical identity of each LAMMPS atom type, as
specified using the "reax/c pair_coeff"_pair_reax_c.html command and
specified using the "reax/c pair_coeff"_pair_reaxc.html command and
the ReaxFF force field file.
The optional keyword {position} writes center-of-mass positions of
@ -158,8 +158,8 @@ more instructions on how to use the accelerated styles effectively.
[Restrictions:]
The fix species currently only works with
"pair_style reax/c"_pair_reax_c.html and it requires that the "pair_style
reax/c"_pair_reax_c.html be invoked. This fix is part of the
"pair_style reax/c"_pair_reaxc.html and it requires that the "pair_style
reax/c"_pair_reaxc.html be invoked. This fix is part of the
USER-REAXC package. It is only enabled if LAMMPS was built with that
package. See the "Making LAMMPS"_Section_start.html#start_3 section
for more info.
@ -170,7 +170,7 @@ It should be possible to extend it to other reactive pair_styles (such as
[Related commands:]
"pair_style reax/c"_pair_reax_c.html, "fix
"pair_style reax/c"_pair_reaxc.html, "fix
reax/bonds"_fix_reax_bonds.html
[Default:]

5
doc/src/improper_cossq.txt
View File

@ -45,12 +45,9 @@ above, or in the data file or restart files read by the
"read_data"_read_data.html or "read_restart"_read_restart.html
commands:
K (energy/radian^2)
K (energy)
X0 (degrees) :ul
X0 is specified in degrees, but LAMMPS converts it to radians
internally; hence the units of K are in energy/radian^2.
:line
Styles with a {gpu}, {intel}, {kk}, {omp}, or {opt} suffix are

5
doc/src/improper_ring.txt
View File

@ -49,12 +49,9 @@ above, or in the data file or restart files read by the
"read_data"_read_data.html or "read_restart"_read_restart.html
commands:
K (energy/radian^2)
K (energy)
theta0 (degrees) :ul
theta0 is specified in degrees, but LAMMPS converts it to radians
internally; hence the units of K are in energy/radian^2.
:line
Styles with a {gpu}, {intel}, {kk}, {omp}, or {opt} suffix are

2
doc/src/lammps.book
View File

@ -469,7 +469,7 @@ pair_peri.html
pair_polymorphic.html
pair_quip.html
pair_reax.html
pair_reax_c.html
pair_reaxc.html
pair_resquared.html
pair_sdk.html
pair_smd_hertz.html

8
doc/src/pair_hybrid.txt
View File

@ -73,7 +73,7 @@ pair_coeff command to assign parameters for the different type pairs.
NOTE: There are two exceptions to this option to list an individual
pair style multiple times. The first is for pair styles implemented
as Fortran libraries: "pair_style meam"_pair_meam.html and "pair_style
reax"_pair_reax.html ("pair_style reax/c"_pair_reax_c.html is OK).
reax"_pair_reax.html ("pair_style reax/c"_pair_reaxc.html is OK).
This is because unlike a C++ class, they can not be instantiated
multiple times, due to the manner in which they were coded in Fortran.
The second is for GPU-enabled pair styles in the GPU package. This is
@ -225,6 +225,12 @@ special_bonds lj/coul 1e-20 1e-20 0.5
pair_hybrid tersoff lj/cut/coul/long 12.0
pair_modify pair tersoff special lj/coul 1.0 1.0 1.0 :pre
For use with the various "compute */tally"_compute_tally.html
computes, the "pair_modify compute/tally"_pair_modify.html
command can be used to selectively turn off processing of
the compute tally styles, for example, if those pair styles
(e.g. manybody styles) do not support this feature.
See the "pair_modify"_pair_modify.html doc page for details on
the specific syntax, requirements and restrictions.

29
doc/src/pair_modify.txt
View File

@ -15,11 +15,13 @@ pair_modify keyword values ... :pre
one or more keyword/value pairs may be listed :ulb,l
keyword = {pair} or {shift} or {mix} or {table} or {table/disp} or {tabinner} or {tabinner/disp} or {tail} or {compute} :l
{pair} values = sub-style N {special} which wt1 wt2 wt3
or sub-style N {compute/tally} flag
sub-style = sub-style of "pair hybrid"_pair_hybrid.html
N = which instance of sub-style (only if sub-style is used multiple times)
{special} which wt1 wt2 wt3 = override {special_bonds} settings (optional)
which = {lj/coul} or {lj} or {coul}
w1,w2,w3 = 1-2, 1-3, and 1-4 weights from 0.0 to 1.0 inclusive
{special} which wt1 wt2 wt3 = override {special_bonds} settings (optional)
which = {lj/coul} or {lj} or {coul}
w1,w2,w3 = 1-2, 1-3, and 1-4 weights from 0.0 to 1.0 inclusive
{compute/tally} flag = {yes} or {no}
{mix} value = {geometric} or {arithmetic} or {sixthpower}
{shift} value = {yes} or {no}
{table} value = N
@ -40,6 +42,7 @@ pair_modify shift yes mix geometric
pair_modify tail yes
pair_modify table 12
pair_modify pair lj/cut compute no
pair_modify pair tersoff compute/tally no
pair_modify pair lj/cut/coul/long 1 special lj/coul 0.0 0.0 0.0 :pre
[Description:]
@ -60,9 +63,12 @@ keywords will be applied to. Note that if the {pair} keyword is not
used, and the pair style is {hybrid} or {hybrid/overlay}, then all the
specified keywords will be applied to all sub-styles.
The {special} keyword can only be used in conjunction with the {pair}
keyword and must directly follow it. It allows to override the
The {special} and {compute/tally} keywords can [only] be used in
conjunction with the {pair} keyword and must directly follow it.
{special} allows to override the
"special_bonds"_special_bonds.html settings for the specified sub-style.
{compute/tally} allows to disable or enable registering
"compute */tally"_compute_tally.html computes for a given sub-style.
More details are given below.
The {mix} keyword affects pair coefficients for interactions between
@ -231,6 +237,14 @@ setting. Substituting 1.0e-10 for 0.0 and 0.9999999999 for 1.0 is
usually a sufficient workaround in this case without causing a
significant error.
The {compute/tally} keyword takes exactly 1 argument ({no} or {yes}),
and allows to selectively disable or enable processing of the various
"compute */tally"_compute_tally.html styles for a given
"pair hybrid or hybrid/overlay"_pair_hybrid.html sub-style.
NOTE: Any "pair_modify pair compute/tally" command must be issued
[before] the corresponding compute style is defined.
:line
[Restrictions:] none
@ -240,8 +254,9 @@ conflicting options. You cannot use {tail} yes with 2d simulations.
[Related commands:]
"pair_style"_pair_style.html, "pair_coeff"_pair_coeff.html,
"thermo_style"_thermo_style.html
"pair_style"_pair_style.html, "pair_style hybrid"_pair_hybrid.html,
pair_coeff"_pair_coeff.html, "thermo_style"_thermo_style.html,
"compute */tally"_compute_tally.html
[Default:]

6
doc/src/pair_reax.txt
View File

@ -36,7 +36,7 @@ supplemental information of the following paper:
the most up-to-date version of ReaxFF as of summer 2010.
WARNING: pair style reax is now deprecated and will soon be retired. Users
should switch to "pair_style reax/c"_pair_reax_c.html. The {reax} style
should switch to "pair_style reax/c"_pair_reaxc.html. The {reax} style
differs from the {reax/c} style in the lo-level implementation details.
The {reax} style is a
Fortran library, linked to LAMMPS. The {reax/c} style was initially
@ -82,7 +82,7 @@ be specified.
Two examples using {pair_style reax} are provided in the examples/reax
sub-directory, along with corresponding examples for
"pair_style reax/c"_pair_reax_c.html. Note that while the energy and force
"pair_style reax/c"_pair_reaxc.html. Note that while the energy and force
calculated by both of these pair styles match very closely, the
contributions due to the valence angles differ slightly due to
the fact that with {pair_style reax/c} the default value of {thb_cutoff_sq}
@ -201,7 +201,7 @@ appropriate units if your simulation doesn't use "real" units.
[Related commands:]
"pair_coeff"_pair_coeff.html, "pair_style reax/c"_pair_reax_c.html,
"pair_coeff"_pair_coeff.html, "pair_style reax/c"_pair_reaxc.html,
"fix_reax_bonds"_fix_reax_bonds.html
[Default:]

10
doc/src/pair_reax_c.txt → doc/src/pair_reaxc.txt
View File

@ -17,6 +17,7 @@ cfile = NULL or name of a control file :ulb,l
zero or more keyword/value pairs may be appended :l
keyword = {checkqeq} or {lgvdw} or {safezone} or {mincap}
{checkqeq} value = {yes} or {no} = whether or not to require qeq/reax fix
{enobonds} value = {yes} or {no} = whether or not to tally energy of atoms with no bonds
{lgvdw} value = {yes} or {no} = whether or not to use a low gradient vdW correction
{safezone} = factor used for array allocation
{mincap} = minimum size for array allocation :pre
@ -127,6 +128,13 @@ recommended value for parameter {thb} is 0.01, which can be set in the
control file. Note: Force field files are different for the original
or lg corrected pair styles, using wrong ffield file generates an error message.
Using the optional keyword {enobonds} with the value {yes}, the energy
of atoms with no bonds (i.e. isolated atoms) is included in the total
potential energy and the per-atom energy of that atom. If the value
{no} is specified then the energy of atoms with no bonds is set to zero.
The latter behavior is usual not desired, as it causes discontinuities
in the potential energy when the bonding of an atom drops to zero.
Optional keywords {safezone} and {mincap} are used for allocating
reax/c arrays. Increasing these values can avoid memory problems, such
as segmentation faults and bondchk failed errors, that could occur under
@ -331,7 +339,7 @@ reax"_pair_reax.html
[Default:]
The keyword defaults are checkqeq = yes, lgvdw = no, safezone = 1.2,
The keyword defaults are checkqeq = yes, enobonds = yes, lgvdw = no, safezone = 1.2,
mincap = 50.
:line

2
doc/src/pair_sdk.txt
View File

@ -134,7 +134,7 @@ respa"_run_style.html command.
[Restrictions:]
All of the lj/sdk pair styles are part of the USER-CG-CMM package.
All of the lj/sdk pair styles are part of the USER-CGSDK package.
The {lj/sdk/coul/long} style also requires the KSPACE package to be
built (which is enabled by default). They are only enabled if LAMMPS
was built with that package. See the "Making

2
doc/src/pair_srp.txt
View File

@ -150,6 +150,8 @@ hybrid"_pair_hybrid.html.
This pair style requires the "newton"_newton.html command to be {on}
for non-bonded interactions.
This pair style is not compatible with "rigid body integrators"_fix_rigid.html
[Related commands:]
"pair_style hybrid"_pair_hybrid.html, "pair_coeff"_pair_coeff.html,

2
doc/src/pairs.txt
View File

@ -73,7 +73,7 @@ Pair Styles :h1
pair_polymorphic
pair_quip
pair_reax
pair_reax_c
pair_reaxc
pair_resquared
pair_sdk
pair_smd_hertz

2
doc/src/python.txt
View File

@ -305,7 +305,7 @@ which corresponds to SELF in the python command. The first line of
the function imports the Python module lammps.py in the python dir of
the distribution. The second line creates a Python object "lmp" which
wraps the instance of LAMMPS that called the function. The
"ptr=lmpptr" argument is what makes that happen. The thrid line
"ptr=lmpptr" argument is what makes that happen. The third line
invokes the command() function in the LAMMPS library interface. It
takes a single string argument which is a LAMMPS input script command
for LAMMPS to execute, the same as if it appeared in your input

2
examples/USER/cg-cmm/README → examples/USER/cgsdk/README
View File

@ -1,4 +1,4 @@
LAMMPS USER-CMM-CG example problems
LAMMPS USER-CGSDK example problems
Each of these sub-directories contains a sample problem for the SDK
coarse grained MD potentials that you can run with LAMMPS.

0
examples/USER/cg-cmm/peg-verlet/data.pegc12e8.gz → examples/USER/cgsdk/peg-verlet/data.pegc12e8.gz
View File

0
examples/USER/cg-cmm/peg-verlet/in.pegc12e8 → examples/USER/cgsdk/peg-verlet/in.pegc12e8
View File

0
examples/USER/cg-cmm/peg-verlet/in.pegc12e8-angle → examples/USER/cgsdk/peg-verlet/in.pegc12e8-angle
View File

0
examples/USER/cg-cmm/peg-verlet/log.pegc12e8 → examples/USER/cgsdk/peg-verlet/log.pegc12e8
View File

0
examples/USER/cg-cmm/peg-verlet/log.pegc12e8-angle → examples/USER/cgsdk/peg-verlet/log.pegc12e8-angle
View File

0
examples/USER/cg-cmm/sds-monolayer/data.sds.gz → examples/USER/cgsdk/sds-monolayer/data.sds.gz
View File

0
examples/USER/cg-cmm/sds-monolayer/in.sds-hybrid → examples/USER/cgsdk/sds-monolayer/in.sds-hybrid
View File

0
examples/USER/cg-cmm/sds-monolayer/in.sds-regular → examples/USER/cgsdk/sds-monolayer/in.sds-regular
View File

0
examples/USER/cg-cmm/sds-monolayer/log.sds-hybrid → examples/USER/cgsdk/sds-monolayer/log.sds-hybrid
View File

0
examples/USER/cg-cmm/sds-monolayer/log.sds-regular → examples/USER/cgsdk/sds-monolayer/log.sds-regular
View File

0
examples/USER/flow_gauss/README → examples/USER/misc/flow_gauss/README
View File

0
examples/USER/flow_gauss/in.GD → examples/USER/misc/flow_gauss/in.GD
View File

205
examples/cmap/log.11Apr17.cmap.g++.1 Normal file
View File

@ -0,0 +1,205 @@
LAMMPS (31 Mar 2017)
# Created by charmm2lammps v1.8.2.6 beta on Thu Mar 3 20:56:57 EST 2016
units real
neigh_modify delay 2 every 1
#newton off
boundary p p p
atom_style full
bond_style harmonic
angle_style charmm
dihedral_style charmmfsw
improper_style harmonic
pair_style lj/charmmfsw/coul/charmmfsh 8 12
pair_modify mix arithmetic
fix cmap all cmap charmm22.cmap
Reading potential file charmm22.cmap with DATE: 2016-09-26
fix_modify cmap energy yes
read_data gagg.data fix cmap crossterm CMAP
orthogonal box = (-34.4147 -36.1348 -39.3491) to (45.5853 43.8652 40.6509)
1 by 1 by 1 MPI processor grid
reading atoms ...
34 atoms
scanning bonds ...
4 = max bonds/atom
scanning angles ...
6 = max angles/atom
scanning dihedrals ...
12 = max dihedrals/atom
scanning impropers ...
1 = max impropers/atom
reading bonds ...
33 bonds
reading angles ...
57 angles
reading dihedrals ...
75 dihedrals
reading impropers ...
7 impropers
4 = max # of 1-2 neighbors
7 = max # of 1-3 neighbors
13 = max # of 1-4 neighbors
16 = max # of special neighbors
special_bonds charmm
fix 1 all nve
#fix 1 all nvt temp 300 300 100.0
#fix 2 all shake 1e-9 500 0 m 1.0
velocity all create 0.0 12345678 dist uniform
thermo 1000
thermo_style custom step ecoul evdwl ebond eangle edihed f_cmap eimp
timestep 2.0
run 100000
Neighbor list info ...
update every 1 steps, delay 2 steps, check yes
max neighbors/atom: 2000, page size: 100000
master list distance cutoff = 14
ghost atom cutoff = 14
binsize = 7, bins = 12 12 12
1 neighbor lists, perpetual/occasional/extra = 1 0 0
(1) pair lj/charmmfsw/coul/charmmfsh, perpetual
attributes: half, newton on
pair build: half/bin/newton
stencil: half/bin/3d/newton
bin: standard
Per MPI rank memory allocation (min/avg/max) = 14.96 | 14.96 | 14.96 Mbytes
Step E_coul E_vdwl E_bond E_angle E_dihed f_cmap E_impro
0 16.287573 -0.85933785 1.2470497 4.8441789 4.5432816 -1.473352 0.10453023
1000 18.816462 -0.84379243 0.78931817 2.7554247 4.4371421 -2.7762038 0.12697656
2000 18.091571 -1.045888 0.72306589 3.0951524 4.6725102 -2.3580092 0.22712496
3000 17.835596 -1.2171641 0.72666403 2.6696491 5.4373798 -2.0737041 0.075101693
4000 16.211232 -0.42713611 0.99472642 3.8961462 5.2009895 -2.5626866 0.17356243
5000 17.72183 -0.57081189 0.90733068 3.4376382 4.5457582 -2.3727543 0.12354518
6000 18.753977 -1.5772499 0.81468321 2.9236782 4.6033216 -2.3380859 0.12835782
7000 18.186024 -0.84205608 0.58996182 3.0329585 4.7221473 -2.5733243 0.10047631
8000 18.214306 -1.1360938 0.72597611 3.7493028 4.7319958 -2.8957969 0.2006046
9000 17.248408 -0.48641993 0.90266229 2.9721743 4.7651056 -2.1473354 0.1302043
10000 17.760655 -1.2968444 0.92384663 3.7007455 4.7378947 -2.2147779 0.06940579
11000 17.633929 -0.57368413 0.84872849 3.4277114 4.285393 -2.236944 0.17204973
12000 18.305835 -1.0675148 0.75879532 2.8853173 4.685027 -2.409087 0.087538866
13000 17.391558 -0.9975291 0.66671947 3.8065638 5.2285578 -2.4198822 0.06253594
14000 17.483387 -0.67727643 0.91966477 3.7317031 4.7770445 -2.6080027 0.11487095
15000 18.131749 -1.1918751 1.0025684 3.1238131 4.789742 -2.2546745 0.13782813
16000 16.972343 -0.43926531 0.60644597 3.7551592 4.8658618 -2.2627659 0.12353145
17000 18.080785 -1.2073565 0.7867072 3.5671106 4.43754 -2.5092904 0.17429146
18000 17.474576 -0.97836065 0.8678524 3.7961537 4.3409032 -1.8922572 0.134048
19000 17.000911 -1.2286864 0.83615834 3.9322908 4.9319492 -2.3281576 0.056689619
20000 17.043286 -0.8506561 0.80966589 3.5087339 4.8603878 -2.3365263 0.096794824
21000 17.314495 -1.1430889 0.95363892 4.2446032 4.2756745 -2.1829483 0.17119518
22000 18.954881 -0.998673 0.58688334 2.71536 4.6634319 -2.6862804 0.20328442
23000 17.160427 -0.97803282 0.86894041 4.0897736 4.3146238 -2.1962289 0.075339092
24000 17.602026 -1.0833323 0.94888776 3.7341878 4.3084335 -2.1640414 0.081493681
25000 17.845584 -1.3432612 0.93497086 3.8911043 4.468032 -2.3475883 0.093204333
26000 17.833261 -1.1020534 0.77931087 3.7628141 4.512381 -2.3134761 0.15568465
27000 17.68607 -1.3222026 1.1985872 3.5817624 4.6360755 -2.3492774 0.08427906
28000 18.326649 -1.2669291 0.74809075 3.2624429 4.4698564 -2.3679076 0.14677293
29000 17.720933 -1.0773886 0.83099482 3.7652834 4.6584594 -2.8255303 0.23092596
30000 18.201999 -1.0168706 1.0637455 3.453095 4.3738593 -2.8063214 0.18658217
31000 17.823502 -1.2685768 0.84805585 3.8600661 4.2195821 -2.1169716 0.12517101
32000 16.883133 -0.62062648 0.84434922 3.5042683 5.1264906 -2.2674699 0.030138165
33000 17.805715 -1.679553 1.2430372 4.314677 4.2523894 -2.3008321 0.18591872
34000 16.723767 -0.54189072 1.1282827 3.8542159 4.3026559 -2.2186336 0.05392425
35000 17.976909 -0.72092075 0.5876319 2.9726396 5.0881439 -2.491692 0.17356291
36000 18.782492 -1.514246 0.63237955 3.2777164 4.6077164 -2.502574 0.082537318
37000 17.247716 -0.6344626 0.79885976 3.452491 4.7618281 -2.3902444 0.11450271
38000 17.996494 -1.6712877 1.0111769 4.1689136 4.46963 -2.4076725 0.11875756
39000 17.586857 -0.74508086 0.95970486 3.7395038 4.6011357 -2.9854953 0.30143284
40000 17.494879 -0.30772446 0.72047991 3.2604877 4.7283734 -2.3812495 0.16399034
41000 15.855772 -0.49642605 0.82496448 4.5139653 4.76884 -2.214141 0.10899661
42000 17.898568 -1.3078863 1.1505144 4.0429873 4.3889581 -2.8696559 0.23336417
43000 19.014372 -1.6325979 1.1553166 3.5660772 4.4047997 -2.9302044 0.13672127
44000 18.250782 -0.97211613 0.72714301 3.2258362 4.7257298 -2.5533613 0.11968073
45000 17.335174 0.24746331 1.0415866 3.3220992 4.5251095 -3.0415216 0.24453084
46000 17.72846 -0.9541418 0.88153841 3.7893452 4.5251883 -2.4003613 0.051809816
47000 18.226762 -0.67057787 0.84352989 3.0609522 4.5449078 -2.4694254 0.073703949
48000 17.838074 -0.88768441 1.3812262 3.5890492 4.5827868 -3.0137515 0.21417113
49000 17.973733 -0.75118705 0.69667886 3.3989025 4.7058886 -2.8243945 0.26665792
50000 17.461583 -0.65040016 0.68943524 2.9374743 5.6971777 -2.4438011 0.1697603
51000 16.79766 -0.010684434 0.89795555 3.959039 4.56763 -2.5101098 0.15048853
52000 17.566543 -0.7262764 0.74354418 3.3423185 4.8426523 -2.4187649 0.16908776
53000 17.964274 -0.9270914 1.065952 3.0397181 4.4682262 -2.2179503 0.07873406
54000 17.941256 -0.5807578 0.76516121 3.7262371 4.6975126 -3.179899 0.24433708
55000 17.079478 -0.48559832 0.95364453 3.0414645 5.2811414 -2.7064882 0.30102814
56000 17.632179 -0.75403299 0.97577942 3.3672363 4.4851336 -2.3683659 0.051117638
57000 16.17128 -0.44699325 0.76341543 4.267716 5.0881056 -2.4122329 0.16671692
58000 16.899276 -0.76481024 1.0400825 3.973493 4.8823309 -2.4270284 0.048716383
59000 18.145412 -0.84968335 0.71698306 3.2024358 4.6115739 -2.2520353 0.19466966
60000 17.578258 -1.0067331 0.72822527 3.5375208 4.9110255 -2.2319607 0.11922362
61000 17.434762 -1.0244393 0.90593099 3.8446915 4.8571191 -2.6228357 0.23259208
62000 17.580489 -1.1135917 0.79577432 3.7043524 4.6058114 -2.351492 0.042904152
63000 18.207335 -1.1512268 0.82684507 3.4114738 4.351069 -2.1878441 0.082922105
64000 18.333083 -1.1182287 0.74058959 3.6905164 4.3226172 -2.7110393 0.14721704
65000 16.271579 -0.7122151 1.0200168 4.6983643 4.3681131 -2.194921 0.12831024
66000 17.316444 -0.5729385 0.85254108 3.5769963 4.5526705 -2.3321328 0.040452643
67000 17.19011 -0.8814312 1.1381258 3.8605789 4.4183813 -2.299607 0.091527355
68000 18.223367 -1.362189 0.74472056 3.259165 4.486512 -2.2181134 0.048952796
69000 17.646348 -0.91647162 0.73990335 3.9313692 5.2663097 -3.3816778 0.27769877
70000 18.173493 -1.3107718 0.96484426 3.219728 4.5045124 -2.3349534 0.082327407
71000 17.0627 -0.58509083 0.85964129 3.8490884 4.437895 -2.1673348 0.24151404
72000 17.809764 -0.35128902 0.65479258 3.3945008 4.6160508 -2.5486166 0.10829531
73000 18.27769 -1.0739758 0.80890957 3.6070901 4.6256762 -2.4576547 0.080025736
74000 18.109437 -1.0691837 0.66679323 3.5923203 4.4825716 -2.5048169 0.21372319
75000 17.914569 -1.3500765 1.2993494 3.362421 4.4160377 -2.1278163 0.19397641
76000 16.563928 -0.16539261 1.0067302 3.5742755 4.8581915 -2.1362429 0.059822408
77000 18.130477 -0.38361279 0.43406954 3.4725995 4.7005855 -2.8836242 0.11958174
78000 16.746204 -1.1732959 0.7455507 3.6296638 5.6344113 -2.459208 0.16099803
79000 18.243999 -1.5850155 1.0108545 3.4727867 4.3367411 -2.316686 0.070480814
80000 16.960715 -0.84100929 0.91604996 3.862215 4.780949 -2.3711596 0.073916605
81000 17.697722 -1.1126605 0.952804 3.7114455 4.4216316 -2.2770085 0.091372066
82000 17.835901 -1.3091474 0.71867629 3.8168122 5.0150205 -2.4730634 0.062592852
83000 19.168418 -1.476938 0.75592316 3.2304519 4.3946471 -2.2991395 0.13083324
84000 17.945778 -1.5223622 1.0859941 3.4334011 5.0286682 -2.7550892 0.2476269
85000 17.950251 -0.85843846 0.86888218 3.3101287 4.5511879 -2.3640013 0.12080834
86000 17.480699 -0.97493649 0.85049761 3.4973085 4.6344922 -2.343121 0.2009677
87000 17.980244 -1.114983 0.88796989 3.4113329 4.3535853 -2.2535412 0.14494917
88000 18.023866 -1.226683 0.62339706 3.7649269 4.5923973 -2.3923523 0.10464375
89000 16.362829 -0.311462 1.0265375 4.0101723 4.4184777 -2.0314129 0.056570704
90000 17.533149 -0.41526788 1.0362029 3.4247412 4.2734431 -2.4776658 0.16960663
91000 17.719099 -1.1956801 1.0069945 3.2380672 4.8982805 -2.2154906 0.12950936
92000 17.762654 -1.170027 0.95814525 3.5217717 4.5405343 -2.5983677 0.15037754
93000 17.393958 -0.45641026 0.6579069 3.6002204 4.5942053 -2.5559641 0.12026544
94000 16.8182 -0.92962066 0.86801362 4.2914398 4.659848 -2.5251987 0.18000415
95000 17.642086 -0.7994896 0.7003756 3.8036697 4.5252487 -2.4166307 0.15686517
96000 18.114292 -1.5102104 1.2635908 3.2764427 5.0659496 -2.2777806 0.054309645
97000 18.575765 -1.6015311 0.69500699 3.1649317 4.9945742 -2.4012125 0.067373724
98000 16.578893 -0.78030229 0.91524222 4.4429655 4.4622392 -2.4052655 0.15355705
99000 17.26063 -0.57832833 0.7098846 3.9000046 4.5576484 -2.5333026 0.25517222
100000 18.377235 -0.89109577 0.68988617 2.8751751 4.4115591 -2.3560731 0.12185212
Loop time of 2.96043 on 1 procs for 100000 steps with 34 atoms
Performance: 5836.990 ns/day, 0.004 hours/ns, 33778.875 timesteps/s
99.9% CPU use with 1 MPI tasks x no OpenMP threads
MPI task timing breakdown:
Section | min time | avg time | max time |%varavg| %total
---------------------------------------------------------------
Pair | 1.074 | 1.074 | 1.074 | 0.0 | 36.28
Bond | 1.6497 | 1.6497 | 1.6497 | 0.0 | 55.72
Neigh | 0.007576 | 0.007576 | 0.007576 | 0.0 | 0.26
Comm | 0.012847 | 0.012847 | 0.012847 | 0.0 | 0.43
Output | 0.0010746 | 0.0010746 | 0.0010746 | 0.0 | 0.04
Modify | 0.16485 | 0.16485 | 0.16485 | 0.0 | 5.57
Other | | 0.05037 | | | 1.70
Nlocal: 34 ave 34 max 34 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Nghost: 0 ave 0 max 0 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Neighs: 395 ave 395 max 395 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Total # of neighbors = 395
Ave neighs/atom = 11.6176
Ave special neighs/atom = 9.52941
Neighbor list builds = 253
Dangerous builds = 0
Total wall time: 0:00:02

