diff --git a/doc/Section_commands.html b/doc/Section_commands.html index 70bc1e8857..5f996268de 100644 --- a/doc/Section_commands.html +++ b/doc/Section_commands.html @@ -399,12 +399,13 @@ potentials. Click on the style itself for a full description: lj/charmm/coul/long/gpulj/charmm/coul/long/optlj/class2lj/class2/coul/cut lj/class2/coul/longlj/cutlj/cut/gpulj/cut/opt lj/cut/coul/cutlj/cut/coul/cut/gpulj/cut/coul/debyelj/cut/coul/long -lj/cut/coul/long/gpulj/cut/coul/long/tip4plj/expandlj/gromacs -lj/gromacs/coul/gromacslj/smoothlj96/cutlj96/cut/gpu -lubricatemeammorsemorse/opt -peri/lpsperi/pmbreaxresquared -softswtabletersoff -tersoff/zblyukawayukawa/colloid +lj/cut/coul/long/gpulj/cut/coul/long/tip4plj/expandlj/expand/gpu +lj/gromacslj/gromacs/coul/gromacslj/smoothlj96/cut +lj96/cut/gpulubricatemeammorse +morse/gpumorse/optperi/lpsperi/pmb +reaxresquaredsoftsw +tabletersofftersoff/zblyukawa +yukawa/colloid

These are pair styles contributed by users, which can be used if @@ -483,7 +484,8 @@ description: Kspace solvers. Click on the style itself for a full description:

- +
ewaldpppmpppm/tip4p +
ewaldpppmpppm/gpu/singlepppm/gpu/double
pppm/tip4p

These are Kspace solvers contributed by users, which can be used if diff --git a/doc/Section_commands.txt b/doc/Section_commands.txt index 1a18b8b9b2..1c58401303 100644 --- a/doc/Section_commands.txt +++ b/doc/Section_commands.txt @@ -611,6 +611,7 @@ potentials. Click on the style itself for a full description: "lj/cut/coul/long/gpu"_pair_lj.html, "lj/cut/coul/long/tip4p"_pair_lj.html, "lj/expand"_pair_lj_expand.html, +"lj/expand/gpu"_pair_lj_expand.html, "lj/gromacs"_pair_gromacs.html, "lj/gromacs/coul/gromacs"_pair_gromacs.html, "lj/smooth"_pair_lj_smooth.html, @@ -619,6 +620,7 @@ potentials. Click on the style itself for a full description: "lubricate"_pair_lubricate.html, "meam"_pair_meam.html, "morse"_pair_morse.html, +"morse/gpu"_pair_morse.html, "morse/opt"_pair_morse.html, "peri/lps"_pair_peri.html, "peri/pmb"_pair_peri.html, @@ -728,6 +730,8 @@ Kspace solvers. Click on the style itself for a full description: "ewald"_kspace_style.html, "pppm"_kspace_style.html, +"pppm/gpu/single"_kspace_style.html, +"pppm/gpu/double"_kspace_style.html, "pppm/tip4p"_kspace_style.html :tb(c=4,ea=c,w=100) These are Kspace solvers contributed by users, which can be used if diff --git a/doc/Section_errors.html b/doc/Section_errors.html index 8cd9ed6e46..b90784e0c0 100644 --- a/doc/Section_errors.html +++ b/doc/Section_errors.html @@ -173,6 +173,10 @@ the bond topologies you have defined. neighbors for each atom. This likely means something is wrong with the bond topologies you have defined. +

