1 // Copyright (c) 2017-2025, Lawrence Livermore National Security, LLC and other CEED contributors. 2 // All Rights Reserved. See the top-level LICENSE and NOTICE files for details. 3 // 4 // SPDX-License-Identifier: BSD-2-Clause 5 // 6 // This file is part of CEED: http://github.com/ceed 7 8 #define CEED_DEBUG_COLOR 12 9 10 #include <ceed.h> 11 #include <ceed/backend.h> 12 #include <ceed/gen-tools.h> 13 #include <ceed/jit-tools.h> 14 #include <cuda_runtime.h> 15 16 #include <iostream> 17 #include <sstream> 18 #include <string> 19 20 #include "../cuda-ref/ceed-cuda-ref.h" 21 #include "../cuda-shared/ceed-cuda-shared.h" 22 #include "../cuda/ceed-cuda-common.h" 23 #include "../cuda/ceed-cuda-compile.h" 24 #include "ceed-cuda-gen.h" 25 26 struct FieldReuse_Cuda { 27 CeedInt index; 28 bool is_input; 29 CeedEvalMode eval_mode; 30 }; 31 32 //------------------------------------------------------------------------------ 33 // Determine type of operator 34 //------------------------------------------------------------------------------ 35 static int CeedOperatorBuildKernelData_Cuda_gen(Ceed ceed, CeedInt num_input_fields, CeedOperatorField *op_input_fields, 36 CeedQFunctionField *qf_input_fields, CeedInt num_output_fields, CeedOperatorField *op_output_fields, 37 CeedQFunctionField *qf_output_fields, CeedInt *max_P, CeedInt *max_P_1d, CeedInt *Q, CeedInt *Q_1d, 38 CeedInt *max_dim, bool *is_all_tensor, bool *use_3d_slices) { 39 // Check if all are tensor 40 *is_all_tensor = true; 41 for (CeedInt i = 0; i < num_input_fields; i++) { 42 CeedBasis basis; 43 44 CeedCallBackend(CeedOperatorFieldGetBasis(op_input_fields[i], &basis)); 45 if (basis != CEED_BASIS_NONE) { 46 bool is_field_tensor; 47 48 CeedCallBackend(CeedBasisIsTensor(basis, &is_field_tensor)); 49 *is_all_tensor = *is_all_tensor && is_field_tensor; 50 } 51 CeedCallBackend(CeedBasisDestroy(&basis)); 52 } 53 for (CeedInt i = 0; i < num_output_fields; i++) { 54 CeedBasis basis; 55 56 CeedCallBackend(CeedOperatorFieldGetBasis(op_output_fields[i], &basis)); 57 if (basis != CEED_BASIS_NONE) { 58 bool is_field_tensor; 59 60 CeedCallBackend(CeedBasisIsTensor(basis, &is_field_tensor)); 61 *is_all_tensor = *is_all_tensor && is_field_tensor; 62 } 63 CeedCallBackend(CeedBasisDestroy(&basis)); 64 } 65 66 // Find max_P, max_P_1d, Q, and Q_1d 67 bool is_all_3d = true; 68 69 *max_P = 0; 70 *max_P_1d = 0; 71 *Q = 0; 72 *Q_1d = 0; 73 for (CeedInt i = 0; i < num_input_fields; i++) { 74 CeedBasis basis; 75 76 CeedCallBackend(CeedOperatorFieldGetBasis(op_input_fields[i], &basis)); 77 if (basis != CEED_BASIS_NONE) { 78 bool is_field_tensor; 79 CeedInt field_dim = 0, field_P = 0, field_P_1d = 0, field_Q = 0, field_Q_1d = 0; 80 81 // Check if 3D 82 CeedCallBackend(CeedBasisGetDimension(basis, &field_dim)); 83 is_all_3d = is_all_3d && (field_dim == 3); 84 *max_dim = CeedIntMax(*max_dim, field_dim); 85 86 // Collect P, P_1d, Q, and Q_1d 87 CeedCallBackend(CeedBasisGetNumNodes(basis, &field_P)); 88 *max_P = CeedIntMax(*max_P, field_P); 89 CeedCallBackend(CeedBasisIsTensor(basis, &is_field_tensor)); 90 if (is_field_tensor) { 91 CeedCallBackend(CeedBasisGetNumNodes1D(basis, &field_P_1d)); 92 *max_P_1d = CeedIntMax(*max_P_1d, field_P_1d); 93 } 94 CeedCallBackend(CeedBasisGetNumQuadraturePoints(basis, &field_Q)); 95 CeedCheck(*Q == 0 || field_Q == *Q, ceed, CEED_ERROR_BACKEND, "Quadrature spaces must be compatible"); 96 *Q = field_Q; 97 if (is_field_tensor) { 98 CeedCallBackend(CeedBasisGetNumQuadraturePoints1D(basis, &field_Q_1d)); 99 CeedCheck(*Q_1d == 0 || field_Q_1d == *Q_1d, ceed, CEED_ERROR_BACKEND, "Quadrature spaces must be compatible"); 100 *Q_1d = field_Q_1d; 101 } 102 } 103 CeedCallBackend(CeedBasisDestroy(&basis)); 104 } 105 for (CeedInt i = 0; i < num_output_fields; i++) { 106 CeedBasis basis; 107 108 CeedCallBackend(CeedOperatorFieldGetBasis(op_output_fields[i], &basis)); 109 if (basis != CEED_BASIS_NONE) { 110 bool is_field_tensor; 111 CeedInt field_dim = 0, field_P = 0, field_P_1d = 0, field_Q = 0, field_Q_1d = 0; 112 113 // Check if 3D 114 CeedCallBackend(CeedBasisGetDimension(basis, &field_dim)); 115 is_all_3d = is_all_3d && (field_dim == 3); 116 *max_dim = CeedIntMax(*max_dim, field_dim); 117 118 // Collect P, P_1d, Q, and Q_1d 119 CeedCallBackend(CeedBasisGetNumNodes(basis, &field_P)); 120 *max_P = CeedIntMax(*max_P, field_P); 121 CeedCallBackend(CeedBasisIsTensor(basis, &is_field_tensor)); 122 if (is_field_tensor) { 123 CeedCallBackend(CeedBasisGetNumNodes1D(basis, &field_P_1d)); 124 *max_P_1d = CeedIntMax(*max_P_1d, field_P_1d); 125 } 126 CeedCallBackend(CeedBasisGetNumQuadraturePoints(basis, &field_Q)); 127 CeedCheck(*Q == 0 || field_Q == *Q, ceed, CEED_ERROR_BACKEND, "Quadrature spaces must be compatible"); 128 *Q = field_Q; 129 if (is_field_tensor) { 130 CeedCallBackend(CeedBasisGetNumQuadraturePoints1D(basis, &field_Q_1d)); 131 CeedCheck(*Q_1d == 0 || field_Q_1d == *Q_1d, ceed, CEED_ERROR_BACKEND, "Quadrature spaces must be compatible"); 132 *Q_1d = field_Q_1d; 133 } 134 } 135 CeedCallBackend(CeedBasisDestroy(&basis)); 136 } 137 138 // Only use 3D collocated gradient parallelization strategy when gradient is computed 139 *use_3d_slices = false; 140 if (is_all_3d && *is_all_tensor) { 141 bool was_grad_found = false; 142 143 for (CeedInt i = 0; i < num_input_fields; i++) { 144 CeedEvalMode eval_mode; 145 146 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[i], &eval_mode)); 147 if (eval_mode == CEED_EVAL_GRAD) { 148 CeedBasis_Cuda_shared *basis_data; 149 CeedBasis basis; 150 151 CeedCallBackend(CeedOperatorFieldGetBasis(op_input_fields[i], &basis)); 152 CeedCallBackend(CeedBasisGetData(basis, &basis_data)); 153 *use_3d_slices = basis_data->d_collo_grad_1d && (was_grad_found ? *use_3d_slices : true); 154 was_grad_found = true; 155 CeedCallBackend(CeedBasisDestroy(&basis)); 156 } 157 } 158 for (CeedInt i = 0; i < num_output_fields; i++) { 159 CeedEvalMode eval_mode; 160 161 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_output_fields[i], &eval_mode)); 