src/gromacs/nbnxm/opencl/nbnxm_ocl.cpp

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014,2015,2016 by the GROMACS development team.
   5  * Copyright (c) 2017,2018,2019,2020,2021, by the GROMACS development team, led by
   6  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   7  * and including many others, as listed in the AUTHORS file in the
   8  * top-level source directory and at http://www.gromacs.org.
   9  *
  10  * GROMACS is free software; you can redistribute it and/or
  11  * modify it under the terms of the GNU Lesser General Public License
  12  * as published by the Free Software Foundation; either version 2.1
  13  * of the License, or (at your option) any later version.
  14  *
  15  * GROMACS is distributed in the hope that it will be useful,
  16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  18  * Lesser General Public License for more details.
  19  *
  20  * You should have received a copy of the GNU Lesser General Public
  21  * License along with GROMACS; if not, see
  22  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  23  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  24  *
  25  * If you want to redistribute modifications to GROMACS, please
  26  * consider that scientific software is very special. Version
  27  * control is crucial - bugs must be traceable. We will be happy to
  28  * consider code for inclusion in the official distribution, but
  29  * derived work must not be called official GROMACS. Details are found
  30  * in the README & COPYING files - if they are missing, get the
  31  * official version at http://www.gromacs.org.
  32  *
  33  * To help us fund GROMACS development, we humbly ask that you cite
  34  * the research papers on the package. Check out http://www.gromacs.org.
  35  */
  36 /*! \internal \file
  37  *  \brief Define OpenCL implementation of nbnxm_gpu.h
  38  *
  39  *  \author Anca Hamuraru <anca@streamcomputing.eu>
  40  *  \author Teemu Virolainen <teemu@streamcomputing.eu>
  41  *  \author Dimitrios Karkoulis <dimitris.karkoulis@gmail.com>
  42  *  \author Szilárd Páll <pall.szilard@gmail.com>
  43  *  \ingroup module_nbnxm
  44  *
  45  *  TODO (psz):
  46  *  - Add a static const cl_uint c_pruneKernelWorkDim / c_nbnxnKernelWorkDim = 3;
  47  *  - Rework the copying of OCL data structures done before every invocation of both
  48  *    nb and prune kernels (using fillin_ocl_structures); also consider at the same
  49  *    time calling clSetKernelArg only on the updated parameters (if tracking changed
  50  *    parameters is feasible);
  51  *  - Consider using the event_wait_list argument to clEnqueueNDRangeKernel to mark
  52  *    dependencies on the kernel launched: e.g. the non-local nb kernel's dependency
  53  *    on the misc_ops_and_local_H2D_done event could be better expressed this way.
  54  *
  55  *  - Consider extracting common sections of the OpenCL and CUDA nbnxn logic, e.g:
  56  *    - in nbnxn_gpu_launch_kernel_pruneonly() the pre- and post-kernel launch logic
  57  *      is identical in the two implementations, so a 3-way split might allow sharing
  58  *      code;
  59  *    -
  60  *
  61  */
  62 #include "gmxpre.h"
  63
  64 #include <assert.h>
  65 #include <stdlib.h>
  66
  67 #if defined(_MSVC)
  68 #    include <limits>
  69 #endif
  70
  71 #include "gromacs/gpu_utils/device_context.h"
  72 #include "gromacs/gpu_utils/gputraits_ocl.h"
  73 #include "gromacs/gpu_utils/oclutils.h"
  74 #include "gromacs/hardware/device_information.h"
  75 #include "gromacs/hardware/hw_info.h"
  76 #include "gromacs/mdtypes/simulation_workload.h"
  77 #include "gromacs/nbnxm/atomdata.h"
  78 #include "gromacs/nbnxm/gpu_common.h"
  79 #include "gromacs/nbnxm/gpu_common_utils.h"
  80 #include "gromacs/nbnxm/gpu_data_mgmt.h"
  81 #include "gromacs/nbnxm/nbnxm.h"
  82 #include "gromacs/nbnxm/nbnxm_gpu.h"
  83 #include "gromacs/nbnxm/pairlist.h"
  84 #include "gromacs/pbcutil/ishift.h"
  85 #include "gromacs/timing/gpu_timing.h"
  86 #include "gromacs/utility/cstringutil.h"
  87 #include "gromacs/utility/fatalerror.h"
  88 #include "gromacs/utility/gmxassert.h"
  89
  90 #include "nbnxm_ocl_types.h"
  91
  92 namespace Nbnxm
  93 {
  94
  95 /*! \brief Convenience constants */
  96 //@{
  97 static constexpr int c_clSize = c_nbnxnGpuClusterSize;
  98 //@}
  99
 100
 101 /*! \brief Validates the input global work size parameter.
 102  */
 103 static inline void validate_global_work_size(const KernelLaunchConfig& config,
 104                                              int                       work_dim,
 105                                              const DeviceInformation*  dinfo)
 106 {
 107     cl_uint device_size_t_size_bits;
 108     cl_uint host_size_t_size_bits;
 109
 110     GMX_ASSERT(dinfo, "Need a valid device info object");
 111
 112     size_t global_work_size[3];
 113     GMX_ASSERT(work_dim <= 3, "Not supporting hyper-grids just yet");
 114     for (int i = 0; i < work_dim; i++)
 115     {
 116         global_work_size[i] = config.blockSize[i] * config.gridSize[i];
 117     }
 118
 119     /* Each component of a global_work_size must not exceed the range given by the
 120        sizeof(device size_t) for the device on which the kernel execution will
 121        be enqueued. See:
 122        https://www.khronos.org/registry/cl/sdk/1.0/docs/man/xhtml/clEnqueueNDRangeKernel.html
 123      */
 124     device_size_t_size_bits = dinfo->adress_bits;
 125     host_size_t_size_bits   = static_cast<cl_uint>(sizeof(size_t) * 8);
 126
 127     /* If sizeof(host size_t) <= sizeof(device size_t)
 128             => global_work_size components will always be valid
 129        else
 130             => get device limit for global work size and
 131             compare it against each component of global_work_size.
