ESPResSo
Extensible Simulation Package for Research on Soft Matter Systems
Loading...
Searching...
No Matches
PackInfoVecSinglePrecisionCUDA.cu
Go to the documentation of this file.
1//======================================================================================================================
2//
3// This file is part of waLBerla. waLBerla is free software: you can
4// redistribute it and/or modify it under the terms of the GNU General Public
5// License as published by the Free Software Foundation, either version 3 of
6// the License, or (at your option) any later version.
7//
8// waLBerla is distributed in the hope that it will be useful, but WITHOUT
9// ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
10// FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
11// for more details.
12//
13// You should have received a copy of the GNU General Public License along
14// with waLBerla (see COPYING.txt). If not, see <http://www.gnu.org/licenses/>.
15//
16//! \\file PackInfoVecSinglePrecisionCUDA.cpp
17//! \\author pystencils
18//======================================================================================================================
19
20// kernel generated with pystencils v1.3.7, lbmpy v1.3.7, sympy v1.12.1, lbmpy_walberla/pystencils_walberla from waLBerla commit f36fa0a68bae59f0b516f6587ea8fa7c24a41141
21
22#include "core/DataTypes.h"
23#include "core/cell/CellInterval.h"
24
25#include "domain_decomposition/IBlock.h"
26
27#include "stencil/Directions.h"
28
29#include "core/DataTypes.h"
30#include "core/cell/CellInterval.h"
31#include "domain_decomposition/IBlock.h"
32#include "stencil/Directions.h"
33
34#include "PackInfoVecSinglePrecisionCUDA.h"
35
36#define FUNC_PREFIX __global__
37
38#if defined(__NVCC__)
39#define RESTRICT __restrict__
40#if defined(__NVCC_DIAG_PRAGMA_SUPPORT__)
41#pragma nv_diagnostic push
42#pragma nv_diag_suppress 177 // unused variable
43#else
44#pragma push
45#pragma diag_suppress 177 // unused variable
46#endif // defined(__NVCC_DIAG_PRAGMA_SUPPORT__)
47#elif defined(__clang__)
48#if defined(__CUDA__)
49#if defined(__CUDA_ARCH__)
50// clang compiling CUDA code in device mode
51#define RESTRICT __restrict__
52#pragma clang diagnostic push
53#pragma clang diagnostic ignored "-Wunused-variable"
54#else
55// clang compiling CUDA code in host mode
56#define RESTRICT __restrict__
57#pragma clang diagnostic push
58#pragma clang diagnostic ignored "-Wunused-variable"
59#endif // defined(__CUDA_ARCH__)
60#endif // defined(__CUDA__)
61#elif defined(__GNUC__) or defined(__GNUG__)
62#define RESTRICT __restrict__
63#pragma GCC diagnostic push
64#pragma GCC diagnostic ignored "-Wunused-variable"
65#elif defined(_MSC_VER)
66#define RESTRICT __restrict
67#else
68#define RESTRICT
69#endif
70
71namespace walberla {
72namespace pystencils {
73
74using walberla::cell::CellInterval;
75using walberla::stencil::Direction;
76
77namespace internal_pack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE {
78static FUNC_PREFIX __launch_bounds__(256) void pack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE(float *RESTRICT _data_buffer, float *RESTRICT const _data_field, int64_t const _size_field_0, int64_t const _size_field_1, int64_t const _size_field_2, int64_t const _stride_field_0, int64_t const _stride_field_1, int64_t const _stride_field_2, int64_t const _stride_field_3) {
79 if (blockDim.x * blockIdx.x + threadIdx.x < _size_field_0 && blockDim.y * blockIdx.y + threadIdx.y < _size_field_1 && blockDim.z * blockIdx.z + threadIdx.z < _size_field_2) {
80 const int64_t ctr_0 = blockDim.x * blockIdx.x + threadIdx.x;
81 const int64_t ctr_1 = blockDim.y * blockIdx.y + threadIdx.y;
82 const int64_t ctr_2 = blockDim.z * blockIdx.z + threadIdx.z;
83 _data_buffer[3 * _size_field_0 * _size_field_1 * ctr_2 + 3 * _size_field_0 * ctr_1 + 3 * ctr_0] = _data_field[_stride_field_0 * ctr_0 + _stride_field_1 * ctr_1 + _stride_field_2 * ctr_2];
84 _data_buffer[3 * _size_field_0 * _size_field_1 * ctr_2 + 3 * _size_field_0 * ctr_1 + 3 * ctr_0 + 1] = _data_field[_stride_field_0 * ctr_0 + _stride_field_1 * ctr_1 + _stride_field_2 * ctr_2 + _stride_field_3];
85 _data_buffer[3 * _size_field_0 * _size_field_1 * ctr_2 + 3 * _size_field_0 * ctr_1 + 3 * ctr_0 + 2] = _data_field[_stride_field_0 * ctr_0 + _stride_field_1 * ctr_1 + _stride_field_2 * ctr_2 + 2 * _stride_field_3];
86 }
87}