162 if (eval_mode == CEED_EVAL_GRAD) { 163 CeedBasis_Cuda_shared *basis_data; 164 CeedBasis basis; 165 166 CeedCallBackend(CeedOperatorFieldGetBasis(op_output_fields[i], &basis)); 167 CeedCallBackend(CeedBasisGetData(basis, &basis_data)); 168 *use_3d_slices = basis_data->d_collo_grad_1d && (was_grad_found ? *use_3d_slices : true); 169 was_grad_found = true; 170 CeedCallBackend(CeedBasisDestroy(&basis)); 171 } 172 } 173 } 174 return CEED_ERROR_SUCCESS; 175 } 176 177 //------------------------------------------------------------------------------ 178 // Setup fields 179 //------------------------------------------------------------------------------ 180 static int CeedOperatorBuildKernelFieldData_Cuda_gen(std::ostringstream &code, CeedOperator_Cuda_gen *data, Tab &tab, CeedInt i, 181 CeedOperatorField op_field, CeedQFunctionField qf_field, FieldReuse_Cuda field_reuse, 182 CeedInt max_dim, CeedInt Q, CeedInt Q_1d, bool is_input, bool is_all_tensor, bool is_at_points, 183 bool use_3d_slices, bool skip_active_load) { 184 bool is_tensor = true, is_active = true; 185 CeedBasis basis; 186 187 CeedCallBackend(CeedOperatorFieldGetBasis(op_field, &basis)); 188 if (basis != CEED_BASIS_NONE) CeedCallBackend(CeedBasisIsTensor(basis, &is_tensor)); 189 { 190 CeedVector vec; 191 192 CeedCallBackend(CeedOperatorFieldGetVector(op_field, &vec)); 193 is_active = vec == CEED_VECTOR_ACTIVE; 194 CeedCallBackend(CeedVectorDestroy(&vec)); 195 } 196 197 const char *field_name; 198 std::string var_suffix = (is_input ? "_in_" : "_out_") + std::to_string(i); 199 std::string P_name = (is_tensor ? "P_1d" : "P") + var_suffix, Q_name = is_tensor ? "Q_1d" : "Q"; 200 std::string option_name = (is_input ? "inputs" : "outputs"); 201 CeedEvalMode eval_mode = CEED_EVAL_NONE; 202 CeedInt elem_size = 0, num_comp = 0, dim = max_dim, P_1d = 0; 203 CeedElemRestriction elem_rstr; 204 CeedBasis_Cuda_shared *basis_data; 205 206 // Field reuse info 207 bool use_previous_field = field_reuse.index != -1; 208 209 CeedCallBackend(CeedOperatorFieldGetName(op_field, &field_name)); 210 code << tab << "// -- " << (is_input ? "Input" : "Output") << " field " << i << ": " << field_name << "\n"; 211 212 // Get field data 213 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_field, &elem_rstr)); 214 if (elem_rstr != CEED_ELEMRESTRICTION_NONE) { 215 CeedCallBackend(CeedElemRestrictionGetElementSize(elem_rstr, &elem_size)); 216 CeedCallBackend(CeedElemRestrictionGetNumComponents(elem_rstr, &num_comp)); 217 } 218 CeedCallBackend(CeedElemRestrictionDestroy(&elem_rstr)); 219 if (basis != CEED_BASIS_NONE) { 220 CeedCallBackend(CeedBasisGetData(basis, &basis_data)); 221 CeedCallBackend(CeedBasisGetDimension(basis, &dim)); 222 if (is_tensor) CeedCallBackend(CeedBasisGetNumNodes1D(basis, &P_1d)); 223 else CeedCallBackend(CeedBasisGetNumNodes(basis, &P_1d)); 224 } 225 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_field, &eval_mode)); 226 227 // Set field constants 228 code << tab << "const CeedInt dim" << var_suffix << " = " << dim << ";\n"; 229 if (is_tensor && !is_all_tensor) { 230 CeedInt P = 0; 231 232 CeedCallBackend(CeedBasisGetNumNodes(basis, &P)); 233 code << tab << "const CeedInt P" << var_suffix << " = " << (basis == CEED_BASIS_NONE ? Q : P) << ";\n"; 234 } 235 code << tab << "const CeedInt " << P_name << " = " << (basis == CEED_BASIS_NONE ? Q_1d : P_1d) << ";\n"; 236 if (eval_mode != CEED_EVAL_WEIGHT) { 237 code << tab << "const CeedInt num_comp" << var_suffix << " = " << num_comp << ";\n"; 238 } 239 240 // Load basis data 241 code << tab << "// EvalMode: " << CeedEvalModes[eval_mode] << "\n"; 242 switch (eval_mode) { 243 case CEED_EVAL_NONE: 244 break; 245 case CEED_EVAL_INTERP: 246 if (is_at_points) { 247 // AtPoints 248 if (!basis_data->d_chebyshev_interp_1d) { 249 CeedSize interp_bytes; 250 CeedScalar *chebyshev_interp_1d; 251 252 interp_bytes = P_1d * Q_1d * sizeof(CeedScalar); 253 CeedCallBackend(CeedCalloc(P_1d * Q_1d, &chebyshev_interp_1d)); 254 CeedCallBackend(CeedBasisGetChebyshevInterp1D(basis, chebyshev_interp_1d)); 255 CeedCallCuda(CeedBasisReturnCeed(basis), cudaMalloc((void **)&basis_data->d_chebyshev_interp_1d, interp_bytes)); 256 CeedCallCuda(CeedBasisReturnCeed(basis), 257 cudaMemcpy(basis_data->d_chebyshev_interp_1d, chebyshev_interp_1d, interp_bytes, cudaMemcpyHostToDevice)); 258 CeedCallBackend(CeedFree(&chebyshev_interp_1d)); 259 } 260 if (is_input) data->B.inputs[i] = basis_data->d_chebyshev_interp_1d; 261 else data->B.outputs[i] = basis_data->d_chebyshev_interp_1d; 262 } else { 263 // Standard quadrature 264 if (is_input) data->B.inputs[i] = basis_data->d_interp_1d; 265 else data->B.outputs[i] = basis_data->d_interp_1d; 266 } 267 if (use_previous_field && !skip_active_load) { 268 std::string reuse_var = "s_B" + ((field_reuse.is_input ? "_in_" : "_out_") + std::to_string(field_reuse.index)); 269 270 code << tab << "CeedScalar *s_B" << var_suffix << " = " << reuse_var << ";\n"; 271 } else { 272 bool is_collocated = false; 273 274 CeedCallBackend(CeedBasisIsCollocated(basis, &is_collocated)); 275 if ((is_active && skip_active_load) || (is_collocated && !is_at_points)) { 276 code << tab << "CeedScalar *s_B" << var_suffix << " = NULL;\n"; 277 } else { 278 code << tab << "__shared__ CeedScalar s_B" << var_suffix << "[" << P_name << "*" << Q_name << "];\n"; 279 code << tab << "LoadMatrix<" << P_name << ", " << Q_name << ">(data, B." << option_name << "[" << i << "], s_B" << var_suffix << ");\n"; 280 } 281 } 282 break; 283 case CEED_EVAL_GRAD: 284 if (is_at_points) { 285 // AtPoints 286 if (!basis_data->d_chebyshev_interp_1d) { 287 CeedSize interp_bytes; 288 CeedScalar *chebyshev_interp_1d; 289 290 interp_bytes = P_1d * Q_1d * sizeof(CeedScalar); 291 CeedCallBackend(CeedCalloc(P_1d * Q_1d, &chebyshev_interp_1d)); 292 CeedCallBackend(CeedBasisGetChebyshevInterp1D(basis, chebyshev_interp_1d)); 293 CeedCallCuda(CeedBasisReturnCeed(basis), cudaMalloc((void **)&basis_data->d_chebyshev_interp_1d, interp_bytes)); 