 132      */
 133     if (host_size_t_size_bits > device_size_t_size_bits)
 134     {
 135         size_t device_limit;
 136
 137         device_limit = (1ULL << device_size_t_size_bits) - 1;
 138
 139         for (int i = 0; i < work_dim; i++)
 140         {
 141             if (global_work_size[i] > device_limit)
 142             {
 143                 gmx_fatal(
 144                         FARGS,
 145                         "Watch out, the input system is too large to simulate!\n"
 146                         "The number of nonbonded work units (=number of super-clusters) exceeds the"
 147                         "device capabilities. Global work size limit exceeded (%zu > %zu)!",
 148                         global_work_size[i],
 149                         device_limit);
 150             }
 151         }
 152     }
 153 }
 154
 155 /* Constant arrays listing non-bonded kernel function names. The arrays are
 156  * organized in 2-dim arrays by: electrostatics and VDW type.
 157  *
 158  *  Note that the row- and column-order of function pointers has to match the
 159  *  order of corresponding enumerated electrostatics and vdw types, resp.,
 160  *  defined in nbnxm_ocl_types.h.
 161  */
 162
 163 /*! \brief Force-only kernel function names. */
 164 static const char* nb_kfunc_noener_noprune_ptr[c_numElecTypes][c_numVdwTypes] = {
 165     { "nbnxn_kernel_ElecCut_VdwLJ_F_opencl",
 166       "nbnxn_kernel_ElecCut_VdwLJCombGeom_F_opencl",
 167       "nbnxn_kernel_ElecCut_VdwLJCombLB_F_opencl",
 168       "nbnxn_kernel_ElecCut_VdwLJFsw_F_opencl",
 169       "nbnxn_kernel_ElecCut_VdwLJPsw_F_opencl",
 170       "nbnxn_kernel_ElecCut_VdwLJEwCombGeom_F_opencl",
 171       "nbnxn_kernel_ElecCut_VdwLJEwCombLB_F_opencl" },
 172     { "nbnxn_kernel_ElecRF_VdwLJ_F_opencl",
 173       "nbnxn_kernel_ElecRF_VdwLJCombGeom_F_opencl",
 174       "nbnxn_kernel_ElecRF_VdwLJCombLB_F_opencl",
 175       "nbnxn_kernel_ElecRF_VdwLJFsw_F_opencl",
 176       "nbnxn_kernel_ElecRF_VdwLJPsw_F_opencl",
 177       "nbnxn_kernel_ElecRF_VdwLJEwCombGeom_F_opencl",
 178       "nbnxn_kernel_ElecRF_VdwLJEwCombLB_F_opencl" },
 179     { "nbnxn_kernel_ElecEwQSTab_VdwLJ_F_opencl",
 180       "nbnxn_kernel_ElecEwQSTab_VdwLJCombGeom_F_opencl",
 181       "nbnxn_kernel_ElecEwQSTab_VdwLJCombLB_F_opencl",
 182       "nbnxn_kernel_ElecEwQSTab_VdwLJFsw_F_opencl",
 183       "nbnxn_kernel_ElecEwQSTab_VdwLJPsw_F_opencl",
 184       "nbnxn_kernel_ElecEwQSTab_VdwLJEwCombGeom_F_opencl",
 185       "nbnxn_kernel_ElecEwQSTab_VdwLJEwCombLB_F_opencl" },
 186     { "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJ_F_opencl",
 187       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJCombGeom_F_opencl",
 188       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJCombLB_F_opencl",
 189       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJFsw_F_opencl",
 190       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJPsw_F_opencl",
 191       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJEwCombGeom_F_opencl",
 192       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJEwCombLB_F_opencl" },
 193     { "nbnxn_kernel_ElecEw_VdwLJ_F_opencl",
 194       "nbnxn_kernel_ElecEw_VdwLJCombGeom_F_opencl",
 195       "nbnxn_kernel_ElecEw_VdwLJCombLB_F_opencl",
 196       "nbnxn_kernel_ElecEw_VdwLJFsw_F_opencl",
 197       "nbnxn_kernel_ElecEw_VdwLJPsw_F_opencl",
 198       "nbnxn_kernel_ElecEw_VdwLJEwCombGeom_F_opencl",
 199       "nbnxn_kernel_ElecEw_VdwLJEwCombLB_F_opencl" },
 200     { "nbnxn_kernel_ElecEwTwinCut_VdwLJ_F_opencl",
 201       "nbnxn_kernel_ElecEwTwinCut_VdwLJCombGeom_F_opencl",
 202       "nbnxn_kernel_ElecEwTwinCut_VdwLJCombLB_F_opencl",
 203       "nbnxn_kernel_ElecEwTwinCut_VdwLJFsw_F_opencl",
 204       "nbnxn_kernel_ElecEwTwinCut_VdwLJPsw_F_opencl",
 205       "nbnxn_kernel_ElecEwTwinCut_VdwLJEwCombGeom_F_opencl",
 206       "nbnxn_kernel_ElecEwTwinCut_VdwLJEwCombLB_F_opencl" }
 207 };
 208
 209 /*! \brief Force + energy kernel function pointers. */
 210 static const char* nb_kfunc_ener_noprune_ptr[c_numElecTypes][c_numVdwTypes] = {
 211     { "nbnxn_kernel_ElecCut_VdwLJ_VF_opencl",
 212       "nbnxn_kernel_ElecCut_VdwLJCombGeom_VF_opencl",
 213       "nbnxn_kernel_ElecCut_VdwLJCombLB_VF_opencl",
 214       "nbnxn_kernel_ElecCut_VdwLJFsw_VF_opencl",
 215       "nbnxn_kernel_ElecCut_VdwLJPsw_VF_opencl",
 216       "nbnxn_kernel_ElecCut_VdwLJEwCombGeom_VF_opencl",
 217       "nbnxn_kernel_ElecCut_VdwLJEwCombLB_VF_opencl" },
 218     { "nbnxn_kernel_ElecRF_VdwLJ_VF_opencl",
 219       "nbnxn_kernel_ElecRF_VdwLJCombGeom_VF_opencl",
