src/gromacs/mdtypes/state_propagator_data_gpu_impl_gpu.cpp

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2019, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*! \internal \file
  36  *
  37  * \brief Definitions of interfaces for GPU state data propagator object.
  38  *
  39  * \author Artem Zhmurov <zhmurov@gmail.com>
  40  *
  41  * \ingroup module_mdtypes
  42  */
  43 #include "gmxpre.h"
  44
  45 #include "config.h"
  46
  47 #if GMX_GPU != GMX_GPU_NONE
  48
  49 #    if GMX_GPU == GMX_GPU_CUDA
  50 #        include "gromacs/gpu_utils/cudautils.cuh"
  51 #    endif
  52 #    include "gromacs/gpu_utils/devicebuffer.h"
  53 #    if GMX_GPU == GMX_GPU_OPENCL
  54 #        include "gromacs/gpu_utils/oclutils.h"
  55 #    endif
  56 #    include "gromacs/math/vectypes.h"
  57 #    include "gromacs/mdtypes/state_propagator_data_gpu.h"
  58 #    include "gromacs/utility/classhelpers.h"
  59
  60 #    include "state_propagator_data_gpu_impl.h"
  61
  62 namespace gmx
  63 {
  64
  65 StatePropagatorDataGpu::Impl::Impl(const void*        pmeStream,
  66                                    const void*        localStream,
  67                                    const void*        nonLocalStream,
  68                                    const void*        deviceContext,
  69                                    GpuApiCallBehavior transferKind,
  70                                    int                paddingSize) :
  71     transferKind_(transferKind),
  72     paddingSize_(paddingSize)
  73 {
  74     static_assert(GMX_GPU != GMX_GPU_NONE,
  75                   "This object should only be constructed on the GPU code-paths.");
  76
  77     // TODO: Refactor when the StreamManager is introduced.
  78     if (GMX_GPU == GMX_GPU_OPENCL)
  79     {
  80         GMX_ASSERT(deviceContext != nullptr, "GPU context should be set in OpenCL builds.");
  81         GMX_ASSERT(pmeStream != nullptr, "GPU PME stream should be set in OpenCL builds.");
  82
  83         // The update stream is set to the PME stream in OpenCL, since PME stream is the only stream created in the PME context.
  84         pmeStream_     = *static_cast<const CommandStream*>(pmeStream);
  85         updateStream_  = *static_cast<const CommandStream*>(pmeStream);
  86         deviceContext_ = *static_cast<const DeviceContext*>(deviceContext);
  87         GMX_UNUSED_VALUE(localStream);
  88         GMX_UNUSED_VALUE(nonLocalStream);
  89     }
  90
  91     if (GMX_GPU == GMX_GPU_CUDA)
  92     {
  93         if (pmeStream != nullptr)
  94         {
  95             pmeStream_ = *static_cast<const CommandStream*>(pmeStream);
  96         }
  97         if (localStream != nullptr)
  98         {
  99             localStream_ = *static_cast<const CommandStream*>(localStream);
 100         }
 101         if (nonLocalStream != nullptr)
 102         {
 103             nonLocalStream_ = *static_cast<const CommandStream*>(nonLocalStream);
 104         }
 105
 106         // TODO: The update stream should be created only when it is needed.
 107 #    if (GMX_GPU == GMX_GPU_CUDA)
 108         cudaError_t stat;
 109         stat = cudaStreamCreate(&updateStream_);
 110         CU_RET_ERR(stat, "CUDA stream creation failed in StatePropagatorDataGpu");
 111 #    endif
 112         GMX_UNUSED_VALUE(deviceContext);
 113     }
 114
 115     // Map the atom locality to the stream that will be used for coordinates,
 116     // velocities and forces transfers. Same streams are used for H2D and D2H copies.
 117     // Note, that nullptr stream is used here to indicate that the copy is not supported.
