ReactionKernelIndexed_4_single_precision_CUDA.h

Go to the documentation of this file.

/*
 * Copyright (C) 2022-2025 The ESPResSo project
 * Copyright (C) 2020-2025 The waLBerla project
 *
 * This file is part of ESPResSo.
 *
 * ESPResSo is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 *
 * ESPResSo is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program.  If not, see <http://www.gnu.org/licenses/>.
 */
 
// kernel generated with pystencils v1.4+1.ge851f4e, lbmpy v1.4+1.ge9efe34,
// sympy v1.12.1, lbmpy_walberla/pystencils_walberla from waLBerla commit
// 007e77e077ad9d22b5eed6f3d3118240993e553c
 
/*
 * Boundary class.
 * Adapted from the waLBerla source file
 * https://i10git.cs.fau.de/walberla/walberla/-/blob/3e54d4f2336e47168ad87e3caaf7b3b082d86ca7/python/pystencils_walberla/templates/Boundary.tmpl.h
 */
 
#pragma once
 
#include <core/DataTypes.h>
 
#include <blockforest/StructuredBlockForest.h>
#include <core/debug/Debug.h>
#include <domain_decomposition/BlockDataID.h>
#include <domain_decomposition/IBlock.h>
#include <field/FlagField.h>
#include <gpu/FieldCopy.h>
#include <gpu/GPUField.h>
#include <gpu/GPUWrapper.h>
 
#include <array>
#include <cassert>
#include <functional>
#include <memory>
#include <vector>
 
#if defined(__clang__)
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Wunused-variable"
#pragma clang diagnostic ignored "-Wunused-parameter"
#elif defined(__GNUC__) or defined(__GNUG__)
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wunused-variable"
#pragma GCC diagnostic ignored "-Wunused-parameter"
#endif
 
#ifdef __GNUC__
#define RESTRICT __restrict__
#elif _MSC_VER
#define RESTRICT __restrict
#else
#define RESTRICT
#endif
 
#ifdef WALBERLA_BUILD_WITH_HALF_PRECISION_SUPPORT
using walberla::half;
#endif
 
namespace walberla {
namespace pystencils {
 
class ReactionKernelIndexed_4_single_precision_CUDA {
public:
  struct IndexInfo {
    int32_t x;
    int32_t y;
    int32_t z;
    IndexInfo(int32_t x_, int32_t y_, int32_t z_) : x(x_), y(y_), z(z_) {}
    bool operator==(const IndexInfo &o) const {
      return x == o.x && y == o.y && z == o.z;
    }
  };
 
  class IndexVectors {
  public:
    using CpuIndexVector = std::vector<IndexInfo>;
 
    enum Type { ALL = 0, INNER = 1, OUTER = 2, NUM_TYPES = 3 };
 
    IndexVectors() = default;
    bool operator==(IndexVectors const &other) const {
      return other.cpuVectors_ == cpuVectors_;
    }
 
    ~IndexVectors() {
      for (auto &gpuVec : gpuVectors_) {
        if (gpuVec) {
          WALBERLA_GPU_CHECK(gpuFree(gpuVec));
        }
      }
    }
    auto &indexVector(Type t) { return cpuVectors_[t]; }
    auto const &indexVector(Type t) const { return cpuVectors_[t]; }
    IndexInfo *pointerCpu(Type t) {
      return cpuVectors_[t].empty() ? nullptr : cpuVectors_[t].data();
    }
 
    IndexInfo *pointerGpu(Type t) { return gpuVectors_[t]; }
    void syncGPU() {
      for (auto &gpuVec : gpuVectors_) {
        if (gpuVec) {
          WALBERLA_GPU_CHECK(gpuFree(gpuVec));
          gpuVec = nullptr;
        }
      }
      gpuVectors_.resize(cpuVectors_.size());
 
      WALBERLA_ASSERT_EQUAL(cpuVectors_.size(), NUM_TYPES);
      for (size_t i = 0; i < cpuVectors_.size(); ++i) {
        auto &gpuVec = gpuVectors_[i];
        auto &cpuVec = cpuVectors_[i];
        if (cpuVec.empty()) {
          continue;
        }
        WALBERLA_GPU_CHECK(
            gpuMalloc(&gpuVec, sizeof(IndexInfo) * cpuVec.size()));
        WALBERLA_GPU_CHECK(gpuMemcpy(gpuVec, cpuVec.data(),
                                     sizeof(IndexInfo) * cpuVec.size(),
                                     gpuMemcpyHostToDevice));
      }
    }
 
  private:
    std::vector<CpuIndexVector> cpuVectors_{NUM_TYPES};
 
    using GpuIndexVector = IndexInfo *;
    std::vector<GpuIndexVector> gpuVectors_;
  };
 
  ReactionKernelIndexed_4_single_precision_CUDA(
      const std::shared_ptr<StructuredBlockForest> &blocks,
      BlockDataID rho_0ID_, BlockDataID rho_1ID_, BlockDataID rho_2ID_,
      BlockDataID rho_3ID_, float order_0, float order_1, float order_2,
      float order_3, float rate_coefficient, float stoech_0, float stoech_1,
      float stoech_2, float stoech_3)
      : rho_0ID(rho_0ID_), rho_1ID(rho_1ID_), rho_2ID(rho_2ID_),
        rho_3ID(rho_3ID_), order_0_(order_0), order_1_(order_1),
        order_2_(order_2), order_3_(order_3),
        rate_coefficient_(rate_coefficient), stoech_0_(stoech_0),
        stoech_1_(stoech_1), stoech_2_(stoech_2), stoech_3_(stoech_3) {
    auto createIdxVector = [](IBlock *const, StructuredBlockStorage *const) {
      return new IndexVectors();
    };
    indexVectorID = blocks->addStructuredBlockData<IndexVectors>(
        createIdxVector,
        "IndexField_ReactionKernelIndexed_4_single_precision_CUDA");
  }
 