 220       "nbnxn_kernel_ElecRF_VdwLJCombLB_VF_opencl",
 221       "nbnxn_kernel_ElecRF_VdwLJFsw_VF_opencl",
 222       "nbnxn_kernel_ElecRF_VdwLJPsw_VF_opencl",
 223       "nbnxn_kernel_ElecRF_VdwLJEwCombGeom_VF_opencl",
 224       "nbnxn_kernel_ElecRF_VdwLJEwCombLB_VF_opencl" },
 225     { "nbnxn_kernel_ElecEwQSTab_VdwLJ_VF_opencl",
 226       "nbnxn_kernel_ElecEwQSTab_VdwLJCombGeom_VF_opencl",
 227       "nbnxn_kernel_ElecEwQSTab_VdwLJCombLB_VF_opencl",
 228       "nbnxn_kernel_ElecEwQSTab_VdwLJFsw_VF_opencl",
 229       "nbnxn_kernel_ElecEwQSTab_VdwLJPsw_VF_opencl",
 230       "nbnxn_kernel_ElecEwQSTab_VdwLJEwCombGeom_VF_opencl",
 231       "nbnxn_kernel_ElecEwQSTab_VdwLJEwCombLB_VF_opencl" },
 232     { "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJ_VF_opencl",
 233       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJCombGeom_VF_opencl",
 234       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJCombLB_VF_opencl",
 235       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJFsw_VF_opencl",
 236       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJPsw_VF_opencl",
 237       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJEwCombGeom_VF_opencl",
 238       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJEwCombLB_VF_opencl" },
 239     { "nbnxn_kernel_ElecEw_VdwLJ_VF_opencl",
 240       "nbnxn_kernel_ElecEw_VdwLJCombGeom_VF_opencl",
 241       "nbnxn_kernel_ElecEw_VdwLJCombLB_VF_opencl",
 242       "nbnxn_kernel_ElecEw_VdwLJFsw_VF_opencl",
 243       "nbnxn_kernel_ElecEw_VdwLJPsw_VF_opencl",
 244       "nbnxn_kernel_ElecEw_VdwLJEwCombGeom_VF_opencl",
 245       "nbnxn_kernel_ElecEw_VdwLJEwCombLB_VF_opencl" },
 246     { "nbnxn_kernel_ElecEwTwinCut_VdwLJ_VF_opencl",
 247       "nbnxn_kernel_ElecEwTwinCut_VdwLJCombGeom_VF_opencl",
 248       "nbnxn_kernel_ElecEwTwinCut_VdwLJCombLB_VF_opencl",
 249       "nbnxn_kernel_ElecEwTwinCut_VdwLJFsw_VF_opencl",
 250       "nbnxn_kernel_ElecEwTwinCut_VdwLJPsw_VF_opencl",
 251       "nbnxn_kernel_ElecEwTwinCut_VdwLJEwCombGeom_VF_opencl",
 252       "nbnxn_kernel_ElecEwTwinCut_VdwLJEwCombLB_VF_opencl" }
 253 };
 254
 255 /*! \brief Force + pruning kernel function pointers. */
 256 static const char* nb_kfunc_noener_prune_ptr[c_numElecTypes][c_numVdwTypes] = {
 257     { "nbnxn_kernel_ElecCut_VdwLJ_F_prune_opencl",
 258       "nbnxn_kernel_ElecCut_VdwLJCombGeom_F_prune_opencl",
 259       "nbnxn_kernel_ElecCut_VdwLJCombLB_F_prune_opencl",
 260       "nbnxn_kernel_ElecCut_VdwLJFsw_F_prune_opencl",
 261       "nbnxn_kernel_ElecCut_VdwLJPsw_F_prune_opencl",
 262       "nbnxn_kernel_ElecCut_VdwLJEwCombGeom_F_prune_opencl",
 263       "nbnxn_kernel_ElecCut_VdwLJEwCombLB_F_prune_opencl" },
 264     { "nbnxn_kernel_ElecRF_VdwLJ_F_prune_opencl",
 265       "nbnxn_kernel_ElecRF_VdwLJCombGeom_F_prune_opencl",
 266       "nbnxn_kernel_ElecRF_VdwLJCombLB_F_prune_opencl",
 267       "nbnxn_kernel_ElecRF_VdwLJFsw_F_prune_opencl",
 268       "nbnxn_kernel_ElecRF_VdwLJPsw_F_prune_opencl",
 269       "nbnxn_kernel_ElecRF_VdwLJEwCombGeom_F_prune_opencl",
 270       "nbnxn_kernel_ElecRF_VdwLJEwCombLB_F_prune_opencl" },
 271     { "nbnxn_kernel_ElecEwQSTab_VdwLJ_F_prune_opencl",
 272       "nbnxn_kernel_ElecEwQSTab_VdwLJCombGeom_F_prune_opencl",
 273       "nbnxn_kernel_ElecEwQSTab_VdwLJCombLB_F_prune_opencl",
 274       "nbnxn_kernel_ElecEwQSTab_VdwLJFsw_F_prune_opencl",
 275       "nbnxn_kernel_ElecEwQSTab_VdwLJPsw_F_prune_opencl",
 276       "nbnxn_kernel_ElecEwQSTab_VdwLJEwCombGeom_F_prune_opencl",
 277       "nbnxn_kernel_ElecEwQSTab_VdwLJEwCombLB_F_prune_opencl" },
 278     { "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJ_F_prune_opencl",
 279       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJCombGeom_F_prune_opencl",
 280       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJCombLB_F_prune_opencl",
 281       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJFsw_F_prune_opencl",
 282       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJPsw_F_prune_opencl",
 283       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJEwCombGeom_F_prune_opencl",
 284       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJEwCombLB_F_prune_opencl" },
 285     { "nbnxn_kernel_ElecEw_VdwLJ_F_prune_opencl",
 286       "nbnxn_kernel_ElecEw_VdwLJCombGeom_F_prune_opencl",
 287       "nbnxn_kernel_ElecEw_VdwLJCombLB_F_prune_opencl",
 288       "nbnxn_kernel_ElecEw_VdwLJFsw_F_prune_opencl",
 289       "nbnxn_kernel_ElecEw_VdwLJPsw_F_prune_opencl",
 290       "nbnxn_kernel_ElecEw_VdwLJEwCombGeom_F_prune_opencl",
 291       "nbnxn_kernel_ElecEw_VdwLJEwCombLB_F_prune_opencl" },