294 CeedCallCuda(CeedBasisReturnCeed(basis), 295 cudaMemcpy(basis_data->d_chebyshev_interp_1d, chebyshev_interp_1d, interp_bytes, cudaMemcpyHostToDevice)); 296 CeedCallBackend(CeedFree(&chebyshev_interp_1d)); 297 } 298 if (is_input) data->B.inputs[i] = basis_data->d_chebyshev_interp_1d; 299 else data->B.outputs[i] = basis_data->d_chebyshev_interp_1d; 300 } else { 301 // Standard quadrature 302 if (is_input) data->B.inputs[i] = basis_data->d_interp_1d; 303 else data->B.outputs[i] = basis_data->d_interp_1d; 304 } 305 if (is_tensor) { 306 if (use_previous_field && !skip_active_load) { 307 std::string reuse_var = "s_B" + ((field_reuse.is_input ? "_in_" : "_out_") + std::to_string(field_reuse.index)); 308 309 code << tab << "CeedScalar *s_B" << var_suffix << " = " << reuse_var << ";\n"; 310 } else { 311 bool is_collocated = false; 312 313 CeedCallBackend(CeedBasisIsCollocated(basis, &is_collocated)); 314 if ((is_active && skip_active_load) || (is_collocated && !is_at_points)) { 315 code << tab << "CeedScalar *s_B" << var_suffix << " = NULL;\n"; 316 } else { 317 code << tab << "__shared__ CeedScalar s_B" << var_suffix << "[" << P_name << "*" << Q_name << "];\n"; 318 code << tab << "LoadMatrix<" << P_name << ", " << Q_name << ">(data, B." << option_name << "[" << i << "], s_B" << var_suffix << ");\n"; 319 } 320 } 321 } 322 if (is_at_points) break; // No G mat for AtPoints 323 if (use_3d_slices) { 324 if (is_input) data->G.inputs[i] = basis_data->d_collo_grad_1d; 325 else data->G.outputs[i] = basis_data->d_collo_grad_1d; 326 if (use_previous_field && field_reuse.eval_mode == CEED_EVAL_GRAD && !skip_active_load) { 327 std::string reuse_var = "s_G" + ((field_reuse.is_input ? "_in_" : "_out_") + std::to_string(field_reuse.index)); 328 329 code << tab << "CeedScalar *s_G" << var_suffix << " = " << reuse_var << ";\n"; 330 } else if (is_active && skip_active_load) { 331 code << tab << "CeedScalar *s_G" << var_suffix << " = NULL;\n"; 332 } else { 333 code << tab << "__shared__ CeedScalar s_G" << var_suffix << "[" << Q_name << "*" << Q_name << "];\n"; 334 code << tab << "LoadMatrix<" << Q_name << ", " << Q_name << ">(data, G." << option_name << "[" << i << "], s_G" << var_suffix << ");\n"; 335 } 336 } else { 337 bool has_collo_grad = basis_data->d_collo_grad_1d; 338 339 if (is_input) data->G.inputs[i] = has_collo_grad ? basis_data->d_collo_grad_1d : basis_data->d_grad_1d; 340 else data->G.outputs[i] = has_collo_grad ? basis_data->d_collo_grad_1d : basis_data->d_grad_1d; 341 if (has_collo_grad) { 342 if (use_previous_field && field_reuse.eval_mode == CEED_EVAL_GRAD && !skip_active_load) { 343 std::string reuse_var = "s_G" + ((field_reuse.is_input ? "_in_" : "_out_") + std::to_string(field_reuse.index)); 344 345 code << tab << "CeedScalar *s_G" << var_suffix << " = " << reuse_var << ";\n"; 346 } else if (is_active && skip_active_load) { 347 code << tab << "CeedScalar *s_G" << var_suffix << " = NULL;\n"; 348 } else { 349 code << tab << "__shared__ CeedScalar s_G" << var_suffix << "[" << Q_name << "*" << Q_name << "];\n"; 350 code << tab << "LoadMatrix<" << Q_name << ", " << Q_name << ">(data, G." << option_name << "[" << i << "], s_G" << var_suffix << ");\n"; 351 } 352 } else { 353 if (use_previous_field && field_reuse.eval_mode == CEED_EVAL_GRAD && !skip_active_load) { 354 std::string reuse_var = "s_G" + ((field_reuse.is_input ? "_in_" : "_out_") + std::to_string(field_reuse.index)); 355 356 code << tab << "CeedScalar *s_G" << var_suffix << " = " << reuse_var << ";\n"; 357 } else if (is_active && skip_active_load) { 358 code << tab << "CeedScalar *s_G" << var_suffix << " = NULL;\n"; 359 } else { 360 code << tab << "__shared__ CeedScalar s_G" << var_suffix << "[" << P_name << "*" << Q_name << (is_tensor ? "" : "*dim") 361 << (is_tensor ? "" : var_suffix) << "];\n"; 362 code << tab << "LoadMatrix<" << P_name << ", " << Q_name << (is_tensor ? "" : "*dim") << (is_tensor ? "" : var_suffix) << ">(data, G." 363 << option_name << "[" << i << "], s_G" << var_suffix << ");\n"; 364 } 365 } 366 } 367 break; 368 case CEED_EVAL_WEIGHT: 369 break; // No action 370 // LCOV_EXCL_START 371 case CEED_EVAL_DIV: 372 case CEED_EVAL_CURL: 373 break; // TODO: Not implemented 374 // LCOV_EXCL_STOP 375 } 376 CeedCallBackend(CeedBasisDestroy(&basis)); 377 return CEED_ERROR_SUCCESS; 378 } 379 380 //------------------------------------------------------------------------------ 381 // Restriction 382 //------------------------------------------------------------------------------ 383 static int CeedOperatorBuildKernelRestriction_Cuda_gen(std::ostringstream &code, CeedOperator_Cuda_gen *data, Tab &tab, CeedInt i, 384 CeedInt field_input_buffer[], CeedOperatorField op_field, CeedQFunctionField qf_field, 385 CeedInt max_dim, CeedInt Q_1d, bool is_input, bool is_all_tensor, bool is_at_points, 386 bool use_3d_slices) { 387 std::string var_suffix = (is_input ? "_in_" : "_out_") + std::to_string(i); 388 std::string P_name = (is_all_tensor ? "P_1d" : "P") + var_suffix; 389 CeedEvalMode eval_mode = CEED_EVAL_NONE; 390 CeedInt elem_size = 0, num_comp = 0; 391 CeedSize l_size; 392 CeedRestrictionType rstr_type = CEED_RESTRICTION_STANDARD; 393 CeedElemRestriction_Cuda *rstr_data; 394 CeedElemRestriction elem_rstr; 395 396 // Get field data 397 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_field, &elem_rstr)); 398 if (elem_rstr != CEED_ELEMRESTRICTION_NONE) { 399 CeedCallBackend(CeedElemRestrictionGetType(elem_rstr, &rstr_type)); 400 CeedCallBackend(CeedElemRestrictionGetElementSize(elem_rstr, &elem_size)); 401 CeedCallBackend(CeedElemRestrictionGetNumComponents(elem_rstr, &num_comp)); 402 CeedCallBackend(CeedElemRestrictionGetData(elem_rstr, &rstr_data)); 403 } 404 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_field, &eval_mode)); 405 406 // Restriction 407 if (is_input) { 408 // Input 409 if (field_input_buffer[i] != i) { 410 std::string buffer_name = "r_e_in_" + std::to_string(field_input_buffer[i]); 411 412 // Restriction was already done for previous input 413 code << tab << "CeedScalar *r_e" << var_suffix << " = " << buffer_name << ";\n"; 414 } else if (eval_mode != CEED_EVAL_WEIGHT && !