  ReactionKernelIndexed_4_single_precision_CUDA(
      BlockDataID indexVectorID_, BlockDataID rho_0ID_, BlockDataID rho_1ID_,
      BlockDataID rho_2ID_, BlockDataID rho_3ID_, float order_0, float order_1,
      float order_2, float order_3, float rate_coefficient, float stoech_0,
      float stoech_1, float stoech_2, float stoech_3)
      : indexVectorID(indexVectorID_), rho_0ID(rho_0ID_), rho_1ID(rho_1ID_),
        rho_2ID(rho_2ID_), rho_3ID(rho_3ID_), order_0_(order_0),
        order_1_(order_1), order_2_(order_2), order_3_(order_3),
        rate_coefficient_(rate_coefficient), stoech_0_(stoech_0),
        stoech_1_(stoech_1), stoech_2_(stoech_2), stoech_3_(stoech_3) {}
 
  void run(IBlock *block, gpuStream_t stream = nullptr);
 
  void operator()(IBlock *block, gpuStream_t stream = nullptr) {
    run(block, stream);
  }
 
  void inner(IBlock *block, gpuStream_t stream = nullptr);
 
  void outer(IBlock *block, gpuStream_t stream = nullptr);
 
  Vector3<double> getForce(IBlock * /*block*/) {
 
    WALBERLA_ABORT(
        "Boundary condition was not generated including force calculation.")
    return Vector3<double>(double_c(0.0));
  }
 
  std::function<void(IBlock *)> getSweep(gpuStream_t stream = nullptr) {
    return [this, stream](IBlock *b) { this->run(b, stream); };
  }
 
  std::function<void(IBlock *)> getInnerSweep(gpuStream_t stream = nullptr) {
    return [this, stream](IBlock *b) { this->inner(b, stream); };
  }
 
  std::function<void(IBlock *)> getOuterSweep(gpuStream_t stream = nullptr) {
    return [this, stream](IBlock *b) { this->outer(b, stream); };
  }
 
  template <typename FlagField_T>
  void fillFromFlagField(const std::shared_ptr<StructuredBlockForest> &blocks,
                         ConstBlockDataID flagFieldID, FlagUID boundaryFlagUID,
                         FlagUID domainFlagUID) {
    for (auto &block : *blocks)
      fillFromFlagField<FlagField_T>(&block, flagFieldID, boundaryFlagUID,
                                     domainFlagUID);
  }
 
  template <typename FlagField_T>
  void fillFromFlagField(IBlock *block, ConstBlockDataID flagFieldID,
                         FlagUID boundaryFlagUID, FlagUID domainFlagUID) {
    auto *indexVectors = block->getData<IndexVectors>(indexVectorID);
    auto &indexVectorAll = indexVectors->indexVector(IndexVectors::ALL);
    auto &indexVectorInner = indexVectors->indexVector(IndexVectors::INNER);
    auto &indexVectorOuter = indexVectors->indexVector(IndexVectors::OUTER);
 
    auto *flagField = block->getData<FlagField_T>(flagFieldID);
 
    if (!(flagField->flagExists(boundaryFlagUID) and
          flagField->flagExists(domainFlagUID)))
      return;
 
    auto boundaryFlag = flagField->getFlag(boundaryFlagUID);
    auto domainFlag = flagField->getFlag(domainFlagUID);
 
    auto inner = flagField->xyzSize();
    inner.expand(cell_idx_t(-1));
 
    indexVectorAll.clear();
    indexVectorInner.clear();
    indexVectorOuter.clear();
 
    auto flagWithGLayers = flagField->xyzSizeWithGhostLayer();
    for (auto it = flagField->beginWithGhostLayerXYZ(); it != flagField->end();
         ++it) {
 
      if (!isFlagSet(it, boundaryFlag))
        continue;
      if (flagWithGLayers.contains(it.x() + cell_idx_c(0),
                                   it.y() + cell_idx_c(0),
                                   it.z() + cell_idx_c(0)) &&
          isFlagSet(it.neighbor(0, 0, 0, 0), domainFlag)) {
 
        auto element = IndexInfo(it.x(), it.y(), it.z(), 0);
 
        indexVectorAll.emplace_back(element);
        if (inner.contains(it.x(), it.y(), it.z()))
          indexVectorInner.emplace_back(element);
        else
          indexVectorOuter.emplace_back(element);
      }
    }
 
    indexVectors->syncGPU();
  }
 
private:
  void run_impl(IBlock *block, IndexVectors::Type type,
                gpuStream_t stream = nullptr);
 
  BlockDataID indexVectorID;
 
public:
  BlockDataID rho_0ID;
  BlockDataID rho_1ID;
  BlockDataID rho_2ID;
  BlockDataID rho_3ID;
  float order_0_;
  float order_1_;
  float order_2_;
  float order_3_;
  float rate_coefficient_;
  float stoech_0_;
  float stoech_1_;
  float stoech_2_;
  float stoech_3_;
};
 
#if defined(__clang__)
#pragma clang diagnostic pop
#elif defined(__GNUC__) or defined(__GNUG__)
#pragma GCC diagnostic pop
#endif
 
} // namespace pystencils
} // namespace walberla