 292     { "nbnxn_kernel_ElecEwTwinCut_VdwLJ_F_prune_opencl",
 293       "nbnxn_kernel_ElecEwTwinCut_VdwLJCombGeom_F_prune_opencl",
 294       "nbnxn_kernel_ElecEwTwinCut_VdwLJCombLB_F_prune_opencl",
 295       "nbnxn_kernel_ElecEwTwinCut_VdwLJFsw_F_prune_opencl",
 296       "nbnxn_kernel_ElecEwTwinCut_VdwLJPsw_F_prune_opencl",
 297       "nbnxn_kernel_ElecEwTwinCut_VdwLJEwCombGeom_F_prune_opencl",
 298       "nbnxn_kernel_ElecEwTwinCut_VdwLJEwCombLB_F_prune_opencl" }
 299 };
 300
 301 /*! \brief Force + energy + pruning kernel function pointers. */
 302 static const char* nb_kfunc_ener_prune_ptr[c_numElecTypes][c_numVdwTypes] = {
 303     { "nbnxn_kernel_ElecCut_VdwLJ_VF_prune_opencl",
 304       "nbnxn_kernel_ElecCut_VdwLJCombGeom_VF_prune_opencl",
 305       "nbnxn_kernel_ElecCut_VdwLJCombLB_VF_prune_opencl",
 306       "nbnxn_kernel_ElecCut_VdwLJFsw_VF_prune_opencl",
 307       "nbnxn_kernel_ElecCut_VdwLJPsw_VF_prune_opencl",
 308       "nbnxn_kernel_ElecCut_VdwLJEwCombGeom_VF_prune_opencl",
 309       "nbnxn_kernel_ElecCut_VdwLJEwCombLB_VF_prune_opencl" },
 310     { "nbnxn_kernel_ElecRF_VdwLJ_VF_prune_opencl",
 311       "nbnxn_kernel_ElecRF_VdwLJCombGeom_VF_prune_opencl",
 312       "nbnxn_kernel_ElecRF_VdwLJCombLB_VF_prune_opencl",
 313       "nbnxn_kernel_ElecRF_VdwLJFsw_VF_prune_opencl",
 314       "nbnxn_kernel_ElecRF_VdwLJPsw_VF_prune_opencl",
 315       "nbnxn_kernel_ElecRF_VdwLJEwCombGeom_VF_prune_opencl",
 316       "nbnxn_kernel_ElecRF_VdwLJEwCombLB_VF_prune_opencl" },
 317     { "nbnxn_kernel_ElecEwQSTab_VdwLJ_VF_prune_opencl",
 318       "nbnxn_kernel_ElecEwQSTab_VdwLJCombGeom_VF_prune_opencl",
 319       "nbnxn_kernel_ElecEwQSTab_VdwLJCombLB_VF_prune_opencl",
 320       "nbnxn_kernel_ElecEwQSTab_VdwLJFsw_VF_prune_opencl",
 321       "nbnxn_kernel_ElecEwQSTab_VdwLJPsw_VF_prune_opencl",
 322       "nbnxn_kernel_ElecEwQSTab_VdwLJEwCombGeom_VF_prune_opencl",
 323       "nbnxn_kernel_ElecEwQSTab_VdwLJEwCombLB_VF_prune_opencl" },
 324     { "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJ_VF_prune_opencl",
 325       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJCombGeom_VF_prune_opencl",
 326       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJCombLB_VF_prune_opencl",
 327       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJFsw_VF_prune_opencl",
 328       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJPsw_VF_prune_opencl",
 329       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJEwCombGeom_VF_prune_opencl",
 330       "nbnxn_kernel_ElecEwQSTabTwinCut_VdwLJEwCombLB_VF_prune_opencl" },
 331     { "nbnxn_kernel_ElecEw_VdwLJ_VF_prune_opencl",
 332       "nbnxn_kernel_ElecEw_VdwLJCombGeom_VF_prune_opencl",
 333       "nbnxn_kernel_ElecEw_VdwLJCombLB_VF_prune_opencl",
 334       "nbnxn_kernel_ElecEw_VdwLJFsw_VF_prune_opencl",
 335       "nbnxn_kernel_ElecEw_VdwLJPsw_VF_prune_opencl",
 336       "nbnxn_kernel_ElecEw_VdwLJEwCombGeom_VF_prune_opencl",
 337       "nbnxn_kernel_ElecEw_VdwLJEwCombLB_VF_prune_opencl" },
 338     { "nbnxn_kernel_ElecEwTwinCut_VdwLJ_VF_prune_opencl",
 339       "nbnxn_kernel_ElecEwTwinCut_VdwLJCombGeom_VF_prune_opencl",
 340       "nbnxn_kernel_ElecEwTwinCut_VdwLJCombLB_VF_prune_opencl",
 341       "nbnxn_kernel_ElecEwTwinCut_VdwLJFsw_VF_prune_opencl",
 342       "nbnxn_kernel_ElecEwTwinCut_VdwLJPsw_VF_prune_opencl",
 343       "nbnxn_kernel_ElecEwTwinCut_VdwLJEwCombGeom_VF_prune_opencl",
 344       "nbnxn_kernel_ElecEwTwinCut_VdwLJEwCombLB_VF_prune_opencl" }
 345 };
 346
 347 /*! \brief Return a pointer to the prune kernel version to be executed at the current invocation.
 348  *
 349  * \param[in] kernel_pruneonly  array of prune kernel objects
 350  * \param[in] firstPrunePass    true if the first pruning pass is being executed
 351  */
 352 static inline cl_kernel selectPruneKernel(cl_kernel kernel_pruneonly[], bool firstPrunePass)
 353 {
 354     cl_kernel* kernelPtr;
 355
 356     if (firstPrunePass)
 357     {
 358         kernelPtr = &(kernel_pruneonly[epruneFirst]);
 359     }
 360     else
 361     {
 362         kernelPtr = &(kernel_pruneonly[epruneRolling]);
 363     }
 364     // TODO: consider creating the prune kernel object here to avoid a
 365     // clCreateKernel for the rolling prune kernel if this is not needed.
 366     return *kernelPtr;
 367 }
 368
 369 /*! \brief Return a pointer to the kernel version to be executed at the current step.
 370  *  OpenCL kernel objects are cached in nb. If the requested kernel is not
 371  *  found in the cache, it will be created and the cache will be updated.