88} // namespace internal_pack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE
89
90namespace internal_unpack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE {
91static FUNC_PREFIX __launch_bounds__(256) void unpack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE(float *RESTRICT const _data_buffer, float *RESTRICT _data_field, int64_t const _size_field_0, int64_t const _size_field_1, int64_t const _size_field_2, int64_t const _stride_field_0, int64_t const _stride_field_1, int64_t const _stride_field_2, int64_t const _stride_field_3) {
92 if (blockDim.x * blockIdx.x + threadIdx.x < _size_field_0 && blockDim.y * blockIdx.y + threadIdx.y < _size_field_1 && blockDim.z * blockIdx.z + threadIdx.z < _size_field_2) {
93 const int64_t ctr_0 = blockDim.x * blockIdx.x + threadIdx.x;
94 const int64_t ctr_1 = blockDim.y * blockIdx.y + threadIdx.y;
95 const int64_t ctr_2 = blockDim.z * blockIdx.z + threadIdx.z;
96 _data_field[_stride_field_0 * ctr_0 + _stride_field_1 * ctr_1 + _stride_field_2 * ctr_2] = _data_buffer[3 * _size_field_0 * _size_field_1 * ctr_2 + 3 * _size_field_0 * ctr_1 + 3 * ctr_0];
97 _data_field[_stride_field_0 * ctr_0 + _stride_field_1 * ctr_1 + _stride_field_2 * ctr_2 + _stride_field_3] = _data_buffer[3 * _size_field_0 * _size_field_1 * ctr_2 + 3 * _size_field_0 * ctr_1 + 3 * ctr_0 + 1];
98 _data_field[_stride_field_0 * ctr_0 + _stride_field_1 * ctr_1 + _stride_field_2 * ctr_2 + 2 * _stride_field_3] = _data_buffer[3 * _size_field_0 * _size_field_1 * ctr_2 + 3 * _size_field_0 * ctr_1 + 3 * ctr_0 + 2];
99 }
100}
101} // namespace internal_unpack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE
102
103void PackInfoVecSinglePrecisionCUDA::pack(Direction dir, unsigned char *byte_buffer, IBlock *block, gpuStream_t stream) {
104 float *buffer = reinterpret_cast<float *>(byte_buffer);
105
106 auto field = block->getData<gpu::GPUField<float>>(fieldID);
107
108 CellInterval ci;
109 field->getSliceBeforeGhostLayer(dir, ci, 1, false);
110
111 switch (dir) {
112 case stencil::SW:
113 case stencil::BW:
114 case stencil::W:
115 case stencil::TW:
116 case stencil::NW:
117 case stencil::BS:
118 case stencil::S:
119 case stencil::TS:
120 case stencil::B:
121 case stencil::C:
122 case stencil::T:
123 case stencil::BN:
124 case stencil::N:
125 case stencil::TN:
126 case stencil::SE:
127 case stencil::BE:
128 case stencil::E:
129 case stencil::TE:
130 case stencil::NE: {
131 float *RESTRICT _data_buffer = buffer;
132 WALBERLA_ASSERT_GREATER_EQUAL(ci.xMin(), -int_c(field->nrOfGhostLayers()))
133 WALBERLA_ASSERT_GREATER_EQUAL(ci.yMin(), -int_c(field->nrOfGhostLayers()))
134 WALBERLA_ASSERT_GREATER_EQUAL(ci.zMin(), -int_c(field->nrOfGhostLayers()))
135 float *RESTRICT const _data_field = field->dataAt(ci.xMin(), ci.yMin(), ci.zMin(), 0);
136 WALBERLA_ASSERT_GREATER_EQUAL(field->xSizeWithGhostLayer(), int64_t(int64_c(ci.xSize()) + 0))
137 const int64_t _size_field_0 = int64_t(int64_c(ci.xSize()) + 0);
138 WALBERLA_ASSERT_GREATER_EQUAL(field->ySizeWithGhostLayer(), int64_t(int64_c(ci.ySize()) + 0))
139 const int64_t _size_field_1 = int64_t(int64_c(ci.ySize()) + 0);
140 WALBERLA_ASSERT_GREATER_EQUAL(field->zSizeWithGhostLayer(), int64_t(int64_c(ci.zSize()) + 0))
141 const int64_t _size_field_2 = int64_t(int64_c(ci.zSize()) + 0);
142 const int64_t _stride_field_0 = int64_t(field->xStride());
143 const int64_t _stride_field_1 = int64_t(field->yStride());
144 const int64_t _stride_field_2 = int64_t(field->zStride());
145 const int64_t _stride_field_3 = int64_t(1 * int64_t(field->fStride()));
146 dim3 _block(uint32_c(((128 < _size_field_0) ? 128 : _size_field_0)), uint32_c(((1024 < ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))) ? 1024 : ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))))), uint32_c(((64 < ((_size_field_2 < ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) ? _size_field_2 : ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))))))) ? 64 : ((_size_field_2 < ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) ? _size_field_2 : ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))))));