 118     xCopyStreams_[AtomLocality::Local]    = updateStream_;
 119     xCopyStreams_[AtomLocality::NonLocal] = nonLocalStream_;
 120     xCopyStreams_[AtomLocality::All]      = updateStream_;
 121
 122     vCopyStreams_[AtomLocality::Local]    = updateStream_;
 123     vCopyStreams_[AtomLocality::NonLocal] = nullptr;
 124     vCopyStreams_[AtomLocality::All]      = updateStream_;
 125
 126     fCopyStreams_[AtomLocality::Local]    = localStream_;
 127     fCopyStreams_[AtomLocality::NonLocal] = nonLocalStream_;
 128     fCopyStreams_[AtomLocality::All]      = updateStream_;
 129 }
 130
 131 StatePropagatorDataGpu::Impl::Impl(const void*        pmeStream,
 132                                    const void*        deviceContext,
 133                                    GpuApiCallBehavior transferKind,
 134                                    int                paddingSize) :
 135     transferKind_(transferKind),
 136     paddingSize_(paddingSize)
 137 {
 138     static_assert(GMX_GPU != GMX_GPU_NONE,
 139                   "This object should only be constructed on the GPU code-paths.");
 140
 141     if (GMX_GPU == GMX_GPU_OPENCL)
 142     {
 143         GMX_ASSERT(deviceContext != nullptr, "GPU context should be set in OpenCL builds.");
 144         deviceContext_ = *static_cast<const DeviceContext*>(deviceContext);
 145     }
 146
 147     GMX_ASSERT(pmeStream != nullptr, "GPU PME stream should be set.");
 148     pmeStream_ = *static_cast<const CommandStream*>(pmeStream);
 149
 150     localStream_    = nullptr;
 151     nonLocalStream_ = nullptr;
 152     updateStream_   = nullptr;
 153
 154
 155     // Only local/all coordinates are allowed to be copied in PME-only rank/ PME tests.
 156     // This it temporary measure to make it safe to use this class in those cases.
 157     xCopyStreams_[AtomLocality::Local]    = pmeStream_;
 158     xCopyStreams_[AtomLocality::NonLocal] = nullptr;
 159     xCopyStreams_[AtomLocality::All]      = pmeStream_;
 160
 161     vCopyStreams_[AtomLocality::Local]    = nullptr;
 162     vCopyStreams_[AtomLocality::NonLocal] = nullptr;
 163     vCopyStreams_[AtomLocality::All]      = nullptr;
 164
 165     fCopyStreams_[AtomLocality::Local]    = nullptr;
 166     fCopyStreams_[AtomLocality::NonLocal] = nullptr;
 167     fCopyStreams_[AtomLocality::All]      = nullptr;
 168 }
 169
 170 StatePropagatorDataGpu::Impl::~Impl() {}
 171
 172 void StatePropagatorDataGpu::Impl::reinit(int numAtomsLocal, int numAtomsAll)
 173 {
 174     numAtomsLocal_ = numAtomsLocal;
 175     numAtomsAll_   = numAtomsAll;
 176
 177     int numAtomsPadded;
 178     if (paddingSize_ > 0)
 179     {
 180         numAtomsPadded = ((numAtomsAll_ + paddingSize_ - 1) / paddingSize_) * paddingSize_;
 181     }
 182     else
 183     {
 184         numAtomsPadded = numAtomsAll_;
 185     }
 186
 187     reallocateDeviceBuffer(&d_x_, DIM * numAtomsPadded, &d_xSize_, &d_xCapacity_, deviceContext_);
 188
 189     const size_t paddingAllocationSize = numAtomsPadded - numAtomsAll_;
 190     if (paddingAllocationSize > 0)
 191     {
 192         // The PME stream is used here because the padding region of d_x_ is only in the PME task.
 193         clearDeviceBufferAsync(&d_x_, DIM * numAtomsAll_, DIM * paddingAllocationSize, pmeStream_);
 194     }
 195
 196     reallocateDeviceBuffer(&d_v_, DIM * numAtomsAll_, &d_vSize_, &d_vCapacity_, deviceContext_);
 197     const int d_fOldCapacity = d_fCapacity_;
 198     reallocateDeviceBuffer(&d_f_, DIM * numAtomsAll_, &d_fSize_, &d_fCapacity_, deviceContext_);
 199     // Clearing of the forces can be done in local stream since the nonlocal stream cannot reach
 200     // the force accumulation stage before syncing with the local stream. Only done in CUDA,
 201     // since the force buffer ops are not implemented in OpenCL.