205
examples/cmap/log.11Apr17.cmap.g++.4 Normal file
View File

@ -0,0 +1,205 @@
LAMMPS (31 Mar 2017)
# Created by charmm2lammps v1.8.2.6 beta on Thu Mar 3 20:56:57 EST 2016
units real
neigh_modify delay 2 every 1
#newton off
boundary p p p
atom_style full
bond_style harmonic
angle_style charmm
dihedral_style charmmfsw
improper_style harmonic
pair_style lj/charmmfsw/coul/charmmfsh 8 12
pair_modify mix arithmetic
fix cmap all cmap charmm22.cmap
Reading potential file charmm22.cmap with DATE: 2016-09-26
fix_modify cmap energy yes
read_data gagg.data fix cmap crossterm CMAP
orthogonal box = (-34.4147 -36.1348 -39.3491) to (45.5853 43.8652 40.6509)
1 by 2 by 2 MPI processor grid
reading atoms ...
34 atoms
scanning bonds ...
4 = max bonds/atom
scanning angles ...
6 = max angles/atom
scanning dihedrals ...
12 = max dihedrals/atom
scanning impropers ...
1 = max impropers/atom
reading bonds ...
33 bonds
reading angles ...
57 angles
reading dihedrals ...
75 dihedrals
reading impropers ...
7 impropers
4 = max # of 1-2 neighbors
7 = max # of 1-3 neighbors
13 = max # of 1-4 neighbors
16 = max # of special neighbors
special_bonds charmm
fix 1 all nve
#fix 1 all nvt temp 300 300 100.0
#fix 2 all shake 1e-9 500 0 m 1.0
velocity all create 0.0 12345678 dist uniform
thermo 1000
thermo_style custom step ecoul evdwl ebond eangle edihed f_cmap eimp
timestep 2.0
run 100000
Neighbor list info ...
update every 1 steps, delay 2 steps, check yes
max neighbors/atom: 2000, page size: 100000
master list distance cutoff = 14
ghost atom cutoff = 14
binsize = 7, bins = 12 12 12
1 neighbor lists, perpetual/occasional/extra = 1 0 0
(1) pair lj/charmmfsw/coul/charmmfsh, perpetual
attributes: half, newton on
pair build: half/bin/newton
stencil: half/bin/3d/newton
bin: standard
Per MPI rank memory allocation (min/avg/max) = 14.94 | 15.57 | 16.2 Mbytes
Step E_coul E_vdwl E_bond E_angle E_dihed f_cmap E_impro
0 16.287573 -0.85933785 1.2470497 4.8441789 4.5432816 -1.473352 0.10453023
1000 18.816462 -0.84379243 0.78931817 2.7554247 4.4371421 -2.7762038 0.12697656
2000 18.091571 -1.045888 0.72306589 3.0951524 4.6725102 -2.3580092 0.22712496
3000 17.835596 -1.2171641 0.72666403 2.6696491 5.4373798 -2.0737041 0.075101693
4000 16.211232 -0.42713611 0.99472642 3.8961462 5.2009895 -2.5626866 0.17356243
5000 17.72183 -0.57081189 0.90733068 3.4376382 4.5457582 -2.3727543 0.12354518
6000 18.753977 -1.5772499 0.81468321 2.9236782 4.6033216 -2.3380859 0.12835782
7000 18.186024 -0.84205609 0.58996181 3.0329584 4.7221473 -2.5733244 0.10047631
8000 18.214306 -1.1360934 0.72597583 3.7493032 4.7319959 -2.8957975 0.20060467
9000 17.248415 -0.48642024 0.90266262 2.9721744 4.7651003 -2.1473349 0.13020438
10000 17.760663 -1.2968458 0.92384687 3.7007432 4.7378917 -2.2147799 0.06940514
11000 17.63395 -0.57366075 0.84871737 3.4276851 4.2853865 -2.2369491 0.17205075
12000 18.305713 -1.0672299 0.75876262 2.8852171 4.6850229 -2.4090072 0.087568888
13000 17.383367 -0.99678627 0.66712651 3.8060954 5.233865 -2.4180629 0.062014239
14000 17.510901 -0.68723297 0.92448551 3.7550867 4.7321218 -2.6059088 0.11504409
15000 18.080165 -1.13316 0.99982253 3.09947 4.8171402 -2.2713372 0.14580371
16000 17.383245 -0.4535296 0.57826268 3.6453593 4.6541138 -2.2434512 0.13285609
17000 17.111153 -0.3414839 0.73667584 3.7485311 4.6262965 -2.6166049 0.12635815
18000 16.862046 -1.3592061 1.2371142 4.4878937 4.2937117 -2.2112584 0.066145125
19000 18.313891 -1.654238 0.90644101 3.3934089 4.550735 -2.1862171 0.081267736
20000 19.083561 -1.3081747 0.56257812 2.7633848 4.6211438 -2.5196707 0.13763071
21000 18.23741 -1.051353 0.64408722 3.1735565 4.6912533 -2.2491947 0.099394904
22000 17.914515 -0.89769621 0.61793801 3.1224992 4.8683543 -2.282475 0.14524537
23000 16.756122 -0.98277883 1.2554905 3.7916115 4.7301443 -2.3094994 0.10226772
24000 16.109857 -0.54593177 0.86934462 4.4293574 4.926985 -2.2652264 0.11414331
25000 18.590559 -1.497327 1.1898361 2.9134403 4.7854107 -2.4437918 0.067416154
26000 18.493391 -1.0533797 0.4889578 3.6563013 4.6171721 -2.3240835 0.11607829
27000 18.646522 -1.1229601 0.67956815 2.7937638 4.8991207 -2.4068997 0.10109147
28000 18.545103 -1.7237438 0.72488022 3.8041665 4.6459974 -2.4339333 0.21943258
29000 17.840505 -1.0909667 0.88133248 3.3698456 5.0311644 -2.5116617 0.08102693
30000 17.649527 -0.65409177 0.86781692 3.24112 4.9903073 -2.6234925 0.14799777
31000 18.156812 -0.77476556 0.83192789 2.9620784 4.9160635 -2.8571635 0.22283201
32000 18.251583 -1.3384075 0.8059007 3.2588176 4.4365328 -2.1875071 0.087883637
33000 17.702785 -0.88311587 0.98573641 3.4645713 4.2650091 -2.0909158 0.14233004
34000 17.123413 -1.4873429 1.0419563 4.2628178 4.6318762 -2.2292095 0.105354
35000 18.162061 -1.0136007 0.82436129 3.6365024 4.5801677 -2.6856989 0.28648222
36000 17.65618 -1.094718 0.8872444 3.5075241 4.6382423 -2.3895134 0.18116961
37000 17.336475 -1.0657995 0.98869254 3.9252927 4.4383632 -2.2048244 0.22285949
38000 17.369467 -0.97623132 0.6712095 4.1349304 4.597754 -2.4088341 0.14608514
39000 18.170206 -1.2344285 0.77546195 3.6451049 4.7482287 -2.9895286 0.25768859
40000 16.210866 -0.81407781 0.99246271 4.2676233 5.0253763 -2.2929865 0.13348624
41000 17.641798 -1.0868157 0.80119513 3.4302526 5.280872 -2.4025406 0.22747391
42000 18.349848 -1.613759 1.1497004 3.7800682 4.3237683 -2.8676401 0.2120425
43000 19.130245 -1.196778 0.71845659 2.9325758 4.3684415 -2.433424 0.12240982
44000 18.061321 -1.2410101 1.0329373 3.0751569 4.7138313 -2.2880904 0.075814461
45000 18.162713 -1.4414622 1.009159 4.2298758 4.589593 -2.8502298 0.21606844
46000 18.591574 -0.99730412 1.0955215 3.3965004 4.359466 -3.1049731 0.17322629
47000 18.380259 -1.2717381 0.72291269 3.3958016 4.6099628 -2.4605065 0.19825185
48000 18.130478 -1.5051279 1.2087492 3.2488529 4.6690881 -2.2518174 0.05633061
49000 16.419912 -0.89320635 0.98926144 4.0388252 4.9919488 -2.1699511 0.15646479
50000 16.453196 -1.0433497 0.778346 4.6078069 4.7320614 -2.3760788 0.17161976
51000 18.245221 -0.89550444 0.9310446 3.0758194 4.3944595 -2.3082379 0.19983428
52000 17.839632 -1.0221781 0.76425017 3.3331547 4.5368437 -2.0988773 0.21098435
53000 18.693035 -1.4231915 0.76333082 3.1612761 4.583242 -2.4485762 0.089191206
54000 16.334672 -0.36309884 1.0200365 4.6700448 4.1628702 -2.1713841 0.11431995
55000 17.33842 -0.61522682 0.89847366 3.4970659 4.673495 -2.4743036 0.068004878
56000 17.790294 -1.0150845 0.73697112 3.6000297 4.5988343 -2.4822509 0.11434632
57000 18.913486 -1.0985507 1.0231848 2.7483267 4.4421755 -2.574424 0.1763388
58000 17.586896 -0.98284126 0.96965633 3.3330357 4.5325543 -2.1936869 0.083230915
59000 17.77788 -1.1649953 0.83092298 3.8004148 4.3940176 -2.3136642 0.017207608
60000 17.013042 -0.21728023 1.1688832 3.5374476 4.5462244 -2.4425301 0.15028297
61000 17.236242 -1.1342147 1.0301086 3.685948 4.6842331 -2.328108 0.070210812
62000 17.529852 -1.2961547 1.0323133 3.4474598 5.1435839 -2.4553423 0.060842687
63000 18.754704 -1.1816999 0.51806039 3.140172 4.5832701 -2.2713213 0.06327871
64000 17.54594 -1.3592836 0.9694558 4.1363258 4.3547729 -2.3818433 0.12634448
65000 16.962312 -0.54192775 0.90321315 4.0788618 4.2008255 -2.1376711 0.039504515
66000 18.078619 -1.3552947 1.0716861 3.3285374 4.7229362 -2.3331115 0.21978698
67000 17.132732 -1.4376876 0.91486534 4.4461852 4.6894176 -2.3655045 0.068150385
68000 18.69286 -1.2856207 0.3895394 3.0620063 4.9922992 -2.3459189 0.079879643
69000 18.329552 -1.1545957 0.88632275 3.1741058 4.4562418 -2.7094867 0.25329613
70000 16.681168 -0.94434373 1.2450393 4.5737944 4.4902996 -2.4581775 0.15313095
71000 17.375032 -1.0514442 1.0741595 3.4896146 4.8407713 -2.5302576 0.13640847
72000 17.833013 -0.9047134 0.87067876 3.1658924 4.8825932 -2.4398117 0.2343991
73000 17.421411 -1.2190741 0.73706811 4.2895 4.6464636 -2.3872727 0.19696525
74000 17.383158 -0.34208984 0.71333984 3.2718891 4.2718495 -2.2484281 0.10827022
75000 17.20885 -1.2710479 1.125102 3.8414467 5.3222741 -2.375505 0.12910797
76000 16.811578 -0.545162 0.59076961 3.9118604 4.8031296 -2.2777895 0.063015508
77000 16.679231 -0.080955983 0.7253398 3.4203454 5.0987608 -2.379614 0.12961874
78000 18.164524 -1.3115525 0.92526408 3.5764487 4.3814882 -2.3712488 0.073436724
79000 17.738686 -1.0697859 1.2186866 3.0593848 4.6551053 -2.2505871 0.075340661
80000 16.767483 -0.84777477 1.03128 4.1982958 4.6992227 -2.4146425 0.079774219
81000 16.257265 0.62803774 0.84032194 3.3873471 5.0961071 -2.7219776 0.20467848
82000 18.232082 -1.2129302 0.50746051 3.9207128 4.5073437 -2.599371 0.094522372
83000 16.618985 -0.60917055 0.8825847 3.805497 4.9560959 -2.2194726 0.14852687
84000 17.90762 -0.82336075 0.90504161 3.0324198 4.7444271 -2.5036073 0.15860682
85000 16.699883 -0.50297228 0.83405307 3.8598996 4.7971968 -2.2427788 0.10338668
86000 16.353038 -0.0096880616 0.80705167 4.0865115 4.5364338 -2.4548873 0.098456203
87000 17.887331 -0.75281219 1.0030148 4.0117123 4.3443074 -2.9774392 0.16190152
88000 18.583708 -1.4867053 0.86324814 3.3971237 4.3526221 -2.221239 0.14459352
89000 17.684828 -1.283764 1.0021118 3.5426808 4.9057005 -2.3921967 0.05844702
90000 17.2597 -0.84306489 0.99797936 3.8896866 4.4315457 -2.5662899 0.18270206
91000 16.705581 -0.44704047 0.75239556 3.470805 4.976868 -2.1894571 0.12312848
92000 17.548071 -1.2222664 0.92898812 4.0813773 4.3432647 -2.1631158 0.14071343
93000 17.163675 -0.94994776 0.96876981 3.9137692 4.4388666 -2.1260232 0.13187968
94000 18.842071 -1.2822113 0.58767049 3.1393475 4.5820965 -2.7264682 0.10406266
95000 18.112287 -1.1011381 0.63546648 3.4672667 4.486275 -2.2991936 0.041589685
96000 17.102713 -0.6877313 0.8389032 3.6892719 4.5676004 -2.1905327 0.13507011
97000 16.778253 -1.2902153 1.1588744 4.2820083 4.9537657 -2.4798159 0.35696636
98000 18.34638 -1.2908146 1.185356 3.0739807 4.4575453 -2.3959144 0.22407922
99000 17.995148 -1.3939639 0.7727299 3.8774144 4.4345458 -2.1142776 0.13550099
100000 18.444746 -1.2456693 0.86061526 3.468696 4.5264336 -2.4239851 0.074369539
Loop time of 2.52011 on 4 procs for 100000 steps with 34 atoms
Performance: 6856.851 ns/day, 0.004 hours/ns, 39680.850 timesteps/s
98.8% CPU use with 4 MPI tasks x no OpenMP threads
MPI task timing breakdown:
Section | min time | avg time | max time |%varavg| %total
---------------------------------------------------------------
Pair | 0.072506 | 0.28131 | 0.69088 | 46.2 | 11.16
Bond | 0.050544 | 0.45307 | 0.9416 | 57.6 | 17.98
Neigh | 0.0060885 | 0.0061619 | 0.0062056 | 0.1 | 0.24
Comm | 0.44686 | 1.3679 | 2.0111 | 53.5 | 54.28
Output | 0.0028057 | 0.0029956 | 0.003264 | 0.3 | 0.12
Modify | 0.028202 | 0.095174 | 0.15782 | 19.8 | 3.78
Other | | 0.3135 | | | 12.44
Nlocal: 8.5 ave 14 max 2 min
Histogram: 1 0 1 0 0 0 0 0 0 2
Nghost: 25.5 ave 32 max 20 min
Histogram: 2 0 0 0 0 0 0 1 0 1
Neighs: 98.75 ave 242 max 31 min
Histogram: 2 0 1 0 0 0 0 0 0 1
Total # of neighbors = 395
Ave neighs/atom = 11.6176
Ave special neighs/atom = 9.52941
Neighbor list builds = 246
Dangerous builds = 0
Total wall time: 0:00:02

145
examples/mscg/log.31Mar17.g++.1 Normal file
View File

@ -0,0 +1,145 @@
LAMMPS (13 Apr 2017)
units real
atom_style full
pair_style zero 10.0
read_data data.meoh
orthogonal box = (-20.6917 -20.6917 -20.6917) to (20.6917 20.6917 20.6917)
1 by 1 by 1 MPI processor grid
reading atoms ...
1000 atoms
0 = max # of 1-2 neighbors
0 = max # of 1-3 neighbors
0 = max # of 1-4 neighbors
1 = max # of special neighbors
pair_coeff * *
thermo 1
thermo_style custom step
# Test 1a: range finder functionality
fix 1 all mscg 1 range on
rerun dump.meoh first 0 last 4500 every 250 dump x y z fx fy fz
Neighbor list info ...
update every 1 steps, delay 10 steps, check yes
max neighbors/atom: 2000, page size: 100000
master list distance cutoff = 12
ghost atom cutoff = 12
binsize = 6, bins = 7 7 7
1 neighbor lists, perpetual/occasional/extra = 1 0 0
(1) pair zero, perpetual
attributes: half, newton on
pair build: half/bin/newton
stencil: half/bin/3d/newton
bin: standard
Per MPI rank memory allocation (min/avg/max) = 5.794 | 5.794 | 5.794 Mbytes
Step
0
250
500
750
1000
1250
1500
1750
2000
2250
2500
2750
3000
3250
3500
3750
4000
4250
4500
Loop time of 0.581537 on 1 procs for 19 steps with 1000 atoms
Performance: 2.823 ns/day, 8.502 hours/ns, 32.672 timesteps/s
99.2% 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
Bond | 0 | 0 | 0 | 0.0 | 0.00
Neigh | 0 | 0 | 0 | 0.0 | 0.00
Comm | 0 | 0 | 0 | 0.0 | 0.00
Output | 0 | 0 | 0 | 0.0 | 0.00
Modify | 0 | 0 | 0 | 0.0 | 0.00
Other | | 0.5815 | | |100.00
Nlocal: 1000 ave 1000 max 1000 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Nghost: 2934 ave 2934 max 2934 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Neighs: 50654 ave 50654 max 50654 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Total # of neighbors = 50654
Ave neighs/atom = 50.654
Ave special neighs/atom = 0
Neighbor list builds = 0
Dangerous builds = 0
print "TEST_1a mscg range finder"
TEST_1a mscg range finder
unfix 1
# Test 1b: force matching functionality
fix 1 all mscg 1
rerun dump.meoh first 0 last 4500 every 250 dump x y z fx fy fz
Per MPI rank memory allocation (min/avg/max) = 5.794 | 5.794 | 5.794 Mbytes
Step
0
250
500
750
1000
1250
1500
1750
2000
2250
2500
2750
3000
3250
3500
3750
4000
4250
4500
Loop time of 0.841917 on 1 procs for 19 steps with 1000 atoms
Performance: 1.950 ns/day, 12.309 hours/ns, 22.568 timesteps/s
99.8% 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
Bond | 0 | 0 | 0 | 0.0 | 0.00
Neigh | 0 | 0 | 0 | 0.0 | 0.00
Comm | 0 | 0 | 0 | 0.0 | 0.00
Output | 0 | 0 | 0 | 0.0 | 0.00
Modify | 0 | 0 | 0 | 0.0 | 0.00
Other | | 0.8419 | | |100.00
Nlocal: 1000 ave 1000 max 1000 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Nghost: 2934 ave 2934 max 2934 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Neighs: 50654 ave 50654 max 50654 min
Histogram: 1 0 0 0 0 0 0 0 0 0
Total # of neighbors = 50654
Ave neighs/atom = 50.654
Ave special neighs/atom = 0
Neighbor list builds = 0
Dangerous builds = 0
print "TEST_1b mscg force matching"
TEST_1b mscg force matching
print TEST_DONE
TEST_DONE
Total wall time: 0:00:01

82
lib/Install.py Normal file
View File

@ -0,0 +1,82 @@
#!/usr/bin/env python
# install.py tool to do a generic build of a library
# soft linked to by many of the lib/Install.py files
# used to automate the steps described in the corresponding lib/README
import sys,commands,os
# help message
help = """
Syntax: python Install.py -m machine -e suffix
specify -m and optionally -e, order does not matter
-m = peform a clean followed by "make -f Makefile.machine"
machine = suffix of a lib/Makefile.* file
-e = set EXTRAMAKE variable in Makefile.machine to Makefile.lammps.suffix
does not alter existing Makefile.machine
"""
# print error message or help
def error(str=None):
if not str: print help
else: print "ERROR",str
sys.exit()
# parse args
args = sys.argv[1:]
nargs = len(args)
if nargs == 0: error()
machine = None
extraflag = 0
iarg = 0
while iarg < nargs:
if args[iarg] == "-m":
if iarg+2 > nargs: error()
machine = args[iarg+1]
iarg += 2
elif args[iarg] == "-e":
if iarg+2 > nargs: error()
extraflag = 1
suffix = args[iarg+1]
iarg += 2
else: error()
# set lib from working dir
cwd = os.getcwd()
lib = os.path.basename(cwd)
# create Makefile.auto as copy of Makefile.machine
# reset EXTRAMAKE if requested
if not os.path.exists("Makefile.%s" % machine):
error("lib/%s/Makefile.%s does not exist" % (lib,machine))
lines = open("Makefile.%s" % machine,'r').readlines()
fp = open("Makefile.auto",'w')
for line in lines:
words = line.split()
if len(words) == 3 and extraflag and \
words[0] == "EXTRAMAKE" and words[1] == '=':
line = line.replace(words[2],"Makefile.lammps.%s" % suffix)
print >>fp,line,
fp.close()
# make the library via Makefile.auto
print "Building lib%s.a ..." % lib
cmd = "make -f Makefile.auto clean; make -f Makefile.auto"
txt = commands.getoutput(cmd)
print txt
if os.path.exists("lib%s.a" % lib): print "Build was successful"
else: error("Build of lib/%s/lib%s.a was NOT successful" % (lib,lib))
if not os.path.exists("Makefile.lammps"):
print "lib/%s/Makefile.lammps was NOT created" % lib

6
lib/README
View File

@ -33,14 +33,16 @@ kokkos Kokkos package for GPU and many-core acceleration
from Kokkos development team (Sandia)
linalg set of BLAS and LAPACK routines needed by USER-ATC package
from Axel Kohlmeyer (Temple U)
poems POEMS rigid-body integration package, POEMS package
from Rudranarayan Mukherjee (RPI)
meam modified embedded atom method (MEAM) potential, MEAM package
from Greg Wagner (Sandia)
molfile hooks to VMD molfile plugins, used by the USER-MOLFILE package
from Axel Kohlmeyer (Temple U) and the VMD development team
mscg hooks to the MSCG library, used by fix_mscg command
from Jacob Wagner and Greg Voth group (U Chicago)
netcdf hooks to a NetCDF library installed on your system
from Lars Pastewka (Karlsruhe Institute of Technology)
poems POEMS rigid-body integration package, POEMS package
from Rudranarayan Mukherjee (RPI)
python hooks to the system Python library, used by the PYTHON package
from the LAMMPS development team
qmmm quantum mechanics/molecular mechanics coupling interface

82
lib/atc/Install.py Normal file
View File

@ -0,0 +1,82 @@
#!/usr/bin/env python
# install.py tool to do a generic build of a library
# soft linked to by many of the lib/Install.py files
# used to automate the steps described in the corresponding lib/README
import sys,commands,os
# help message
help = """
Syntax: python Install.py -m machine -e suffix
specify -m and optionally -e, order does not matter
-m = peform a clean followed by "make -f Makefile.machine"
machine = suffix of a lib/Makefile.* file
-e = set EXTRAMAKE variable in Makefile.machine to Makefile.lammps.suffix
does not alter existing Makefile.machine
"""
# print error message or help
def error(str=None):
if not str: print help
else: print "ERROR",str
sys.exit()
# parse args
args = sys.argv[1:]
nargs = len(args)
if nargs == 0: error()
machine = None
extraflag = 0
iarg = 0
while iarg < nargs:
if args[iarg] == "-m":
if iarg+2 > nargs: error()
machine = args[iarg+1]
iarg += 2
elif args[iarg] == "-e":
if iarg+2 > nargs: error()
extraflag = 1
suffix = args[iarg+1]
iarg += 2
else: error()
# set lib from working dir
cwd = os.getcwd()
lib = os.path.basename(cwd)
# create Makefile.auto as copy of Makefile.machine
# reset EXTRAMAKE if requested
if not os.path.exists("Makefile.%s" % machine):
error("lib/%s/Makefile.%s does not exist" % (lib,machine))
lines = open("Makefile.%s" % machine,'r').readlines()
fp = open("Makefile.auto",'w')
for line in lines:
words = line.split()
if len(words) == 3 and extraflag and \
words[0] == "EXTRAMAKE" and words[1] == '=':
line = line.replace(words[2],"Makefile.lammps.%s" % suffix)
print >>fp,line,
fp.close()
# make the library via Makefile.auto
print "Building lib%s.a ..." % lib
cmd = "make -f Makefile.auto clean; make -f Makefile.auto"
txt = commands.getoutput(cmd)
print txt
if os.path.exists("lib%s.a" % lib): print "Build was successful"
else: error("Build of lib/%s/lib%s.a was NOT successful" % (lib,lib))
if not os.path.exists("Makefile.lammps"):
print "lib/%s/Makefile.lammps was NOT created" % lib

25
lib/atc/README
View File

@ -15,6 +15,11 @@ links against when using the USER-ATC package.
This library must be built with a C++ compiler, before LAMMPS is
built, so LAMMPS can link against it.
You can type "make lib-atc" from the src directory to see help on how
to build this library via make commands, or you can do the same thing
by typing "python Install.py" from within this directory, or you can
do it manually by following the instructions below.
Build the library using one of the provided Makefile.* files or create
your own, specific to your compiler and system. For example:
@ -44,16 +49,16 @@ user-atc_SYSINC = leave blank for this package
user-atc_SYSLIB = BLAS and LAPACK libraries needed by this package
user-atc_SYSPATH = path(s) to where those libraries are
You have several choices for these settings:
You have 3 choices for these settings:
If the 2 libraries are already installed on your system, the settings
in Makefile.lammps.installed should work.
a) If the 2 libraries are already installed on your system, the
settings in Makefile.lammps.installed should work.
If they are not, you can install them yourself, and speficy the
appropriate settings accordingly.
b) If they are not, you can install them yourself, and specify the
appropriate settings accordingly in a Makefile.lammps.* file
and set the EXTRAMAKE setting in Makefile.* to that file.
If you want to use the minimalist version of these libraries provided
with LAMMPS in lib/linalg, then the settings in Makefile.lammps.linalg
should work. Note that in this case you also need to build the
linear-algebra in lib/linalg; see the lib/linalg/README for more
details.
c) Use the minimalist version of these libraries provided with LAMMPS
in lib/linalg, by using Makefile.lammps.linalg. In this case you also
need to build the library in lib/linalg; see the lib/linalg/README
file for more details.

82
lib/awpmd/Install.py Normal file
View File

@ -0,0 +1,82 @@
#!/usr/bin/env python
# install.py tool to do a generic build of a library
# soft linked to by many of the lib/Install.py files
# used to automate the steps described in the corresponding lib/README
import sys,commands,os
# help message
help = """
Syntax: python Install.py -m machine -e suffix
specify -m and optionally -e, order does not matter
-m = peform a clean followed by "make -f Makefile.machine"
machine = suffix of a lib/Makefile.* file
-e = set EXTRAMAKE variable in Makefile.machine to Makefile.lammps.suffix
does not alter existing Makefile.machine
"""
# print error message or help
def error(str=None):
if not str: print help
else: print "ERROR",str
sys.exit()
# parse args
args = sys.argv[1:]
nargs = len(args)
if nargs == 0: error()
machine = None
extraflag = 0
iarg = 0
while iarg < nargs:
if args[iarg] == "-m":
if iarg+2 > nargs: error()
machine = args[iarg+1]
iarg += 2
elif args[iarg] == "-e":
if iarg+2 > nargs: error()
extraflag = 1
suffix = args[iarg+1]
iarg += 2
else: error()
# set lib from working dir
cwd = os.getcwd()
lib = os.path.basename(cwd)
# create Makefile.auto as copy of Makefile.machine
# reset EXTRAMAKE if requested
if not os.path.exists("Makefile.%s" % machine):
error("lib/%s/Makefile.%s does not exist" % (lib,machine))
lines = open("Makefile.%s" % machine,'r').readlines()
fp = open("Makefile.auto",'w')
for line in lines:
words = line.split()
if len(words) == 3 and extraflag and \
words[0] == "EXTRAMAKE" and words[1] == '=':
line = line.replace(words[2],"Makefile.lammps.%s" % suffix)
print >>fp,line,
fp.close()
# make the library via Makefile.auto
print "Building lib%s.a ..." % lib
cmd = "make -f Makefile.auto clean; make -f Makefile.auto"
txt = commands.getoutput(cmd)
print txt
if os.path.exists("lib%s.a" % lib): print "Build was successful"
else: error("Build of lib/%s/lib%s.a was NOT successful" % (lib,lib))
if not os.path.exists("Makefile.lammps"):
print "lib/%s/Makefile.lammps was NOT created" % lib

25
lib/awpmd/README
View File

@ -19,6 +19,11 @@ links against when using the USER-AWPMD package.
This library must be built with a C++ compiler, before LAMMPS is
built, so LAMMPS can link against it.
You can type "make lib-awpmd" from the src directory to see help on
how to build this library via make commands, or you can do the same
thing by typing "python Install.py" from within this directory, or you
can do it manually by following the instructions below.
Build the library using one of the provided Makefile.* files or create
your own, specific to your compiler and system. For example:
@ -47,16 +52,16 @@ user-awpmd_SYSINC = leave blank for this package
user-awpmd_SYSLIB = BLAS and LAPACK libraries needed by this package
user-awpmd_SYSPATH = path(s) to where those libraries are
You have several choices for these settings:
You have 3 choices for these settings:
If the 2 libraries are already installed on your system, the settings
in Makefile.lammps.installed should work.
a) If the 2 libraries are already installed on your system, the
settings in Makefile.lammps.installed should work.
If they are not, you can install them yourself, and speficy the
appropriate settings accordingly.
b) If they are not, you can install them yourself, and specify the
appropriate settings accordingly in a Makefile.lammps.* file
and set the EXTRAMAKE setting in Makefile.* to that file.
If you want to use the minimalist version of these libraries provided
with LAMMPS in lib/linalg, then the settings in Makefile.lammps.linalg
should work. Note that in this case you also need to build the
linear-algebra in lib/linalg; see the lib/linalg/README for more
details.
c) Use the minimalist version of these libraries provided with LAMMPS
in lib/linalg, by using Makefile.lammps.linalg. In this case you also
need to build the library in lib/linalg; see the lib/linalg/README
file for more details.

