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

Did some more cleanups

Browse Source
This commit is contained in:
Trung Nguyen
2019-04-17 12:04:31 -05:00
parent 1f43efc111
commit 4a4297591e
2 changed files with 14 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
lib/gpu/lal_device.cpp
View File

@ -634,7 +634,7 @@ void DeviceT::output_kspace_times(UCL_Timer &time_in,
if (screen && times[6]>0.0) { if (screen && times[6]>0.0) {
fprintf(screen,"\n\n-------------------------------------"); fprintf(screen,"\n\n-------------------------------------");
fprintf(screen,"--------------------------------\n"); fprintf(screen,"--------------------------------\n");
fprintf(screen," Device Time Info (average): "); fprintf(screen," Device Time Info (average) for kspace: ");
fprintf(screen,"\n-------------------------------------"); fprintf(screen,"\n-------------------------------------");
fprintf(screen,"--------------------------------\n"); fprintf(screen,"--------------------------------\n");

24
lib/gpu/lal_device.h
View File

@ -57,8 +57,9 @@ class Device {
/** \param charge True if charges need to be stored /** \param charge True if charges need to be stored
* \param rot True if quaternions need to be stored * \param rot True if quaternions need to be stored
* \param nlocal Total number of local particles to allocate memory for * \param nlocal Total number of local particles to allocate memory for
* \param host_nlocal Initial number of host particles to allocate memory for
* \param nall Total number of local+ghost particles * \param nall Total number of local+ghost particles
* \param maxspecial Maximum mumber of special bonded atoms per atom
* \param vel True if velocities need to be stored
* *
* Returns: * Returns:
* - 0 if successfull * - 0 if successfull
@ -66,7 +67,7 @@ class Device {
* - -3 if there is an out of memory error * - -3 if there is an out of memory error
* - -4 if the GPU library was not compiled for GPU * - -4 if the GPU library was not compiled for GPU
* - -5 Double precision is not supported on card **/ * - -5 Double precision is not supported on card **/
int init(Answer<numtyp,acctyp> &a, const bool charge, const bool rot, int init(Answer<numtyp,acctyp> &ans, const bool charge, const bool rot,
const int nlocal, const int nall, const int maxspecial, const int nlocal, const int nall, const int maxspecial,
const bool vel=false); const bool vel=false);
@ -82,12 +83,13 @@ class Device {
* - -5 Double precision is not supported on card **/ * - -5 Double precision is not supported on card **/
int init(Answer<numtyp,acctyp> &ans, const int nlocal, const int nall); int init(Answer<numtyp,acctyp> &ans, const int nlocal, const int nall);
/// Initialize neighbor list storage and build /// Initialize the neighbor list storage
/** \param charge True if charges need to be stored /** \param charge True if charges need to be stored
* \param rot True if quaternions need to be stored * \param rot True if quaternions need to be stored
* \param nlocal Total number of local particles to allocate memory for * \param nlocal Total number of local particles to allocate memory for
* \param host_nlocal Initial number of host particles to allocate memory for * \param host_nlocal Initial number of host particles to allocate memory for
* \param nall Total number of local+ghost particles * \param nall Total number of local+ghost particles
* \param maxspecial Maximum mumber of special bonded atoms per atom
* \param gpu_host 0 if host will not perform force calculations, * \param gpu_host 0 if host will not perform force calculations,
* 1 if gpu_nbor is true, and host needs a half nbor list, * 1 if gpu_nbor is true, and host needs a half nbor list,
* 2 if gpu_nbor is true, and host needs a full nbor list * 2 if gpu_nbor is true, and host needs a full nbor list
@ -104,10 +106,10 @@ class Device {
* - -4 if the GPU library was not compiled for GPU * - -4 if the GPU library was not compiled for GPU
* - -5 Double precision is not supported on card **/ * - -5 Double precision is not supported on card **/
int init_nbor(Neighbor *nbor, const int nlocal, int init_nbor(Neighbor *nbor, const int nlocal,
const int host_nlocal, const int nall, const int host_nlocal, const int nall,
const int maxspecial, const int gpu_host, const int maxspecial, const int gpu_host,
const int max_nbors, const double cell_size, const int max_nbors, const double cell_size,
const bool pre_cut, const int threads_per_atom); const bool pre_cut, const int threads_per_atom);
/// Output a message for pair_style acceleration with device stats /// Output a message for pair_style acceleration with device stats
void init_message(FILE *screen, const char *name, void init_message(FILE *screen, const char *name,
@ -191,7 +193,7 @@ class Device {
/// Return host memory usage in bytes /// Return host memory usage in bytes
double host_memory_usage() const; double host_memory_usage() const;
/// Return the number of procs sharing a device (size of device commincator) /// Return the number of procs sharing a device (size of device communicator)
inline int procs_per_gpu() const { return _procs_per_gpu; } inline int procs_per_gpu() const { return _procs_per_gpu; }
/// Return the number of threads per proc /// Return the number of threads per proc
inline int num_threads() const { return _nthreads; } inline int num_threads() const { return _nthreads; }
@ -278,12 +280,12 @@ class Device {
/// Atom Data /// Atom Data
Atom<numtyp,acctyp> atom; Atom<numtyp,acctyp> atom;
// --------------------------- NBOR DATA ---------------------------- // --------------------------- NBOR SHARED KERNELS ----------------
/// Neighbor Data /// Shared kernels for neighbor lists
NeighborShared _neighbor_shared; NeighborShared _neighbor_shared;
// ------------------------ LONG RANGE DATA ------------------------- // ------------------------ LONG RANGE DATA -----------------------
// Long Range Data // Long Range Data
int _long_range_precompute; int _long_range_precompute;