 202     if (GMX_GPU == GMX_GPU_CUDA && d_fCapacity_ != d_fOldCapacity)
 203     {
 204         clearDeviceBufferAsync(&d_f_, 0, d_fCapacity_, localStream_);
 205     }
 206 }
 207
 208 std::tuple<int, int> StatePropagatorDataGpu::Impl::getAtomRangesFromAtomLocality(AtomLocality atomLocality)
 209 {
 210     int atomsStartAt   = 0;
 211     int numAtomsToCopy = 0;
 212     switch (atomLocality)
 213     {
 214         case AtomLocality::All:
 215             atomsStartAt   = 0;
 216             numAtomsToCopy = numAtomsAll_;
 217             break;
 218         case AtomLocality::Local:
 219             atomsStartAt   = 0;
 220             numAtomsToCopy = numAtomsLocal_;
 221             break;
 222         case AtomLocality::NonLocal:
 223             atomsStartAt   = numAtomsLocal_;
 224             numAtomsToCopy = numAtomsAll_ - numAtomsLocal_;
 225             break;
 226         default:
 227             GMX_RELEASE_ASSERT(false,
 228                                "Wrong range of atoms requested in GPU state data manager. Should "
 229                                "be All, Local or NonLocal.");
 230     }
 231     GMX_ASSERT(atomsStartAt >= 0,
 232                "The first elemtnt to copy has negative index. Probably, the GPU propagator state "
 233                "was not initialized.");
 234     GMX_ASSERT(numAtomsToCopy >= 0,
 235                "Number of atoms to copy is negative. Probably, the GPU propagator state was not "
 236                "initialized.");
 237     return std::make_tuple(atomsStartAt, numAtomsToCopy);
 238 }
 239
 240 void StatePropagatorDataGpu::Impl::copyToDevice(DeviceBuffer<float>                  d_data,
 241                                                 const gmx::ArrayRef<const gmx::RVec> h_data,
 242                                                 int                                  dataSize,
 243                                                 AtomLocality                         atomLocality,
 244                                                 CommandStream                        commandStream)
 245 {
 246
 247     GMX_UNUSED_VALUE(dataSize);
 248
 249     GMX_ASSERT(dataSize >= 0, "Trying to copy to device buffer before it was allocated.");
 250
 251     int atomsStartAt, numAtomsToCopy;
 252     std::tie(atomsStartAt, numAtomsToCopy) = getAtomRangesFromAtomLocality(atomLocality);
 253
 254     int elementsStartAt   = atomsStartAt * DIM;
 255     int numElementsToCopy = numAtomsToCopy * DIM;
 256
 257     if (numAtomsToCopy != 0)
 258     {
 259         GMX_ASSERT(elementsStartAt + numElementsToCopy <= dataSize,
 260                    "The device allocation is smaller than requested copy range.");
 261         GMX_ASSERT(atomsStartAt + numAtomsToCopy <= h_data.ssize(),
 262                    "The host buffer is smaller than the requested copy range.");
 263
 264         copyToDeviceBuffer(&d_data, reinterpret_cast<const float*>(&h_data.data()[atomsStartAt]),
 265                            elementsStartAt, numElementsToCopy, commandStream, transferKind_, nullptr);
 266     }
 267 }
 268
 269 void StatePropagatorDataGpu::Impl::copyFromDevice(gmx::ArrayRef<gmx::RVec> h_data,
 270                                                   DeviceBuffer<float>      d_data,
 271                                                   int                      dataSize,
 272                                                   AtomLocality             atomLocality,
 273                                                   CommandStream            commandStream)
 274 {
 275
 276     GMX_UNUSED_VALUE(dataSize);
 277
 278     GMX_ASSERT(dataSize >= 0, "Trying to copy from device buffer before it was allocated.");
 279
 280     int atomsStartAt, numAtomsToCopy;
 281     std::tie(atomsStartAt, numAtomsToCopy) = getAtomRangesFromAtomLocality(atomLocality);
 282
 283     int elementsStartAt   = atomsStartAt * DIM;
 284     int numElementsToCopy = numAtomsToCopy * DIM;
 285
 286     if (numAtomsToCopy != 0)