82
lib/colvars/Install.py Normal file
View File

@ -0,0 +1,82 @@
#!/usr/bin/env python
# install.py tool to do a generic build of a library
# soft linked to by many of the lib/Install.py files
# used to automate the steps described in the corresponding lib/README
import sys,commands,os
# help message
help = """
Syntax: python Install.py -m machine -e suffix
specify -m and optionally -e, order does not matter
-m = peform a clean followed by "make -f Makefile.machine"
machine = suffix of a lib/Makefile.* file
-e = set EXTRAMAKE variable in Makefile.machine to Makefile.lammps.suffix
does not alter existing Makefile.machine
"""
# print error message or help
def error(str=None):
if not str: print help
else: print "ERROR",str
sys.exit()
# parse args
args = sys.argv[1:]
nargs = len(args)
if nargs == 0: error()
machine = None
extraflag = 0
iarg = 0
while iarg < nargs:
if args[iarg] == "-m":
if iarg+2 > nargs: error()
machine = args[iarg+1]
iarg += 2
elif args[iarg] == "-e":
if iarg+2 > nargs: error()
extraflag = 1
suffix = args[iarg+1]
iarg += 2
else: error()
# set lib from working dir
cwd = os.getcwd()
lib = os.path.basename(cwd)
# create Makefile.auto as copy of Makefile.machine
# reset EXTRAMAKE if requested
if not os.path.exists("Makefile.%s" % machine):
error("lib/%s/Makefile.%s does not exist" % (lib,machine))
lines = open("Makefile.%s" % machine,'r').readlines()
fp = open("Makefile.auto",'w')
for line in lines:
words = line.split()
if len(words) == 3 and extraflag and \
words[0] == "EXTRAMAKE" and words[1] == '=':
line = line.replace(words[2],"Makefile.lammps.%s" % suffix)
print >>fp,line,
fp.close()
# make the library via Makefile.auto
print "Building lib%s.a ..." % lib
cmd = "make -f Makefile.auto clean; make -f Makefile.auto"
txt = commands.getoutput(cmd)
print txt
if os.path.exists("lib%s.a" % lib): print "Build was successful"
else: error("Build of lib/%s/lib%s.a was NOT successful" % (lib,lib))
if not os.path.exists("Makefile.lammps"):
print "lib/%s/Makefile.lammps was NOT created" % lib

5
lib/colvars/README
View File

@ -35,6 +35,11 @@ links against when using the USER-COLVARS package.
This library must be built with a C++ compiler, before LAMMPS is
built, so LAMMPS can link against it.
You can type "make lib-colvars" from the src directory to see help on
how to build this library via make commands, or you can do the same
thing by typing "python Install.py" from within this directory, or you
can do it manually by following the instructions below.
Build the library using one of the provided Makefile.* files or create
your own, specific to your compiler and system. For example:

146
lib/gpu/Install.py Normal file
View File

@ -0,0 +1,146 @@
#!/usr/bin/env python
# Install.py tool to build the GPU library
# used to automate the steps described in the README file in this dir
import sys,os,re,commands
# help message
help = """
Syntax: python Install.py -i isuffix -h hdir -a arch -p precision -e esuffix -m -o osuffix
specify one or more options, order does not matter
copies an existing Makefile.isuffix in lib/gpu to Makefile.auto
optionally edits these variables in Makefile.auto:
CUDA_HOME, CUDA_ARCH, CUDA_PRECISION, EXTRAMAKE
optionally uses Makefile.auto to build the GPU library -> libgpu.a
and to copy a Makefile.lammps.esuffix -> Makefile.lammps
optionally copies Makefile.auto to a new Makefile.osuffix
-i = use Makefile.isuffix as starting point, copy to Makefile.auto
default isuffix = linux
-h = set CUDA_HOME variable in Makefile.auto to hdir
hdir = path to NVIDIA Cuda software, e.g. /usr/local/cuda
-a = set CUDA_ARCH variable in Makefile.auto to arch
use arch = ?? for K40 (Tesla)
use arch = 37 for dual K80 (Tesla)
use arch = 60 for P100 (Pascal)
-p = set CUDA_PRECISION variable in Makefile.auto to precision
use precision = double or mixed or single
-e = set EXTRAMAKE variable in Makefile.auto to Makefile.lammps.esuffix
-m = make the GPU library using Makefile.auto
first performs a "make clean"
produces libgpu.a if successful
also copies EXTRAMAKE file -> Makefile.lammps
-e can set which Makefile.lammps.esuffix file is copied
-o = copy final Makefile.auto to Makefile.osuffix
"""
# print error message or help
def error(str=None):
if not str: print help
else: print "ERROR",str
sys.exit()
# parse args
args = sys.argv[1:]
nargs = len(args)
if nargs == 0: error()
isuffix = "linux"
hflag = aflag = pflag = eflag = 0
makeflag = 0
outflag = 0
iarg = 0
while iarg < nargs:
if args[iarg] == "-i":
if iarg+2 > nargs: error()
isuffix = args[iarg+1]
iarg += 2
elif args[iarg] == "-h":
if iarg+2 > nargs: error()
hflag = 1
hdir = args[iarg+1]
iarg += 2
elif args[iarg] == "-a":
if iarg+2 > nargs: error()
aflag = 1
arch = args[iarg+1]
iarg += 2
elif args[iarg] == "-p":
if iarg+2 > nargs: error()
pflag = 1
precision = args[iarg+1]
iarg += 2
elif args[iarg] == "-e":
if iarg+2 > nargs: error()
eflag = 1
lmpsuffix = args[iarg+1]
iarg += 2
elif args[iarg] == "-m":
makeflag = 1
iarg += 1
elif args[iarg] == "-o":
if iarg+2 > nargs: error()
outflag = 1
osuffix = args[iarg+1]
iarg += 2
else: error()
if pflag:
if precision == "double": precstr = "-D_DOUBLE_DOUBLE"
elif precision == "mixed": precstr = "-D_SINGLE_DOUBLE"
elif precision == "single": precstr = "-D_SINGLE_SINGLE"
else: error("Invalid precision setting")
# create Makefile.auto
# reset EXTRAMAKE, CUDA_HOME, CUDA_ARCH, CUDA_PRECISION if requested
if not os.path.exists("Makefile.%s" % isuffix):
error("lib/gpu/Makefile.%s does not exist" % isuffix)
lines = open("Makefile.%s" % isuffix,'r').readlines()
fp = open("Makefile.auto",'w')
for line in lines:
words = line.split()
if len(words) != 3:
print >>fp,line,
continue
if hflag and words[0] == "CUDA_HOME" and words[1] == '=':
line = line.replace(words[2],hdir)
if aflag and words[0] == "CUDA_ARCH" and words[1] == '=':
line = line.replace(words[2],"-arch=sm_%s" % arch)
if pflag and words[0] == "CUDA_PRECISION" and words[1] == '=':
line = line.replace(words[2],precstr)
if eflag and words[0] == "EXTRAMAKE" and words[1] == '=':
line = line.replace(words[2],"Makefile.lammps.%s" % lmpsuffix)
print >>fp,line,
fp.close()
# perform make
# make operations copies EXTRAMAKE file to Makefile.lammps
if makeflag:
print "Building libgpu.a ..."
cmd = "rm -f libgpu.a"
commands.getoutput(cmd)
cmd = "make -f Makefile.auto clean; make -f Makefile.auto"
commands.getoutput(cmd)
if not os.path.exists("libgpu.a"):
error("Build of lib/gpu/libgpu.a was NOT successful")
if not os.path.exists("Makefile.lammps"):
error("lib/gpu/Makefile.lammps was NOT created")
# copy new Makefile.auto to Makefile.osuffix
if outflag:
print "Creating new Makefile.%s" % osuffix
cmd = "cp Makefile.auto Makefile.%s" % osuffix
commands.getoutput(cmd)

40
lib/gpu/Nvidia.makefile
View File

@ -43,8 +43,8 @@ OBJS = $(OBJ_DIR)/lal_atom.o $(OBJ_DIR)/lal_ans.o \
$(OBJ_DIR)/lal_coul_long.o $(OBJ_DIR)/lal_coul_long_ext.o \
$(OBJ_DIR)/lal_morse.o $(OBJ_DIR)/lal_morse_ext.o \
$(OBJ_DIR)/lal_charmm_long.o $(OBJ_DIR)/lal_charmm_long_ext.o \
$(OBJ_DIR)/lal_cg_cmm.o $(OBJ_DIR)/lal_cg_cmm_ext.o \
$(OBJ_DIR)/lal_cg_cmm_long.o $(OBJ_DIR)/lal_cg_cmm_long_ext.o \
$(OBJ_DIR)/lal_lj_sdk.o $(OBJ_DIR)/lal_lj_sdk_ext.o \
$(OBJ_DIR)/lal_lj_sdk_long.o $(OBJ_DIR)/lal_lj_sdk_long_ext.o \
$(OBJ_DIR)/lal_eam.o $(OBJ_DIR)/lal_eam_ext.o \
$(OBJ_DIR)/lal_eam_fs_ext.o $(OBJ_DIR)/lal_eam_alloy_ext.o \
$(OBJ_DIR)/lal_buck.o $(OBJ_DIR)/lal_buck_ext.o \
@ -98,8 +98,8 @@ CBNS = $(OBJ_DIR)/device.cubin $(OBJ_DIR)/device_cubin.h \
$(OBJ_DIR)/coul_long.cubin $(OBJ_DIR)/coul_long_cubin.h \
$(OBJ_DIR)/morse.cubin $(OBJ_DIR)/morse_cubin.h \
$(OBJ_DIR)/charmm_long.cubin $(OBJ_DIR)/charmm_long_cubin.h \
$(OBJ_DIR)/cg_cmm.cubin $(OBJ_DIR)/cg_cmm_cubin.h \
$(OBJ_DIR)/cg_cmm_long.cubin $(OBJ_DIR)/cg_cmm_long_cubin.h \
$(OBJ_DIR)/lj_sdk.cubin $(OBJ_DIR)/lj_sdk_cubin.h \
$(OBJ_DIR)/lj_sdk_long.cubin $(OBJ_DIR)/lj_sdk_long_cubin.h \
$(OBJ_DIR)/eam.cubin $(OBJ_DIR)/eam_cubin.h \
$(OBJ_DIR)/buck.cubin $(OBJ_DIR)/buck_cubin.h \
$(OBJ_DIR)/buck_coul_long.cubin $(OBJ_DIR)/buck_coul_long_cubin.h \
@ -391,29 +391,29 @@ $(OBJ_DIR)/lal_lj_expand.o: $(ALL_H) lal_lj_expand.h lal_lj_expand.cpp $(OBJ_DIR
$(OBJ_DIR)/lal_lj_expand_ext.o: $(ALL_H) lal_lj_expand.h lal_lj_expand_ext.cpp lal_base_atomic.h
$(CUDR) -o $@ -c lal_lj_expand_ext.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/cg_cmm.cubin: lal_cg_cmm.cu lal_precision.h lal_preprocessor.h
$(CUDA) --cubin -DNV_KERNEL -o $@ lal_cg_cmm.cu
$(OBJ_DIR)/lj_sdk.cubin: lal_lj_sdk.cu lal_precision.h lal_preprocessor.h
$(CUDA) --cubin -DNV_KERNEL -o $@ lal_lj_sdk.cu
$(OBJ_DIR)/cg_cmm_cubin.h: $(OBJ_DIR)/cg_cmm.cubin $(OBJ_DIR)/cg_cmm.cubin
$(BIN2C) -c -n cg_cmm $(OBJ_DIR)/cg_cmm.cubin > $(OBJ_DIR)/cg_cmm_cubin.h
$(OBJ_DIR)/lj_sdk_cubin.h: $(OBJ_DIR)/lj_sdk.cubin $(OBJ_DIR)/lj_sdk.cubin
$(BIN2C) -c -n lj_sdk $(OBJ_DIR)/lj_sdk.cubin > $(OBJ_DIR)/lj_sdk_cubin.h
$(OBJ_DIR)/lal_cg_cmm.o: $(ALL_H) lal_cg_cmm.h lal_cg_cmm.cpp $(OBJ_DIR)/cg_cmm_cubin.h $(OBJ_DIR)/lal_base_atomic.o
$(CUDR) -o $@ -c lal_cg_cmm.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/lal_lj_sdk.o: $(ALL_H) lal_lj_sdk.h lal_lj_sdk.cpp $(OBJ_DIR)/lj_sdk_cubin.h $(OBJ_DIR)/lal_base_atomic.o
$(CUDR) -o $@ -c lal_lj_sdk.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/lal_cg_cmm_ext.o: $(ALL_H) lal_cg_cmm.h lal_cg_cmm_ext.cpp lal_base_atomic.h
$(CUDR) -o $@ -c lal_cg_cmm_ext.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/lal_lj_sdk_ext.o: $(ALL_H) lal_lj_sdk.h lal_lj_sdk_ext.cpp lal_base_atomic.h
$(CUDR) -o $@ -c lal_lj_sdk_ext.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/cg_cmm_long.cubin: lal_cg_cmm_long.cu lal_precision.h lal_preprocessor.h
$(CUDA) --cubin -DNV_KERNEL -o $@ lal_cg_cmm_long.cu
$(OBJ_DIR)/lj_sdk_long.cubin: lal_lj_sdk_long.cu lal_precision.h lal_preprocessor.h
$(CUDA) --cubin -DNV_KERNEL -o $@ lal_lj_sdk_long.cu
$(OBJ_DIR)/cg_cmm_long_cubin.h: $(OBJ_DIR)/cg_cmm_long.cubin $(OBJ_DIR)/cg_cmm_long.cubin
$(BIN2C) -c -n cg_cmm_long $(OBJ_DIR)/cg_cmm_long.cubin > $(OBJ_DIR)/cg_cmm_long_cubin.h
$(OBJ_DIR)/lj_sdk_long_cubin.h: $(OBJ_DIR)/lj_sdk_long.cubin $(OBJ_DIR)/lj_sdk_long.cubin
$(BIN2C) -c -n lj_sdk_long $(OBJ_DIR)/lj_sdk_long.cubin > $(OBJ_DIR)/lj_sdk_long_cubin.h
$(OBJ_DIR)/lal_cg_cmm_long.o: $(ALL_H) lal_cg_cmm_long.h lal_cg_cmm_long.cpp $(OBJ_DIR)/cg_cmm_long_cubin.h $(OBJ_DIR)/lal_base_atomic.o
$(CUDR) -o $@ -c lal_cg_cmm_long.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/lal_lj_sdk_long.o: $(ALL_H) lal_lj_sdk_long.h lal_lj_sdk_long.cpp $(OBJ_DIR)/lj_sdk_long_cubin.h $(OBJ_DIR)/lal_base_atomic.o
$(CUDR) -o $@ -c lal_lj_sdk_long.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/lal_cg_cmm_long_ext.o: $(ALL_H) lal_cg_cmm_long.h lal_cg_cmm_long_ext.cpp lal_base_charge.h
$(CUDR) -o $@ -c lal_cg_cmm_long_ext.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/lal_lj_sdk_long_ext.o: $(ALL_H) lal_lj_sdk_long.h lal_lj_sdk_long_ext.cpp lal_base_charge.h
$(CUDR) -o $@ -c lal_lj_sdk_long_ext.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/eam.cubin: lal_eam.cu lal_precision.h lal_preprocessor.h
$(CUDA) --cubin -DNV_KERNEL -o $@ lal_eam.cu

32
lib/gpu/Opencl.makefile
View File

@ -32,8 +32,8 @@ OBJS = $(OBJ_DIR)/lal_atom.o $(OBJ_DIR)/lal_answer.o \
$(OBJ_DIR)/lal_coul_long.o $(OBJ_DIR)/lal_coul_long_ext.o \
$(OBJ_DIR)/lal_morse.o $(OBJ_DIR)/lal_morse_ext.o \
$(OBJ_DIR)/lal_charmm_long.o $(OBJ_DIR)/lal_charmm_long_ext.o \
$(OBJ_DIR)/lal_cg_cmm.o $(OBJ_DIR)/lal_cg_cmm_ext.o \
$(OBJ_DIR)/lal_cg_cmm_long.o $(OBJ_DIR)/lal_cg_cmm_long_ext.o \
$(OBJ_DIR)/lal_lj_sdk.o $(OBJ_DIR)/lal_lj_sdk_ext.o \
$(OBJ_DIR)/lal_lj_sdk_long.o $(OBJ_DIR)/lal_lj_sdk_long_ext.o \
$(OBJ_DIR)/lal_eam.o $(OBJ_DIR)/lal_eam_ext.o \
$(OBJ_DIR)/lal_eam_fs_ext.o $(OBJ_DIR)/lal_eam_alloy_ext.o \
$(OBJ_DIR)/lal_buck.o $(OBJ_DIR)/lal_buck_ext.o \
@ -75,8 +75,8 @@ KERS = $(OBJ_DIR)/device_cl.h $(OBJ_DIR)/atom_cl.h \
$(OBJ_DIR)/lj_coul_long_cl.h $(OBJ_DIR)/lj_dsf_cl.h \
$(OBJ_DIR)/lj_class2_long_cl.h \
$(OBJ_DIR)/coul_long_cl.h $(OBJ_DIR)/morse_cl.h \
$(OBJ_DIR)/charmm_long_cl.h $(OBJ_DIR)/cg_cmm_cl.h \
$(OBJ_DIR)/cg_cmm_long_cl.h $(OBJ_DIR)/neighbor_gpu_cl.h \
$(OBJ_DIR)/charmm_long_cl.h $(OBJ_DIR)/lj_sdk_cl.h \
$(OBJ_DIR)/lj_sdk_long_cl.h $(OBJ_DIR)/neighbor_gpu_cl.h \
$(OBJ_DIR)/eam_cl.h $(OBJ_DIR)/buck_cl.h \
$(OBJ_DIR)/buck_coul_cl.h $(OBJ_DIR)/buck_coul_long_cl.h \
$(OBJ_DIR)/table_cl.h $(OBJ_DIR)/yukawa_cl.h \
@ -273,23 +273,23 @@ $(OBJ_DIR)/lal_lj_expand.o: $(ALL_H) lal_lj_expand.h lal_lj_expand.cpp $(OBJ_DI
$(OBJ_DIR)/lal_lj_expand_ext.o: $(ALL_H) lal_lj_expand.h lal_lj_expand_ext.cpp lal_base_atomic.h
$(OCL) -o $@ -c lal_lj_expand_ext.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/cg_cmm_cl.h: lal_cg_cmm.cu $(PRE1_H)
$(BSH) ./geryon/file_to_cstr.sh cg_cmm $(PRE1_H) lal_cg_cmm.cu $(OBJ_DIR)/cg_cmm_cl.h;
$(OBJ_DIR)/lj_sdk_cl.h: lal_lj_sdk.cu $(PRE1_H)
$(BSH) ./geryon/file_to_cstr.sh lj_sdk $(PRE1_H) lal_lj_sdk.cu $(OBJ_DIR)/lj_sdk_cl.h;
$(OBJ_DIR)/lal_cg_cmm.o: $(ALL_H) lal_cg_cmm.h lal_cg_cmm.cpp $(OBJ_DIR)/cg_cmm_cl.h $(OBJ_DIR)/cg_cmm_cl.h $(OBJ_DIR)/lal_base_atomic.o
$(OCL) -o $@ -c lal_cg_cmm.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/lal_lj_sdk.o: $(ALL_H) lal_lj_sdk.h lal_lj_sdk.cpp $(OBJ_DIR)/lj_sdk_cl.h $(OBJ_DIR)/lj_sdk_cl.h $(OBJ_DIR)/lal_base_atomic.o
$(OCL) -o $@ -c lal_lj_sdk.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/lal_cg_cmm_ext.o: $(ALL_H) lal_cg_cmm.h lal_cg_cmm_ext.cpp lal_base_atomic.h
$(OCL) -o $@ -c lal_cg_cmm_ext.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/lal_lj_sdk_ext.o: $(ALL_H) lal_lj_sdk.h lal_lj_sdk_ext.cpp lal_base_atomic.h
$(OCL) -o $@ -c lal_lj_sdk_ext.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/cg_cmm_long_cl.h: lal_cg_cmm_long.cu $(PRE1_H)
$(BSH) ./geryon/file_to_cstr.sh cg_cmm_long $(PRE1_H) lal_cg_cmm_long.cu $(OBJ_DIR)/cg_cmm_long_cl.h;
$(OBJ_DIR)/lj_sdk_long_cl.h: lal_lj_sdk_long.cu $(PRE1_H)
$(BSH) ./geryon/file_to_cstr.sh lj_sdk_long $(PRE1_H) lal_lj_sdk_long.cu $(OBJ_DIR)/lj_sdk_long_cl.h;
$(OBJ_DIR)/lal_cg_cmm_long.o: $(ALL_H) lal_cg_cmm_long.h lal_cg_cmm_long.cpp $(OBJ_DIR)/cg_cmm_long_cl.h $(OBJ_DIR)/cg_cmm_long_cl.h $(OBJ_DIR)/lal_base_atomic.o
$(OCL) -o $@ -c lal_cg_cmm_long.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/lal_lj_sdk_long.o: $(ALL_H) lal_lj_sdk_long.h lal_lj_sdk_long.cpp $(OBJ_DIR)/lj_sdk_long_cl.h $(OBJ_DIR)/lj_sdk_long_cl.h $(OBJ_DIR)/lal_base_atomic.o
$(OCL) -o $@ -c lal_lj_sdk_long.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/lal_cg_cmm_long_ext.o: $(ALL_H) lal_cg_cmm_long.h lal_cg_cmm_long_ext.cpp lal_base_charge.h
$(OCL) -o $@ -c lal_cg_cmm_long_ext.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/lal_lj_sdk_long_ext.o: $(ALL_H) lal_lj_sdk_long.h lal_lj_sdk_long_ext.cpp lal_base_charge.h
$(OCL) -o $@ -c lal_lj_sdk_long_ext.cpp -I$(OBJ_DIR)
$(OBJ_DIR)/eam_cl.h: lal_eam.cu $(PRE1_H)
$(BSH) ./geryon/file_to_cstr.sh eam $(PRE1_H) lal_eam.cu $(OBJ_DIR)/eam_cl.h;

11
lib/gpu/README
View File

@ -17,6 +17,11 @@ links against when using the GPU package.
This library must be built with a C++ compiler, before LAMMPS is
built, so LAMMPS can link against it.
You can type "make lib-gpu" from the src directory to see help on how
to build this library via make commands, or you can do the same thing
by typing "python Install.py" from within this directory, or you can
do it manually by following the instructions below.
Build the library using one of the provided Makefile.* files or create
your own, specific to your compiler and system. For example:
@ -164,9 +169,9 @@ this directory).
The gpu library supports 3 precision modes as determined by
the CUDA_PRECISION variable:
CUDA_PREC = -D_SINGLE_SINGLE # Single precision for all calculations
CUDA_PREC = -D_DOUBLE_DOUBLE # Double precision for all calculations
CUDA_PREC = -D_SINGLE_DOUBLE # Accumulation of forces, etc. in double
CUDA_PRECISION = -D_SINGLE_SINGLE # Single precision for all calculations
CUDA_PRECISION = -D_DOUBLE_DOUBLE # Double precision for all calculations
CUDA_PRECISION = -D_SINGLE_DOUBLE # Accumulation of forces, etc. in double
NOTE: PPPM acceleration can only be run on GPUs with compute capability>=1.1.
You will get the error "GPU library not compiled for this accelerator."

34
lib/gpu/lal_cg_cmm.cpp → lib/gpu/lal_lj_sdk.cpp
View File

@ -1,5 +1,5 @@
/***************************************************************************
cg_cmm.cpp
lj_sdk.cpp
-------------------
W. Michael Brown (ORNL)
@ -14,14 +14,14 @@
***************************************************************************/
#if defined(USE_OPENCL)
#include "cg_cmm_cl.h"
#include "lj_sdk_cl.h"
#elif defined(USE_CUDART)
const char *cg_cmm=0;
const char *lj_sdk=0;
#else
#include "cg_cmm_cubin.h"
#include "lj_sdk_cubin.h"
#endif
#include "lal_cg_cmm.h"
#include "lal_lj_sdk.h"
#include <cassert>
using namespace LAMMPS_AL;
#define CGCMMT CGCMM<numtyp, acctyp>
@ -53,33 +53,33 @@ int CGCMMT::init(const int ntypes, double **host_cutsq,
const double gpu_split, FILE *_screen) {
int success;
success=this->init_atomic(nlocal,nall,max_nbors,maxspecial,cell_size,gpu_split,
_screen,cg_cmm,"k_cg_cmm");
_screen,lj_sdk,"k_lj_sdk");
if (success!=0)
return success;
// If atom type constants fit in shared memory use fast kernel
int cmm_types=ntypes;
int sdk_types=ntypes;
shared_types=false;
int max_shared_types=this->device->max_shared_types();
if (cmm_types<=max_shared_types && this->_block_size>=max_shared_types) {
cmm_types=max_shared_types;
if (sdk_types<=max_shared_types && this->_block_size>=max_shared_types) {
sdk_types=max_shared_types;
shared_types=true;
}
_cmm_types=cmm_types;
_sdk_types=sdk_types;
// Allocate a host write buffer for data initialization
UCL_H_Vec<numtyp> host_write(cmm_types*cmm_types*32,*(this->ucl_device),
UCL_H_Vec<numtyp> host_write(sdk_types*sdk_types*32,*(this->ucl_device),
UCL_WRITE_ONLY);
for (int i=0; i<cmm_types*cmm_types; i++)
for (int i=0; i<sdk_types*sdk_types; i++)
host_write[i]=0.0;
lj1.alloc(cmm_types*cmm_types,*(this->ucl_device),UCL_READ_ONLY);
this->atom->type_pack4(ntypes,cmm_types,lj1,host_write,host_cutsq,
lj1.alloc(sdk_types*sdk_types,*(this->ucl_device),UCL_READ_ONLY);
this->atom->type_pack4(ntypes,sdk_types,lj1,host_write,host_cutsq,
host_cg_type,host_lj1,host_lj2);
lj3.alloc(cmm_types*cmm_types,*(this->ucl_device),UCL_READ_ONLY);
this->atom->type_pack4(ntypes,cmm_types,lj3,host_write,host_lj3,host_lj4,
lj3.alloc(sdk_types*sdk_types,*(this->ucl_device),UCL_READ_ONLY);
this->atom->type_pack4(ntypes,sdk_types,lj3,host_write,host_lj3,host_lj4,
host_offset);
UCL_H_Vec<double> dview;
@ -143,7 +143,7 @@ void CGCMMT::loop(const bool _eflag, const bool _vflag) {
} else {
this->k_pair.set_size(GX,BX);
this->k_pair.run(&this->atom->x, &lj1, &lj3,
&_cmm_types, &sp_lj, &this->nbor->dev_nbor,
&_sdk_types, &sp_lj, &this->nbor->dev_nbor,
&this->_nbor_data->begin(), &this->ans->force,
&this->ans->engv, &eflag, &vflag, &ainum,
&nbor_pitch, &this->_threads_per_atom);

6
lib/gpu/lal_cg_cmm.cu → lib/gpu/lal_lj_sdk.cu
View File

@ -1,5 +1,5 @@
// **************************************************************************
// cg_cmm.cu
// lj_sdk.cu
// -------------------
// W. Michael Brown (ORNL)
//
@ -24,7 +24,7 @@ texture<int4,1> pos_tex;
#define pos_tex x_
#endif
__kernel void k_cg_cmm(const __global numtyp4 *restrict x_,
__kernel void k_lj_sdk(const __global numtyp4 *restrict x_,
const __global numtyp4 *restrict lj1,
const __global numtyp4 *restrict lj3,
const int lj_types,
@ -116,7 +116,7 @@ __kernel void k_cg_cmm(const __global numtyp4 *restrict x_,
} // if ii
}
__kernel void k_cg_cmm_fast(const __global numtyp4 *restrict x_,
__kernel void k_lj_sdk_fast(const __global numtyp4 *restrict x_,
const __global numtyp4 *restrict lj1_in,
const __global numtyp4 *restrict lj3_in,
const __global numtyp *restrict sp_lj_in,

4
lib/gpu/lal_cg_cmm.h → lib/gpu/lal_lj_sdk.h
View File

@ -1,5 +1,5 @@
/***************************************************************************
cg_cmm.h
lj_sdk.h
-------------------
W. Michael Brown (ORNL)
@ -67,7 +67,7 @@ class CGCMM : public BaseAtomic<numtyp, acctyp> {
bool shared_types;
/// Number of atom types
int _cmm_types;
int _sdk_types;
private:
bool _allocated;

14
lib/gpu/lal_cg_cmm_ext.cpp → lib/gpu/lal_lj_sdk_ext.cpp
View File

@ -1,5 +1,5 @@
/***************************************************************************
cg_cmm.h
lj_sdk.h
-------------------
W. Michael Brown (ORNL)
@ -17,7 +17,7 @@
#include <cassert>
#include <math.h>
#include "lal_cg_cmm.h"
#include "lal_lj_sdk.h"
using namespace std;
using namespace LAMMPS_AL;
@ -27,7 +27,7 @@ static CGCMM<PRECISION,ACC_PRECISION> CMMMF;
// ---------------------------------------------------------------------------
// Allocate memory on host and device and copy constants to device
// ---------------------------------------------------------------------------
int cmm_gpu_init(const int ntypes, double **cutsq, int **cg_types,
int sdk_gpu_init(const int ntypes, double **cutsq, int **cg_types,
double **host_lj1, double **host_lj2, double **host_lj3,
double **host_lj4, double **offset, double *special_lj,
const int inum, const int nall, const int max_nbors,
@ -89,11 +89,11 @@ int cmm_gpu_init(const int ntypes, double **cutsq, int **cg_types,
return init_ok;
}
void cmm_gpu_clear() {
void sdk_gpu_clear() {
CMMMF.clear();
}
int** cmm_gpu_compute_n(const int ago, const int inum_full,
int** sdk_gpu_compute_n(const int ago, const int inum_full,
const int nall, double **host_x, int *host_type,
double *sublo, double *subhi, tagint *tag, int **nspecial,
tagint **special, const bool eflag, const bool vflag,
@ -105,7 +105,7 @@ int** cmm_gpu_compute_n(const int ago, const int inum_full,
vatom, host_start, ilist, jnum, cpu_time, success);
}
void cmm_gpu_compute(const int ago, const int inum_full, const int nall,
void sdk_gpu_compute(const int ago, const int inum_full, const int nall,
double **host_x, int *host_type, int *ilist, int *numj,
int **firstneigh, const bool eflag, const bool vflag,
const bool eatom, const bool vatom, int &host_start,
@ -114,7 +114,7 @@ void cmm_gpu_compute(const int ago, const int inum_full, const int nall,
firstneigh,eflag,vflag,eatom,vatom,host_start,cpu_time,success);
}
double cmm_gpu_bytes() {
double sdk_gpu_bytes() {
return CMMMF.host_memory_usage();
}

12
lib/gpu/lal_cg_cmm_long.cpp → lib/gpu/lal_lj_sdk_long.cpp
View File

@ -1,5 +1,5 @@
/***************************************************************************
cg_cmm_long.cpp
lj_sdk_long.cpp
-------------------
W. Michael Brown (ORNL)
@ -14,14 +14,14 @@
***************************************************************************/
#if defined(USE_OPENCL)
#include "cg_cmm_long_cl.h"
#include "lj_sdk_long_cl.h"
#elif defined(USE_CUDART)
const char *cg_cmm_long=0;
const char *lj_sdk_long=0;
#else
#include "cg_cmm_long_cubin.h"
#include "lj_sdk_long_cubin.h"
#endif
#include "lal_cg_cmm_long.h"
#include "lal_lj_sdk_long.h"
#include <cassert>
using namespace LAMMPS_AL;
#define CGCMMLongT CGCMMLong<numtyp, acctyp>
@ -58,7 +58,7 @@ int CGCMMLongT::init(const int ntypes, double **host_cutsq,
const double g_ewald) {
int success;
success=this->init_atomic(nlocal,nall,max_nbors,maxspecial,cell_size,gpu_split,
_screen,cg_cmm_long,"k_cg_cmm_long");
_screen,lj_sdk_long,"k_lj_sdk_long");
if (success!=0)
return success;