147 dim3 _grid(uint32_c(((_size_field_0) % (((128 < _size_field_0) ? 128 : _size_field_0)) == 0 ? (int64_t)(_size_field_0) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)) : ((int64_t)(_size_field_0) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))) + 1)), uint32_c(((_size_field_1) % (((1024 < ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))) ? 1024 : ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))))) == 0 ? (int64_t)(_size_field_1) / (int64_t)(((1024 < ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))) ? 1024 : ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))))) : ((int64_t)(_size_field_1) / (int64_t)(((1024 < ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))) ? 1024 : ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) + 1)), uint32_c(((_size_field_2) % (((64 < ((_size_field_2 < ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) ? _size_field_2 : ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))))))) ? 64 : ((_size_field_2 < ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) ? _size_field_2 : ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))))) == 0 ? (int64_t)(_size_field_2) / (int64_t)(((64 < ((_size_field_2 < ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) ? _size_field_2 : ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))))))) ? 64 : ((_size_field_2 < ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) ? _size_field_2 : ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))))) : ((int64_t)(_size_field_2) / (int64_t)(((64 < ((_size_field_2 < ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) ? _size_field_2 : ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))))))) ? 64 : ((_size_field_2 < ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) ? _size_field_2 : ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))))))))) + 1)));
148 internal_pack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::pack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE<<<_grid, _block, 0, stream>>>(_data_buffer, _data_field, _size_field_0, _size_field_1, _size_field_2, _stride_field_0, _stride_field_1, _stride_field_2, _stride_field_3);
149 break;
150 }
151
152 default:
153 return;
154 }
155}
156
157void PackInfoVecSinglePrecisionCUDA::unpack(Direction dir, unsigned char *byte_buffer, IBlock *block, gpuStream_t stream) {
158 float *buffer = reinterpret_cast<float *>(byte_buffer);
159
160 auto field = block->getData<gpu::GPUField<float>>(fieldID);
161
162 CellInterval ci;
163 field->getGhostRegion(dir, ci, 1, false);
164 auto communciationDirection = stencil::inverseDir[dir];
165
166 switch (communciationDirection) {
167 case stencil::SW:
168 case stencil::BW:
169 case stencil::W:
170 case stencil::TW:
171 case stencil::NW:
172 case stencil::BS:
173 case stencil::S:
174 case stencil::TS:
175 case stencil::B:
176 case stencil::C:
177 case stencil::T:
178 case stencil::BN:
179 case stencil::N:
180 case stencil::TN:
181 case stencil::SE:
182 case stencil::BE:
183 case stencil::E:
184 case stencil::TE:
185 case stencil::NE: {
186 float *RESTRICT const _data_buffer = buffer;
187 WALBERLA_ASSERT_GREATER_EQUAL(ci.xMin(), -int_c(field->nrOfGhostLayers()))
188 WALBERLA_ASSERT_GREATER_EQUAL(ci.yMin(), -int_c(field->nrOfGhostLayers()))
189 WALBERLA_ASSERT_GREATER_EQUAL(ci.zMin(), -int_c(field->nrOfGhostLayers()))
190 float *RESTRICT _data_field = field->dataAt(ci.xMin(), ci.yMin(), ci.zMin(), 0);
191 WALBERLA_ASSERT_GREATER_EQUAL(field->xSizeWithGhostLayer(), int64_t(int64_c(ci.xSize()) + 0))
192 const int64_t _size_field_0 = int64_t(int64_c(ci.xSize()) + 0);
193 WALBERLA_ASSERT_GREATER_EQUAL(field->ySizeWithGhostLayer(), int64_t(int64_c(ci.ySize()) + 0))
194 const int64_t _size_field_1 = int64_t(int64_c(ci.ySize()) + 0);
195 WALBERLA_ASSERT_GREATER_EQUAL(field->zSizeWithGhostLayer(), int64_t(int64_c(ci.zSize()) + 0))
196 const int64_t _size_field_2 = int64_t(int64_c(ci.zSize()) + 0);
197 const int64_t _stride_field_0 = int64_t(field->xStride());
198 const int64_t _stride_field_1 = int64_t(field->yStride());
199 const int64_t _stride_field_2 = int64_t(field->zStride());
200 const int64_t _stride_field_3 = int64_t(1 * int64_t(field->fStride()));
201 dim3 _block(uint32_c(((128 < _size_field_0) ? 128 : _size_field_0)), uint32_c(((1024 < ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))) ? 1024 : ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))))), uint32_c(((64 < ((_size_field_2 < ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) ? _size_field_2 : ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))))))) ? 64 : ((_size_field_2 < ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) ? _size_field_2 : ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))))));