 287     {
 288         GMX_ASSERT(elementsStartAt + numElementsToCopy <= dataSize,
 289                    "The device allocation is smaller than requested copy range.");
 290         GMX_ASSERT(atomsStartAt + numAtomsToCopy <= h_data.ssize(),
 291                    "The host buffer is smaller than the requested copy range.");
 292
 293         copyFromDeviceBuffer(reinterpret_cast<float*>(&h_data.data()[atomsStartAt]), &d_data,
 294                              elementsStartAt, numElementsToCopy, commandStream, transferKind_, nullptr);
 295     }
 296 }
 297
 298 DeviceBuffer<float> StatePropagatorDataGpu::Impl::getCoordinates()
 299 {
 300     return d_x_;
 301 }
 302
 303 void StatePropagatorDataGpu::Impl::copyCoordinatesToGpu(const gmx::ArrayRef<const gmx::RVec> h_x,
 304                                                         AtomLocality atomLocality)
 305 {
 306     GMX_ASSERT(atomLocality < AtomLocality::Count, "Wrong atom locality.");
 307     CommandStream commandStream = xCopyStreams_[atomLocality];
 308     GMX_ASSERT(commandStream != nullptr,
 309                "No stream is valid for copying positions with given atom locality.");
 310
 311     copyToDevice(d_x_, h_x, d_xSize_, atomLocality, commandStream);
 312
 313     // markEvent is skipped in OpenCL as:
 314     //   - it's not needed, copy is done in the same stream as the only consumer task (PME)
 315     //   - we don't consume the events in OpenCL which is not allowed by GpuEventSynchronizer (would leak memory).
 316     // TODO: remove this by adding an event-mark free flavor of this function
 317     if (GMX_GPU == GMX_GPU_CUDA)
 318     {
 319         xReadyOnDevice_[atomLocality].markEvent(commandStream);
 320     }
 321 }
 322
 323 GpuEventSynchronizer*
 324 StatePropagatorDataGpu::Impl::getCoordinatesReadyOnDeviceEvent(AtomLocality atomLocality,
 325                                                                const SimulationWorkload& simulationWork,
 326                                                                const StepWorkload&       stepWork)
 327 {
 328     // The provider of the coordinates may be different for local atoms. If the update is offloaded
 329     // and this is not a neighbor search step, then the consumer needs to wait for the update
 330     // to complete. Otherwise, the coordinates are copied from the host and we need to wait for
 331     // the copy event. Non-local coordinates are always provided by the H2D copy.
 332     //
 333     // TODO: This should be reconsidered to support the halo exchange.
 334     //
 335     // In OpenCL no events are used as coordinate sync is not necessary
 336     if (GMX_GPU == GMX_GPU_OPENCL)
 337     {
 338         return nullptr;
 339     }
 340     if (atomLocality == AtomLocality::Local && simulationWork.useGpuUpdate && !stepWork.doNeighborSearch)
 341     {
 342         return &xUpdatedOnDevice_;
 343     }
 344     else
 345     {
 346         return &xReadyOnDevice_[atomLocality];
 347     }
 348 }
 349
 350 void StatePropagatorDataGpu::Impl::waitCoordinatesCopiedToDevice(AtomLocality atomLocality)
 351 {
 352     GMX_ASSERT(atomLocality < AtomLocality::Count, "Wrong atom locality.");
 353     xReadyOnDevice_[atomLocality].waitForEvent();
 354 }
 355
 356 GpuEventSynchronizer* StatePropagatorDataGpu::Impl::xUpdatedOnDevice()
 357 {
 358     return &xUpdatedOnDevice_;
 359 }
 360
 361 void StatePropagatorDataGpu::Impl::copyCoordinatesFromGpu(gmx::ArrayRef<gmx::RVec> h_x, AtomLocality atomLocality)
 362 {
 363     GMX_ASSERT(atomLocality < AtomLocality::Count, "Wrong atom locality.");
 364     CommandStream commandStream = xCopyStreams_[atomLocality];
 365     GMX_ASSERT(commandStream != nullptr,
 366                "No stream is valid for copying positions with given atom locality.");
 367
 368     copyFromDevice(h_x, d_x_, d_xSize_, atomLocality, commandStream);
 369     // Note: unlike copyCoordinatesToGpu this is not used in OpenCL, and the conditional is not needed.