 372  */
 373 static inline cl_kernel
 374 select_nbnxn_kernel(NbnxmGpu* nb, enum ElecType elecType, enum VdwType vdwType, bool bDoEne, bool bDoPrune)
 375 {
 376     const char* kernel_name_to_run;
 377     cl_kernel*  kernel_ptr;
 378     cl_int      cl_error;
 379
 380     const int elecTypeIdx = static_cast<int>(elecType);
 381     const int vdwTypeIdx  = static_cast<int>(vdwType);
 382
 383     GMX_ASSERT(elecTypeIdx < c_numElecTypes,
 384                "The electrostatics type requested is not implemented in the OpenCL kernels.");
 385     GMX_ASSERT(vdwTypeIdx < c_numVdwTypes,
 386                "The VdW type requested is not implemented in the OpenCL kernels.");
 387
 388     if (bDoEne)
 389     {
 390         if (bDoPrune)
 391         {
 392             kernel_name_to_run = nb_kfunc_ener_prune_ptr[elecTypeIdx][vdwTypeIdx];
 393             kernel_ptr         = &(nb->kernel_ener_prune_ptr[elecTypeIdx][vdwTypeIdx]);
 394         }
 395         else
 396         {
 397             kernel_name_to_run = nb_kfunc_ener_noprune_ptr[elecTypeIdx][vdwTypeIdx];
 398             kernel_ptr         = &(nb->kernel_ener_noprune_ptr[elecTypeIdx][vdwTypeIdx]);
 399         }
 400     }
 401     else
 402     {
 403         if (bDoPrune)
 404         {
 405             kernel_name_to_run = nb_kfunc_noener_prune_ptr[elecTypeIdx][vdwTypeIdx];
 406             kernel_ptr         = &(nb->kernel_noener_prune_ptr[elecTypeIdx][vdwTypeIdx]);
 407         }
 408         else
 409         {
 410             kernel_name_to_run = nb_kfunc_noener_noprune_ptr[elecTypeIdx][vdwTypeIdx];
 411             kernel_ptr         = &(nb->kernel_noener_noprune_ptr[elecTypeIdx][vdwTypeIdx]);
 412         }
 413     }
 414
 415     if (nullptr == kernel_ptr[0])
 416     {
 417         *kernel_ptr = clCreateKernel(nb->dev_rundata->program, kernel_name_to_run, &cl_error);
 418         GMX_ASSERT(cl_error == CL_SUCCESS,
 419                    ("clCreateKernel failed: " + ocl_get_error_string(cl_error)
 420                     + " for kernel named " + kernel_name_to_run)
 421                            .c_str());
 422     }
 423
 424     return *kernel_ptr;
 425 }
 426
 427 /*! \brief Calculates the amount of shared memory required by the nonbonded kernel in use.
 428  */
 429 static inline int calc_shmem_required_nonbonded(enum VdwType vdwType, bool bPrefetchLjParam)
 430 {
 431     int shmem;
 432
 433     /* size of shmem (force-buffers/xq/atom type preloading) */
 434     /* NOTE: with the default kernel on sm3.0 we need shmem only for pre-loading */
 435     /* i-atom x+q in shared memory */
 436     shmem = c_nbnxnGpuNumClusterPerSupercluster * c_clSize * sizeof(float) * 4; /* xqib */
 437     /* cj in shared memory, for both warps separately
 438      * TODO: in the "nowarp kernels we load cj only once  so the factor 2 is not needed.
 439      */
 440     shmem += 2 * c_nbnxnGpuJgroupSize * sizeof(int); /* cjs  */
 441     if (bPrefetchLjParam)
 442     {
 443         if (useLjCombRule(vdwType))
 444         {
 445             /* i-atom LJ combination parameters in shared memory */
 446             shmem += c_nbnxnGpuNumClusterPerSupercluster * c_clSize * 2
 447                      * sizeof(float); /* atib abused for ljcp, float2 */
 448         }
 449         else
 450         {
 451             /* i-atom types in shared memory */
 452             shmem += c_nbnxnGpuNumClusterPerSupercluster * c_clSize * sizeof(int); /* atib */
 453         }
 454     }
 455     /* force reduction buffers in shared memory */
 456     shmem += c_clSize * c_clSize * 3 * sizeof(float); /* f_buf */
 457     /* Warp vote. In fact it must be * number of warps in block.. */
 458     shmem += sizeof(cl_uint) * 2; /* warp_any */
 459     return shmem;
 460 }
 461
 462 /*! \brief Initializes data structures that are going to be sent to the OpenCL device.
 463  *
 464  *  The device can't use the same data structures as the host for two main reasons:
 465  *  - OpenCL restrictions (pointers are not accepted inside data structures)
 466  *  - some host side fields are not needed for the OpenCL kernels.
 467  *
 468  *  This function is called before the launch of both nbnxn and prune kernels.
 469  */
 470 static void fillin_ocl_structures(NBParamGpu* nbp, cl_nbparam_params_t* nbparams_params)
 471 {
 472     nbparams_params->coulomb_tab_scale = nbp->coulomb_tab_scale;
 473     nbparams_params->c_rf              = nbp->c_rf;
 474     nbparams_params->dispersion_shift  = nbp->dispersion_shift;
 475     nbparams_params->elecType          = nbp->elecType;
 476     nbparams_params->epsfac            = nbp->epsfac;
 477     nbparams_params->ewaldcoeff_lj     = nbp->ewaldcoeff_lj;
 478     nbparams_params->ewald_beta        = nbp->ewald_beta;
 479     nbparams_params->rcoulomb_sq       = nbp->rcoulomb_sq;
 480     nbparams_params->repulsion_shift   = nbp->repulsion_shift;
 481     nbparams_params->rlistOuter_sq     = nbp->rlistOuter_sq;
 482     nbparams_params->rvdw_sq           = nbp->rvdw_sq;
 483     nbparams_params->rlistInner_sq     = nbp->rlistInner_sq;
 484     nbparams_params->rvdw_switch       = nbp->rvdw_switch;
 485     nbparams_params->sh_ewald          = nbp->sh_ewald;
 486     nbparams_params->sh_lj_ewald       = nbp->sh_lj_ewald;
 487     nbparams_params->two_k_rf          = nbp->two_k_rf;
 488     nbparams_params->vdwType           = nbp->vdwType;
 489     nbparams_params->vdw_switch        = nbp->vdw_switch;
 490 }
 491
 492 /*! \brief Launch GPU kernel
 493
 494    As we execute nonbonded workload in separate queues, before launching
 495    the kernel we need to make sure that he following operations have completed:
 496    - atomdata allocation and related H2D transfers (every nstlist step);
 497    - pair list H2D transfer (every nstlist step);
 498    - shift vector H2D transfer (every nstlist step);
 499    - force (+shift force and energy) output clearing (every step).
 500
 501    These operations are issued in the local queue at the beginning of the step
 502    and therefore always complete before the local kernel launch. The non-local
 503    kernel is launched after the local on the same device/context, so this is
 504    inherently scheduled after the operations in the local stream (including the
 505    above "misc_ops").