6
lib/gpu/lal_cg_cmm_long.cu → lib/gpu/lal_lj_sdk_long.cu
View File

@ -1,5 +1,5 @@
// **************************************************************************
// cg_cmm_long.cu
// lj_sdk_long.cu
// -------------------
// W. Michael Brown (ORNL)
//
@ -29,7 +29,7 @@ texture<int2> q_tex;
#define q_tex q_
#endif
__kernel void k_cg_cmm_long(const __global numtyp4 *restrict x_,
__kernel void k_lj_sdk_long(const __global numtyp4 *restrict x_,
const __global numtyp4 *restrict lj1,
const __global numtyp4 *restrict lj3,
const int lj_types,
@ -154,7 +154,7 @@ __kernel void k_cg_cmm_long(const __global numtyp4 *restrict x_,
} // if ii
}
__kernel void k_cg_cmm_long_fast(const __global numtyp4 *restrict x_,
__kernel void k_lj_sdk_long_fast(const __global numtyp4 *restrict x_,
const __global numtyp4 *restrict lj1_in,
const __global numtyp4 *restrict lj3_in,
const __global numtyp *restrict sp_lj_in,

2
lib/gpu/lal_cg_cmm_long.h → lib/gpu/lal_lj_sdk_long.h
View File

@ -1,5 +1,5 @@
/***************************************************************************
cg_cmm_long.h
lj_sdk_long.h
-------------------
W. Michael Brown (ORNL)

14
lib/gpu/lal_cg_cmm_long_ext.cpp → lib/gpu/lal_lj_sdk_long_ext.cpp
View File

@ -1,5 +1,5 @@
/***************************************************************************
cg_cmm_long.h
lj_sdk_long.h
-------------------
W. Michael Brown (ORNL)
@ -17,7 +17,7 @@
#include <cassert>
#include <math.h>
#include "lal_cg_cmm_long.h"
#include "lal_lj_sdk_long.h"
using namespace std;
using namespace LAMMPS_AL;
@ -27,7 +27,7 @@ static CGCMMLong<PRECISION,ACC_PRECISION> CMMLMF;
// ---------------------------------------------------------------------------
// Allocate memory on host and device and copy constants to device
// ---------------------------------------------------------------------------
int cmml_gpu_init(const int ntypes, double **cutsq, int **cg_type,
int sdkl_gpu_init(const int ntypes, double **cutsq, int **cg_type,
double **host_lj1, double **host_lj2, double **host_lj3,
double **host_lj4, double **offset, double *special_lj,
const int inum, const int nall, const int max_nbors,
@ -93,11 +93,11 @@ int cmml_gpu_init(const int ntypes, double **cutsq, int **cg_type,
return init_ok;
}
void cmml_gpu_clear() {
void sdkl_gpu_clear() {
CMMLMF.clear();
}
int** cmml_gpu_compute_n(const int ago, const int inum_full,
int** sdkl_gpu_compute_n(const int ago, const int inum_full,
const int nall, double **host_x, int *host_type,
double *sublo, double *subhi, tagint *tag, int **nspecial,
tagint **special, const bool eflag, const bool vflag,
@ -111,7 +111,7 @@ int** cmml_gpu_compute_n(const int ago, const int inum_full,
host_q,boxlo,prd);
}
void cmml_gpu_compute(const int ago, const int inum_full, const int nall,
void sdkl_gpu_compute(const int ago, const int inum_full, const int nall,
double **host_x, int *host_type, int *ilist, int *numj,
int **firstneigh, const bool eflag, const bool vflag,
const bool eatom, const bool vatom, int &host_start,
@ -122,7 +122,7 @@ void cmml_gpu_compute(const int ago, const int inum_full, const int nall,
host_q,nlocal,boxlo,prd);
}
double cmml_gpu_bytes() {
double sdkl_gpu_bytes() {
return CMMLMF.host_memory_usage();
}

82
lib/h5md/Install.py Normal file
View File

@ -0,0 +1,82 @@
#!/usr/bin/env python
# install.py tool to do a generic build of a library
# soft linked to by many of the lib/Install.py files
# used to automate the steps described in the corresponding lib/README
import sys,commands,os
# help message
help = """
Syntax: python Install.py -m machine -e suffix
specify -m and optionally -e, order does not matter
-m = peform a clean followed by "make -f Makefile.machine"
machine = suffix of a lib/Makefile.* file
-e = set EXTRAMAKE variable in Makefile.machine to Makefile.lammps.suffix
does not alter existing Makefile.machine
"""
# print error message or help
def error(str=None):
if not str: print help
else: print "ERROR",str
sys.exit()
# parse args
args = sys.argv[1:]
nargs = len(args)
if nargs == 0: error()
machine = None
extraflag = 0
iarg = 0
while iarg < nargs:
if args[iarg] == "-m":
if iarg+2 > nargs: error()
machine = args[iarg+1]
iarg += 2
elif args[iarg] == "-e":
if iarg+2 > nargs: error()
extraflag = 1
suffix = args[iarg+1]
iarg += 2
else: error()
# set lib from working dir
cwd = os.getcwd()
lib = os.path.basename(cwd)
# create Makefile.auto as copy of Makefile.machine
# reset EXTRAMAKE if requested
if not os.path.exists("Makefile.%s" % machine):
error("lib/%s/Makefile.%s does not exist" % (lib,machine))
lines = open("Makefile.%s" % machine,'r').readlines()
fp = open("Makefile.auto",'w')
for line in lines:
words = line.split()
if len(words) == 3 and extraflag and \
words[0] == "EXTRAMAKE" and words[1] == '=':
line = line.replace(words[2],"Makefile.lammps.%s" % suffix)
print >>fp,line,
fp.close()
# make the library via Makefile.auto
print "Building lib%s.a ..." % lib
cmd = "make -f Makefile.auto clean; make -f Makefile.auto"
txt = commands.getoutput(cmd)
print txt
if os.path.exists("lib%s.a" % lib): print "Build was successful"
else: error("Build of lib/%s/lib%s.a was NOT successful" % (lib,lib))
if not os.path.exists("Makefile.lammps"):
print "lib/%s/Makefile.lammps was NOT created" % lib

2
lib/h5md/Makefile → lib/h5md/Makefile.h5cc
View File

@ -19,7 +19,7 @@ build/ch5md.o: src/ch5md.c | build
$(CC) $(INC) $(CFLAGS) -c $< -o $@
Makefile.lammps:
cp Makefile.lammps.empty $@
cp $(EXTRAMAKE) $@
.PHONY: all lib clean

15
lib/h5md/README
View File

@ -3,6 +3,11 @@ LAMMPS under its own BSD license; see below. This library is used
when the USER-H5MD package is included in a LAMMPS build and the dump
h5md command is invoked in a LAMMPS input script.
You can type "make lib-h5md" from the src directory to see help on how
to build this library via make commands, or you can do the same thing
by typing "python Install.py" from within this directory, or you can
do it manually by following the instructions below.
---------------------
ch5md : Read and write H5MD files in C
@ -17,8 +22,14 @@ molecular data, whose development is found at <http://nongnu.org/h5md/>.
ch5md is developped by Pierre de Buyl and is released under the 3-clause BSD
license that can be found in the file LICENSE.
To use the h5md dump style in lammps, execute make in this directory then 'make
yes-user-h5md' in the src directory of lammps. Rebuild lammps.
To use the h5md dump style in lammps, execute
make -f Makefile.h5cc
in this directory then
make yes-user-h5md
in the src directory of LAMMPS to rebuild LAMMPS.
Note that you must have the h5cc compiler installed to use
Makefile.h5cc. It should be part
If HDF5 is not in a standard system location, edit Makefile.lammps accordingly.

23
lib/kokkos/CHANGELOG.md
View File

@ -1,5 +1,28 @@
# Change Log
## [2.03.00](https://github.com/kokkos/kokkos/tree/2.03.00) (2017-04-25)
[Full Changelog](https://github.com/kokkos/kokkos/compare/2.02.15...2.03.00)
**Implemented enhancements:**
- UnorderedMap: make it accept Devices or MemorySpaces [\#711](https://github.com/kokkos/kokkos/issues/711)
- sort to accept DynamicView and \[begin,end\) indices [\#691](https://github.com/kokkos/kokkos/issues/691)
- ENABLE Macros should only be used via \#ifdef or \#if defined [\#675](https://github.com/kokkos/kokkos/issues/675)
- Remove impl/Kokkos\_Synchronic\_\* [\#666](https://github.com/kokkos/kokkos/issues/666)
- Turning off IVDEP for Intel 14. [\#638](https://github.com/kokkos/kokkos/issues/638)
- Using an installed Kokkos in a target application using CMake [\#633](https://github.com/kokkos/kokkos/issues/633)
- Create Kokkos Bill of Materials [\#632](https://github.com/kokkos/kokkos/issues/632)
- MDRangePolicy and tagged evaluators [\#547](https://github.com/kokkos/kokkos/issues/547)
- Add PGI support [\#289](https://github.com/kokkos/kokkos/issues/289)
**Fixed bugs:**
- Output from PerTeam fails [\#733](https://github.com/kokkos/kokkos/issues/733)
- Cuda: architecture flag not added to link line [\#688](https://github.com/kokkos/kokkos/issues/688)
- Getting large chunks of memory for a thread team in a universal way [\#664](https://github.com/kokkos/kokkos/issues/664)
- Kokkos RNG normal\(\) function hangs for small seed value [\#655](https://github.com/kokkos/kokkos/issues/655)
- Kokkos Tests Errors on Shepard/HSW Builds [\#644](https://github.com/kokkos/kokkos/issues/644)
## [2.02.15](https://github.com/kokkos/kokkos/tree/2.02.15) (2017-02-10)
[Full Changelog](https://github.com/kokkos/kokkos/compare/2.02.07...2.02.15)

11
lib/kokkos/CMakeLists.txt
View File

@ -98,10 +98,10 @@ TRIBITS_ADD_OPTION_AND_DEFINE(
)
TRIBITS_ADD_OPTION_AND_DEFINE(
Kokkos_ENABLE_QTHREAD
KOKKOS_HAVE_QTHREAD
"Enable QTHREAD support in Kokkos."
"${TPL_ENABLE_QTHREAD}"
Kokkos_ENABLE_Qthreads
KOKKOS_HAVE_QTHREADS
"Enable Qthreads support in Kokkos."
"${TPL_ENABLE_QTHREADS}"
)
TRIBITS_ADD_OPTION_AND_DEFINE(
@ -110,7 +110,7 @@ TRIBITS_ADD_OPTION_AND_DEFINE(
"Enable C++11 support in Kokkos."
"${${PROJECT_NAME}_ENABLE_CXX11}"
)
TRIBITS_ADD_OPTION_AND_DEFINE(
Kokkos_ENABLE_HWLOC
KOKKOS_HAVE_HWLOC
@ -213,4 +213,3 @@ TRIBITS_EXCLUDE_FILES(
)
TRIBITS_PACKAGE_POSTPROCESS()