 370     xReadyOnHost_[atomLocality].markEvent(commandStream);
 371 }
 372
 373 void StatePropagatorDataGpu::Impl::waitCoordinatesReadyOnHost(AtomLocality atomLocality)
 374 {
 375     xReadyOnHost_[atomLocality].waitForEvent();
 376 }
 377
 378
 379 DeviceBuffer<float> StatePropagatorDataGpu::Impl::getVelocities()
 380 {
 381     return d_v_;
 382 }
 383
 384 void StatePropagatorDataGpu::Impl::copyVelocitiesToGpu(const gmx::ArrayRef<const gmx::RVec> h_v,
 385                                                        AtomLocality atomLocality)
 386 {
 387     GMX_ASSERT(atomLocality < AtomLocality::Count, "Wrong atom locality.");
 388     CommandStream commandStream = vCopyStreams_[atomLocality];
 389     GMX_ASSERT(commandStream != nullptr,
 390                "No stream is valid for copying velocities with given atom locality.");
 391
 392     copyToDevice(d_v_, h_v, d_vSize_, atomLocality, commandStream);
 393     vReadyOnDevice_[atomLocality].markEvent(commandStream);
 394 }
 395
 396 GpuEventSynchronizer* StatePropagatorDataGpu::Impl::getVelocitiesReadyOnDeviceEvent(AtomLocality atomLocality)
 397 {
 398     return &vReadyOnDevice_[atomLocality];
 399 }
 400
 401
 402 void StatePropagatorDataGpu::Impl::copyVelocitiesFromGpu(gmx::ArrayRef<gmx::RVec> h_v, AtomLocality atomLocality)
 403 {
 404     GMX_ASSERT(atomLocality < AtomLocality::Count, "Wrong atom locality.");
 405     CommandStream commandStream = vCopyStreams_[atomLocality];
 406     GMX_ASSERT(commandStream != nullptr,
 407                "No stream is valid for copying velocities with given atom locality.");
 408
 409     copyFromDevice(h_v, d_v_, d_vSize_, atomLocality, commandStream);
 410     vReadyOnHost_[atomLocality].markEvent(commandStream);
 411 }
 412
 413 void StatePropagatorDataGpu::Impl::waitVelocitiesReadyOnHost(AtomLocality atomLocality)
 414 {
 415     vReadyOnHost_[atomLocality].waitForEvent();
 416 }
 417
 418
 419 DeviceBuffer<float> StatePropagatorDataGpu::Impl::getForces()
 420 {
 421     return d_f_;
 422 }
 423
 424 void StatePropagatorDataGpu::Impl::copyForcesToGpu(const gmx::ArrayRef<const gmx::RVec> h_f,
 425                                                    AtomLocality atomLocality)
 426 {
 427     GMX_ASSERT(atomLocality < AtomLocality::Count, "Wrong atom locality.");
 428     CommandStream commandStream = fCopyStreams_[atomLocality];
 429     GMX_ASSERT(commandStream != nullptr,
 430                "No stream is valid for copying forces with given atom locality.");
 431
 432     copyToDevice(d_f_, h_f, d_fSize_, atomLocality, commandStream);
 433     fReadyOnDevice_[atomLocality].markEvent(commandStream);
 434 }
 435
 436 GpuEventSynchronizer* StatePropagatorDataGpu::Impl::getForcesReadyOnDeviceEvent(AtomLocality atomLocality,
 437                                                                                 bool useGpuFBufferOps)
 438 {
 439     if ((atomLocality == AtomLocality::Local || atomLocality == AtomLocality::NonLocal) && useGpuFBufferOps)
 440     {
 441         return &fReducedOnDevice_;
 442     }
 443     else
 444     {
 445         return &fReadyOnDevice_[atomLocality];
 446     }
 447 }
 448
 449 GpuEventSynchronizer* StatePropagatorDataGpu::Impl::fReducedOnDevice()
 450 {
 451     return &fReducedOnDevice_;
 452 }
 453
 454 void StatePropagatorDataGpu::Impl::copyForcesFromGpu(gmx::ArrayRef<gmx::RVec> h_f, AtomLocality atomLocality)