 506    However, for the sake of having a future-proof implementation, we use the
 507    misc_ops_done event to record the point in time when the above  operations
 508    are finished and synchronize with this event in the non-local stream.
 509  */
 510 void gpu_launch_kernel(NbnxmGpu* nb, const gmx::StepWorkload& stepWork, const Nbnxm::InteractionLocality iloc)
 511 {
 512     NBAtomData*         adat         = nb->atdat;
 513     NBParamGpu*         nbp          = nb->nbparam;
 514     gpu_plist*          plist        = nb->plist[iloc];
 515     Nbnxm::GpuTimers*   timers       = nb->timers;
 516     const DeviceStream& deviceStream = *nb->deviceStreams[iloc];
 517
 518     bool bDoTime = nb->bDoTime;
 519
 520     cl_nbparam_params_t nbparams_params;
 521
 522     /* Don't launch the non-local kernel if there is no work to do.
 523        Doing the same for the local kernel is more complicated, since the
 524        local part of the force array also depends on the non-local kernel.
 525        So to avoid complicating the code and to reduce the risk of bugs,
 526        we always call the local kernel and later (not in
 527        this function) the stream wait, local f copyback and the f buffer
 528        clearing. All these operations, except for the local interaction kernel,
 529        are needed for the non-local interactions. The skip of the local kernel
 530        call is taken care of later in this function. */
 531     if (canSkipNonbondedWork(*nb, iloc))
 532     {
 533         plist->haveFreshList = false;
 534
 535         return;
 536     }
 537
 538     if (nbp->useDynamicPruning && plist->haveFreshList)
 539     {
 540         /* Prunes for rlistOuter and rlistInner, sets plist->haveFreshList=false
 541            (that's the way the timing accounting can distinguish between
 542            separate prune kernel and combined force+prune).
 543          */
 544         Nbnxm::gpu_launch_kernel_pruneonly(nb, iloc, 1);
 545     }
 546
 547     if (plist->nsci == 0)
 548     {
 549         /* Don't launch an empty local kernel (is not allowed with OpenCL).
 550          */
 551         return;
 552     }
 553
 554     /* beginning of timed nonbonded calculation section */
 555     if (bDoTime)
 556     {
 557         timers->interaction[iloc].nb_k.openTimingRegion(deviceStream);
 558     }
 559
 560     /* kernel launch config */
 561
 562     KernelLaunchConfig config;
 563     config.sharedMemorySize = calc_shmem_required_nonbonded(nbp->vdwType, nb->bPrefetchLjParam);
 564     config.blockSize[0]     = c_clSize;
 565     config.blockSize[1]     = c_clSize;
 566     config.gridSize[0]      = plist->nsci;
 567
 568     validate_global_work_size(config, 3, &nb->deviceContext_->deviceInfo());
 569
 570     if (debug)
 571     {
 572         fprintf(debug,
 573                 "Non-bonded GPU launch configuration:\n\tLocal work size: %zux%zux%zu\n\t"
 574                 "Global work size : %zux%zu\n\t#Super-clusters/clusters: %d/%d (%d)\n",
 575                 config.blockSize[0],
 576                 config.blockSize[1],
 577                 config.blockSize[2],
 578                 config.blockSize[0] * config.gridSize[0],
 579                 config.blockSize[1] * config.gridSize[1],
 580                 plist->nsci * c_nbnxnGpuNumClusterPerSupercluster,
 581                 c_nbnxnGpuNumClusterPerSupercluster,
 582                 plist->na_c);
 583     }
 584
 585     fillin_ocl_structures(nbp, &nbparams_params);
 586
 587     auto* timingEvent = bDoTime ? timers->interaction[iloc].nb_k.fetchNextEvent() : nullptr;
 588     constexpr char kernelName[] = "k_calc_nb";
 589     const auto     kernel =
 590             select_nbnxn_kernel(nb,
 591                                 nbp->elecType,
 592                                 nbp->vdwType,
 593                                 stepWork.computeEnergy,
 594                                 (plist->haveFreshList && !nb->timers->interaction[iloc].didPrune));
 595
 596
 597     // The OpenCL kernel takes int as second to last argument because bool is
 598     // not supported as a kernel argument type (sizeof(bool) is implementation defined).
 599     const int computeFshift = static_cast<int>(stepWork.computeVirial);
 600     if (useLjCombRule(nb->nbparam->vdwType))
 601     {
 602         const auto kernelArgs = prepareGpuKernelArguments(kernel,
 603                                                           config,
 604                                                           &nbparams_params,
 605                                                           &adat->xq,
 606                                                           &adat->f,
 607                                                           &adat->eLJ,
 608                                                           &adat->eElec,
 609                                                           &adat->fShift,
 610                                                           &adat->ljComb,
 611                                                           &adat->shiftVec,
 612                                                           &nbp->nbfp,
 613                                                           &nbp->nbfp_comb,
 614                                                           &nbp->coulomb_tab,
 615                                                           &plist->sci,
 616                                                           &plist->cj4,
 617                                                           &plist->excl,
 618                                                           &computeFshift);
 619
 620         launchGpuKernel(kernel, config, deviceStream, timingEvent, kernelName, kernelArgs);
 621     }
 622     else
 623     {
 624         const auto kernelArgs = prepareGpuKernelArguments(kernel,
 625                                                           config,
 626                                                           &adat->numTypes,
 627                                                           &nbparams_params,
 628                                                           &adat->xq,
 629                                                           &adat->f,
 630                                                           &adat->eLJ,
 631                                                           &adat->eElec,
 632                                                           &adat->fShift,
 633                                                           &adat->atomTypes,
 634                                                           &adat->shiftVec,
 635                                                           &nbp->nbfp,
 636                                                           &nbp->nbfp_comb,
 637                                                           &nbp->coulomb_tab,
 638                                                           &plist->sci,
 639                                                           &plist->cj4,
 640                                                           &plist->excl,
 641                                                           &computeFshift);
 642         launchGpuKernel(kernel, config, deviceStream, timingEvent, kernelName, kernelArgs);
 643     }
 644
 645     if (bDoTime)
 646     {
 647         timers->interaction[iloc].nb_k.closeTimingRegion(deviceStream);
 648     }
 649 }
 650
 651
 652 /*! \brief Calculates the amount of shared memory required by the prune kernel.
 653  *
 654  *  Note that for the sake of simplicity we use the CUDA terminology "shared memory"
 655  *  for OpenCL local memory.