611
lib/kokkos/Makefile.kokkos
View File

@ -1,39 +1,38 @@
# Default settings common options
# Default settings common options.
#LAMMPS specific settings:
KOKKOS_PATH=../../lib/kokkos
CXXFLAGS=$(CCFLAGS)
#Options: OpenMP,Serial,Pthreads,Cuda
# Options: Cuda,OpenMP,Pthreads,Qthreads,Serial
KOKKOS_DEVICES ?= "OpenMP"
#KOKKOS_DEVICES ?= "Pthreads"
#Options: KNC,SNB,HSW,Kepler,Kepler30,Kepler32,Kepler35,Kepler37,Maxwell,Maxwell50,Maxwell52,Maxwell53,Pascal61,ARMv80,ARMv81,ARMv8-ThunderX,BGQ,Power7,Power8,Power9,KNL,BDW,SKX
# Options: KNC,SNB,HSW,Kepler,Kepler30,Kepler32,Kepler35,Kepler37,Maxwell,Maxwell50,Maxwell52,Maxwell53,Pascal60,Pascal61,ARMv80,ARMv81,ARMv8-ThunderX,BGQ,Power7,Power8,Power9,KNL,BDW,SKX
KOKKOS_ARCH ?= ""
#Options: yes,no
# Options: yes,no
KOKKOS_DEBUG ?= "no"
#Options: hwloc,librt,experimental_memkind
# Options: hwloc,librt,experimental_memkind
KOKKOS_USE_TPLS ?= ""
#Options: c++11,c++1z
# Options: c++11,c++1z
KOKKOS_CXX_STANDARD ?= "c++11"
#Options: aggressive_vectorization,disable_profiling
# Options: aggressive_vectorization,disable_profiling
KOKKOS_OPTIONS ?= ""
#Default settings specific options
#Options: force_uvm,use_ldg,rdc,enable_lambda
# Default settings specific options.
# Options: force_uvm,use_ldg,rdc,enable_lambda
KOKKOS_CUDA_OPTIONS ?= "enable_lambda"
# Check for general settings
# Check for general settings.
KOKKOS_INTERNAL_ENABLE_DEBUG := $(strip $(shell echo $(KOKKOS_DEBUG) | grep "yes" | wc -l))
KOKKOS_INTERNAL_ENABLE_CXX11 := $(strip $(shell echo $(KOKKOS_CXX_STANDARD) | grep "c++11" | wc -l))
KOKKOS_INTERNAL_ENABLE_CXX1Z := $(strip $(shell echo $(KOKKOS_CXX_STANDARD) | grep "c++1z" | wc -l))
# Check for external libraries
# Check for external libraries.
KOKKOS_INTERNAL_USE_HWLOC := $(strip $(shell echo $(KOKKOS_USE_TPLS) | grep "hwloc" | wc -l))
KOKKOS_INTERNAL_USE_LIBRT := $(strip $(shell echo $(KOKKOS_USE_TPLS) | grep "librt" | wc -l))
KOKKOS_INTERNAL_USE_MEMKIND := $(strip $(shell echo $(KOKKOS_USE_TPLS) | grep "experimental_memkind" | wc -l))
# Check for advanced settings
# Check for advanced settings.
KOKKOS_INTERNAL_OPT_RANGE_AGGRESSIVE_VECTORIZATION := $(strip $(shell echo $(KOKKOS_OPTIONS) | grep "aggressive_vectorization" | wc -l))
KOKKOS_INTERNAL_DISABLE_PROFILING := $(strip $(shell echo $(KOKKOS_OPTIONS) | grep "disable_profiling" | wc -l))
KOKKOS_INTERNAL_CUDA_USE_LDG := $(strip $(shell echo $(KOKKOS_CUDA_OPTIONS) | grep "use_ldg" | wc -l))
@ -41,21 +40,21 @@ KOKKOS_INTERNAL_CUDA_USE_UVM := $(strip $(shell echo $(KOKKOS_CUDA_OPTIONS) | gr
KOKKOS_INTERNAL_CUDA_USE_RELOC := $(strip $(shell echo $(KOKKOS_CUDA_OPTIONS) | grep "rdc" | wc -l))
KOKKOS_INTERNAL_CUDA_USE_LAMBDA := $(strip $(shell echo $(KOKKOS_CUDA_OPTIONS) | grep "enable_lambda" | wc -l))
# Check for Kokkos Host Execution Spaces one of which must be on
# Check for Kokkos Host Execution Spaces one of which must be on.
KOKKOS_INTERNAL_USE_OPENMP := $(strip $(shell echo $(KOKKOS_DEVICES) | grep OpenMP | wc -l))
KOKKOS_INTERNAL_USE_PTHREADS := $(strip $(shell echo $(KOKKOS_DEVICES) | grep Pthread | wc -l))
KOKKOS_INTERNAL_USE_QTHREADS := $(strip $(shell echo $(KOKKOS_DEVICES) | grep Qthreads | wc -l))
KOKKOS_INTERNAL_USE_SERIAL := $(strip $(shell echo $(KOKKOS_DEVICES) | grep Serial | wc -l))
KOKKOS_INTERNAL_USE_QTHREAD := $(strip $(shell echo $(KOKKOS_DEVICES) | grep Qthread | wc -l))
ifeq ($(KOKKOS_INTERNAL_USE_OPENMP), 0)
ifeq ($(KOKKOS_INTERNAL_USE_PTHREADS), 0)
KOKKOS_INTERNAL_USE_SERIAL := 1
ifeq ($(KOKKOS_INTERNAL_USE_QTHREADS), 0)
KOKKOS_INTERNAL_USE_SERIAL := 1
endif
endif
endif
# Check for other Execution Spaces
# Check for other Execution Spaces.
KOKKOS_INTERNAL_USE_CUDA := $(strip $(shell echo $(KOKKOS_DEVICES) | grep Cuda | wc -l))
ifeq ($(KOKKOS_INTERNAL_USE_CUDA), 1)
@ -64,27 +63,25 @@ ifeq ($(KOKKOS_INTERNAL_USE_CUDA), 1)
KOKKOS_INTERNAL_COMPILER_NVCC_VERSION := $(shell nvcc --version 2>&1 | grep release | cut -d' ' -f5 | cut -d',' -f1 | tr -d .)
endif
# Check OS
# Check OS.
KOKKOS_OS := $(shell uname -s)
KOKKOS_INTERNAL_OS_CYGWIN := $(shell uname -s | grep CYGWIN | wc -l)
KOKKOS_INTERNAL_OS_LINUX := $(shell uname -s | grep Linux | wc -l)
KOKKOS_INTERNAL_OS_DARWIN := $(shell uname -s | grep Darwin | wc -l)
# Check compiler
KOKKOS_INTERNAL_COMPILER_INTEL := $(shell $(CXX) --version 2>&1 | grep "Intel Corporation" | wc -l)
KOKKOS_INTERNAL_COMPILER_PGI := $(shell $(CXX) --version 2>&1 | grep PGI | wc -l)
KOKKOS_INTERNAL_COMPILER_XL := $(shell $(CXX) -qversion 2>&1 | grep XL | wc -l)
KOKKOS_INTERNAL_COMPILER_CRAY := $(shell $(CXX) -craype-verbose 2>&1 | grep "CC-" | wc -l)
KOKKOS_INTERNAL_COMPILER_NVCC := $(shell $(CXX) --version 2>&1 | grep "nvcc" | wc -l)
# Check compiler.
KOKKOS_INTERNAL_COMPILER_INTEL := $(shell $(CXX) --version 2>&1 | grep "Intel Corporation" | wc -l)
KOKKOS_INTERNAL_COMPILER_PGI := $(shell $(CXX) --version 2>&1 | grep PGI | wc -l)
KOKKOS_INTERNAL_COMPILER_XL := $(shell $(CXX) -qversion 2>&1 | grep XL | wc -l)
KOKKOS_INTERNAL_COMPILER_CRAY := $(shell $(CXX) -craype-verbose 2>&1 | grep "CC-" | wc -l)
KOKKOS_INTERNAL_COMPILER_NVCC := $(shell $(CXX) --version 2>&1 | grep "nvcc" | wc -l)
ifneq ($(OMPI_CXX),)
KOKKOS_INTERNAL_COMPILER_NVCC := $(shell $(OMPI_CXX) --version 2>&1 | grep "nvcc" | wc -l)
endif
ifneq ($(MPICH_CXX),)
KOKKOS_INTERNAL_COMPILER_NVCC := $(shell $(MPICH_CXX) --version 2>&1 | grep "nvcc" | wc -l)
endif
KOKKOS_INTERNAL_COMPILER_CLANG := $(shell $(CXX) --version 2>&1 | grep "clang" | wc -l)
KOKKOS_INTERNAL_COMPILER_CLANG := $(shell $(CXX) --version 2>&1 | grep "clang" | wc -l)
ifeq ($(KOKKOS_INTERNAL_COMPILER_CLANG), 2)
KOKKOS_INTERNAL_COMPILER_CLANG = 1
@ -95,17 +92,17 @@ endif
ifeq ($(KOKKOS_INTERNAL_COMPILER_CLANG), 1)
KOKKOS_INTERNAL_COMPILER_CLANG_VERSION := $(shell clang --version | grep version | cut -d ' ' -f3 | tr -d '.')
ifeq ($(KOKKOS_INTERNAL_USE_CUDA), 1)
ifeq ($(shell test $(KOKKOS_INTERNAL_COMPILER_CLANG_VERSION) -lt 400; echo $$?),0)
$(error Compiling Cuda code directly with Clang requires version 4.0.0 or higher)
$(error Compiling Cuda code directly with Clang requires version 4.0.0 or higher)
endif
KOKKOS_INTERNAL_CUDA_USE_LAMBDA := 1
endif
endif
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
KOKKOS_INTERNAL_OPENMP_FLAG := -mp
KOKKOS_INTERNAL_OPENMP_FLAG := -mp
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_CLANG), 1)
KOKKOS_INTERNAL_OPENMP_FLAG := -fopenmp=libomp
@ -114,7 +111,7 @@ else
KOKKOS_INTERNAL_OPENMP_FLAG := -qsmp=omp
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_CRAY), 1)
# OpenMP is turned on by default in Cray compiler environment
# OpenMP is turned on by default in Cray compiler environment.
KOKKOS_INTERNAL_OPENMP_FLAG :=
else
KOKKOS_INTERNAL_OPENMP_FLAG := -fopenmp
@ -138,9 +135,9 @@ else
endif
endif
# Check for Kokkos Architecture settings
# Check for Kokkos Architecture settings.
#Intel based
# Intel based.
KOKKOS_INTERNAL_USE_ARCH_KNC := $(strip $(shell echo $(KOKKOS_ARCH) | grep KNC | wc -l))
KOKKOS_INTERNAL_USE_ARCH_SNB := $(strip $(shell echo $(KOKKOS_ARCH) | grep SNB | wc -l))
KOKKOS_INTERNAL_USE_ARCH_HSW := $(strip $(shell echo $(KOKKOS_ARCH) | grep HSW | wc -l))
@ -148,8 +145,8 @@ KOKKOS_INTERNAL_USE_ARCH_BDW := $(strip $(shell echo $(KOKKOS_ARCH) | grep BDW |
KOKKOS_INTERNAL_USE_ARCH_SKX := $(strip $(shell echo $(KOKKOS_ARCH) | grep SKX | wc -l))
KOKKOS_INTERNAL_USE_ARCH_KNL := $(strip $(shell echo $(KOKKOS_ARCH) | grep KNL | wc -l))
#NVIDIA based
NVCC_WRAPPER := $(KOKKOS_PATH)/config/nvcc_wrapper
# NVIDIA based.
NVCC_WRAPPER := $(KOKKOS_PATH)/config/nvcc_wrapper
KOKKOS_INTERNAL_USE_ARCH_KEPLER30 := $(strip $(shell echo $(KOKKOS_ARCH) | grep Kepler30 | wc -l))
KOKKOS_INTERNAL_USE_ARCH_KEPLER32 := $(strip $(shell echo $(KOKKOS_ARCH) | grep Kepler32 | wc -l))
KOKKOS_INTERNAL_USE_ARCH_KEPLER35 := $(strip $(shell echo $(KOKKOS_ARCH) | grep Kepler35 | wc -l))
@ -170,46 +167,46 @@ KOKKOS_INTERNAL_USE_ARCH_NVIDIA := $(strip $(shell echo $(KOKKOS_INTERNAL_USE_AR
+ $(KOKKOS_INTERNAL_USE_ARCH_MAXWELL53) | bc))
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_NVIDIA), 0)
KOKKOS_INTERNAL_USE_ARCH_MAXWELL50 := $(strip $(shell echo $(KOKKOS_ARCH) | grep Maxwell | wc -l))
KOKKOS_INTERNAL_USE_ARCH_KEPLER35 := $(strip $(shell echo $(KOKKOS_ARCH) | grep Kepler | wc -l))
KOKKOS_INTERNAL_USE_ARCH_NVIDIA := $(strip $(shell echo $(KOKKOS_INTERNAL_USE_ARCH_KEPLER30) \
+ $(KOKKOS_INTERNAL_USE_ARCH_KEPLER32) \
+ $(KOKKOS_INTERNAL_USE_ARCH_KEPLER35) \
+ $(KOKKOS_INTERNAL_USE_ARCH_KEPLER37) \
+ $(KOKKOS_INTERNAL_USE_ARCH_PASCAL61) \
+ $(KOKKOS_INTERNAL_USE_ARCH_PASCAL60) \
+ $(KOKKOS_INTERNAL_USE_ARCH_MAXWELL50) \
+ $(KOKKOS_INTERNAL_USE_ARCH_MAXWELL52) \
+ $(KOKKOS_INTERNAL_USE_ARCH_MAXWELL53) | bc))
KOKKOS_INTERNAL_USE_ARCH_MAXWELL50 := $(strip $(shell echo $(KOKKOS_ARCH) | grep Maxwell | wc -l))
KOKKOS_INTERNAL_USE_ARCH_KEPLER35 := $(strip $(shell echo $(KOKKOS_ARCH) | grep Kepler | wc -l))
KOKKOS_INTERNAL_USE_ARCH_NVIDIA := $(strip $(shell echo $(KOKKOS_INTERNAL_USE_ARCH_KEPLER30) \
+ $(KOKKOS_INTERNAL_USE_ARCH_KEPLER32) \
+ $(KOKKOS_INTERNAL_USE_ARCH_KEPLER35) \
+ $(KOKKOS_INTERNAL_USE_ARCH_KEPLER37) \
+ $(KOKKOS_INTERNAL_USE_ARCH_PASCAL61) \
+ $(KOKKOS_INTERNAL_USE_ARCH_PASCAL60) \
+ $(KOKKOS_INTERNAL_USE_ARCH_MAXWELL50) \
+ $(KOKKOS_INTERNAL_USE_ARCH_MAXWELL52) \
+ $(KOKKOS_INTERNAL_USE_ARCH_MAXWELL53) | bc))
endif
#ARM based
# ARM based.
KOKKOS_INTERNAL_USE_ARCH_ARMV80 := $(strip $(shell echo $(KOKKOS_ARCH) | grep ARMv80 | wc -l))
KOKKOS_INTERNAL_USE_ARCH_ARMV81 := $(strip $(shell echo $(KOKKOS_ARCH) | grep ARMv81 | wc -l))
KOKKOS_INTERNAL_USE_ARCH_ARMV8_THUNDERX := $(strip $(shell echo $(KOKKOS_ARCH) | grep ARMv8-ThunderX | wc -l))
#IBM based
# IBM based.
KOKKOS_INTERNAL_USE_ARCH_BGQ := $(strip $(shell echo $(KOKKOS_ARCH) | grep BGQ | wc -l))
KOKKOS_INTERNAL_USE_ARCH_POWER7 := $(strip $(shell echo $(KOKKOS_ARCH) | grep Power7 | wc -l))
KOKKOS_INTERNAL_USE_ARCH_POWER8 := $(strip $(shell echo $(KOKKOS_ARCH) | grep Power8 | wc -l))
KOKKOS_INTERNAL_USE_ARCH_POWER9 := $(strip $(shell echo $(KOKKOS_ARCH) | grep Power9 | wc -l))
KOKKOS_INTERNAL_USE_ARCH_IBM := $(strip $(shell echo $(KOKKOS_INTERNAL_USE_ARCH_BGQ)+$(KOKKOS_INTERNAL_USE_ARCH_POWER7)+$(KOKKOS_INTERNAL_USE_ARCH_POWER8)+$(KOKKOS_INTERNAL_USE_ARCH_POWER9) | bc))
#AMD based
# AMD based.
KOKKOS_INTERNAL_USE_ARCH_AMDAVX := $(strip $(shell echo $(KOKKOS_ARCH) | grep AMDAVX | wc -l))
#Any AVX?
# Any AVX?
KOKKOS_INTERNAL_USE_ARCH_AVX := $(strip $(shell echo $(KOKKOS_INTERNAL_USE_ARCH_SNB)+$(KOKKOS_INTERNAL_USE_ARCH_AMDAVX) | bc ))
KOKKOS_INTERNAL_USE_ARCH_AVX2 := $(strip $(shell echo $(KOKKOS_INTERNAL_USE_ARCH_HSW)+$(KOKKOS_INTERNAL_USE_ARCH_BDW) | bc ))
KOKKOS_INTERNAL_USE_ARCH_AVX512MIC := $(strip $(shell echo $(KOKKOS_INTERNAL_USE_ARCH_KNL) | bc ))
KOKKOS_INTERNAL_USE_ARCH_AVX512XEON := $(strip $(shell echo $(KOKKOS_INTERNAL_USE_ARCH_SKX) | bc ))
# Decide what ISA level we are able to support
KOKKOS_INTERNAL_USE_ISA_X86_64 := $(strip $(shell echo $(KOKKOS_INTERNAL_USE_ARCH_SNB)+$(KOKKOS_INTERNAL_USE_ARCH_HSW)+$(KOKKOS_INTERNAL_USE_ARCH_BDW)+$(KOKKOS_INTERNAL_USE_ARCH_KNL)+$(KOKKOS_INTERNAL_USE_ARCH_SKX) | bc ))
KOKKOS_INTERNAL_USE_ISA_KNC := $(strip $(shell echo $(KOKKOS_INTERNAL_USE_ARCH_KNC) | bc ))
KOKKOS_INTERNAL_USE_ISA_POWERPCLE := $(strip $(shell echo $(KOKKOS_INTERNAL_USE_ARCH_POWER8)+$(KOKKOS_INTERNAL_USE_ARCH_POWER9) | bc ))
# Decide what ISA level we are able to support.
KOKKOS_INTERNAL_USE_ISA_X86_64 := $(strip $(shell echo $(KOKKOS_INTERNAL_USE_ARCH_SNB)+$(KOKKOS_INTERNAL_USE_ARCH_HSW)+$(KOKKOS_INTERNAL_USE_ARCH_BDW)+$(KOKKOS_INTERNAL_USE_ARCH_KNL)+$(KOKKOS_INTERNAL_USE_ARCH_SKX) | bc ))
KOKKOS_INTERNAL_USE_ISA_KNC := $(strip $(shell echo $(KOKKOS_INTERNAL_USE_ARCH_KNC) | bc ))
KOKKOS_INTERNAL_USE_ISA_POWERPCLE := $(strip $(shell echo $(KOKKOS_INTERNAL_USE_ARCH_POWER8)+$(KOKKOS_INTERNAL_USE_ARCH_POWER9) | bc ))
#Incompatible flags?
# Incompatible flags?
KOKKOS_INTERNAL_USE_ARCH_MULTIHOST := $(strip $(shell echo "$(KOKKOS_INTERNAL_USE_ARCH_AVX)+$(KOKKOS_INTERNAL_USE_ARCH_AVX2)+$(KOKKOS_INTERNAL_USE_ARCH_KNC)+$(KOKKOS_INTERNAL_USE_ARCH_IBM)+$(KOKKOS_INTERNAL_USE_ARCH_AMDAVX)+$(KOKKOS_INTERNAL_USE_ARCH_ARMV80)+$(KOKKOS_INTERNAL_USE_ARCH_ARMV81)+$(KOKKOS_INTERNAL_USE_ARCH_ARMV8_THUNDERX)>1" | bc ))
KOKKOS_INTERNAL_USE_ARCH_MULTIGPU := $(strip $(shell echo "$(KOKKOS_INTERNAL_USE_ARCH_NVIDIA)>1" | bc))
@ -220,7 +217,7 @@ ifeq ($(KOKKOS_INTERNAL_USE_ARCH_MULTIGPU), 1)
$(error Defined Multiple GPU architectures: KOKKOS_ARCH=$(KOKKOS_ARCH) )
endif
#Generating the list of Flags
# Generating the list of Flags.
KOKKOS_CPPFLAGS = -I./ -I$(KOKKOS_PATH)/core/src -I$(KOKKOS_PATH)/containers/src -I$(KOKKOS_PATH)/algorithms/src
@ -233,98 +230,96 @@ KOKKOS_CXXFLAGS =
KOKKOS_LIBS = -lkokkos -ldl
KOKKOS_LDFLAGS = -L$(shell pwd)
KOKKOS_SRC =
KOKKOS_SRC =
KOKKOS_HEADERS =
#Generating the KokkosCore_config.h file
# Generating the KokkosCore_config.h file.
tmp := $(shell echo "/* ---------------------------------------------" > KokkosCore_config.tmp)
tmp := $(shell echo "Makefile constructed configuration:" >> KokkosCore_config.tmp)
tmp := $(shell date >> KokkosCore_config.tmp)
tmp := $(shell echo "----------------------------------------------*/" >> KokkosCore_config.tmp)
tmp := $(shell echo "/* Execution Spaces */" >> KokkosCore_config.tmp)
ifeq ($(KOKKOS_INTERNAL_USE_CUDA), 1)
tmp := $(shell echo "\#define KOKKOS_HAVE_CUDA 1" >> KokkosCore_config.tmp )
endif
ifeq ($(KOKKOS_INTERNAL_USE_OPENMP), 1)
tmp := $(shell echo '\#define KOKKOS_HAVE_OPENMP 1' >> KokkosCore_config.tmp)
tmp := $(shell echo '\#define KOKKOS_HAVE_OPENMP 1' >> KokkosCore_config.tmp)
endif
ifeq ($(KOKKOS_INTERNAL_USE_PTHREADS), 1)
tmp := $(shell echo "\#define KOKKOS_HAVE_PTHREAD 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_HAVE_PTHREAD 1" >> KokkosCore_config.tmp )
endif
ifeq ($(KOKKOS_INTERNAL_USE_QTHREADS), 1)
tmp := $(shell echo "\#define KOKKOS_HAVE_QTHREADS 1" >> KokkosCore_config.tmp )
endif
ifeq ($(KOKKOS_INTERNAL_USE_SERIAL), 1)
tmp := $(shell echo "\#define KOKKOS_HAVE_SERIAL 1" >> KokkosCore_config.tmp )
endif
ifeq ($(KOKKOS_INTERNAL_USE_CUDA), 1)
tmp := $(shell echo "\#define KOKKOS_HAVE_CUDA 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_HAVE_SERIAL 1" >> KokkosCore_config.tmp )
endif
ifeq ($(KOKKOS_INTERNAL_USE_ISA_X86_64), 1)
tmp := $(shell echo "\#ifndef __CUDA_ARCH__" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_USE_ISA_X86_64" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#endif" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#ifndef __CUDA_ARCH__" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_USE_ISA_X86_64" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#endif" >> KokkosCore_config.tmp )
endif
ifeq ($(KOKKOS_INTERNAL_USE_ISA_KNC), 1)
tmp := $(shell echo "\#ifndef __CUDA_ARCH__" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_USE_ISA_KNC" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#endif" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#ifndef __CUDA_ARCH__" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_USE_ISA_KNC" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#endif" >> KokkosCore_config.tmp )
endif
ifeq ($(KOKKOS_INTERNAL_USE_ISA_POWERPCLE), 1)
tmp := $(shell echo "\#ifndef __CUDA_ARCH__" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_USE_ISA_POWERPCLE" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#endif" >> KokkosCore_config.tmp )
endif
ifeq ($(KOKKOS_INTERNAL_USE_QTHREAD), 1)
KOKKOS_CPPFLAGS += -I$(QTHREAD_PATH)/include
KOKKOS_LDFLAGS += -L$(QTHREAD_PATH)/lib
tmp := $(shell echo "\#define KOKKOS_HAVE_QTHREAD 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#ifndef __CUDA_ARCH__" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_USE_ISA_POWERPCLE" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#endif" >> KokkosCore_config.tmp )
endif
tmp := $(shell echo "/* General Settings */" >> KokkosCore_config.tmp)
ifeq ($(KOKKOS_INTERNAL_ENABLE_CXX11), 1)
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_CXX11_FLAG)
tmp := $(shell echo "\#define KOKKOS_HAVE_CXX11 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_CXX11_FLAG)
tmp := $(shell echo "\#define KOKKOS_HAVE_CXX11 1" >> KokkosCore_config.tmp )
endif
ifeq ($(KOKKOS_INTERNAL_ENABLE_CXX1Z), 1)
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_CXX1Z_FLAG)
tmp := $(shell echo "\#define KOKKOS_HAVE_CXX11 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_HAVE_CXX1Z 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_CXX1Z_FLAG)
tmp := $(shell echo "\#define KOKKOS_HAVE_CXX11 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_HAVE_CXX1Z 1" >> KokkosCore_config.tmp )
endif
ifeq ($(KOKKOS_INTERNAL_ENABLE_DEBUG), 1)
ifeq ($(KOKKOS_INTERNAL_COMPILER_NVCC), 1)
KOKKOS_CXXFLAGS += -lineinfo
KOKKOS_CXXFLAGS += -lineinfo
endif
KOKKOS_CXXFLAGS += -g
KOKKOS_LDFLAGS += -g -ldl
tmp := $(shell echo "\#define KOKKOS_ENABLE_DEBUG_BOUNDS_CHECK 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_HAVE_DEBUG 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += -g
KOKKOS_LDFLAGS += -g -ldl
tmp := $(shell echo "\#define KOKKOS_ENABLE_DEBUG_BOUNDS_CHECK 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_HAVE_DEBUG 1" >> KokkosCore_config.tmp )
endif
ifeq ($(KOKKOS_INTERNAL_USE_HWLOC), 1)
KOKKOS_CPPFLAGS += -I$(HWLOC_PATH)/include
KOKKOS_LDFLAGS += -L$(HWLOC_PATH)/lib
KOKKOS_LIBS += -lhwloc
tmp := $(shell echo "\#define KOKKOS_HAVE_HWLOC 1" >> KokkosCore_config.tmp )
KOKKOS_CPPFLAGS += -I$(HWLOC_PATH)/include
KOKKOS_LDFLAGS += -L$(HWLOC_PATH)/lib
KOKKOS_LIBS += -lhwloc
tmp := $(shell echo "\#define KOKKOS_HAVE_HWLOC 1" >> KokkosCore_config.tmp )
endif
ifeq ($(KOKKOS_INTERNAL_USE_LIBRT), 1)
tmp := $(shell echo "\#define KOKKOS_USE_LIBRT 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define PREC_TIMER 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_USE_LIBRT 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define PREC_TIMER 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOSP_ENABLE_RTLIB 1" >> KokkosCore_config.tmp )
KOKKOS_LIBS += -lrt
KOKKOS_LIBS += -lrt
endif
ifeq ($(KOKKOS_INTERNAL_USE_MEMKIND), 1)
KOKKOS_CPPFLAGS += -I$(MEMKIND_PATH)/include
KOKKOS_LDFLAGS += -L$(MEMKIND_PATH)/lib
KOKKOS_LIBS += -lmemkind
KOKKOS_LDFLAGS += -L$(MEMKIND_PATH)/lib
KOKKOS_LIBS += -lmemkind
tmp := $(shell echo "\#define KOKKOS_HAVE_HBWSPACE 1" >> KokkosCore_config.tmp )
endif
@ -341,262 +336,286 @@ endif
tmp := $(shell echo "/* Cuda Settings */" >> KokkosCore_config.tmp)
ifeq ($(KOKKOS_INTERNAL_USE_CUDA), 1)
ifeq ($(KOKKOS_INTERNAL_CUDA_USE_LDG), 1)
tmp := $(shell echo "\#define KOKKOS_CUDA_USE_LDG_INTRINSIC 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_CUDA_USE_LDG_INTRINSIC 1" >> KokkosCore_config.tmp )
endif
ifeq ($(KOKKOS_INTERNAL_CUDA_USE_UVM), 1)
tmp := $(shell echo "\#define KOKKOS_CUDA_USE_UVM 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_USE_CUDA_UVM 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_CUDA_USE_UVM 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_USE_CUDA_UVM 1" >> KokkosCore_config.tmp )
endif
ifeq ($(KOKKOS_INTERNAL_CUDA_USE_RELOC), 1)
tmp := $(shell echo "\#define KOKKOS_CUDA_USE_RELOCATABLE_DEVICE_CODE 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += --relocatable-device-code=true
KOKKOS_LDFLAGS += --relocatable-device-code=true
tmp := $(shell echo "\#define KOKKOS_CUDA_USE_RELOCATABLE_DEVICE_CODE 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += --relocatable-device-code=true
KOKKOS_LDFLAGS += --relocatable-device-code=true
endif
ifeq ($(KOKKOS_INTERNAL_CUDA_USE_LAMBDA), 1)
ifeq ($(KOKKOS_INTERNAL_COMPILER_NVCC), 1)
ifeq ($(shell test $(KOKKOS_INTERNAL_COMPILER_NVCC_VERSION) -gt 70; echo $$?),0)
tmp := $(shell echo "\#define KOKKOS_CUDA_USE_LAMBDA 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += -expt-extended-lambda
tmp := $(shell echo "\#define KOKKOS_CUDA_USE_LAMBDA 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += -expt-extended-lambda
else
$(warning Warning: Cuda Lambda support was requested but NVCC version is too low. This requires NVCC for Cuda version 7.5 or higher. Disabling Lambda support now.)
endif
endif
ifeq ($(KOKKOS_INTERNAL_COMPILER_CLANG), 1)
tmp := $(shell echo "\#define KOKKOS_CUDA_USE_LAMBDA 1" >> KokkosCore_config.tmp )
endif
endif
endif
#Add Architecture flags
# Add Architecture flags.
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_ARMV80), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_ARMV80 1" >> KokkosCore_config.tmp )
ifeq ($(KOKKOS_INTERNAL_COMPILER_CRAY), 1)
KOKKOS_CXXFLAGS +=
KOKKOS_LDFLAGS +=
tmp := $(shell echo "\#define KOKKOS_ARCH_ARMV80 1" >> KokkosCore_config.tmp )
ifeq ($(KOKKOS_INTERNAL_COMPILER_CRAY), 1)
KOKKOS_CXXFLAGS +=
KOKKOS_LDFLAGS +=
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
KOKKOS_CXXFLAGS +=
KOKKOS_LDFLAGS +=
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
KOKKOS_CXXFLAGS +=
KOKKOS_LDFLAGS +=
else
KOKKOS_CXXFLAGS += -march=armv8-a
KOKKOS_LDFLAGS += -march=armv8-a
endif
KOKKOS_CXXFLAGS += -march=armv8-a
KOKKOS_LDFLAGS += -march=armv8-a
endif
endif
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_ARMV81), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_ARMV81 1" >> KokkosCore_config.tmp )
ifeq ($(KOKKOS_INTERNAL_COMPILER_CRAY), 1)
KOKKOS_CXXFLAGS +=
KOKKOS_LDFLAGS +=
tmp := $(shell echo "\#define KOKKOS_ARCH_ARMV81 1" >> KokkosCore_config.tmp )
ifeq ($(KOKKOS_INTERNAL_COMPILER_CRAY), 1)
KOKKOS_CXXFLAGS +=
KOKKOS_LDFLAGS +=
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
KOKKOS_CXXFLAGS +=
KOKKOS_LDFLAGS +=
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
KOKKOS_CXXFLAGS +=
KOKKOS_LDFLAGS +=
else
KOKKOS_CXXFLAGS += -march=armv8.1-a
KOKKOS_LDFLAGS += -march=armv8.1-a
endif
KOKKOS_CXXFLAGS += -march=armv8.1-a
KOKKOS_LDFLAGS += -march=armv8.1-a
endif
endif
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_ARMV8_THUNDERX), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_ARMV80 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_ARMV8_THUNDERX 1" >> KokkosCore_config.tmp )
ifeq ($(KOKKOS_INTERNAL_COMPILER_CRAY), 1)
KOKKOS_CXXFLAGS +=
KOKKOS_LDFLAGS +=
tmp := $(shell echo "\#define KOKKOS_ARCH_ARMV80 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_ARMV8_THUNDERX 1" >> KokkosCore_config.tmp )
ifeq ($(KOKKOS_INTERNAL_COMPILER_CRAY), 1)
KOKKOS_CXXFLAGS +=
KOKKOS_LDFLAGS +=
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
KOKKOS_CXXFLAGS +=
KOKKOS_LDFLAGS +=
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
KOKKOS_CXXFLAGS +=
KOKKOS_LDFLAGS +=
else
KOKKOS_CXXFLAGS += -march=armv8-a -mtune=thunderx
KOKKOS_LDFLAGS += -march=armv8-a -mtune=thunderx
endif
KOKKOS_CXXFLAGS += -march=armv8-a -mtune=thunderx
KOKKOS_LDFLAGS += -march=armv8-a -mtune=thunderx
endif
endif
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_AVX), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_AVX 1" >> KokkosCore_config.tmp )
ifeq ($(KOKKOS_INTERNAL_COMPILER_INTEL), 1)
KOKKOS_CXXFLAGS += -mavx
KOKKOS_LDFLAGS += -mavx
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_CRAY), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_AVX 1" >> KokkosCore_config.tmp )
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
KOKKOS_CXXFLAGS += -tp=sandybridge
KOKKOS_LDFLAGS += -tp=sandybridge
else
# Assume that this is a really a GNU compiler
KOKKOS_CXXFLAGS += -mavx
KOKKOS_LDFLAGS += -mavx
endif
endif
endif
ifeq ($(KOKKOS_INTERNAL_COMPILER_INTEL), 1)
KOKKOS_CXXFLAGS += -mavx
KOKKOS_LDFLAGS += -mavx
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_CRAY), 1)
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
KOKKOS_CXXFLAGS += -tp=sandybridge
KOKKOS_LDFLAGS += -tp=sandybridge
else
# Assume that this is a really a GNU compiler.
KOKKOS_CXXFLAGS += -mavx
KOKKOS_LDFLAGS += -mavx
endif
endif
endif
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_POWER8), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_POWER8 1" >> KokkosCore_config.tmp )
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_POWER8 1" >> KokkosCore_config.tmp )
else
# Assume that this is a really a GNU compiler or it could be XL on P8
KOKKOS_CXXFLAGS += -mcpu=power8 -mtune=power8
KOKKOS_LDFLAGS += -mcpu=power8 -mtune=power8
endif
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
else
# Assume that this is a really a GNU compiler or it could be XL on P8.
KOKKOS_CXXFLAGS += -mcpu=power8 -mtune=power8
KOKKOS_LDFLAGS += -mcpu=power8 -mtune=power8
endif
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_POWER9), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_POWER9 1" >> KokkosCore_config.tmp )
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_POWER9 1" >> KokkosCore_config.tmp )
else
# Assume that this is a really a GNU compiler or it could be XL on P9
KOKKOS_CXXFLAGS += -mcpu=power9 -mtune=power9
KOKKOS_LDFLAGS += -mcpu=power9 -mtune=power9
endif
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
else
# Assume that this is a really a GNU compiler or it could be XL on P9.
KOKKOS_CXXFLAGS += -mcpu=power9 -mtune=power9
KOKKOS_LDFLAGS += -mcpu=power9 -mtune=power9
endif
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_AVX2), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_AVX2 1" >> KokkosCore_config.tmp )
ifeq ($(KOKKOS_INTERNAL_COMPILER_INTEL), 1)
KOKKOS_CXXFLAGS += -xCORE-AVX2
KOKKOS_LDFLAGS += -xCORE-AVX2
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_CRAY), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_AVX2 1" >> KokkosCore_config.tmp )
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
KOKKOS_CXXFLAGS += -tp=haswell
KOKKOS_LDFLAGS += -tp=haswell
else
# Assume that this is a really a GNU compiler
KOKKOS_CXXFLAGS += -march=core-avx2 -mtune=core-avx2
KOKKOS_LDFLAGS += -march=core-avx2 -mtune=core-avx2
endif
endif
endif
ifeq ($(KOKKOS_INTERNAL_COMPILER_INTEL), 1)
KOKKOS_CXXFLAGS += -xCORE-AVX2
KOKKOS_LDFLAGS += -xCORE-AVX2
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_CRAY), 1)
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
KOKKOS_CXXFLAGS += -tp=haswell
KOKKOS_LDFLAGS += -tp=haswell
else
# Assume that this is a really a GNU compiler.
KOKKOS_CXXFLAGS += -march=core-avx2 -mtune=core-avx2
KOKKOS_LDFLAGS += -march=core-avx2 -mtune=core-avx2
endif
endif
endif
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_AVX512MIC), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_AVX512MIC 1" >> KokkosCore_config.tmp )
ifeq ($(KOKKOS_INTERNAL_COMPILER_INTEL), 1)
KOKKOS_CXXFLAGS += -xMIC-AVX512
KOKKOS_LDFLAGS += -xMIC-AVX512
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_CRAY), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_AVX512MIC 1" >> KokkosCore_config.tmp )
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
ifeq ($(KOKKOS_INTERNAL_COMPILER_INTEL), 1)
KOKKOS_CXXFLAGS += -xMIC-AVX512
KOKKOS_LDFLAGS += -xMIC-AVX512
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_CRAY), 1)
else
# Asssume that this is really a GNU compiler
KOKKOS_CXXFLAGS += -march=knl
KOKKOS_LDFLAGS += -march=knl
endif
endif
endif
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
else
# Asssume that this is really a GNU compiler.
KOKKOS_CXXFLAGS += -march=knl
KOKKOS_LDFLAGS += -march=knl
endif
endif
endif
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_AVX512XEON), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_AVX512XEON 1" >> KokkosCore_config.tmp )
ifeq ($(KOKKOS_INTERNAL_COMPILER_INTEL), 1)
KOKKOS_CXXFLAGS += -xCORE-AVX512
KOKKOS_LDFLAGS += -xCORE-AVX512
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_CRAY), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_AVX512XEON 1" >> KokkosCore_config.tmp )
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
ifeq ($(KOKKOS_INTERNAL_COMPILER_INTEL), 1)
KOKKOS_CXXFLAGS += -xCORE-AVX512
KOKKOS_LDFLAGS += -xCORE-AVX512
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_CRAY), 1)
else
# Nothing here yet
KOKKOS_CXXFLAGS += -march=skylake-avx512
KOKKOS_LDFLAGS += -march=skylake-avx512
endif
endif
endif
else
ifeq ($(KOKKOS_INTERNAL_COMPILER_PGI), 1)
else
# Nothing here yet.
KOKKOS_CXXFLAGS += -march=skylake-avx512
KOKKOS_LDFLAGS += -march=skylake-avx512
endif
endif
endif
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_KNC), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_KNC 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += -mmic
KOKKOS_LDFLAGS += -mmic
tmp := $(shell echo "\#define KOKKOS_ARCH_KNC 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += -mmic
KOKKOS_LDFLAGS += -mmic
endif
#Figure out the architecture flag for Cuda
# Figure out the architecture flag for Cuda.
ifeq ($(KOKKOS_INTERNAL_USE_CUDA), 1)
ifeq ($(KOKKOS_INTERNAL_COMPILER_NVCC), 1)
KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG=-arch
endif
ifeq ($(KOKKOS_INTERNAL_COMPILER_CLANG), 1)
KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG=-x cuda --cuda-gpu-arch
KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG=--cuda-gpu-arch
KOKKOS_CXXFLAGS += -x cuda
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_KEPLER30), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_KEPLER 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_KEPLER30 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_30
tmp := $(shell echo "\#define KOKKOS_ARCH_KEPLER 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_KEPLER30 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_30
KOKKOS_LDFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_30
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_KEPLER32), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_KEPLER 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_KEPLER32 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_32
tmp := $(shell echo "\#define KOKKOS_ARCH_KEPLER 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_KEPLER32 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_32
KOKKOS_LDFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_32
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_KEPLER35), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_KEPLER 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_KEPLER35 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_35
tmp := $(shell echo "\#define KOKKOS_ARCH_KEPLER 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_KEPLER35 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_35
KOKKOS_LDFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_35
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_KEPLER37), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_KEPLER 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_KEPLER37 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_37
tmp := $(shell echo "\#define KOKKOS_ARCH_KEPLER 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_KEPLER37 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_37
KOKKOS_LDFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_37
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_MAXWELL50), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_MAXWELL 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_MAXWELL50 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_50
tmp := $(shell echo "\#define KOKKOS_ARCH_MAXWELL 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_MAXWELL50 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_50
KOKKOS_LDFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_50
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_MAXWELL52), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_MAXWELL 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_MAXWELL52 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_52
tmp := $(shell echo "\#define KOKKOS_ARCH_MAXWELL 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_MAXWELL52 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_52
KOKKOS_LDFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_52
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_MAXWELL53), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_MAXWELL 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_MAXWELL53 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_53
tmp := $(shell echo "\#define KOKKOS_ARCH_MAXWELL 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_MAXWELL53 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_53
KOKKOS_LDFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_53
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_PASCAL61), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_PASCAL 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_PASCAL61 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_61
tmp := $(shell echo "\#define KOKKOS_ARCH_PASCAL 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_PASCAL61 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_61
KOKKOS_LDFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_61
endif
ifeq ($(KOKKOS_INTERNAL_USE_ARCH_PASCAL60), 1)
tmp := $(shell echo "\#define KOKKOS_ARCH_PASCAL 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_PASCAL60 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_60
tmp := $(shell echo "\#define KOKKOS_ARCH_PASCAL 1" >> KokkosCore_config.tmp )
tmp := $(shell echo "\#define KOKKOS_ARCH_PASCAL60 1" >> KokkosCore_config.tmp )
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_60
KOKKOS_LDFLAGS += $(KOKKOS_INTERNAL_COMPILER_CUDA_ARCH_FLAG)=sm_60
endif
endif
KOKKOS_INTERNAL_LS_CONFIG := $(shell ls KokkosCore_config.h)
ifeq ($(KOKKOS_INTERNAL_LS_CONFIG), KokkosCore_config.h)
KOKKOS_INTERNAL_NEW_CONFIG := $(strip $(shell diff KokkosCore_config.h KokkosCore_config.tmp | grep define | wc -l))
KOKKOS_INTERNAL_NEW_CONFIG := $(strip $(shell diff KokkosCore_config.h KokkosCore_config.tmp | grep define | wc -l))
else
KOKKOS_INTERNAL_NEW_CONFIG := 1
KOKKOS_INTERNAL_NEW_CONFIG := 1
endif
ifneq ($(KOKKOS_INTERNAL_NEW_CONFIG), 0)
tmp := $(shell cp KokkosCore_config.tmp KokkosCore_config.h)
tmp := $(shell cp KokkosCore_config.tmp KokkosCore_config.h)
endif
KOKKOS_HEADERS += $(wildcard $(KOKKOS_PATH)/core/src/*.hpp)
@ -609,53 +628,57 @@ KOKKOS_SRC += $(wildcard $(KOKKOS_PATH)/core/src/impl/*.cpp)
KOKKOS_SRC += $(wildcard $(KOKKOS_PATH)/containers/src/impl/*.cpp)
ifeq ($(KOKKOS_INTERNAL_USE_CUDA), 1)
KOKKOS_SRC += $(wildcard $(KOKKOS_PATH)/core/src/Cuda/*.cpp)
KOKKOS_HEADERS += $(wildcard $(KOKKOS_PATH)/core/src/Cuda/*.hpp)
KOKKOS_CXXFLAGS += -I$(CUDA_PATH)/include
KOKKOS_LDFLAGS += -L$(CUDA_PATH)/lib64
KOKKOS_LIBS += -lcudart -lcuda
endif
ifeq ($(KOKKOS_INTERNAL_USE_PTHREADS), 1)
KOKKOS_LIBS += -lpthread
KOKKOS_SRC += $(wildcard $(KOKKOS_PATH)/core/src/Threads/*.cpp)
KOKKOS_HEADERS += $(wildcard $(KOKKOS_PATH)/core/src/Threads/*.hpp)
endif
ifeq ($(KOKKOS_INTERNAL_USE_QTHREAD), 1)
KOKKOS_LIBS += -lqthread
KOKKOS_SRC += $(wildcard $(KOKKOS_PATH)/core/src/Qthread/*.cpp)
KOKKOS_HEADERS += $(wildcard $(KOKKOS_PATH)/core/src/Qthread/*.hpp)
KOKKOS_SRC += $(wildcard $(KOKKOS_PATH)/core/src/Cuda/*.cpp)
KOKKOS_HEADERS += $(wildcard $(KOKKOS_PATH)/core/src/Cuda/*.hpp)
KOKKOS_CXXFLAGS += -I$(CUDA_PATH)/include
KOKKOS_LDFLAGS += -L$(CUDA_PATH)/lib64
KOKKOS_LIBS += -lcudart -lcuda
endif
ifeq ($(KOKKOS_INTERNAL_USE_OPENMP), 1)
KOKKOS_SRC += $(wildcard $(KOKKOS_PATH)/core/src/OpenMP/*.cpp)
KOKKOS_HEADERS += $(wildcard $(KOKKOS_PATH)/core/src/OpenMP/*.hpp)
ifeq ($(KOKKOS_INTERNAL_COMPILER_NVCC), 1)
KOKKOS_CXXFLAGS += -Xcompiler $(KOKKOS_INTERNAL_OPENMP_FLAG)
else
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_OPENMP_FLAG)
endif
KOKKOS_LDFLAGS += $(KOKKOS_INTERNAL_OPENMP_FLAG)
KOKKOS_SRC += $(wildcard $(KOKKOS_PATH)/core/src/OpenMP/*.cpp)
KOKKOS_HEADERS += $(wildcard $(KOKKOS_PATH)/core/src/OpenMP/*.hpp)
ifeq ($(KOKKOS_INTERNAL_COMPILER_NVCC), 1)
KOKKOS_CXXFLAGS += -Xcompiler $(KOKKOS_INTERNAL_OPENMP_FLAG)
else
KOKKOS_CXXFLAGS += $(KOKKOS_INTERNAL_OPENMP_FLAG)
endif
KOKKOS_LDFLAGS += $(KOKKOS_INTERNAL_OPENMP_FLAG)
endif
#Explicitly set the GCC Toolchain for Clang
ifeq ($(KOKKOS_INTERNAL_USE_PTHREADS), 1)
KOKKOS_SRC += $(wildcard $(KOKKOS_PATH)/core/src/Threads/*.cpp)
KOKKOS_HEADERS += $(wildcard $(KOKKOS_PATH)/core/src/Threads/*.hpp)
KOKKOS_LIBS += -lpthread
endif
ifeq ($(KOKKOS_INTERNAL_USE_QTHREADS), 1)
KOKKOS_SRC += $(wildcard $(KOKKOS_PATH)/core/src/Qthreads/*.cpp)
KOKKOS_HEADERS += $(wildcard $(KOKKOS_PATH)/core/src/Qthreads/*.hpp)
KOKKOS_CPPFLAGS += -I$(QTHREADS_PATH)/include
KOKKOS_LDFLAGS += -L$(QTHREADS_PATH)/lib
KOKKOS_LIBS += -lqthread
endif
# Explicitly set the GCC Toolchain for Clang.
ifeq ($(KOKKOS_INTERNAL_COMPILER_CLANG), 1)
KOKKOS_INTERNAL_GCC_PATH = $(shell which g++)
KOKKOS_INTERNAL_GCC_TOOLCHAIN = $(KOKKOS_INTERNAL_GCC_PATH:/bin/g++=)
KOKKOS_CXXFLAGS += --gcc-toolchain=$(KOKKOS_INTERNAL_GCC_TOOLCHAIN) -DKOKKOS_CUDA_CLANG_WORKAROUND -DKOKKOS_CUDA_USE_LDG_INTRINSIC
KOKKOS_LDFLAGS += --gcc-toolchain=$(KOKKOS_INTERNAL_GCC_TOOLCHAIN)
KOKKOS_INTERNAL_GCC_PATH = $(shell which g++)
KOKKOS_INTERNAL_GCC_TOOLCHAIN = $(KOKKOS_INTERNAL_GCC_PATH:/bin/g++=)
KOKKOS_CXXFLAGS += --gcc-toolchain=$(KOKKOS_INTERNAL_GCC_TOOLCHAIN) -DKOKKOS_CUDA_CLANG_WORKAROUND -DKOKKOS_CUDA_USE_LDG_INTRINSIC
KOKKOS_LDFLAGS += --gcc-toolchain=$(KOKKOS_INTERNAL_GCC_TOOLCHAIN)
endif
#With Cygwin functions such as fdopen and fileno are not defined
#when strict ansi is enabled. strict ansi gets enabled with --std=c++11
#though. So we hard undefine it here. Not sure if that has any bad side effects
#This is needed for gtest actually, not for Kokkos itself!
# With Cygwin functions such as fdopen and fileno are not defined
# when strict ansi is enabled. strict ansi gets enabled with --std=c++11
# though. So we hard undefine it here. Not sure if that has any bad side effects
# This is needed for gtest actually, not for Kokkos itself!
ifeq ($(KOKKOS_INTERNAL_OS_CYGWIN), 1)
KOKKOS_CXXFLAGS += -U__STRICT_ANSI__
endif
# Setting up dependencies
# Setting up dependencies.
KokkosCore_config.h:

13
lib/kokkos/Makefile.targets
View File

@ -18,6 +18,8 @@ Kokkos_Serial_Task.o: $(KOKKOS_CPP_DEPENDS) $(KOKKOS_PATH)/core/src/impl/Kokkos_
$(CXX) $(KOKKOS_CPPFLAGS) $(KOKKOS_CXXFLAGS) $(CXXFLAGS) -c $(KOKKOS_PATH)/core/src/impl/Kokkos_Serial_Task.cpp
Kokkos_TaskQueue.o: $(KOKKOS_CPP_DEPENDS) $(KOKKOS_PATH)/core/src/impl/Kokkos_TaskQueue.cpp
$(CXX) $(KOKKOS_CPPFLAGS) $(KOKKOS_CXXFLAGS) $(CXXFLAGS) -c $(KOKKOS_PATH)/core/src/impl/Kokkos_TaskQueue.cpp
Kokkos_HostThreadTeam.o: $(KOKKOS_CPP_DEPENDS) $(KOKKOS_PATH)/core/src/impl/Kokkos_HostThreadTeam.cpp
$(CXX) $(KOKKOS_CPPFLAGS) $(KOKKOS_CXXFLAGS) $(CXXFLAGS) -c $(KOKKOS_PATH)/core/src/impl/Kokkos_HostThreadTeam.cpp
Kokkos_spinwait.o: $(KOKKOS_CPP_DEPENDS) $(KOKKOS_PATH)/core/src/impl/Kokkos_spinwait.cpp
$(CXX) $(KOKKOS_CPPFLAGS) $(KOKKOS_CXXFLAGS) $(CXXFLAGS) -c $(KOKKOS_PATH)/core/src/impl/Kokkos_spinwait.cpp
Kokkos_Profiling_Interface.o: $(KOKKOS_CPP_DEPENDS) $(KOKKOS_PATH)/core/src/impl/Kokkos_Profiling_Interface.cpp
@ -43,11 +45,11 @@ Kokkos_ThreadsExec.o: $(KOKKOS_CPP_DEPENDS) $(KOKKOS_PATH)/core/src/Threads/Kokk
$(CXX) $(KOKKOS_CPPFLAGS) $(KOKKOS_CXXFLAGS) $(CXXFLAGS) -c $(KOKKOS_PATH)/core/src/Threads/Kokkos_ThreadsExec.cpp
endif
ifeq ($(KOKKOS_INTERNAL_USE_QTHREAD), 1)
Kokkos_QthreadExec.o: $(KOKKOS_CPP_DEPENDS) $(KOKKOS_PATH)/core/src/Qthread/Kokkos_QthreadExec.cpp
$(CXX) $(KOKKOS_CPPFLAGS) $(KOKKOS_CXXFLAGS) $(CXXFLAGS) -c $(KOKKOS_PATH)/core/src/Qthread/Kokkos_QthreadExec.cpp
Kokkos_Qthread_TaskPolicy.o: $(KOKKOS_CPP_DEPENDS) $(KOKKOS_PATH)/core/src/Qthread/Kokkos_Qthread_TaskPolicy.cpp
$(CXX) $(KOKKOS_CPPFLAGS) $(KOKKOS_CXXFLAGS) $(CXXFLAGS) -c $(KOKKOS_PATH)/core/src/Qthread/Kokkos_Qthread_TaskPolicy.cpp
ifeq ($(KOKKOS_INTERNAL_USE_QTHREADS), 1)
Kokkos_QthreadsExec.o: $(KOKKOS_CPP_DEPENDS) $(KOKKOS_PATH)/core/src/Qthreads/Kokkos_QthreadsExec.cpp
$(CXX) $(KOKKOS_CPPFLAGS) $(KOKKOS_CXXFLAGS) $(CXXFLAGS) -c $(KOKKOS_PATH)/core/src/Qthreads/Kokkos_QthreadsExec.cpp
Kokkos_Qthreads_Task.o: $(KOKKOS_CPP_DEPENDS) $(KOKKOS_PATH)/core/src/Qthreads/Kokkos_Qthreads_Task.cpp
$(CXX) $(KOKKOS_CPPFLAGS) $(KOKKOS_CXXFLAGS) $(CXXFLAGS) -c $(KOKKOS_PATH)/core/src/Qthreads/Kokkos_Qthreads_Task.cpp
endif
ifeq ($(KOKKOS_INTERNAL_USE_OPENMP), 1)
@ -59,4 +61,3 @@ endif
Kokkos_HBWSpace.o: $(KOKKOS_CPP_DEPENDS) $(KOKKOS_PATH)/core/src/impl/Kokkos_HBWSpace.cpp
$(CXX) $(KOKKOS_CPPFLAGS) $(KOKKOS_CXXFLAGS) $(CXXFLAGS) -c $(KOKKOS_PATH)/core/src/impl/Kokkos_HBWSpace.cpp

14
lib/kokkos/README
View File

@ -45,31 +45,39 @@ Primary tested compilers on X86 are:
GCC 4.8.4
GCC 4.9.2
GCC 5.1.0
GCC 5.2.0
Intel 14.0.4
Intel 15.0.2
Intel 16.0.1
Intel 17.0.098
Intel 17.1.132
Clang 3.5.2
Clang 3.6.1
Clang 3.7.1
Clang 3.8.1
Clang 3.9.0
PGI 17.1
Primary tested compilers on Power 8 are:
GCC 5.4.0 (OpenMP,Serial)
IBM XL 13.1.3 (OpenMP, Serial) (There is a workaround in place to avoid a compiler bug)
Primary tested compilers on Intel KNL are:
GCC 6.2.0
Intel 16.2.181 (with gcc 4.7.2)
Intel 17.0.098 (with gcc 4.7.2)
Intel 17.1.132 (with gcc 4.9.3)
Intel 17.2.174 (with gcc 4.9.3)
Intel 18.0.061 (beta) (with gcc 4.9.3)
Secondary tested compilers are:
CUDA 7.0 (with gcc 4.7.2)
CUDA 7.5 (with gcc 4.7.2)
CUDA 7.0 (with gcc 4.8.4)
CUDA 7.5 (with gcc 4.8.4)
CUDA 8.0 (with gcc 5.3.0 on X86 and gcc 5.4.0 on Power8)
CUDA/Clang 8.0 using Clang/Trunk compiler
Other compilers working:
X86:
PGI 15.4
Cygwin 2.1.0 64bit with gcc 4.9.3
Known non-working combinations:

2
lib/kokkos/algorithms/cmake/Dependencies.cmake
View File

@ -1,5 +1,5 @@
TRIBITS_PACKAGE_DEFINE_DEPENDENCIES(
LIB_REQUIRED_PACKAGES KokkosCore
LIB_REQUIRED_PACKAGES KokkosCore KokkosContainers
LIB_OPTIONAL_TPLS Pthread CUDA HWLOC
TEST_OPTIONAL_TPLS CUSPARSE
)

8
lib/kokkos/algorithms/src/Kokkos_Random.hpp
View File

@ -547,7 +547,7 @@ namespace Kokkos {
KOKKOS_INLINE_FUNCTION
Random_XorShift64 (uint64_t state, int state_idx = 0)
: state_(state),state_idx_(state_idx){}
: state_(state==0?uint64_t(1318319):state),state_idx_(state_idx){}
KOKKOS_INLINE_FUNCTION
uint32_t urand() {
@ -719,6 +719,9 @@ namespace Kokkos {
}
void init(uint64_t seed, int num_states) {
if(seed==0)
seed = uint64_t(1318319);
num_states_ = num_states;
locks_ = lock_type("Kokkos::Random_XorShift64::locks",num_states_);
@ -968,8 +971,9 @@ namespace Kokkos {
inline
void init(uint64_t seed, int num_states) {
if(seed==0)
seed = uint64_t(1318319);
num_states_ = num_states;
locks_ = int_view_type("Kokkos::Random_XorShift1024::locks",num_states_);
state_ = state_data_type("Kokkos::Random_XorShift1024::state",num_states_);
p_ = int_view_type("Kokkos::Random_XorShift1024::p",num_states_);

325
lib/kokkos/algorithms/src/Kokkos_Sort.hpp
View File

@ -53,69 +53,122 @@ namespace Kokkos {
namespace Impl {
template<class ValuesViewType, int Rank=ValuesViewType::Rank>
template< class DstViewType , class SrcViewType
, int Rank = DstViewType::Rank >
struct CopyOp;
template<class ValuesViewType>
struct CopyOp<ValuesViewType,1> {
template<class DstType, class SrcType>
template< class DstViewType , class SrcViewType >
struct CopyOp<DstViewType,SrcViewType,1> {
KOKKOS_INLINE_FUNCTION
static void copy(DstType& dst, size_t i_dst,
SrcType& src, size_t i_src ) {
static void copy(DstViewType const& dst, size_t i_dst,
SrcViewType const& src, size_t i_src ) {
dst(i_dst) = src(i_src);
}
};
template<class ValuesViewType>
struct CopyOp<ValuesViewType,2> {
template<class DstType, class SrcType>
template< class DstViewType , class SrcViewType >
struct CopyOp<DstViewType,SrcViewType,2> {
KOKKOS_INLINE_FUNCTION
static void copy(DstType& dst, size_t i_dst,
SrcType& src, size_t i_src ) {
for(int j = 0;j< (int) dst.dimension_1(); j++)
static void copy(DstViewType const& dst, size_t i_dst,
SrcViewType const& src, size_t i_src ) {
for(int j = 0;j< (int) dst.extent(1); j++)
dst(i_dst,j) = src(i_src,j);
}
};
template<class ValuesViewType>
struct CopyOp<ValuesViewType,3> {
template<class DstType, class SrcType>
template< class DstViewType , class SrcViewType >
struct CopyOp<DstViewType,SrcViewType,3> {
KOKKOS_INLINE_FUNCTION
static void copy(DstType& dst, size_t i_dst,
SrcType& src, size_t i_src ) {
for(int j = 0; j<dst.dimension_1(); j++)
for(int k = 0; k<dst.dimension_2(); k++)
static void copy(DstViewType const& dst, size_t i_dst,
SrcViewType const& src, size_t i_src ) {
for(int j = 0; j<dst.extent(1); j++)
for(int k = 0; k<dst.extent(2); k++)
dst(i_dst,j,k) = src(i_src,j,k);
}
};
}
template<class KeyViewType, class BinSortOp, class ExecutionSpace = typename KeyViewType::execution_space,
class SizeType = typename KeyViewType::memory_space::size_type>
//----------------------------------------------------------------------------
template< class KeyViewType
, class BinSortOp
, class Space = typename KeyViewType::device_type
, class SizeType = typename KeyViewType::memory_space::size_type
>
class BinSort {
public:
template<class ValuesViewType, class PermuteViewType, class CopyOp>
struct bin_sort_sort_functor {
typedef ExecutionSpace execution_space;
typedef typename ValuesViewType::non_const_type values_view_type;
typedef typename ValuesViewType::const_type const_values_view_type;
Kokkos::View<typename values_view_type::const_data_type,typename values_view_type::array_layout,
typename values_view_type::memory_space,Kokkos::MemoryTraits<Kokkos::RandomAccess> > values;
values_view_type sorted_values;
typename PermuteViewType::const_type sort_order;
bin_sort_sort_functor(const_values_view_type values_, values_view_type sorted_values_, PermuteViewType sort_order_):
values(values_),sorted_values(sorted_values_),sort_order(sort_order_) {}
template< class DstViewType , class SrcViewType >
struct copy_functor {
typedef typename SrcViewType::const_type src_view_type ;
typedef Impl::CopyOp< DstViewType , src_view_type > copy_op ;
DstViewType dst_values ;
src_view_type src_values ;
int dst_offset ;
copy_functor( DstViewType const & dst_values_
, int const & dst_offset_
, SrcViewType const & src_values_
)
: dst_values( dst_values_ )
, src_values( src_values_ )
, dst_offset( dst_offset_ )
{}
KOKKOS_INLINE_FUNCTION
void operator() (const int& i) const {
//printf("Sort: %i %i\n",i,sort_order(i));
CopyOp::copy(sorted_values,i,values,sort_order(i));
void operator() (const int& i) const {
// printf("copy: dst(%i) src(%i)\n",i+dst_offset,i);
copy_op::copy(dst_values,i+dst_offset,src_values,i);
}
};
typedef ExecutionSpace execution_space;
template< class DstViewType
, class PermuteViewType
, class SrcViewType
>
struct copy_permute_functor {
// If a Kokkos::View then can generate constant random access
// otherwise can only use the constant type.
typedef typename std::conditional
< Kokkos::is_view< SrcViewType >::value
, Kokkos::View< typename SrcViewType::const_data_type
, typename SrcViewType::array_layout
, typename SrcViewType::device_type
, Kokkos::MemoryTraits<Kokkos::RandomAccess>
>
, typename SrcViewType::const_type
>::type src_view_type ;
typedef typename PermuteViewType::const_type perm_view_type ;
typedef Impl::CopyOp< DstViewType , src_view_type > copy_op ;
DstViewType dst_values ;
perm_view_type sort_order ;
src_view_type src_values ;
copy_permute_functor( DstViewType const & dst_values_
, PermuteViewType const & sort_order_
, SrcViewType const & src_values_
)
: dst_values( dst_values_ )
, sort_order( sort_order_ )
, src_values( src_values_ )
{}
KOKKOS_INLINE_FUNCTION
void operator() (const int& i) const {
// printf("copy_permute: dst(%i) src(%i)\n",i,sort_order(i));
copy_op::copy(dst_values,i,src_values,sort_order(i));
}
};
typedef typename Space::execution_space execution_space;
typedef BinSortOp bin_op_type;
struct bin_count_tag {};
@ -124,84 +177,137 @@ public:
struct bin_sort_bins_tag {};
public:
typedef SizeType size_type;
typedef size_type value_type;
typedef Kokkos::View<size_type*, execution_space> offset_type;
typedef Kokkos::View<const int*, execution_space> bin_count_type;
typedef Kokkos::View<size_type*, Space> offset_type;
typedef Kokkos::View<const int*, Space> bin_count_type;
typedef typename KeyViewType::const_type const_key_view_type ;
typedef Kokkos::View<typename KeyViewType::const_data_type,
typename KeyViewType::array_layout,
typename KeyViewType::memory_space> const_key_view_type;
typedef Kokkos::View<typename KeyViewType::const_data_type,
typename KeyViewType::array_layout,
typename KeyViewType::memory_space,
Kokkos::MemoryTraits<Kokkos::RandomAccess> > const_rnd_key_view_type;
// If a Kokkos::View then can generate constant random access
// otherwise can only use the constant type.
typedef typename std::conditional
< Kokkos::is_view< KeyViewType >::value
, Kokkos::View< typename KeyViewType::const_data_type,
typename KeyViewType::array_layout,
typename KeyViewType::device_type,
Kokkos::MemoryTraits<Kokkos::RandomAccess> >
, const_key_view_type
>::type const_rnd_key_view_type;
typedef typename KeyViewType::non_const_value_type non_const_key_scalar;
typedef typename KeyViewType::const_value_type const_key_scalar;
typedef Kokkos::View<int*, Space, Kokkos::MemoryTraits<Kokkos::Atomic> > bin_count_atomic_type ;
private:
const_key_view_type keys;
const_rnd_key_view_type keys_rnd;
public:
BinSortOp bin_op;
offset_type bin_offsets;
BinSortOp bin_op ;
offset_type bin_offsets ;
bin_count_atomic_type bin_count_atomic ;
bin_count_type bin_count_const ;
offset_type sort_order ;
Kokkos::View<int*, ExecutionSpace, Kokkos::MemoryTraits<Kokkos::Atomic> > bin_count_atomic;
bin_count_type bin_count_const;
offset_type sort_order;
bool sort_within_bins;
int range_begin ;
int range_end ;
bool sort_within_bins ;
public:
// Constructor: takes the keys, the binning_operator and optionally whether to sort within bins (default false)
BinSort(const_key_view_type keys_, BinSortOp bin_op_,
bool sort_within_bins_ = false)
:keys(keys_),keys_rnd(keys_), bin_op(bin_op_) {
BinSort() {}
bin_count_atomic = Kokkos::View<int*, ExecutionSpace >("Kokkos::SortImpl::BinSortFunctor::bin_count",bin_op.max_bins());
//----------------------------------------
// Constructor: takes the keys, the binning_operator and optionally whether to sort within bins (default false)
BinSort( const_key_view_type keys_
, int range_begin_
, int range_end_
, BinSortOp bin_op_
, bool sort_within_bins_ = false
)
: keys(keys_)
, keys_rnd(keys_)
, bin_op(bin_op_)
, bin_offsets()
, bin_count_atomic()
, bin_count_const()
, sort_order()
, range_begin( range_begin_ )
, range_end( range_end_ )
, sort_within_bins( sort_within_bins_ )
{
bin_count_atomic = Kokkos::View<int*, Space >("Kokkos::SortImpl::BinSortFunctor::bin_count",bin_op.max_bins());
bin_count_const = bin_count_atomic;
bin_offsets = offset_type("Kokkos::SortImpl::BinSortFunctor::bin_offsets",bin_op.max_bins());
sort_order = offset_type("PermutationVector",keys.dimension_0());
sort_within_bins = sort_within_bins_;
sort_order = offset_type("PermutationVector",range_end-range_begin);
}
BinSort( const_key_view_type keys_
, BinSortOp bin_op_
, bool sort_within_bins_ = false
)
: BinSort( keys_ , 0 , keys_.extent(0), bin_op_ , sort_within_bins_ ) {}
//----------------------------------------
// Create the permutation vector, the bin_offset array and the bin_count array. Can be called again if keys changed
void create_permute_vector() {
Kokkos::parallel_for (Kokkos::RangePolicy<ExecutionSpace,bin_count_tag> (0,keys.dimension_0()),*this);
Kokkos::parallel_scan(Kokkos::RangePolicy<ExecutionSpace,bin_offset_tag> (0,bin_op.max_bins()) ,*this);
const size_t len = range_end - range_begin ;
Kokkos::parallel_for (Kokkos::RangePolicy<execution_space,bin_count_tag> (0,len),*this);
Kokkos::parallel_scan(Kokkos::RangePolicy<execution_space,bin_offset_tag> (0,bin_op.max_bins()) ,*this);
Kokkos::deep_copy(bin_count_atomic,0);
Kokkos::parallel_for (Kokkos::RangePolicy<ExecutionSpace,bin_binning_tag> (0,keys.dimension_0()),*this);
Kokkos::parallel_for (Kokkos::RangePolicy<execution_space,bin_binning_tag> (0,len),*this);
if(sort_within_bins)
Kokkos::parallel_for (Kokkos::RangePolicy<ExecutionSpace,bin_sort_bins_tag>(0,bin_op.max_bins()) ,*this);
Kokkos::parallel_for (Kokkos::RangePolicy<execution_space,bin_sort_bins_tag>(0,bin_op.max_bins()) ,*this);
}
// Sort a view with respect ot the first dimension using the permutation array
template<class ValuesViewType>
void sort(ValuesViewType values) {
ValuesViewType sorted_values = ValuesViewType("Copy",
values.dimension_0(),
values.dimension_1(),
values.dimension_2(),
values.dimension_3(),
values.dimension_4(),
values.dimension_5(),
values.dimension_6(),
values.dimension_7());
void sort( ValuesViewType const & values)
{
typedef
Kokkos::View< typename ValuesViewType::data_type,
typename ValuesViewType::array_layout,
typename ValuesViewType::device_type >
scratch_view_type ;
parallel_for(values.dimension_0(),
bin_sort_sort_functor<ValuesViewType, offset_type,
Impl::CopyOp<ValuesViewType> >(values,sorted_values,sort_order));
const size_t len = range_end - range_begin ;
deep_copy(values,sorted_values);
scratch_view_type
sorted_values("Scratch",
len,
values.extent(1),
values.extent(2),
values.extent(3),
values.extent(4),
values.extent(5),
values.extent(6),
values.extent(7));
{
copy_permute_functor< scratch_view_type /* DstViewType */
, offset_type /* PermuteViewType */
, ValuesViewType /* SrcViewType */
>
functor( sorted_values , sort_order , values );
parallel_for( Kokkos::RangePolicy<execution_space>(0,len),functor);
}
{
copy_functor< ValuesViewType , scratch_view_type >
functor( values , range_begin , sorted_values );
parallel_for( Kokkos::RangePolicy<execution_space>(0,len),functor);
}
}
// Get the permutation vector
@ -217,9 +323,11 @@ public:
bin_count_type get_bin_count() const {return bin_count_const;}
public:
KOKKOS_INLINE_FUNCTION
void operator() (const bin_count_tag& tag, const int& i) const {
bin_count_atomic(bin_op.bin(keys,i))++;
const int j = range_begin + i ;
bin_count_atomic(bin_op.bin(keys,j))++;
}
KOKKOS_INLINE_FUNCTION
@ -232,10 +340,11 @@ public:
KOKKOS_INLINE_FUNCTION
void operator() (const bin_binning_tag& tag, const int& i) const {
const int bin = bin_op.bin(keys,i);
const int j = range_begin + i ;
const int bin = bin_op.bin(keys,j);
const int count = bin_count_atomic(bin)++;
sort_order(bin_offsets(bin) + count) = i;
sort_order(bin_offsets(bin) + count) = j ;
}
KOKKOS_INLINE_FUNCTION
@ -262,13 +371,19 @@ public:
}
};
//----------------------------------------------------------------------------
template<class KeyViewType>
struct BinOp1D {
const int max_bins_;
const double mul_;
int max_bins_;
double mul_;
typename KeyViewType::const_value_type range_;
typename KeyViewType::const_value_type min_;
BinOp1D():max_bins_(0),mul_(0.0),
range_(typename KeyViewType::const_value_type()),
min_(typename KeyViewType::const_value_type()) {}
//Construct BinOp with number of bins, minimum value and maxuimum value
BinOp1D(int max_bins__, typename KeyViewType::const_value_type min,
typename KeyViewType::const_value_type max )
@ -302,12 +417,14 @@ struct BinOp3D {
typename KeyViewType::non_const_value_type range_[3];
typename KeyViewType::non_const_value_type min_[3];
BinOp3D() {}
BinOp3D(int max_bins__[], typename KeyViewType::const_value_type min[],
typename KeyViewType::const_value_type max[] )
{
max_bins_[0] = max_bins__[0]+1;
max_bins_[1] = max_bins__[1]+1;
max_bins_[2] = max_bins__[2]+1;
max_bins_[0] = max_bins__[0];
max_bins_[1] = max_bins__[1];
max_bins_[2] = max_bins__[2];
mul_[0] = 1.0*max_bins__[0]/(max[0]-min[0]);
mul_[1] = 1.0*max_bins__[1]/(max[1]-min[1]);
mul_[2] = 1.0*max_bins__[2]/(max[2]-min[2]);
@ -364,7 +481,7 @@ bool try_std_sort(ViewType view) {
possible = possible && (ViewType::Rank == 1);
possible = possible && (stride[0] == 1);
if(possible) {
std::sort(view.ptr_on_device(),view.ptr_on_device()+view.dimension_0());
std::sort(view.data(),view.data()+view.extent(0));
}
return possible;
}
@ -386,7 +503,8 @@ struct min_max_functor {
}
template<class ViewType>
void sort(ViewType view, bool always_use_kokkos_sort = false) {
void sort( ViewType const & view , bool const always_use_kokkos_sort = false)
{
if(!always_use_kokkos_sort) {
if(Impl::try_std_sort(view)) return;
}
@ -394,14 +512,37 @@ void sort(ViewType view, bool always_use_kokkos_sort = false) {
Kokkos::Experimental::MinMaxScalar<typename ViewType::non_const_value_type> result;
Kokkos::Experimental::MinMax<typename ViewType::non_const_value_type> reducer(result);
parallel_reduce(Kokkos::RangePolicy<typename ViewType::execution_space>(0,view.dimension_0()),
parallel_reduce(Kokkos::RangePolicy<typename ViewType::execution_space>(0,view.extent(0)),
Impl::min_max_functor<ViewType>(view),reducer);
if(result.min_val == result.max_val) return;
BinSort<ViewType, CompType> bin_sort(view,CompType(view.dimension_0()/2,result.min_val,result.max_val),true);
BinSort<ViewType, CompType> bin_sort(view,CompType(view.extent(0)/2,result.min_val,result.max_val),true);
bin_sort.create_permute_vector();
bin_sort.sort(view);
}
template<class ViewType>
void sort( ViewType view
, size_t const begin
, size_t const end
)
{
typedef Kokkos::RangePolicy<typename ViewType::execution_space> range_policy ;
typedef BinOp1D<ViewType> CompType;
Kokkos::Experimental::MinMaxScalar<typename ViewType::non_const_value_type> result;
Kokkos::Experimental::MinMax<typename ViewType::non_const_value_type> reducer(result);
parallel_reduce( range_policy( begin , end )
, Impl::min_max_functor<ViewType>(view),reducer );
if(result.min_val == result.max_val) return;
BinSort<ViewType, CompType>
bin_sort(view,begin,end,CompType((end-begin)/2,result.min_val,result.max_val),true);
bin_sort.create_permute_vector();
bin_sort.sort(view);
}
}
#endif

67
lib/kokkos/algorithms/unit_tests/TestSort.hpp
View File

@ -44,6 +44,7 @@
#include <gtest/gtest.h>
#include<Kokkos_Core.hpp>
#include<Kokkos_DynamicView.hpp>
#include<Kokkos_Random.hpp>
#include<Kokkos_Sort.hpp>
@ -192,17 +193,81 @@ void test_3D_sort(unsigned int n) {
double epsilon = 1e-10;
unsigned int equal_sum = (ratio > (1.0-epsilon)) && (ratio < (1.0+epsilon)) ? 1 : 0;
printf("3D Sort Sum: %f %f Fails: %u\n",sum_before,sum_after,sort_fails);
if ( sort_fails )
printf("3D Sort Sum: %f %f Fails: %u\n",sum_before,sum_after,sort_fails);
ASSERT_EQ(sort_fails,0);
ASSERT_EQ(equal_sum,1);
}
//----------------------------------------------------------------------------
template<class ExecutionSpace, typename KeyType>
void test_dynamic_view_sort(unsigned int n )
{
typedef typename ExecutionSpace::memory_space memory_space ;
typedef Kokkos::Experimental::DynamicView<KeyType*,ExecutionSpace> KeyDynamicViewType;
typedef Kokkos::View<KeyType*,ExecutionSpace> KeyViewType;
const size_t upper_bound = 2 * n ;
typename KeyDynamicViewType::memory_pool
pool( memory_space() , 2 * n * sizeof(KeyType) );
KeyDynamicViewType keys("Keys",pool,upper_bound);
keys.resize_serial(n);
KeyViewType keys_view("KeysTmp", n );
// Test sorting array with all numbers equal
Kokkos::deep_copy(keys_view,KeyType(1));
Kokkos::Experimental::deep_copy(keys,keys_view);
Kokkos::sort(keys, 0 /* begin */ , n /* end */ );
Kokkos::Random_XorShift64_Pool<ExecutionSpace> g(1931);
Kokkos::fill_random(keys_view,g,Kokkos::Random_XorShift64_Pool<ExecutionSpace>::generator_type::MAX_URAND);
Kokkos::Experimental::deep_copy(keys,keys_view);
double sum_before = 0.0;
double sum_after = 0.0;
unsigned int sort_fails = 0;
Kokkos::parallel_reduce(n,sum<ExecutionSpace, KeyType>(keys_view),sum_before);
Kokkos::sort(keys, 0 /* begin */ , n /* end */ );
Kokkos::Experimental::deep_copy( keys_view , keys );
Kokkos::parallel_reduce(n,sum<ExecutionSpace, KeyType>(keys_view),sum_after);
Kokkos::parallel_reduce(n-1,is_sorted_struct<ExecutionSpace, KeyType>(keys_view),sort_fails);
double ratio = sum_before/sum_after;
double epsilon = 1e-10;
unsigned int equal_sum = (ratio > (1.0-epsilon)) && (ratio < (1.0+epsilon)) ? 1 : 0;
if ( sort_fails != 0 || equal_sum != 1 ) {
std::cout << " N = " << n
<< " ; sum_before = " << sum_before
<< " ; sum_after = " << sum_after
<< " ; ratio = " << ratio
<< std::endl ;
}
ASSERT_EQ(sort_fails,0);
ASSERT_EQ(equal_sum,1);
}
//----------------------------------------------------------------------------
template<class ExecutionSpace, typename KeyType>
void test_sort(unsigned int N)
{
test_1D_sort<ExecutionSpace,KeyType>(N*N*N, true);
test_1D_sort<ExecutionSpace,KeyType>(N*N*N, false);
test_3D_sort<ExecutionSpace,KeyType>(N);
test_dynamic_view_sort<ExecutionSpace,KeyType>(N*N);
}
}

5
lib/kokkos/bin/nvcc_wrapper
View File

@ -140,6 +140,9 @@ do
#strip of pedantic because it produces endless warnings about #LINE added by the preprocessor
-pedantic|-Wpedantic|-ansi)
;;
#strip of -Woverloaded-virtual to avoid "cc1: warning: command line option ‘-Woverloaded-virtual’ is valid for C++/ObjC++ but not for C"
-Woverloaded-virtual)
;;
#strip -Xcompiler because we add it
-Xcompiler)
if [ $first_xcompiler_arg -eq 1 ]; then
@ -190,7 +193,7 @@ do
object_files_xlinker="$object_files_xlinker -Xlinker $1"
;;
#Handle object files which always need to use "-Xlinker": -x cu applies to all input files, so give them to linker, except if only linking
*.dylib)
@*|*.dylib)
object_files="$object_files -Xlinker $1"
object_files_xlinker="$object_files_xlinker -Xlinker $1"
;;

3
lib/kokkos/cmake/deps/QTHREAD.cmake → lib/kokkos/cmake/deps/QTHREADS.cmake
View File