 455 {
 456     GMX_ASSERT(atomLocality < AtomLocality::Count, "Wrong atom locality.");
 457     CommandStream commandStream = fCopyStreams_[atomLocality];
 458     GMX_ASSERT(commandStream != nullptr,
 459                "No stream is valid for copying forces with given atom locality.");
 460
 461     copyFromDevice(h_f, d_f_, d_fSize_, atomLocality, commandStream);
 462     fReadyOnHost_[atomLocality].markEvent(commandStream);
 463 }
 464
 465 void StatePropagatorDataGpu::Impl::waitForcesReadyOnHost(AtomLocality atomLocality)
 466 {
 467     fReadyOnHost_[atomLocality].waitForEvent();
 468 }
 469
 470 void* StatePropagatorDataGpu::Impl::getUpdateStream()
 471 {
 472     return &updateStream_;
 473 }
 474
 475 int StatePropagatorDataGpu::Impl::numAtomsLocal()
 476 {
 477     return numAtomsLocal_;
 478 }
 479
 480 int StatePropagatorDataGpu::Impl::numAtomsAll()
 481 {
 482     return numAtomsAll_;
 483 }
 484
 485
 486 StatePropagatorDataGpu::StatePropagatorDataGpu(const void*        pmeStream,
 487                                                const void*        localStream,
 488                                                const void*        nonLocalStream,
 489                                                const void*        deviceContext,
 490                                                GpuApiCallBehavior transferKind,
 491                                                int                paddingSize) :
 492     impl_(new Impl(pmeStream, localStream, nonLocalStream, deviceContext, transferKind, paddingSize))
 493 {
 494 }
 495
 496 StatePropagatorDataGpu::StatePropagatorDataGpu(const void*        pmeStream,
 497                                                const void*        deviceContext,
 498                                                GpuApiCallBehavior transferKind,
 499                                                int                paddingSize) :
 500     impl_(new Impl(pmeStream, deviceContext, transferKind, paddingSize))
 501 {
 502 }
 503
 504 StatePropagatorDataGpu::StatePropagatorDataGpu(StatePropagatorDataGpu&& /* other */) noexcept = default;
 505
 506 StatePropagatorDataGpu& StatePropagatorDataGpu::operator=(StatePropagatorDataGpu&& /* other */) noexcept = default;
 507
 508 StatePropagatorDataGpu::~StatePropagatorDataGpu() = default;
 509
 510
 511 void StatePropagatorDataGpu::reinit(int numAtomsLocal, int numAtomsAll)
 512 {
 513     return impl_->reinit(numAtomsLocal, numAtomsAll);
 514 }
 515
 516 std::tuple<int, int> StatePropagatorDataGpu::getAtomRangesFromAtomLocality(AtomLocality atomLocality)
 517 {
 518     return impl_->getAtomRangesFromAtomLocality(atomLocality);
 519 }
 520
 521
 522 DeviceBuffer<float> StatePropagatorDataGpu::getCoordinates()
 523 {
 524     return impl_->getCoordinates();
 525 }
 526
 527 void StatePropagatorDataGpu::copyCoordinatesToGpu(const gmx::ArrayRef<const gmx::RVec> h_x,
 528                                                   AtomLocality                         atomLocality)
 529 {
 530     return impl_->copyCoordinatesToGpu(h_x, atomLocality);
 531 }
 532
 533 GpuEventSynchronizer*
 534 StatePropagatorDataGpu::getCoordinatesReadyOnDeviceEvent(AtomLocality              atomLocality,
 535                                                          const SimulationWorkload& simulationWork,