Accelerated style in input script but no fix gpu + +
GPU acceleration requires fix gpu in the input script. +
All angle coeffs are not set
All angle coefficients must be set in the data file or by the @@ -1240,9 +1244,9 @@ non-periodic z dimension. unless you use the kspace_modify command to define a 2d slab with a non-periodic z dimension. -
Cannot use pair hybrid with multiple GPU pair styles +
Cannot use pair hybrid with GPU neighbor builds -
Self-explanatory. +
See documentation for fix gpu.
Cannot use pair tail corrections with 2d simulations @@ -1843,7 +1847,7 @@ does not exist.
Self-explanatory. -
Could not find or initialize a specified accelerator device +
Could not find/initialize a specified accelerator device
Your GPU setup is invalid. @@ -2123,6 +2127,10 @@ model. used. Most likely, one or more atoms have been blown out of the simulation box to a great distance. +
Double precision is not supported on this accelerator. + +
In this case, you must compile the GPU library for single precision. +
Dump cfg and fix not computed at compatible times
The fix must produce per-atom quantities on timesteps that dump cfg @@ -2355,6 +2363,10 @@ smaller simulation or on more processors.
Self-explanatory. +
Fix gpu split must be positive for hybrid pair styles. + +
See documentation for fix gpu. +
Fix ID for compute atom/molecule does not exist
Self-explanatory. @@ -3227,6 +3239,11 @@ this fix.
This is the way the fix must be defined in your input script. +
GPU library not compiled for this accelerator + +
The GPU library was not built for your accelerator. Check the arch flag in +lib/gpu. +
Gmask function in equal-style variable formula
Gmask is per-atom operation. @@ -3509,7 +3526,7 @@ simulation box.
Eigensolve for rigid body was not sufficiently accurate. -
Insufficient memory on accelerator (or no fix gpu) +
Insufficient memory on accelerator.
Self-explanatory. @@ -4587,10 +4604,6 @@ contain the same atom.
Any rigid body defined by the fix rigid command must contain 2 or more atoms. -
Out of memory on GPGPU - -
You are attempting to run with too many atoms on the GPU. -
Out of range atoms - cannot compute PPPM
One or more atoms are attempting to map their charge to a PPPM grid diff --git a/doc/Section_errors.txt b/doc/Section_errors.txt index d94e8a9be7..0e2b2e804b 100644 --- a/doc/Section_errors.txt +++ b/doc/Section_errors.txt @@ -170,6 +170,10 @@ An inconsistency was detected when computing the number of 1-4 neighbors for each atom. This likely means something is wrong with the bond topologies you have defined. :dd +{Accelerated style in input script but no fix gpu} :dt + +GPU acceleration requires fix gpu in the input script. :dd + {All angle coeffs are not set} :dt All angle coefficients must be set in the data file or by the @@ -1237,9 +1241,9 @@ For kspace style pppm, all 3 dimensions must have periodic boundaries unless you use the kspace_modify command to define a 2d slab with a non-periodic z dimension. :dd -{Cannot use pair hybrid with multiple GPU pair styles} :dt +{Cannot use pair hybrid with GPU neighbor builds} :dt -Self-explanatory. :dd +See documentation for fix gpu. :dd {Cannot use pair tail corrections with 2d simulations} :dt @@ -1840,7 +1844,7 @@ The compute ID for computing temperature does not exist. :dd Self-explanatory. :dd -{Could not find or initialize a specified accelerator device} :dt +{Could not find/initialize a specified accelerator device} :dt Your GPU setup is invalid. :dd @@ -2120,6 +2124,10 @@ The domain has become extremely large so that neighbor bins cannot be used. Most likely, one or more atoms have been blown out of the simulation box to a great distance. :dd +{Double precision is not supported on this accelerator.} :dt + +In this case, you must compile the GPU library for single precision. :dd + {Dump cfg and fix not computed at compatible times} :dt The fix must produce per-atom quantities on timesteps that dump cfg @@ -2352,6 +2360,10 @@ This is not allowed. Make your SRD bin size smaller. :dd Self-explanatory. :dd +{Fix gpu split must be positive for hybrid pair styles.} :dt + +See documentation for fix gpu. :dd + {Fix ID for compute atom/molecule does not exist} :dt Self-explanatory. :dd @@ -3224,6 +3236,11 @@ When using a "*" in the restart file name, no matching file was found. :dd This is the way the fix must be defined in your input script. :dd +{GPU library not compiled for this accelerator} :dt + +The GPU library was not built for your accelerator. Check the arch flag in +lib/gpu. :dd + {Gmask function in equal-style variable formula} :dt Gmask is per-atom operation. :dd @@ -3506,7 +3523,7 @@ Eigensolve for rigid body was not sufficiently accurate. :dd Eigensolve for rigid body was not sufficiently accurate. :dd -{Insufficient memory on accelerator (or no fix gpu)} :dt +{Insufficient memory on accelerator. } :dt Self-explanatory. :dd @@ -4584,10 +4601,6 @@ contain the same atom. :dd Any rigid body defined by the fix rigid command must contain 2 or more atoms. :dd -{Out of memory on GPGPU} :dt - -You are attempting to run with too many atoms on the GPU. :dd - {Out of range atoms - cannot compute PPPM} :dt One or more atoms are attempting to map their charge to a PPPM grid diff --git a/doc/Section_intro.html b/doc/Section_intro.html index f9b00bb689..bce1a9d718 100644 --- a/doc/Section_intro.html +++ b/doc/Section_intro.html @@ -505,6 +505,14 @@ the list.
+ + + + + + + + diff --git a/doc/Section_intro.txt b/doc/Section_intro.txt index e4c26c8aab..a8e46df996 100644 --- a/doc/Section_intro.txt +++ b/doc/Section_intro.txt @@ -490,6 +490,14 @@ the list. :link(sjp,http://www.sandia.gov/~sjplimp) +pppm GPU single and double : Mike Brown (ORNL) +pair_style lj/cut/expand : Inderaj Bains (NVIDIA) +temperature accelerated dynamics (TAD) : Aidan Thompson (Sandia) +pair reax/c and fix qeq/reax : Metin Aktulga (Purdue, now LBNL) +DREIDING force field, pair_style hbond/dreiding, etc : Tod Pascal (CalTech) +fix adapt and compute ti for thermodynamic integreation for free energies : Sai Jayaraman (Sandia) +pair born and pair gauss : Sai Jayaraman (Sandia) +stochastic rotation dynamics (SRD) via fix srd : Jemery Lechman (Sandia) and Pieter in 't Veld (BASF) ipp Perl script tool : Reese Jones (Sandia) eam_database and createatoms tools : Xiaowang Zhou (Sandia) electron force field (eFF) : Andres Jaramillo-Botero and Julius Su (Caltech) diff --git a/doc/Section_start.html b/doc/Section_start.html index a83aaa0ad5..08287e3377 100644 --- a/doc/Section_start.html +++ b/doc/Section_start.html @@ -994,143 +994,130 @@ processing units (GPUs). We plan to add more over time. Currently, they only support NVIDIA GPU cards. To use them you need to install certain NVIDIA CUDA software on your system:

-
pppm GPU single and double Mike Brown (ORNL)
pair_style lj/cut/expand Inderaj Bains (NVIDIA)
temperature accelerated dynamics (TAD) Aidan Thompson (Sandia)
pair reax/c and fix qeq/reax Metin Aktulga (Purdue, now LBNL)
DREIDING force field, pair_style hbond/dreiding, etc Tod Pascal (CalTech)
fix adapt and compute ti for thermodynamic integreation for free energies Sai Jayaraman (Sandia)
pair born and pair gauss Sai Jayaraman (Sandia)
stochastic rotation dynamics (SRD) via fix srd Jemery Lechman (Sandia) and Pieter in 't Veld (BASF)
ipp Perl script tool Reese Jones (Sandia)
eam_database and createatoms tools Xiaowang Zhou (Sandia)
electron force field (eFF) Andres Jaramillo-Botero and Julius Su (Caltech)