2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
5 * Copyright (c) 2001-2012, The GROMACS development team,
6 * check out http://www.gromacs.org for more information.
7 * Copyright (c) 2012,2013, by the GROMACS development team, led by
8 * David van der Spoel, Berk Hess, Erik Lindahl, and including many
9 * others, as listed in the AUTHORS file in the top-level source
10 * directory and at http://www.gromacs.org.
12 * GROMACS is free software; you can redistribute it and/or
13 * modify it under the terms of the GNU Lesser General Public License
14 * as published by the Free Software Foundation; either version 2.1
15 * of the License, or (at your option) any later version.
17 * GROMACS is distributed in the hope that it will be useful,
18 * but WITHOUT ANY WARRANTY; without even the implied warranty of
19 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
20 * Lesser General Public License for more details.
22 * You should have received a copy of the GNU Lesser General Public
23 * License along with GROMACS; if not, see
24 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
25 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
27 * If you want to redistribute modifications to GROMACS, please
28 * consider that scientific software is very special. Version
29 * control is crucial - bugs must be traceable. We will be happy to
30 * consider code for inclusion in the official distribution, but
31 * derived work must not be called official GROMACS. Details are found
32 * in the README & COPYING files - if they are missing, get the
33 * official version at http://www.gromacs.org.
35 * To help us fund GROMACS development, we humbly ask that you cite
36 * the research papers on the package. Check out http://www.gromacs.org.
39 #ifndef NBNXN_CUDA_TYPES_H
40 #define NBNXN_CUDA_TYPES_H
42 #include "types/nbnxn_pairlist.h"
43 #include "types/nbnxn_cuda_types_ext.h"
44 #include "../../gmxlib/cuda_tools/cudautils.cuh"
50 /** Types of electrostatics implementations available in the CUDA non-bonded
51 * force kernels. These represent both the electrostatics types implemented
52 * by the kernels (cut-off, RF, and Ewald - a subset of what's defined in
53 * enums.h) as well as encode implementation details analytical/tabulated
54 * and single or twin cut-off (for Ewald kernels).
55 * Note that the cut-off and RF kernels have only analytical flavor and unlike
56 * in the CPU kernels, the tabulated kernels are ATM Ewald-only.
58 * The order of pointers to different electrostatic kernels defined in
59 * nbnxn_cuda.cu by the nb_default_kfunc_ptr and nb_legacy_kfunc_ptr arrays
60 * should match the order of enumerated types below. */
62 eelCuCUT, eelCuRF, eelCuEWALD_TAB, eelCuEWALD_TAB_TWIN, eelCuEWALD_ANA, eelCuEWALD_ANA_TWIN, eelCuNR
65 /** Kernel flavors with different set of optimizations: default for CUDA <=v4.1
66 * compilers and legacy for earlier, 3.2 and 4.0 CUDA compilers. */
68 eNbnxnCuKDefault, eNbnxnCuKLegacy, eNbnxnCuKNR
71 #define NBNXN_KVER_OLD(k) (k == eNbnxnCuKOld)
72 #define NBNXN_KVER_LEGACY(k) (k == eNbnxnCuKLegacy)
73 #define NBNXN_KVER_DEFAULT(k) (k == eNbnxnCuKDefault)
75 /* Non-bonded kernel versions. */
77 /* All structs prefixed with "cu_" hold data used in GPU calculations and
78 * are passed to the kernels, except cu_timers_t. */
79 typedef struct cu_plist cu_plist_t;
80 typedef struct cu_atomdata cu_atomdata_t;
81 typedef struct cu_nbparam cu_nbparam_t;
82 typedef struct cu_timers cu_timers_t;
83 typedef struct nb_staging nb_staging_t;
86 /** Staging area for temporary data. The energies get downloaded here first,
87 * before getting added to the CPU-side aggregate values.