 536                                                          const StepWorkload&       stepWork)
 537 {
 538     return impl_->getCoordinatesReadyOnDeviceEvent(atomLocality, simulationWork, stepWork);
 539 }
 540
 541 void StatePropagatorDataGpu::waitCoordinatesCopiedToDevice(AtomLocality atomLocality)
 542 {
 543     return impl_->waitCoordinatesCopiedToDevice(atomLocality);
 544 }
 545
 546 GpuEventSynchronizer* StatePropagatorDataGpu::xUpdatedOnDevice()
 547 {
 548     return impl_->xUpdatedOnDevice();
 549 }
 550
 551 void StatePropagatorDataGpu::copyCoordinatesFromGpu(gmx::ArrayRef<RVec> h_x, AtomLocality atomLocality)
 552 {
 553     return impl_->copyCoordinatesFromGpu(h_x, atomLocality);
 554 }
 555
 556 void StatePropagatorDataGpu::waitCoordinatesReadyOnHost(AtomLocality atomLocality)
 557 {
 558     return impl_->waitCoordinatesReadyOnHost(atomLocality);
 559 }
 560
 561
 562 DeviceBuffer<float> StatePropagatorDataGpu::getVelocities()
 563 {
 564     return impl_->getVelocities();
 565 }
 566
 567 void StatePropagatorDataGpu::copyVelocitiesToGpu(const gmx::ArrayRef<const gmx::RVec> h_v,
 568                                                  AtomLocality                         atomLocality)
 569 {
 570     return impl_->copyVelocitiesToGpu(h_v, atomLocality);
 571 }
 572
 573 GpuEventSynchronizer* StatePropagatorDataGpu::getVelocitiesReadyOnDeviceEvent(AtomLocality atomLocality)
 574 {
 575     return impl_->getVelocitiesReadyOnDeviceEvent(atomLocality);
 576 }
 577
 578 void StatePropagatorDataGpu::copyVelocitiesFromGpu(gmx::ArrayRef<RVec> h_v, AtomLocality atomLocality)
 579 {
 580     return impl_->copyVelocitiesFromGpu(h_v, atomLocality);
 581 }
 582
 583 void StatePropagatorDataGpu::waitVelocitiesReadyOnHost(AtomLocality atomLocality)
 584 {
 585     return impl_->waitVelocitiesReadyOnHost(atomLocality);
 586 }
 587
 588
 589 DeviceBuffer<float> StatePropagatorDataGpu::getForces()
 590 {
 591     return impl_->getForces();
 592 }
 593
 594 void StatePropagatorDataGpu::copyForcesToGpu(const gmx::ArrayRef<const gmx::RVec> h_f, AtomLocality atomLocality)
 595 {
 596     return impl_->copyForcesToGpu(h_f, atomLocality);
 597 }
 598
 599 GpuEventSynchronizer* StatePropagatorDataGpu::getForcesReadyOnDeviceEvent(AtomLocality atomLocality,
 600                                                                           bool useGpuFBufferOps)
 601 {
 602     return impl_->getForcesReadyOnDeviceEvent(atomLocality, useGpuFBufferOps);
 603 }
 604
 605 GpuEventSynchronizer* StatePropagatorDataGpu::fReducedOnDevice()
 606 {
 607     return impl_->fReducedOnDevice();
 608 }
 609
 610 void StatePropagatorDataGpu::copyForcesFromGpu(gmx::ArrayRef<RVec> h_f, AtomLocality atomLocality)
 611 {
 612     return impl_->copyForcesFromGpu(h_f, atomLocality);
 613 }
 614
 615 void StatePropagatorDataGpu::waitForcesReadyOnHost(AtomLocality atomLocality)
 616 {
 617     return impl_->waitForcesReadyOnHost(atomLocality);
 618 }
 619
 620
 621 void* StatePropagatorDataGpu::getUpdateStream()
 622 {
 623     return impl_->getUpdateStream();
 624 }
 625
 626 int StatePropagatorDataGpu::numAtomsLocal()
 627 {
 628     return impl_->numAtomsLocal();
 629 }
 630
 631 int StatePropagatorDataGpu::numAtomsAll()
 632 {
 633     return impl_->numAtomsAll();
 634 }
 635
 636 } // namespace gmx
 637
 638 #endif // GMX_GPU == GMX_GPU_NONE