 656  *
 657  * \param[in] num_threads_z cj4 concurrency equal to the number of threads/work items in the 3-rd
 658  * dimension. \returns   the amount of local memory in bytes required by the pruning kernel
 659  */
 660 static inline int calc_shmem_required_prune(const int num_threads_z)
 661 {
 662     int shmem;
 663
 664     /* i-atom x in shared memory (for convenience we load all 4 components including q) */
 665     shmem = c_nbnxnGpuNumClusterPerSupercluster * c_clSize * sizeof(float) * 4;
 666     /* cj in shared memory, for each warp separately
 667      * Note: only need to load once per wavefront, but to keep the code simple,
 668      * for now we load twice on AMD.
 669      */
 670     shmem += num_threads_z * c_nbnxnGpuClusterpairSplit * c_nbnxnGpuJgroupSize * sizeof(int);
 671     /* Warp vote, requires one uint per warp/32 threads per block. */
 672     shmem += sizeof(cl_uint) * 2 * num_threads_z;
 673
 674     return shmem;
 675 }
 676
 677 /*! \brief
 678  * Launch the pairlist prune only kernel for the given locality.
 679  * \p numParts tells in how many parts, i.e. calls the list will be pruned.
 680  */
 681 void gpu_launch_kernel_pruneonly(NbnxmGpu* nb, const InteractionLocality iloc, const int numParts)
 682 {
 683     NBAtomData*         adat         = nb->atdat;
 684     NBParamGpu*         nbp          = nb->nbparam;
 685     gpu_plist*          plist        = nb->plist[iloc];
 686     Nbnxm::GpuTimers*   timers       = nb->timers;
 687     const DeviceStream& deviceStream = *nb->deviceStreams[iloc];
 688     bool                bDoTime      = nb->bDoTime;
 689
 690     if (plist->haveFreshList)
 691     {
 692         GMX_ASSERT(numParts == 1, "With first pruning we expect 1 part");
 693
 694         /* Set rollingPruningNumParts to signal that it is not set */
 695         plist->rollingPruningNumParts = 0;
 696         plist->rollingPruningPart     = 0;
 697     }
 698     else
 699     {
 700         if (plist->rollingPruningNumParts == 0)
 701         {
 702             plist->rollingPruningNumParts = numParts;
 703         }
 704         else
 705         {
 706             GMX_ASSERT(numParts == plist->rollingPruningNumParts,
 707                        "It is not allowed to change numParts in between list generation steps");
 708         }
 709     }
 710
 711     /* Use a local variable for part and update in plist, so we can return here
 712      * without duplicating the part increment code.
 713      */
 714     int part = plist->rollingPruningPart;
 715
 716     plist->rollingPruningPart++;
 717     if (plist->rollingPruningPart >= plist->rollingPruningNumParts)
 718     {
 719         plist->rollingPruningPart = 0;
 720     }
 721
 722     /* Compute the number of list entries to prune in this pass */
 723     int numSciInPart = (plist->nsci - part) / numParts;
 724
 725     /* Don't launch the kernel if there is no work to do. */
 726     if (numSciInPart <= 0)
 727     {
 728         plist->haveFreshList = false;
 729
 730         return;
 731     }
 732
 733     GpuRegionTimer* timer = nullptr;
 734     if (bDoTime)
 735     {
 736         timer = &(plist->haveFreshList ? timers->interaction[iloc].prune_k
 737                                        : timers->interaction[iloc].rollingPrune_k);
 738     }
 739
 740     /* beginning of timed prune calculation section */
 741     if (bDoTime)
 742     {
 743         timer->openTimingRegion(deviceStream);
 744     }
 745
 746     /* Kernel launch config:
 747      * - The thread block dimensions match the size of i-clusters, j-clusters,
 748      *   and j-cluster concurrency, in x, y, and z, respectively.
 749      * - The 1D block-grid contains as many blocks as super-clusters.
 750      */
 751     int num_threads_z = c_pruneKernelJ4Concurrency;
 752     /* kernel launch config */
 753     KernelLaunchConfig config;
 754     config.sharedMemorySize = calc_shmem_required_prune(num_threads_z);
 755     config.blockSize[0]     = c_clSize;
 756     config.blockSize[1]     = c_clSize;
 757     config.blockSize[2]     = num_threads_z;
 758     config.gridSize[0]      = numSciInPart;
 759
 760     validate_global_work_size(config, 3, &nb->deviceContext_->deviceInfo());
 761
 762     if (debug)
 763     {
 764         fprintf(debug,
 765                 "Pruning GPU kernel launch configuration:\n\tLocal work size: %zux%zux%zu\n\t"
 766                 "\tGlobal work size: %zux%zu\n\t#Super-clusters/clusters: %d/%d (%d)\n"
 767                 "\tShMem: %zu\n",
 768                 config.blockSize[0],
 769                 config.blockSize[1],
 770                 config.blockSize[2],
 771                 config.blockSize[0] * config.gridSize[0],
 772                 config.blockSize[1] * config.gridSize[1],
 773                 plist->nsci * c_nbnxnGpuNumClusterPerSupercluster,
 774                 c_nbnxnGpuNumClusterPerSupercluster,
 775                 plist->na_c,
 776                 config.sharedMemorySize);
 777     }
 778
 779     cl_nbparam_params_t nbparams_params;
 780     fillin_ocl_structures(nbp, &nbparams_params);
 781
 782     auto*          timingEvent  = bDoTime ? timer->fetchNextEvent() : nullptr;
 783     constexpr char kernelName[] = "k_pruneonly";
 784     const auto     pruneKernel  = selectPruneKernel(nb->kernel_pruneonly, plist->haveFreshList);
 785     const auto     kernelArgs   = prepareGpuKernelArguments(pruneKernel,
 786                                                       config,
 787                                                       &nbparams_params,
 788                                                       &adat->xq,
 789                                                       &adat->shiftVec,
 790                                                       &plist->sci,
 791                                                       &plist->cj4,
 792                                                       &plist->imask,
 793                                                       &numParts,
 794                                                       &part);
 795     launchGpuKernel(pruneKernel, config, deviceStream, timingEvent, kernelName, kernelArgs);
 796
 797     if (plist->haveFreshList)
 798     {
 799         plist->haveFreshList = false;
 800         /* Mark that pruning has been done */
 801         nb->timers->interaction[iloc].didPrune = true;
 802     }
 803     else
 804     {
 805         /* Mark that rolling pruning has been done */
 806         nb->timers->interaction[iloc].didRollingPrune = true;
 807     }
 808
 809     if (bDoTime)
 810     {
 811         timer->closeTimingRegion(deviceStream);
 812     }
 813 }
 814
 815 /*! \brief
 816  * Launch asynchronously the download of nonbonded forces from the GPU
 817  * (and energies/shift forces if required).