202 dim3 _grid(uint32_c(((_size_field_0) % (((128 < _size_field_0) ? 128 : _size_field_0)) == 0 ? (int64_t)(_size_field_0) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)) : ((int64_t)(_size_field_0) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))) + 1)), uint32_c(((_size_field_1) % (((1024 < ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))) ? 1024 : ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))))) == 0 ? (int64_t)(_size_field_1) / (int64_t)(((1024 < ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))) ? 1024 : ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))))) : ((int64_t)(_size_field_1) / (int64_t)(((1024 < ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))) ? 1024 : ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) + 1)), uint32_c(((_size_field_2) % (((64 < ((_size_field_2 < ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) ? _size_field_2 : ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))))))) ? 64 : ((_size_field_2 < ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) ? _size_field_2 : ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))))) == 0 ? (int64_t)(_size_field_2) / (int64_t)(((64 < ((_size_field_2 < ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) ? _size_field_2 : ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))))))) ? 64 : ((_size_field_2 < ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) ? _size_field_2 : ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))))) : ((int64_t)(_size_field_2) / (int64_t)(((64 < ((_size_field_2 < ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) ? _size_field_2 : ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))))))) ? 64 : ((_size_field_2 < ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0))))))) ? _size_field_2 : ((int64_t)(256) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0) * ((_size_field_1 < 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))) ? _size_field_1 : 2 * ((int64_t)(128) / (int64_t)(((128 < _size_field_0) ? 128 : _size_field_0)))))))))) + 1)));
203 internal_unpack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::unpack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE<<<_grid, _block, 0, stream>>>(_data_buffer, _data_field, _size_field_0, _size_field_1, _size_field_2, _stride_field_0, _stride_field_1, _stride_field_2, _stride_field_3);
204 break;
205 }
206
207 default:
208 return;
209 }
210}
211
212uint_t PackInfoVecSinglePrecisionCUDA::size(stencil::Direction dir, IBlock *block) {
213 auto field = block->getData<gpu::GPUField<float>>(fieldID);
214
215 CellInterval ci;
216 field->getGhostRegion(dir, ci, 1, false);
217
218 uint_t elementsPerCell = 0;
219
220 switch (dir) {
221 case stencil::SW:
222 case stencil::BW:
223 case stencil::W:
224 case stencil::TW:
225 case stencil::NW:
226 case stencil::BS:
227 case stencil::S:
228 case stencil::TS:
229 case stencil::B:
230 case stencil::C:
231 case stencil::T:
232 case stencil::BN:
233 case stencil::N:
234 case stencil::TN:
235 case stencil::SE:
236 case stencil::BE:
237 case stencil::E:
238 case stencil::TE:
239 case stencil::NE:
240 elementsPerCell = 3;
241 break;
242
243 default:
244 elementsPerCell = 0;
245 }
246 return ci.numCells() * elementsPerCell * sizeof(float);
247}
248
249} // namespace pystencils
250} // namespace walberla
FUNC_PREFIX
#define FUNC_PREFIX
\file AdvectiveFluxKernel_double_precision.cpp \author pystencils
Definition AdvectiveFluxKernel_double_precision.cpp:28
RESTRICT
#define RESTRICT
\file AdvectiveFluxKernel_double_precision.h \author pystencils
Definition AdvectiveFluxKernel_double_precision.h:42
walberla::pystencils::PackInfoVecSinglePrecisionCUDA::unpack
void unpack(stencil::Direction dir, unsigned char *buffer, IBlock *block, gpuStream_t stream) override
Definition PackInfoVecSinglePrecisionCUDA.cu:157
walberla::pystencils::PackInfoVecSinglePrecisionCUDA::size
uint_t size(stencil::Direction dir, IBlock *block) override
Definition PackInfoVecSinglePrecisionCUDA.cu:212
walberla::pystencils::PackInfoVecSinglePrecisionCUDA::pack
void pack(stencil::Direction dir, unsigned char *buffer, IBlock *block, gpuStream_t stream) override
Definition PackInfoVecSinglePrecisionCUDA.cu:103
stream
cudaStream_t stream[1]
CUDA streams for parallel computing on CPU and GPU.