@ -63,8 +63,7 @@
# Source: https://code.google.com/p/qthreads
#
TRIBITS_TPL_FIND_INCLUDE_DIRS_AND_LIBRARIES( QTHREAD
TRIBITS_TPL_FIND_INCLUDE_DIRS_AND_LIBRARIES( QTHREADS
REQUIRED_HEADERS qthread.h
REQUIRED_LIBS_NAMES "qthread"
)

3
lib/kokkos/cmake/tpls/FindTPLQTHREAD.cmake → lib/kokkos/cmake/tpls/FindTPLQTHREADS.cmake
View File

@ -63,8 +63,7 @@
# Source: https://code.google.com/p/qthreads
#
TRIBITS_TPL_FIND_INCLUDE_DIRS_AND_LIBRARIES( QTHREAD
TRIBITS_TPL_FIND_INCLUDE_DIRS_AND_LIBRARIES( QTHREADS
REQUIRED_HEADERS qthread.h
REQUIRED_LIBS_NAMES "qthread"
)

15
lib/kokkos/config/kokkos_dev/config-core-all.sh
View File

@ -6,7 +6,7 @@
#-----------------------------------------------------------------------------
# Building on 'kokkos-dev.sandia.gov' with enabled capabilities:
#
# Cuda, OpenMP, Threads, Qthread, hwloc
# Cuda, OpenMP, Threads, Qthreads, hwloc
#
# module loaded on 'kokkos-dev.sandia.gov' for this build
#
@ -82,13 +82,13 @@ CMAKE_CONFIGURE="${CMAKE_CONFIGURE} -D Trilinos_ENABLE_OpenMP:BOOL=ON"
CMAKE_CONFIGURE="${CMAKE_CONFIGURE} -D Kokkos_ENABLE_OpenMP:BOOL=ON"
#-----------------------------------------------------------------------------
# Qthread
# Qthreads
QTHREAD_BASE_DIR="/home/projects/qthreads/2014-07-08/host/gnu/4.7.3"
QTHREADS_BASE_DIR="/home/projects/qthreads/2014-07-08/host/gnu/4.7.3"
CMAKE_CONFIGURE="${CMAKE_CONFIGURE} -D TPL_ENABLE_QTHREAD:BOOL=ON"
CMAKE_CONFIGURE="${CMAKE_CONFIGURE} -D QTHREAD_INCLUDE_DIRS:FILEPATH=${QTHREAD_BASE_DIR}/include"
CMAKE_CONFIGURE="${CMAKE_CONFIGURE} -D QTHREAD_LIBRARY_DIRS:FILEPATH=${QTHREAD_BASE_DIR}/lib"
CMAKE_CONFIGURE="${CMAKE_CONFIGURE} -D TPL_ENABLE_QTHREADS:BOOL=ON"
CMAKE_CONFIGURE="${CMAKE_CONFIGURE} -D QTHREADS_INCLUDE_DIRS:FILEPATH=${QTHREADS_BASE_DIR}/include"
CMAKE_CONFIGURE="${CMAKE_CONFIGURE} -D QTHREADS_LIBRARY_DIRS:FILEPATH=${QTHREADS_BASE_DIR}/lib"
#-----------------------------------------------------------------------------
# C++11
@ -108,6 +108,3 @@ rm -rf CMake* Trilinos* packages Dart* Testing cmake_install.cmake MakeFile*
echo cmake ${CMAKE_CONFIGURE} ${TRILINOS_SOURCE_DIR}
cmake ${CMAKE_CONFIGURE} ${TRILINOS_SOURCE_DIR}
#-----------------------------------------------------------------------------

3
lib/kokkos/config/master_history.txt
View File

@ -4,4 +4,5 @@ tag: 2.01.10 date: 09:27:2016 master: e4119325 develop: e6cda11e
tag: 2.02.00 date: 10:30:2016 master: 6c90a581 develop: ca3dd56e
tag: 2.02.01 date: 11:01:2016 master: 9c698c86 develop: b0072304
tag: 2.02.07 date: 12:16:2016 master: 4b4cc4ba develop: 382c0966
tag: 2.02.15 date: 02:10:2017 master: 8c64cd93 develop: 28dea8b6
tag: 2.02.15 date: 02:10:2017 master: 8c64cd93 develop: 28dea8b6
tag: 2.03.00 date: 04:25:2017 master: 120d9ce7 develop: 015ba641

814
lib/kokkos/config/test_all_sandia
View File

File diff suppressed because it is too large Load Diff

143
lib/kokkos/containers/src/Kokkos_DynamicView.hpp
View File

@ -60,7 +60,7 @@ class DynamicView : public Kokkos::ViewTraits< DataType , P ... >
{
public:
typedef ViewTraits< DataType , P ... > traits ;
typedef Kokkos::ViewTraits< DataType , P ... > traits ;
private:
@ -123,30 +123,41 @@ public:
enum { Rank = 1 };
KOKKOS_INLINE_FUNCTION constexpr size_t size() const
KOKKOS_INLINE_FUNCTION
size_t size() const noexcept
{
return
Kokkos::Impl::MemorySpaceAccess
< Kokkos::Impl::ActiveExecutionMemorySpace
, typename traits::memory_space
>::accessible
? // Runtime size is at the end of the chunk pointer array
(*reinterpret_cast<const uintptr_t*>( m_chunks + m_chunk_max ))
<< m_chunk_shift
: 0 ;
uintptr_t n = 0 ;
if ( Kokkos::Impl::MemorySpaceAccess
< Kokkos::Impl::ActiveExecutionMemorySpace
, typename traits::memory_space
>::accessible ) {
n = *reinterpret_cast<const uintptr_t*>( m_chunks + m_chunk_max );
}
#if defined( KOKKOS_ACTIVE_EXECUTION_MEMORY_SPACE_HOST )
else {
Kokkos::Impl::DeepCopy< Kokkos::HostSpace
, typename traits::memory_space
, Kokkos::HostSpace::execution_space >
( & n
, reinterpret_cast<const uintptr_t*>( m_chunks + m_chunk_max )
, sizeof(uintptr_t) );
}
#endif
return n << m_chunk_shift ;
}
template< typename iType >
KOKKOS_INLINE_FUNCTION constexpr
KOKKOS_INLINE_FUNCTION
size_t extent( const iType & r ) const
{ return r == 0 ? size() : 1 ; }
template< typename iType >
KOKKOS_INLINE_FUNCTION constexpr
KOKKOS_INLINE_FUNCTION
size_t extent_int( const iType & r ) const
{ return r == 0 ? size() : 1 ; }
KOKKOS_INLINE_FUNCTION constexpr size_t dimension_0() const { return size(); }
KOKKOS_INLINE_FUNCTION size_t dimension_0() const { return size(); }
KOKKOS_INLINE_FUNCTION constexpr size_t dimension_1() const { return 1 ; }
KOKKOS_INLINE_FUNCTION constexpr size_t dimension_2() const { return 1 ; }
KOKKOS_INLINE_FUNCTION constexpr size_t dimension_3() const { return 1 ; }
@ -270,10 +281,18 @@ public:
}
/** \brief Resizing in serial can grow or shrink the array size, */
template< typename IntType >
inline
void resize_serial( size_t n )
typename std::enable_if
< std::is_integral<IntType>::value &&
Kokkos::Impl::MemorySpaceAccess< Kokkos::HostSpace
, typename traits::memory_space
>::accessible
>::type
resize_serial( IntType const & n )
{
DynamicView::template verify_space< Kokkos::Impl::ActiveExecutionMemorySpace >::check();
typedef typename traits::value_type value_type ;
typedef value_type * pointer_type ;
const uintptr_t NC = ( n + m_chunk_mask ) >> m_chunk_shift ;
@ -286,8 +305,8 @@ public:
if ( *pc < NC ) {
while ( *pc < NC ) {
m_chunks[*pc] =
m_pool.allocate( sizeof(traits::value_type) << m_chunk_shift );
m_chunks[*pc] = reinterpret_cast<pointer_type>
( m_pool.allocate( sizeof(value_type) << m_chunk_shift ) );
++*pc ;
}
}
@ -295,12 +314,90 @@ public:
while ( NC + 1 <= *pc ) {
--*pc ;
m_pool.deallocate( m_chunks[*pc]
, sizeof(traits::value_type) << m_chunk_shift );
, sizeof(value_type) << m_chunk_shift );
m_chunks[*pc] = 0 ;
}
}
}
//----------------------------------------
struct ResizeSerial {
memory_pool m_pool ;
typename traits::value_type ** m_chunks ;
uintptr_t * m_pc ;
uintptr_t m_nc ;
unsigned m_chunk_shift ;
KOKKOS_INLINE_FUNCTION
void operator()( int ) const
{
typedef typename traits::value_type value_type ;
typedef value_type * pointer_type ;
if ( *m_pc < m_nc ) {
while ( *m_pc < m_nc ) {
m_chunks[*m_pc] = reinterpret_cast<pointer_type>
( m_pool.allocate( sizeof(value_type) << m_chunk_shift ) );
++*m_pc ;
}
}
else {
while ( m_nc + 1 <= *m_pc ) {
--*m_pc ;
m_pool.deallocate( m_chunks[*m_pc]
, sizeof(value_type) << m_chunk_shift );
m_chunks[*m_pc] = 0 ;
}
}
}
ResizeSerial( memory_pool const & arg_pool
, typename traits::value_type ** arg_chunks
, uintptr_t * arg_pc
, uintptr_t arg_nc
, unsigned arg_chunk_shift
)
: m_pool( arg_pool )
, m_chunks( arg_chunks )
, m_pc( arg_pc )
, m_nc( arg_nc )
, m_chunk_shift( arg_chunk_shift )
{}
};
template< typename IntType >
inline
typename std::enable_if
< std::is_integral<IntType>::value &&
! Kokkos::Impl::MemorySpaceAccess< Kokkos::HostSpace
, typename traits::memory_space
>::accessible
>::type
resize_serial( IntType const & n )
{
const uintptr_t NC = ( n + m_chunk_mask ) >> m_chunk_shift ;
if ( m_chunk_max < NC ) {
Kokkos::abort("DynamicView::resize_serial exceeded maximum size");
}
// Must dispatch kernel
typedef Kokkos::RangePolicy< typename traits::execution_space > Range ;
uintptr_t * const pc =
reinterpret_cast<uintptr_t*>( m_chunks + m_chunk_max );
Kokkos::Impl::ParallelFor<ResizeSerial,Range>
closure( ResizeSerial( m_pool, m_chunks, pc, NC, m_chunk_shift )
, Range(0,1) );
closure.execute();
traits::execution_space::fence();
}
//----------------------------------------------------------------------
~DynamicView() = default ;
@ -311,15 +408,17 @@ public:
DynamicView & operator = ( const DynamicView & ) = default ;
template< class RT , class ... RP >
KOKKOS_INLINE_FUNCTION
DynamicView( const DynamicView<RT,RP...> & rhs )
: m_pool( rhs.m_pool )
, m_track( rhs.m_track )
, m_chunks( rhs.m_chunks )
, m_chunks( (typename traits::value_type **) rhs.m_chunks )
, m_chunk_shift( rhs.m_chunk_shift )
, m_chunk_mask( rhs.m_chunk_mask )
, m_chunk_max( rhs.m_chunk_max )
{
typedef typename DynamicView<RT,RP...>::traits SrcTraits ;
typedef Kokkos::Impl::ViewMapping< traits , SrcTraits , void > Mapping ;
static_assert( Mapping::is_assignable , "Incompatible DynamicView copy construction" );
}
//----------------------------------------------------------------------
@ -400,8 +499,6 @@ public:
, m_chunk_mask( ( 1 << m_chunk_shift ) - 1 )
, m_chunk_max( ( arg_size_max + m_chunk_mask ) >> m_chunk_shift )
{
DynamicView::template verify_space< Kokkos::Impl::ActiveExecutionMemorySpace >::check();
// A functor to deallocate all of the chunks upon final destruction
typedef typename traits::memory_space memory_space ;

33
lib/kokkos/containers/src/Kokkos_UnorderedMap.hpp
View File

@ -230,16 +230,17 @@ public:
typedef typename Impl::remove_const<declared_value_type>::type value_type;
typedef typename Impl::add_const<value_type>::type const_value_type;
typedef Device execution_space;
typedef Device device_type;
typedef typename Device::execution_space execution_space;
typedef Hasher hasher_type;
typedef EqualTo equal_to_type;
typedef uint32_t size_type;
//map_types
typedef UnorderedMap<declared_key_type,declared_value_type,execution_space,hasher_type,equal_to_type> declared_map_type;
typedef UnorderedMap<key_type,value_type,execution_space,hasher_type,equal_to_type> insertable_map_type;
typedef UnorderedMap<const_key_type,value_type,execution_space,hasher_type,equal_to_type> modifiable_map_type;
typedef UnorderedMap<const_key_type,const_value_type,execution_space,hasher_type,equal_to_type> const_map_type;
typedef UnorderedMap<declared_key_type,declared_value_type,device_type,hasher_type,equal_to_type> declared_map_type;
typedef UnorderedMap<key_type,value_type,device_type,hasher_type,equal_to_type> insertable_map_type;
typedef UnorderedMap<const_key_type,value_type,device_type,hasher_type,equal_to_type> modifiable_map_type;
typedef UnorderedMap<const_key_type,const_value_type,device_type,hasher_type,equal_to_type> const_map_type;
static const bool is_set = std::is_same<void,value_type>::value;
static const bool has_const_key = std::is_same<const_key_type,declared_key_type>::value;
@ -264,18 +265,18 @@ private:
typedef typename Impl::if_c< is_set, int, declared_value_type>::type impl_value_type;
typedef typename Impl::if_c< is_insertable_map
, View< key_type *, execution_space>
, View< const key_type *, execution_space, MemoryTraits<RandomAccess> >
, View< key_type *, device_type>
, View< const key_type *, device_type, MemoryTraits<RandomAccess> >
>::type key_type_view;
typedef typename Impl::if_c< is_insertable_map || is_modifiable_map
, View< impl_value_type *, execution_space>
, View< const impl_value_type *, execution_space, MemoryTraits<RandomAccess> >
, View< impl_value_type *, device_type>
, View< const impl_value_type *, device_type, MemoryTraits<RandomAccess> >
>::type value_type_view;
typedef typename Impl::if_c< is_insertable_map
, View< size_type *, execution_space>
, View< const size_type *, execution_space, MemoryTraits<RandomAccess> >
, View< size_type *, device_type>
, View< const size_type *, device_type, MemoryTraits<RandomAccess> >
>::type size_type_view;
typedef typename Impl::if_c< is_insertable_map
@ -285,7 +286,7 @@ private:
enum { modified_idx = 0, erasable_idx = 1, failed_insert_idx = 2 };
enum { num_scalars = 3 };
typedef View< int[num_scalars], LayoutLeft, execution_space> scalars_view;
typedef View< int[num_scalars], LayoutLeft, device_type> scalars_view;
public:
//! \name Public member functions
@ -757,7 +758,7 @@ public:
Kokkos::deep_copy(tmp.m_available_indexes, src.m_available_indexes);
typedef Kokkos::Impl::DeepCopy< typename execution_space::memory_space, typename SDevice::memory_space > raw_deep_copy;
typedef Kokkos::Impl::DeepCopy< typename device_type::memory_space, typename SDevice::memory_space > raw_deep_copy;
raw_deep_copy(tmp.m_hash_lists.ptr_on_device(), src.m_hash_lists.ptr_on_device(), sizeof(size_type)*src.m_hash_lists.dimension_0());
raw_deep_copy(tmp.m_next_index.ptr_on_device(), src.m_next_index.ptr_on_device(), sizeof(size_type)*src.m_next_index.dimension_0());
@ -781,21 +782,21 @@ private: // private member functions
void set_flag(int flag) const
{
typedef Kokkos::Impl::DeepCopy< typename execution_space::memory_space, Kokkos::HostSpace > raw_deep_copy;
typedef Kokkos::Impl::DeepCopy< typename device_type::memory_space, Kokkos::HostSpace > raw_deep_copy;
const int true_ = true;
raw_deep_copy(m_scalars.ptr_on_device() + flag, &true_, sizeof(int));
}
void reset_flag(int flag) const
{
typedef Kokkos::Impl::DeepCopy< typename execution_space::memory_space, Kokkos::HostSpace > raw_deep_copy;
typedef Kokkos::Impl::DeepCopy< typename device_type::memory_space, Kokkos::HostSpace > raw_deep_copy;
const int false_ = false;
raw_deep_copy(m_scalars.ptr_on_device() + flag, &false_, sizeof(int));
}
bool get_flag(int flag) const
{
typedef Kokkos::Impl::DeepCopy< Kokkos::HostSpace, typename execution_space::memory_space > raw_deep_copy;
typedef Kokkos::Impl::DeepCopy< Kokkos::HostSpace, typename device_type::memory_space > raw_deep_copy;
int result = false;
raw_deep_copy(&result, m_scalars.ptr_on_device() + flag, sizeof(int));
return result;

63
lib/kokkos/containers/unit_tests/CMakeLists.txt
View File

@ -3,38 +3,49 @@ INCLUDE_DIRECTORIES(${CMAKE_CURRENT_BINARY_DIR})
INCLUDE_DIRECTORIES(REQUIRED_DURING_INSTALLATION_TESTING ${CMAKE_CURRENT_SOURCE_DIR})
INCLUDE_DIRECTORIES(${CMAKE_CURRENT_SOURCE_DIR}/../src )
SET(SOURCES
UnitTestMain.cpp
TestCuda.cpp
)
SET(LIBRARIES kokkoscore)
IF(Kokkos_ENABLE_Pthread)
LIST( APPEND SOURCES
TestThreads.cpp
)
ENDIF()
IF(Kokkos_ENABLE_Serial)
LIST( APPEND SOURCES
TestSerial.cpp
)
ENDIF()
IF(Kokkos_ENABLE_OpenMP)
LIST( APPEND SOURCES
TestOpenMP.cpp
)
ENDIF()
TRIBITS_ADD_EXECUTABLE_AND_TEST(
UnitTest
SOURCES ${SOURCES}
UnitTest_Threads
SOURCES TestThreads.cpp UnitTestMain.cpp
COMM serial mpi
NUM_MPI_PROCS 1
FAIL_REGULAR_EXPRESSION " FAILED "
TESTONLYLIBS kokkos_gtest
)
ENDIF()
IF(Kokkos_ENABLE_Serial)
TRIBITS_ADD_EXECUTABLE_AND_TEST(
UnitTest_Serial
SOURCES TestSerial.cpp UnitTestMain.cpp
COMM serial mpi
NUM_MPI_PROCS 1
FAIL_REGULAR_EXPRESSION " FAILED "
TESTONLYLIBS kokkos_gtest
)
ENDIF()
IF(Kokkos_ENABLE_OpenMP)
TRIBITS_ADD_EXECUTABLE_AND_TEST(
UnitTest_OpenMP
SOURCES TestOpenMP.cpp UnitTestMain.cpp
COMM serial mpi
NUM_MPI_PROCS 1
FAIL_REGULAR_EXPRESSION " FAILED "
TESTONLYLIBS kokkos_gtest
)
ENDIF()
IF(Kokkos_ENABLE_Cuda)
TRIBITS_ADD_EXECUTABLE_AND_TEST(
UnitTest_Cuda
SOURCES TestCuda.cpp UnitTestMain.cpp
COMM serial mpi
NUM_MPI_PROCS 1
FAIL_REGULAR_EXPRESSION " FAILED "
TESTONLYLIBS kokkos_gtest
)
ENDIF()

3
lib/kokkos/containers/unit_tests/TestDynamicView.hpp
View File

@ -64,6 +64,7 @@ struct TestDynamicView
typedef Kokkos::Experimental::MemoryPool<typename Space::device_type> memory_pool_type;
typedef Kokkos::Experimental::DynamicView<Scalar*,Space> view_type;
typedef typename view_type::const_type const_view_type ;
typedef typename Kokkos::TeamPolicy<execution_space>::member_type member_type ;
typedef double value_type;
@ -136,6 +137,8 @@ struct TestDynamicView
view_type da("A",pool,arg_total_size);
const_view_type ca(da);
// printf("TestDynamicView::run(%d) construct test functor\n",arg_total_size);
TestDynamicView functor(da,arg_total_size);

4
lib/kokkos/core/cmake/Dependencies.cmake
View File

@ -1,6 +1,6 @@
TRIBITS_PACKAGE_DEFINE_DEPENDENCIES(
LIB_OPTIONAL_TPLS Pthread CUDA HWLOC QTHREAD DLlib
LIB_OPTIONAL_TPLS Pthread CUDA HWLOC QTHREADS DLlib
TEST_OPTIONAL_TPLS CUSPARSE
)
TRIBITS_TPL_TENTATIVELY_ENABLE(DLlib)
TRIBITS_TPL_TENTATIVELY_ENABLE(DLlib)

2
lib/kokkos/core/cmake/KokkosCore_config.h.in
View File

@ -30,7 +30,7 @@
#cmakedefine KOKKOS_HAVE_PTHREAD
#cmakedefine KOKKOS_HAVE_SERIAL
#cmakedefine KOKKOS_HAVE_QTHREAD
#cmakedefine KOKKOS_HAVE_QTHREADS
#cmakedefine KOKKOS_HAVE_Winthread
#cmakedefine KOKKOS_HAVE_OPENMP
#cmakedefine KOKKOS_HAVE_HWLOC

1
lib/kokkos/core/perf_test/Makefile
View File

@ -60,4 +60,3 @@ clean: kokkos-clean
gtest-all.o:$(GTEST_PATH)/gtest/gtest-all.cc
$(CXX) $(KOKKOS_CPPFLAGS) $(KOKKOS_CXXFLAGS) $(CXXFLAGS) $(EXTRA_INC) -c $(GTEST_PATH)/gtest/gtest-all.cc

10
lib/kokkos/core/perf_test/PerfTestCuda.cpp
View File

@ -52,6 +52,8 @@
#include <impl/Kokkos_Timer.hpp>
#include <PerfTestMDRange.hpp>
#include <PerfTestHexGrad.hpp>
#include <PerfTestBlasKernels.hpp>
#include <PerfTestGramSchmidt.hpp>
@ -72,6 +74,14 @@ class cuda : public ::testing::Test {
}
};
//TEST_F( cuda, mdrange_lr ) {
// EXPECT_NO_THROW( (run_test_mdrange<Kokkos::Cuda , Kokkos::LayoutRight>( 5, 8, "Kokkos::Cuda" )) );
//}
//TEST_F( cuda, mdrange_ll ) {
// EXPECT_NO_THROW( (run_test_mdrange<Kokkos::Cuda , Kokkos::LayoutLeft>( 5, 8, "Kokkos::Cuda" )) );
//}
TEST_F( cuda, hexgrad )
{
EXPECT_NO_THROW( run_test_hexgrad< Kokkos::Cuda >( 10 , 20, "Kokkos::Cuda" ) );

336
lib/kokkos/core/perf_test/PerfTestDriver.hpp
View File

@ -60,6 +60,342 @@ namespace Test {
enum { NUMBER_OF_TRIALS = 5 };
template< class DeviceType , class LayoutType >
void run_test_mdrange( int exp_beg , int exp_end, const char deviceTypeName[], int range_offset = 0, int tile_offset = 0 )
// exp_beg = 6 => 2^6 = 64 is starting range length
{
#define MDRANGE_PERFORMANCE_OUTPUT_VERBOSE 0
std::string label_mdrange ;
label_mdrange.append( "\"MDRange< double , " );
label_mdrange.append( deviceTypeName );
label_mdrange.append( " >\"" );
std::string label_range_col2 ;
label_range_col2.append( "\"RangeColTwo< double , " );
label_range_col2.append( deviceTypeName );
label_range_col2.append( " >\"" );
std::string label_range_col_all ;
label_range_col_all.append( "\"RangeColAll< double , " );
label_range_col_all.append( deviceTypeName );
label_range_col_all.append( " >\"" );
if ( std::is_same<LayoutType, Kokkos::LayoutRight>::value) {
std::cout << "--------------------------------------------------------------\n"
<< "Performance tests for MDRange Layout Right"
<< "\n--------------------------------------------------------------" << std::endl;
} else {
std::cout << "--------------------------------------------------------------\n"
<< "Performance tests for MDRange Layout Left"
<< "\n--------------------------------------------------------------" << std::endl;
}
for (int i = exp_beg ; i < exp_end ; ++i) {
const int range_length = (1<<i) + range_offset;
std::cout << "\n--------------------------------------------------------------\n"
<< "--------------------------------------------------------------\n"
<< "MDRange Test: range bounds: " << range_length << " , " << range_length << " , " << range_length
<< "\n--------------------------------------------------------------\n"
<< "--------------------------------------------------------------\n";
// << std::endl;
int t0_min = 0, t1_min = 0, t2_min = 0;
double seconds_min = 0.0;
// Test 1: The MDRange in full
{
int t0 = 1, t1 = 1, t2 = 1;
int counter = 1;
#if !defined(KOKKOS_HAVE_CUDA)
int min_bnd = 8;
int tfast = range_length;
#else
int min_bnd = 2;
int tfast = 32;
#endif
while ( tfast >= min_bnd ) {
int tmid = min_bnd;
while ( tmid < tfast ) {
t0 = min_bnd;
t1 = tmid;
t2 = tfast;
int t2_rev = min_bnd;
int t1_rev = tmid;
int t0_rev = tfast;
#if defined(KOKKOS_HAVE_CUDA)
//Note: Product of tile sizes must be < 1024 for Cuda
if ( t0*t1*t2 >= 1024 ) {
printf(" Exceeded Cuda tile limits; onto next range set\n\n");
break;
}
#endif
// Run 1 with tiles LayoutRight style
double seconds_1 = 0;
{ seconds_1 = MultiDimRangePerf3D< DeviceType , double , LayoutType >::test_multi_index(range_length,range_length,range_length, t0, t1, t2) ; }
#if MDRANGE_PERFORMANCE_OUTPUT_VERBOSE
std::cout << label_mdrange
<< " , " << t0 << " , " << t1 << " , " << t2
<< " , " << seconds_1
<< std::endl ;
#endif
if ( counter == 1 ) {
seconds_min = seconds_1;
t0_min = t0;
t1_min = t1;
t2_min = t2;
}
else {
if ( seconds_1 < seconds_min )
{
seconds_min = seconds_1;
t0_min = t0;
t1_min = t1;
t2_min = t2;
}
}
// Run 2 with tiles LayoutLeft style - reverse order of tile dims
double seconds_1rev = 0;
{ seconds_1rev = MultiDimRangePerf3D< DeviceType , double , LayoutType >::test_multi_index(range_length,range_length,range_length, t0_rev, t1_rev, t2_rev) ; }
#if MDRANGE_PERFORMANCE_OUTPUT_VERBOSE
std::cout << label_mdrange
<< " , " << t0_rev << " , " << t1_rev << " , " << t2_rev
<< " , " << seconds_1rev
<< std::endl ;
#endif
if ( seconds_1rev < seconds_min )
{
seconds_min = seconds_1rev;
t0_min = t0_rev;
t1_min = t1_rev;
t2_min = t2_rev;
}
++counter;
tmid <<= 1;
} //end inner while
tfast >>=1;
} //end outer while
std::cout << "\n"
<< "--------------------------------------------------------------\n"
<< label_mdrange
<< "\n Min values "
<< "\n Range length per dim (3D): " << range_length
<< "\n TileDims: " << t0_min << " , " << t1_min << " , " << t2_min
<< "\n Min time: " << seconds_min
<< "\n---------------------------------------------------------------"
<< std::endl ;
} //end scope
#if !defined(KOKKOS_HAVE_CUDA)
double seconds_min_c = 0.0;
int t0c_min = 0, t1c_min = 0, t2c_min = 0;
int counter = 1;
{
int min_bnd = 8;
// Test 1_c: MDRange with 0 for 'inner' tile dim; this case will utilize the full span in that direction, should be similar to Collapse<2>
if ( std::is_same<LayoutType, Kokkos::LayoutRight>::value ) {
for ( unsigned int T0 = min_bnd; T0 < static_cast<unsigned int>(range_length); T0<<=1 ) {
for ( unsigned int T1 = min_bnd; T1 < static_cast<unsigned int>(range_length); T1<<=1 ) {
double seconds_c = 0;
{ seconds_c = MultiDimRangePerf3D< DeviceType , double , LayoutType >::test_multi_index(range_length,range_length,range_length, T0, T1, 0) ; }
#if MDRANGE_PERFORMANCE_OUTPUT_VERBOSE
std::cout << " MDRange LR with '0' tile - collapse-like \n"
<< label_mdrange
<< " , " << T0 << " , " << T1 << " , " << range_length
<< " , " << seconds_c
<< std::endl ;
#endif
t2c_min = range_length;
if ( counter == 1 ) {
seconds_min_c = seconds_c;
t0c_min = T0;
t1c_min = T1;
}
else {
if ( seconds_c < seconds_min_c )
{
seconds_min_c = seconds_c;
t0c_min = T0;
t1c_min = T1;
}
}
++counter;
}
}
}
else {
for ( unsigned int T1 = min_bnd; T1 <= static_cast<unsigned int>(range_length); T1<<=1 ) {
for ( unsigned int T2 = min_bnd; T2 <= static_cast<unsigned int>(range_length); T2<<=1 ) {
double seconds_c = 0;
{ seconds_c = MultiDimRangePerf3D< DeviceType , double , LayoutType >::test_multi_index(range_length,range_length,range_length, 0, T1, T2) ; }
#if MDRANGE_PERFORMANCE_OUTPUT_VERBOSE
std::cout << " MDRange LL with '0' tile - collapse-like \n"
<< label_mdrange
<< " , " <<range_length << " < " << T1 << " , " << T2
<< " , " << seconds_c
<< std::endl ;
#endif
t0c_min = range_length;
if ( counter == 1 ) {
seconds_min_c = seconds_c;
t1c_min = T1;
t2c_min = T2;
}
else {
if ( seconds_c < seconds_min_c )
{
seconds_min_c = seconds_c;
t1c_min = T1;
t2c_min = T2;
}
}
++counter;
}
}
}
std::cout
// << "--------------------------------------------------------------\n"
<< label_mdrange
<< " Collapse<2> style: "
<< "\n Min values "
<< "\n Range length per dim (3D): " << range_length
<< "\n TileDims: " << t0c_min << " , " << t1c_min << " , " << t2c_min
<< "\n Min time: " << seconds_min_c
<< "\n---------------------------------------------------------------"
<< std::endl ;
} //end scope test 2
#endif
// Test 2: RangePolicy Collapse2 style
double seconds_2 = 0;
{ seconds_2 = RangePolicyCollapseTwo< DeviceType , double , LayoutType >::test_index_collapse_two(range_length,range_length,range_length) ; }
std::cout << label_range_col2
<< " , " << range_length
<< " , " << seconds_2
<< std::endl ;
// Test 3: RangePolicy Collapse all style - not necessary, always slow
/*
double seconds_3 = 0;
{ seconds_3 = RangePolicyCollapseAll< DeviceType , double , LayoutType >::test_collapse_all(range_length,range_length,range_length) ; }
std::cout << label_range_col_all
<< " , " << range_length
<< " , " << seconds_3
<< "\n---------------------------------------------------------------"
<< std::endl ;
*/
// Compare fastest times... will never be collapse all so ignore it
// seconds_min = tiled MDRange
// seconds_min_c = collapse<2>-like MDRange (tiledim = span for fast dim) - only for non-Cuda, else tile too long
// seconds_2 = collapse<2>-style RangePolicy
// seconds_3 = collapse<3>-style RangePolicy
#if !defined(KOKKOS_HAVE_CUDA)
if ( seconds_min < seconds_min_c ) {
if ( seconds_min < seconds_2 ) {
std::cout << "--------------------------------------------------------------\n"
<< " Fastest run: MDRange tiled\n"
<< " Time: " << seconds_min
<< " Difference: " << seconds_2 - seconds_min
<< " Other times: \n"
<< " MDrange collapse-like (tiledim = span on fast dim) type: " << seconds_min_c << "\n"
<< " Collapse2 Range Policy: " << seconds_2 << "\n"
<< "\n--------------------------------------------------------------"
<< "\n--------------------------------------------------------------"
//<< "\n\n"
<< std::endl;
}
else if ( seconds_min > seconds_2 ) {
std::cout << " Fastest run: Collapse2 RangePolicy\n"
<< " Time: " << seconds_2
<< " Difference: " << seconds_min - seconds_2
<< " Other times: \n"
<< " MDrange Tiled: " << seconds_min << "\n"
<< " MDrange collapse-like (tiledim = span on fast dim) type: " << seconds_min_c << "\n"
<< "\n--------------------------------------------------------------"
<< "\n--------------------------------------------------------------"
//<< "\n\n"
<< std::endl;
}
}
else if ( seconds_min > seconds_min_c ) {
if ( seconds_min_c < seconds_2 ) {
std::cout << "--------------------------------------------------------------\n"
<< " Fastest run: MDRange collapse-like (tiledim = span on fast dim) type\n"
<< " Time: " << seconds_min_c
<< " Difference: " << seconds_2 - seconds_min_c
<< " Other times: \n"
<< " MDrange Tiled: " << seconds_min << "\n"
<< " Collapse2 Range Policy: " << seconds_2 << "\n"
<< "\n--------------------------------------------------------------"
<< "\n--------------------------------------------------------------"
//<< "\n\n"
<< std::endl;
}
else if ( seconds_min_c > seconds_2 ) {
std::cout << " Fastest run: Collapse2 RangePolicy\n"
<< " Time: " << seconds_2
<< " Difference: " << seconds_min_c - seconds_2
<< " Other times: \n"
<< " MDrange Tiled: " << seconds_min << "\n"
<< " MDrange collapse-like (tiledim = span on fast dim) type: " << seconds_min_c << "\n"
<< "\n--------------------------------------------------------------"
<< "\n--------------------------------------------------------------"
//<< "\n\n"
<< std::endl;
}
} // end else if
#else
if ( seconds_min < seconds_2 ) {
std::cout << "--------------------------------------------------------------\n"
<< " Fastest run: MDRange tiled\n"
<< " Time: " << seconds_min
<< " Difference: " << seconds_2 - seconds_min
<< " Other times: \n"
<< " Collapse2 Range Policy: " << seconds_2 << "\n"
<< "\n--------------------------------------------------------------"
<< "\n--------------------------------------------------------------"
//<< "\n\n"
<< std::endl;
}
else if ( seconds_min > seconds_2 ) {
std::cout << " Fastest run: Collapse2 RangePolicy\n"
<< " Time: " << seconds_2
<< " Difference: " << seconds_min - seconds_2
<< " Other times: \n"
<< " MDrange Tiled: " << seconds_min << "\n"
<< "\n--------------------------------------------------------------"
<< "\n--------------------------------------------------------------"
//<< "\n\n"
<< std::endl;
}
#endif
} //end for
#undef MDRANGE_PERFORMANCE_OUTPUT_VERBOSE
}
template< class DeviceType >