 818  */
 819 void gpu_launch_cpyback(NbnxmGpu*                nb,
 820                         struct nbnxn_atomdata_t* nbatom,
 821                         const gmx::StepWorkload& stepWork,
 822                         const AtomLocality       atomLocality)
 823 {
 824     GMX_ASSERT(nb, "Need a valid nbnxn_gpu object");
 825
 826     cl_int gmx_unused cl_error;
 827
 828     /* determine interaction locality from atom locality */
 829     const InteractionLocality iloc = gpuAtomToInteractionLocality(atomLocality);
 830     GMX_ASSERT(iloc == InteractionLocality::Local
 831                        || (iloc == InteractionLocality::NonLocal && nb->bNonLocalStreamDoneMarked == false),
 832                "Non-local stream is indicating that the copy back event is enqueued at the "
 833                "beginning of the copy back function.");
 834
 835     NBAtomData*         adat         = nb->atdat;
 836     Nbnxm::GpuTimers*   timers       = nb->timers;
 837     bool                bDoTime      = nb->bDoTime;
 838     const DeviceStream& deviceStream = *nb->deviceStreams[iloc];
 839
 840     /* don't launch non-local copy-back if there was no non-local work to do */
 841     if ((iloc == InteractionLocality::NonLocal) && !haveGpuShortRangeWork(*nb, iloc))
 842     {
 843         /* TODO An alternative way to signal that non-local work is
 844            complete is to use a clEnqueueMarker+clEnqueueBarrier
 845            pair. However, the use of bNonLocalStreamDoneMarked has the
 846            advantage of being local to the host, so probably minimizes
 847            overhead. Curiously, for NVIDIA OpenCL with an empty-domain
 848            test case, overall simulation performance was higher with
 849            the API calls, but this has not been tested on AMD OpenCL,
 850            so could be worth considering in future. */
 851         nb->bNonLocalStreamDoneMarked = false;
 852         return;
 853     }
 854
 855     /* local/nonlocal offset and length used for xq and f */
 856     auto atomsRange = getGpuAtomRange(adat, atomLocality);
 857
 858     /* beginning of timed D2H section */
 859     if (bDoTime)
 860     {
 861         timers->xf[atomLocality].nb_d2h.openTimingRegion(deviceStream);
 862     }
 863
 864     /* With DD the local D2H transfer can only start after the non-local
 865        has been launched. */
 866     if (iloc == InteractionLocality::Local && nb->bNonLocalStreamDoneMarked)
 867     {
 868         nb->nonlocal_done.enqueueWaitEvent(deviceStream);
 869         nb->bNonLocalStreamDoneMarked = false;
 870     }
 871
 872     /* DtoH f */
 873     GMX_ASSERT(sizeof(*nbatom->out[0].f.data()) == sizeof(float),
 874                "The host force buffer should be in single precision to match device data size.");
 875     copyFromDeviceBuffer(reinterpret_cast<Float3*>(nbatom->out[0].f.data()) + atomsRange.begin(),
 876                          &adat->f,
 877                          atomsRange.begin(),
 878                          atomsRange.size(),
 879                          deviceStream,
 880                          GpuApiCallBehavior::Async,
 881                          bDoTime ? timers->xf[atomLocality].nb_d2h.fetchNextEvent() : nullptr);
 882
 883     /* kick off work */
 884     cl_error = clFlush(deviceStream.stream());
 885     GMX_ASSERT(cl_error == CL_SUCCESS, ("clFlush failed: " + ocl_get_error_string(cl_error)).c_str());
 886
 887     /* After the non-local D2H is launched the nonlocal_done event can be
 888        recorded which signals that the local D2H can proceed. This event is not
 889        placed after the non-local kernel because we first need the non-local
 890        data back first. */
 891     if (iloc == InteractionLocality::NonLocal)
 892     {
 893         nb->nonlocal_done.markEvent(deviceStream);
 894         nb->bNonLocalStreamDoneMarked = true;
 895     }
 896
 897     /* only transfer energies in the local stream */
 898     if (iloc == InteractionLocality::Local)
 899     {
 900         /* DtoH fshift when virial is needed */
 901         if (stepWork.computeVirial)
 902         {
 903             static_assert(
 904                     sizeof(*nb->nbst.fShift) == sizeof(Float3),
 905                     "Sizes of host- and device-side shift vector elements should be the same.");
 906             copyFromDeviceBuffer(nb->nbst.fShift,
 907                                  &adat->fShift,
 908                                  0,
 909                                  SHIFTS,
 910                                  deviceStream,
 911                                  GpuApiCallBehavior::Async,
 912                                  bDoTime ? timers->xf[atomLocality].nb_d2h.fetchNextEvent() : nullptr);
 913         }
 914
 915         /* DtoH energies */
 916         if (stepWork.computeEnergy)
 917         {
 918             static_assert(sizeof(*nb->nbst.eLJ) == sizeof(float),
 919                           "Sizes of host- and device-side LJ energy terms should be the same.");
 920             copyFromDeviceBuffer(nb->nbst.eLJ,
 921                                  &adat->eLJ,
 922                                  0,
 923                                  1,
 924                                  deviceStream,
 925                                  GpuApiCallBehavior::Async,
 926                                  bDoTime ? timers->xf[atomLocality].nb_d2h.fetchNextEvent() : nullptr);
 927             static_assert(sizeof(*nb->nbst.eElec) == sizeof(float),
 928                           "Sizes of host- and device-side electrostatic energy terms should be the "
 929                           "same.");
 930             copyFromDeviceBuffer(nb->nbst.eElec,
 931                                  &adat->eElec,
 932                                  0,
 933                                  1,
 934                                  deviceStream,
 935                                  GpuApiCallBehavior::Async,
 936                                  bDoTime ? timers->xf[atomLocality].nb_d2h.fetchNextEvent() : nullptr);
 937         }
 938     }
 939
 940     if (bDoTime)
 941     {
 942         timers->xf[atomLocality].nb_d2h.closeTimingRegion(deviceStream);
 943     }
 944 }
 945
 946 } // namespace Nbnxm