91 float *e_lj; /**< LJ energy */
92 float *e_el; /**< electrostatic energy */
93 float3 *fshift; /**< shift forces */
96 /** Nonbonded atom data -- both inputs and outputs. */
99 int natoms; /**< number of atoms */
100 int natoms_local; /**< number of local atoms */
101 int nalloc; /**< allocation size for the atom data (xq, f) */
103 float4 *xq; /**< atom coordinates + charges, size natoms */
104 float3 *f; /**< force output array, size natoms */
105 /* TODO: try float2 for the energies */
106 float *e_lj, /**< LJ energy output, size 1 */
107 *e_el; /**< Electrostatics energy input, size 1 */
109 float3 *fshift; /**< shift forces */
111 int ntypes; /**< number of atom types */
112 int *atom_types; /**< atom type indices, size natoms */
114 float3 *shift_vec; /**< shifts */
115 bool bShiftVecUploaded; /**< true if the shift vector has been uploaded */
118 /** Parameters required for the CUDA nonbonded calculations. */
121 int eeltype; /**< type of electrostatics */
123 float epsfac; /**< charge multiplication factor */
124 float c_rf, /**< Reaction-field/plain cutoff electrostatics const. */
125 two_k_rf; /**< Reaction-field electrostatics constant */
126 float ewald_beta; /**< Ewald/PME parameter */
127 float sh_ewald; /**< Ewald/PME correction term */
128 float rvdw_sq; /**< VdW cut-off */
129 float rcoulomb_sq; /**< Coulomb cut-off */
130 float rlist_sq; /**< pair-list cut-off */
131 float sh_invrc6; /**< LJ potential correction term */
133 float *nbfp; /**< nonbonded parameter table with C6/C12 pairs */
135 /* Ewald Coulomb force table data */
136 int coulomb_tab_size; /**< table size (s.t. it fits in texture cache) */
137 float coulomb_tab_scale; /**< table scale/spacing */
138 float *coulomb_tab; /**< pointer to the table in the device memory */
141 /** Pair list data */
144 int na_c; /**< number of atoms per cluster */
146 int nsci; /**< size of sci, # of i clusters in the list */
147 int sci_nalloc; /**< allocation size of sci */
148 nbnxn_sci_t *sci; /**< list of i-cluster ("super-clusters") */
150 int ncj4; /**< total # of 4*j clusters */
151 int cj4_nalloc; /**< allocation size of cj4 */
152 nbnxn_cj4_t *cj4; /**< 4*j cluster list, contains j cluster number
153 and index into the i cluster list */
154 nbnxn_excl_t *excl; /**< atom interaction bits */
155 int nexcl; /**< count for excl */
156 int excl_nalloc; /**< allocation size of excl */
158 bool bDoPrune; /**< true if pair-list pruning needs to be
159 done during the current step */
162 /** CUDA events used for timing GPU kernels and H2D/D2H transfers.
163 * The two-sized arrays hold the local and non-local values and should always
164 * be indexed with eintLocal/eintNonlocal.
168 cudaEvent_t start_atdat; /**< start event for atom data transfer (every PS step) */
169 cudaEvent_t stop_atdat; /**< stop event for atom data transfer (every PS step) */
170 cudaEvent_t start_nb_h2d[2]; /**< start events for x/q H2D transfers (l/nl, every step) */
171 cudaEvent_t stop_nb_h2d[2]; /**< stop events for x/q H2D transfers (l/nl, every step) */
172 cudaEvent_t start_nb_d2h[2]; /**< start events for f D2H transfer (l/nl, every step) */
173 cudaEvent_t stop_nb_d2h[2]; /**< stop events for f D2H transfer (l/nl, every step) */
174 cudaEvent_t start_pl_h2d[2]; /**< start events for pair-list H2D transfers (l/nl, every PS step) */
175 cudaEvent_t stop_pl_h2d[2]; /**< start events for pair-list H2D transfers (l/nl, every PS step) */
176 cudaEvent_t start_nb_k[2]; /**< start event for non-bonded kernels (l/nl, every step) */
177 cudaEvent_t stop_nb_k[2]; /**< stop event non-bonded kernels (l/nl, every step) */
180 /** Main data structure for CUDA nonbonded force calculations. */
183 cuda_dev_info_t *dev_info; /**< CUDA device information */
184 int kernel_ver; /**< The version of the kernel to be executed on the
185 device in use, possible values: eNbnxnCuK* */
186 bool bUseTwoStreams; /**< true if doing both local/non-local NB work on GPU */
187 bool bUseStreamSync; /**< true if the standard cudaStreamSynchronize is used
188 and not memory polling-based waiting */
189 cu_atomdata_t *atdat; /**< atom data */
190 cu_nbparam_t *nbparam; /**< parameters required for the non-bonded calc. */
191 cu_plist_t *plist[2]; /**< pair-list data structures (local and non-local) */
192 nb_staging_t nbst; /**< staging area where fshift/energies get downloaded */
194 cudaStream_t stream[2]; /**< local and non-local GPU streams */
196 /** events used for synchronization */
197 cudaEvent_t nonlocal_done; /**< event triggered when the non-local non-bonded kernel
198 is done (and the local transfer can proceed) */
199 cudaEvent_t misc_ops_done; /**< event triggered when the operations that precede the
200 main force calculations are done (e.g. buffer 0-ing) */
202 /* NOTE: With current CUDA versions (<=5.0) timing doesn't work with multiple
203 * concurrent streams, so we won't time if both l/nl work is done on GPUs.
204 * Timer init/uninit is still done even with timing off so only the condition
205 * setting bDoTime needs to be change if this CUDA "feature" gets fixed. */
206 bool bDoTime; /**< True if event-based timing is enabled. */
207 cu_timers_t *timers; /**< CUDA event-based timers. */
208 wallclock_gpu_t *timings; /**< Timing data. */
215 #endif /* NBNXN_CUDA_TYPES_H */