Definition common_cuda.cu:34
block
static double * block(double *p, std::size_t index, std::size_t size)
Definition elc.cpp:172
walberla::pystencils::internal_pack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::_stride_field_1
static FUNC_PREFIX double *RESTRICT const int64_t const int64_t const int64_t const int64_t const int64_t const _stride_field_1
Definition PackInfoVecDoublePrecisionCUDA.cu:78
walberla::pystencils::internal_pack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::_size_field_2
static FUNC_PREFIX double *RESTRICT const int64_t const int64_t const int64_t const _size_field_2
Definition PackInfoVecDoublePrecisionCUDA.cu:78
walberla::pystencils::internal_pack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::__launch_bounds__
static FUNC_PREFIX __launch_bounds__(256) void pack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE(double *RESTRICT _data_buffer
walberla::pystencils::internal_pack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::_data_field
static FUNC_PREFIX double *RESTRICT const _data_field
Definition PackInfoVecDoublePrecisionCUDA.cu:78
walberla::pystencils::internal_pack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::_size_field_1
static FUNC_PREFIX double *RESTRICT const int64_t const int64_t const _size_field_1
Definition PackInfoVecDoublePrecisionCUDA.cu:78
walberla::pystencils::internal_pack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::_stride_field_2
static FUNC_PREFIX double *RESTRICT const int64_t const int64_t const int64_t const int64_t const int64_t const int64_t const _stride_field_2
Definition PackInfoVecDoublePrecisionCUDA.cu:78
walberla::pystencils::internal_pack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::_stride_field_0
static FUNC_PREFIX double *RESTRICT const int64_t const int64_t const int64_t const int64_t const _stride_field_0
Definition PackInfoVecDoublePrecisionCUDA.cu:78
walberla::pystencils::internal_pack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::_size_field_0
static FUNC_PREFIX double *RESTRICT const int64_t const _size_field_0
Definition PackInfoVecDoublePrecisionCUDA.cu:78
walberla::pystencils::internal_unpack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::_size_field_0
static FUNC_PREFIX double *RESTRICT int64_t const _size_field_0
Definition PackInfoVecDoublePrecisionCUDA.cu:91
walberla::pystencils::internal_unpack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::_stride_field_2
static FUNC_PREFIX double *RESTRICT int64_t const int64_t const int64_t const int64_t const int64_t const int64_t const _stride_field_2
Definition PackInfoVecDoublePrecisionCUDA.cu:91
walberla::pystencils::internal_unpack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::_data_field
static FUNC_PREFIX double *RESTRICT _data_field
Definition PackInfoVecDoublePrecisionCUDA.cu:91
walberla::pystencils::internal_unpack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::_size_field_1
static FUNC_PREFIX double *RESTRICT int64_t const int64_t const _size_field_1
Definition PackInfoVecDoublePrecisionCUDA.cu:91
walberla::pystencils::internal_unpack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::__launch_bounds__
static FUNC_PREFIX __launch_bounds__(256) void unpack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE(double *RESTRICT const _data_buffer
walberla::pystencils::internal_unpack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::_stride_field_0
static FUNC_PREFIX double *RESTRICT int64_t const int64_t const int64_t const int64_t const _stride_field_0
Definition PackInfoVecDoublePrecisionCUDA.cu:91
walberla::pystencils::internal_unpack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::_stride_field_1
static FUNC_PREFIX double *RESTRICT int64_t const int64_t const int64_t const int64_t const int64_t const _stride_field_1
Definition PackInfoVecDoublePrecisionCUDA.cu:91
walberla::pystencils::internal_unpack_SW_BW_W_TW_NW_BS_S_TS_B_C_T_BN_N_TN_SE_BE_E_TE_NE::_size_field_2
static FUNC_PREFIX double *RESTRICT int64_t const int64_t const int64_t const _size_field_2
Definition PackInfoVecDoublePrecisionCUDA.cu:91
walberla
\file PackInfoPdfDoublePrecision.cpp \author pystencils
Definition EKWalberla.hpp:36