10
lib/kokkos/core/perf_test/PerfTestHost.cpp
View File

@ -66,6 +66,8 @@ const char TestHostDeviceName[] = "Kokkos::Serial" ;
#include <impl/Kokkos_Timer.hpp>
#include <PerfTestMDRange.hpp>
#include <PerfTestHexGrad.hpp>
#include <PerfTestBlasKernels.hpp>
#include <PerfTestGramSchmidt.hpp>
@ -102,6 +104,14 @@ protected:
}
};
//TEST_F( host, mdrange_lr ) {
// EXPECT_NO_THROW( (run_test_mdrange<TestHostDevice , Kokkos::LayoutRight> (5, 8, TestHostDeviceName) ) );
//}
//TEST_F( host, mdrange_ll ) {
// EXPECT_NO_THROW( (run_test_mdrange<TestHostDevice , Kokkos::LayoutLeft> (5, 8, TestHostDeviceName) ) );
//}
TEST_F( host, hexgrad ) {
EXPECT_NO_THROW(run_test_hexgrad< TestHostDevice>( 10, 20, TestHostDeviceName ));
}

564
lib/kokkos/core/perf_test/PerfTestMDRange.hpp Normal file
View File

@ -0,0 +1,564 @@
/*
//@HEADER
// ************************************************************************
//
// Kokkos v. 2.0
// Copyright (2014) Sandia Corporation
//
// Under the terms of Contract DE-AC04-94AL85000 with Sandia Corporation,
// the U.S. Government retains certain rights in this software.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
// 1. Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
//
// 2. Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
//
// 3. Neither the name of the Corporation nor the names of the
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY SANDIA CORPORATION "AS IS" AND ANY
// EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
// PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL SANDIA CORPORATION OR THE
// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
// LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
// NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
// SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
// Questions? Contact H. Carter Edwards (hcedwar@sandia.gov)
//
// ************************************************************************
//@HEADER
*/
namespace Test {
template< class DeviceType
, typename ScalarType = double
, typename TestLayout = Kokkos::LayoutRight
>
struct MultiDimRangePerf3D
{
typedef DeviceType execution_space;
typedef typename execution_space::size_type size_type;
using iterate_type = Kokkos::Experimental::Iterate;
typedef Kokkos::View<ScalarType***, TestLayout, DeviceType> view_type;
typedef typename view_type::HostMirror host_view_type;
view_type A;
view_type B;
const long irange;
const long jrange;
const long krange;
MultiDimRangePerf3D(const view_type & A_, const view_type & B_, const long &irange_, const long &jrange_, const long &krange_)
: A(A_), B(B_), irange(irange_), jrange(jrange_), krange(krange_)
{}
KOKKOS_INLINE_FUNCTION
void operator()(const long i, const long j, const long k) const
{
A(i,j,k) = 0.25*(ScalarType)( B(i+2,j,k) + B(i+1,j,k)
+ B(i,j+2,k) + B(i,j+1,k)
+ B(i,j,k+2) + B(i,j,k+1)
+ B(i,j,k) );
}
struct InitZeroTag {};
// struct InitViewTag {};
struct Init
{
Init(const view_type & input_, const long &irange_, const long &jrange_, const long &krange_)
: input(input_), irange(irange_), jrange(jrange_), krange(krange_) {}
KOKKOS_INLINE_FUNCTION
void operator()(const long i, const long j, const long k) const
{
input(i,j,k) = 1.0;
}
KOKKOS_INLINE_FUNCTION
void operator()(const InitZeroTag&, const long i, const long j, const long k) const
{
input(i,j,k) = 0;
}
view_type input;
const long irange;
const long jrange;
const long krange;
};
static double test_multi_index(const unsigned int icount, const unsigned int jcount, const unsigned int kcount, const unsigned int Ti = 1, const unsigned int Tj = 1, const unsigned int Tk = 1, const long iter = 1)
{
//This test performs multidim range over all dims
view_type Atest("Atest", icount, jcount, kcount);
view_type Btest("Btest", icount+2, jcount+2, kcount+2);
typedef MultiDimRangePerf3D<execution_space,ScalarType,TestLayout> FunctorType;
double dt_min = 0;
// LayoutRight
if ( std::is_same<TestLayout, Kokkos::LayoutRight>::value ) {
Kokkos::Experimental::MDRangePolicy<Kokkos::Experimental::Rank<3, iterate_type::Right, iterate_type::Right>, execution_space > policy_initA({{0,0,0}},{{icount,jcount,kcount}},{{Ti,Tj,Tk}});
Kokkos::Experimental::MDRangePolicy<Kokkos::Experimental::Rank<3, iterate_type::Right, iterate_type::Right>, execution_space > policy_initB({{0,0,0}},{{icount+2,jcount+2,kcount+2}},{{Ti,Tj,Tk}});
typedef typename Kokkos::Experimental::MDRangePolicy<Kokkos::Experimental::Rank<3, iterate_type::Right, iterate_type::Right>, execution_space > MDRangeType;
using tile_type = typename MDRangeType::tile_type;
using point_type = typename MDRangeType::point_type;
Kokkos::Experimental::MDRangePolicy<Kokkos::Experimental::Rank<3, iterate_type::Right, iterate_type::Right>, execution_space > policy(point_type{{0,0,0}},point_type{{icount,jcount,kcount}},tile_type{{Ti,Tj,Tk}} );
Kokkos::Experimental::md_parallel_for( policy_initA, Init(Atest, icount, jcount, kcount) );
execution_space::fence();
Kokkos::Experimental::md_parallel_for( policy_initB, Init(Btest, icount+2, jcount+2, kcount+2) );
execution_space::fence();
for (int i = 0; i < iter; ++i)
{
Kokkos::Timer timer;
Kokkos::Experimental::md_parallel_for( policy, FunctorType(Atest, Btest, icount, jcount, kcount) );
execution_space::fence();
const double dt = timer.seconds();
if ( 0 == i ) dt_min = dt ;
else dt_min = dt < dt_min ? dt : dt_min ;
//Correctness check - only the first run
if ( 0 == i )
{
long numErrors = 0;
host_view_type Ahost("Ahost", icount, jcount, kcount);
Kokkos::deep_copy(Ahost, Atest);
host_view_type Bhost("Bhost", icount+2, jcount+2, kcount+2);
Kokkos::deep_copy(Bhost, Btest);
// On KNL, this may vectorize - add print statement to prevent
// Also, compare against epsilon, as vectorization can change bitwise answer
for ( long l = 0; l < static_cast<long>(icount); ++l ) {
for ( long j = 0; j < static_cast<long>(jcount); ++j ) {
for ( long k = 0; k < static_cast<long>(kcount); ++k ) {
ScalarType check = 0.25*(ScalarType)( Bhost(l+2,j,k) + Bhost(l+1,j,k)
+ Bhost(l,j+2,k) + Bhost(l,j+1,k)
+ Bhost(l,j,k+2) + Bhost(l,j,k+1)
+ Bhost(l,j,k) );
if ( Ahost(l,j,k) - check != 0 ) {
++numErrors;
std::cout << " Correctness error at index: " << l << ","<<j<<","<<k<<"\n"
<< " multi Ahost = " << Ahost(l,j,k) << " expected = " << check
<< " multi Bhost(ijk) = " << Bhost(l,j,k)
<< " multi Bhost(l+1jk) = " << Bhost(l+1,j,k)
<< " multi Bhost(l+2jk) = " << Bhost(l+2,j,k)
<< " multi Bhost(ij+1k) = " << Bhost(l,j+1,k)
<< " multi Bhost(ij+2k) = " << Bhost(l,j+2,k)
<< " multi Bhost(ijk+1) = " << Bhost(l,j,k+1)
<< " multi Bhost(ijk+2) = " << Bhost(l,j,k+2)
<< std::endl;
//exit(-1);
}
} } }
if ( numErrors != 0 ) { std::cout << "LR multi: errors " << numErrors << " range product " << icount*jcount*kcount << " LL " << jcount*kcount << " LR " << icount*jcount << std::endl; }
//else { std::cout << " multi: No errors!" << std::endl; }
}
} //end for
}
// LayoutLeft
else {
Kokkos::Experimental::MDRangePolicy<Kokkos::Experimental::Rank<3,iterate_type::Left,iterate_type::Left>, execution_space > policy_initA({{0,0,0}},{{icount,jcount,kcount}},{{Ti,Tj,Tk}});
Kokkos::Experimental::MDRangePolicy<Kokkos::Experimental::Rank<3,iterate_type::Left,iterate_type::Left>, execution_space > policy_initB({{0,0,0}},{{icount+2,jcount+2,kcount+2}},{{Ti,Tj,Tk}});
//typedef typename Kokkos::Experimental::MDRangePolicy<Kokkos::Experimental::Rank<3, iterate_type::Left, iterate_type::Left>, execution_space > MDRangeType;
//using tile_type = typename MDRangeType::tile_type;
//using point_type = typename MDRangeType::point_type;
//Kokkos::Experimental::MDRangePolicy<Kokkos::Experimental::Rank<3, iterate_type::Left, iterate_type::Left>, execution_space > policy(point_type{{0,0,0}},point_type{{icount,jcount,kcount}},tile_type{{Ti,Tj,Tk}} );
Kokkos::Experimental::MDRangePolicy<Kokkos::Experimental::Rank<3, iterate_type::Left, iterate_type::Left>, execution_space > policy({{0,0,0}},{{icount,jcount,kcount}},{{Ti,Tj,Tk}} );
Kokkos::Experimental::md_parallel_for( policy_initA, Init(Atest, icount, jcount, kcount) );
execution_space::fence();
Kokkos::Experimental::md_parallel_for( policy_initB, Init(Btest, icount+2, jcount+2, kcount+2) );
execution_space::fence();
for (int i = 0; i < iter; ++i)
{
Kokkos::Timer timer;
Kokkos::Experimental::md_parallel_for( policy, FunctorType(Atest, Btest, icount, jcount, kcount) );
execution_space::fence();
const double dt = timer.seconds();
if ( 0 == i ) dt_min = dt ;
else dt_min = dt < dt_min ? dt : dt_min ;
//Correctness check - only the first run
if ( 0 == i )
{
long numErrors = 0;
host_view_type Ahost("Ahost", icount, jcount, kcount);
Kokkos::deep_copy(Ahost, Atest);
host_view_type Bhost("Bhost", icount+2, jcount+2, kcount+2);
Kokkos::deep_copy(Bhost, Btest);
// On KNL, this may vectorize - add print statement to prevent
// Also, compare against epsilon, as vectorization can change bitwise answer
for ( long l = 0; l < static_cast<long>(icount); ++l ) {
for ( long j = 0; j < static_cast<long>(jcount); ++j ) {
for ( long k = 0; k < static_cast<long>(kcount); ++k ) {
ScalarType check = 0.25*(ScalarType)( Bhost(l+2,j,k) + Bhost(l+1,j,k)
+ Bhost(l,j+2,k) + Bhost(l,j+1,k)
+ Bhost(l,j,k+2) + Bhost(l,j,k+1)
+ Bhost(l,j,k) );
if ( Ahost(l,j,k) - check != 0 ) {
++numErrors;
std::cout << " Correctness error at index: " << l << ","<<j<<","<<k<<"\n"
<< " multi Ahost = " << Ahost(l,j,k) << " expected = " << check
<< " multi Bhost(ijk) = " << Bhost(l,j,k)
<< " multi Bhost(l+1jk) = " << Bhost(l+1,j,k)
<< " multi Bhost(l+2jk) = " << Bhost(l+2,j,k)
<< " multi Bhost(ij+1k) = " << Bhost(l,j+1,k)
<< " multi Bhost(ij+2k) = " << Bhost(l,j+2,k)
<< " multi Bhost(ijk+1) = " << Bhost(l,j,k+1)
<< " multi Bhost(ijk+2) = " << Bhost(l,j,k+2)
<< std::endl;
//exit(-1);
}
} } }
if ( numErrors != 0 ) { std::cout << " LL multi run: errors " << numErrors << " range product " << icount*jcount*kcount << " LL " << jcount*kcount << " LR " << icount*jcount << std::endl; }
//else { std::cout << " multi: No errors!" << std::endl; }
}
} //end for
}
return dt_min;
}
};
template< class DeviceType
, typename ScalarType = double
, typename TestLayout = Kokkos::LayoutRight
>
struct RangePolicyCollapseTwo
{
// RangePolicy for 3D range, but will collapse only 2 dims => like Rank<2> for multi-dim; unroll 2 dims in one-dim
typedef DeviceType execution_space;
typedef typename execution_space::size_type size_type;
typedef TestLayout layout;
using iterate_type = Kokkos::Experimental::Iterate;
typedef Kokkos::View<ScalarType***, TestLayout, DeviceType> view_type;
typedef typename view_type::HostMirror host_view_type;
view_type A;
view_type B;
const long irange;
const long jrange;
const long krange;
RangePolicyCollapseTwo(view_type & A_, const view_type & B_, const long &irange_, const long &jrange_, const long &krange_)
: A(A_), B(B_) , irange(irange_), jrange(jrange_), krange(krange_)
{}
KOKKOS_INLINE_FUNCTION
void operator()(const long r) const
{
if ( std::is_same<TestLayout, Kokkos::LayoutRight>::value )
{
//id(i,j,k) = k + j*Nk + i*Nk*Nj = k + Nk*(j + i*Nj) = k + Nk*r
//r = j + i*Nj
long i = int(r / jrange);
long j = int( r - i*jrange);
for (int k = 0; k < krange; ++k) {
A(i,j,k) = 0.25*(ScalarType)( B(i+2,j,k) + B(i+1,j,k)
+ B(i,j+2,k) + B(i,j+1,k)
+ B(i,j,k+2) + B(i,j,k+1)
+ B(i,j,k) );
}
}
else if ( std::is_same<TestLayout, Kokkos::LayoutLeft>::value )
{
//id(i,j,k) = i + j*Ni + k*Ni*Nj = i + Ni*(j + k*Nj) = i + Ni*r
//r = j + k*Nj
long k = int(r / jrange);
long j = int( r - k*jrange);
for (int i = 0; i < irange; ++i) {
A(i,j,k) = 0.25*(ScalarType)( B(i+2,j,k) + B(i+1,j,k)
+ B(i,j+2,k) + B(i,j+1,k)
+ B(i,j,k+2) + B(i,j,k+1)
+ B(i,j,k) );
}
}
}
struct Init
{
view_type input;
const long irange;
const long jrange;
const long krange;
Init(const view_type & input_, const long &irange_, const long &jrange_, const long &krange_)
: input(input_), irange(irange_), jrange(jrange_), krange(krange_) {}
KOKKOS_INLINE_FUNCTION
void operator()(const long r) const
{
if ( std::is_same<TestLayout, Kokkos::LayoutRight>::value )
{
long i = int(r / jrange);
long j = int( r - i*jrange);
for (int k = 0; k < krange; ++k) {
input(i,j,k) = 1;
}
}
else if ( std::is_same<TestLayout, Kokkos::LayoutLeft>::value )
{
long k = int(r / jrange);
long j = int( r - k*jrange);
for (int i = 0; i < irange; ++i) {
input(i,j,k) = 1;
}
}
}
};
static double test_index_collapse_two(const unsigned int icount, const unsigned int jcount, const unsigned int kcount, const long iter = 1)
{
// This test refers to collapsing two dims while using the RangePolicy
view_type Atest("Atest", icount, jcount, kcount);
view_type Btest("Btest", icount+2, jcount+2, kcount+2);
typedef RangePolicyCollapseTwo<execution_space,ScalarType,TestLayout> FunctorType;
long collapse_index_rangeA = 0;
long collapse_index_rangeB = 0;
if ( std::is_same<TestLayout, Kokkos::LayoutRight>::value ) {
collapse_index_rangeA = icount*jcount;
collapse_index_rangeB = (icount+2)*(jcount+2);
// std::cout << " LayoutRight " << std::endl;
} else if ( std::is_same<TestLayout, Kokkos::LayoutLeft>::value ) {
collapse_index_rangeA = kcount*jcount;
collapse_index_rangeB = (kcount+2)*(jcount+2);
// std::cout << " LayoutLeft " << std::endl;
} else {
std::cout << " LayoutRight or LayoutLeft required - will pass 0 as range instead " << std::endl;
exit(-1);
}
Kokkos::RangePolicy<execution_space> policy(0, (collapse_index_rangeA) );
Kokkos::RangePolicy<execution_space> policy_initB(0, (collapse_index_rangeB) );
double dt_min = 0;
Kokkos::parallel_for( policy, Init(Atest,icount,jcount,kcount) );
execution_space::fence();
Kokkos::parallel_for( policy_initB, Init(Btest,icount+2,jcount+2,kcount+2) );
execution_space::fence();
for (int i = 0; i < iter; ++i)
{
Kokkos::Timer timer;
Kokkos::parallel_for(policy, FunctorType(Atest, Btest, icount, jcount, kcount));
execution_space::fence();
const double dt = timer.seconds();
if ( 0 == i ) dt_min = dt ;
else dt_min = dt < dt_min ? dt : dt_min ;
//Correctness check - first iteration only
if ( 0 == i )
{
long numErrors = 0;
host_view_type Ahost("Ahost", icount, jcount, kcount);
Kokkos::deep_copy(Ahost, Atest);
host_view_type Bhost("Bhost", icount+2, jcount+2, kcount+2);
Kokkos::deep_copy(Bhost, Btest);
// On KNL, this may vectorize - add print statement to prevent
// Also, compare against epsilon, as vectorization can change bitwise answer
for ( long l = 0; l < static_cast<long>(icount); ++l ) {
for ( long j = 0; j < static_cast<long>(jcount); ++j ) {
for ( long k = 0; k < static_cast<long>(kcount); ++k ) {
ScalarType check = 0.25*(ScalarType)( Bhost(l+2,j,k) + Bhost(l+1,j,k)
+ Bhost(l,j+2,k) + Bhost(l,j+1,k)
+ Bhost(l,j,k+2) + Bhost(l,j,k+1)
+ Bhost(l,j,k) );
if ( Ahost(l,j,k) - check != 0 ) {
++numErrors;
std::cout << " Correctness error at index: " << l << ","<<j<<","<<k<<"\n"
<< " flat Ahost = " << Ahost(l,j,k) << " expected = " << check << std::endl;
//exit(-1);
}
} } }
if ( numErrors != 0 ) { std::cout << " RP collapse2: errors " << numErrors << " range product " << icount*jcount*kcount << " LL " << jcount*kcount << " LR " << icount*jcount << std::endl; }
//else { std::cout << " RP collapse2: Pass! " << std::endl; }
}
}
return dt_min;
}
};
template< class DeviceType
, typename ScalarType = double
, typename TestLayout = Kokkos::LayoutRight
>
struct RangePolicyCollapseAll
{
// RangePolicy for 3D range, but will collapse all dims
typedef DeviceType execution_space;
typedef typename execution_space::size_type size_type;
typedef TestLayout layout;
typedef Kokkos::View<ScalarType***, TestLayout, DeviceType> view_type;
typedef typename view_type::HostMirror host_view_type;
view_type A;
view_type B;
const long irange;
const long jrange;
const long krange;
RangePolicyCollapseAll(view_type & A_, const view_type & B_, const long &irange_, const long &jrange_, const long &krange_)
: A(A_), B(B_), irange(irange_), jrange(jrange_), krange(krange_)
{}
KOKKOS_INLINE_FUNCTION
void operator()(const long r) const
{
if ( std::is_same<TestLayout, Kokkos::LayoutRight>::value )
{
long i = int(r / (jrange*krange));
long j = int(( r - i*jrange*krange)/krange);
long k = int(r - i*jrange*krange - j*krange);
A(i,j,k) = 0.25*(ScalarType)( B(i+2,j,k) + B(i+1,j,k)
+ B(i,j+2,k) + B(i,j+1,k)
+ B(i,j,k+2) + B(i,j,k+1)
+ B(i,j,k) );
}
else if ( std::is_same<TestLayout, Kokkos::LayoutLeft>::value )
{
long k = int(r / (irange*jrange));
long j = int(( r - k*irange*jrange)/irange);
long i = int(r - k*irange*jrange - j*irange);
A(i,j,k) = 0.25*(ScalarType)( B(i+2,j,k) + B(i+1,j,k)
+ B(i,j+2,k) + B(i,j+1,k)
+ B(i,j,k+2) + B(i,j,k+1)
+ B(i,j,k) );
}
}
struct Init
{
view_type input;
const long irange;
const long jrange;
const long krange;
Init(const view_type & input_, const long &irange_, const long &jrange_, const long &krange_)
: input(input_), irange(irange_), jrange(jrange_), krange(krange_) {}
KOKKOS_INLINE_FUNCTION
void operator()(const long r) const
{
if ( std::is_same<TestLayout, Kokkos::LayoutRight>::value )
{
long i = int(r / (jrange*krange));
long j = int(( r - i*jrange*krange)/krange);
long k = int(r - i*jrange*krange - j*krange);
input(i,j,k) = 1;
}
else if ( std::is_same<TestLayout, Kokkos::LayoutLeft>::value )
{
long k = int(r / (irange*jrange));
long j = int(( r - k*irange*jrange)/irange);
long i = int(r - k*irange*jrange - j*irange);
input(i,j,k) = 1;
}
}
};
static double test_collapse_all(const unsigned int icount, const unsigned int jcount, const unsigned int kcount, const long iter = 1)
{
//This test refers to collapsing all dims using the RangePolicy
view_type Atest("Atest", icount, jcount, kcount);
view_type Btest("Btest", icount+2, jcount+2, kcount+2);
typedef RangePolicyCollapseAll<execution_space,ScalarType,TestLayout> FunctorType;
const long flat_index_range = icount*jcount*kcount;
Kokkos::RangePolicy<execution_space> policy(0, flat_index_range );
Kokkos::RangePolicy<execution_space> policy_initB(0, (icount+2)*(jcount+2)*(kcount+2) );
double dt_min = 0;
Kokkos::parallel_for( policy, Init(Atest,icount,jcount,kcount) );
execution_space::fence();
Kokkos::parallel_for( policy_initB, Init(Btest,icount+2,jcount+2,kcount+2) );
execution_space::fence();
for (int i = 0; i < iter; ++i)
{
Kokkos::Timer timer;
Kokkos::parallel_for(policy, FunctorType(Atest, Btest, icount, jcount, kcount));
execution_space::fence();
const double dt = timer.seconds();
if ( 0 == i ) dt_min = dt ;
else dt_min = dt < dt_min ? dt : dt_min ;
//Correctness check - first iteration only
if ( 0 == i )
{
long numErrors = 0;
host_view_type Ahost("Ahost", icount, jcount, kcount);
Kokkos::deep_copy(Ahost, Atest);
host_view_type Bhost("Bhost", icount+2, jcount+2, kcount+2);
Kokkos::deep_copy(Bhost, Btest);
// On KNL, this may vectorize - add print statement to prevent
// Also, compare against epsilon, as vectorization can change bitwise answer
for ( long l = 0; l < static_cast<long>(icount); ++l ) {
for ( long j = 0; j < static_cast<long>(jcount); ++j ) {
for ( long k = 0; k < static_cast<long>(kcount); ++k ) {
ScalarType check = 0.25*(ScalarType)( Bhost(l+2,j,k) + Bhost(l+1,j,k)
+ Bhost(l,j+2,k) + Bhost(l,j+1,k)
+ Bhost(l,j,k+2) + Bhost(l,j,k+1)
+ Bhost(l,j,k) );
if ( Ahost(l,j,k) - check != 0 ) {
++numErrors;
std::cout << " Callapse ALL Correctness error at index: " << l << ","<<j<<","<<k<<"\n"
<< " flat Ahost = " << Ahost(l,j,k) << " expected = " << check << std::endl;
//exit(-1);
}
} } }
if ( numErrors != 0 ) { std::cout << " RP collapse all: errors " << numErrors << " range product " << icount*jcount*kcount << " LL " << jcount*kcount << " LR " << icount*jcount << std::endl; }
//else { std::cout << " RP collapse all: Pass! " << std::endl; }
}
}
return dt_min;
}
};
} //end namespace Test

12
lib/kokkos/core/src/CMakeLists.txt
View File

@ -92,13 +92,13 @@ LIST(APPEND SOURCES ${SOURCES_CUDA} )
INSTALL(FILES ${HEADERS_CUDA} DESTINATION ${TRILINOS_INCDIR}/Cuda/)
#-----------------------------------------------------------------------------
FILE(GLOB HEADERS_QTHREAD Qthread/*.hpp)
FILE(GLOB SOURCES_QTHREAD Qthread/*.cpp)
FILE(GLOB HEADERS_QTHREADS Qthreads/*.hpp)
FILE(GLOB SOURCES_QTHREADS Qthreads/*.cpp)
LIST(APPEND HEADERS_PRIVATE ${HEADERS_QTHREAD} )
LIST(APPEND SOURCES ${SOURCES_QTHREAD} )
LIST(APPEND HEADERS_PRIVATE ${HEADERS_QTHREADS} )
LIST(APPEND SOURCES ${SOURCES_QTHREADS} )
INSTALL(FILES ${HEADERS_QTHREAD} DESTINATION ${TRILINOS_INCDIR}/Qthread/)
INSTALL(FILES ${HEADERS_QTHREADS} DESTINATION ${TRILINOS_INCDIR}/Qthreads/)
#-----------------------------------------------------------------------------
@ -109,5 +109,3 @@ TRIBITS_ADD_LIBRARY(
SOURCES ${SOURCES}
DEPLIBS
)

Some files were not shown because too many files have changed in this diff Show More