((eval_mode == CEED_EVAL_NONE) && use_3d_slices && is_at_points)) { 415 if (eval_mode == CEED_EVAL_NONE && rstr_type != CEED_RESTRICTION_POINTS) { 416 // No basis action, so r_e_in_* in also r_q_in_* and needs to be allocated 417 code << tab << "CeedScalar r_e" << var_suffix << "[num_comp" << var_suffix << "*" << P_name << "];\n"; 418 } else if (rstr_type != CEED_RESTRICTION_POINTS) { 419 // Otherwise we're using the scratch space 420 code << tab << "CeedScalar *r_e" << var_suffix << " = r_e_scratch;\n"; 421 } 422 switch (rstr_type) { 423 case CEED_RESTRICTION_STANDARD: { 424 CeedInt comp_stride; 425 426 CeedCallBackend(CeedElemRestrictionGetLVectorSize(elem_rstr, &l_size)); 427 code << tab << "const CeedInt l_size" << var_suffix << " = " << l_size << ";\n"; 428 CeedCallBackend(CeedElemRestrictionGetCompStride(elem_rstr, &comp_stride)); 429 code << tab << "const CeedInt comp_stride" << var_suffix << " = " << comp_stride << ";\n"; 430 data->indices.inputs[i] = (CeedInt *)rstr_data->d_offsets; 431 code << tab << "ReadLVecStandard" << (is_all_tensor ? max_dim : 1) << "d<num_comp" << var_suffix << ", comp_stride" << var_suffix << ", " 432 << P_name << ">(data, l_size" << var_suffix << ", elem, indices.inputs[" << i << "], d" << var_suffix << ", r_e" << var_suffix 433 << ");\n"; 434 break; 435 } 436 case CEED_RESTRICTION_STRIDED: { 437 bool has_backend_strides; 438 CeedInt num_elem; 439 440 CeedCallBackend(CeedElemRestrictionHasBackendStrides(elem_rstr, &has_backend_strides)); 441 CeedCallBackend(CeedElemRestrictionGetNumElements(elem_rstr, &num_elem)); 442 CeedInt strides[3] = {1, elem_size * num_elem, elem_size}; 443 444 if (!has_backend_strides) { 445 CeedCallBackend(CeedElemRestrictionGetStrides(elem_rstr, strides)); 446 } 447 code << tab << "const CeedInt strides" << var_suffix << "_0 = " << strides[0] << ", strides" << var_suffix << "_1 = " << strides[1] 448 << ", strides" << var_suffix << "_2 = " << strides[2] << ";\n"; 449 code << tab << "ReadLVecStrided" << (is_all_tensor ? max_dim : 1) << "d<num_comp" << var_suffix << ", " << P_name << ", strides" 450 << var_suffix << "_0, strides" << var_suffix << "_1, strides" << var_suffix << "_2>(data, elem, d" << var_suffix << ", r_e" 451 << var_suffix << ");\n"; 452 break; 453 } 454 case CEED_RESTRICTION_POINTS: { 455 CeedInt comp_stride; 456 457 CeedCallBackend(CeedElemRestrictionGetCompStride(elem_rstr, &comp_stride)); 458 code << tab << "const CeedInt comp_stride" << var_suffix << " = " << comp_stride << ";\n"; 459 data->indices.inputs[i] = (CeedInt *)rstr_data->d_offsets; 460 break; 461 } 462 // LCOV_EXCL_START 463 case CEED_RESTRICTION_ORIENTED: 464 case CEED_RESTRICTION_CURL_ORIENTED: 465 break; // TODO: Not implemented 466 // LCOV_EXCL_STOP 467 } 468 } 469 } else { 470 // Output 471 switch (rstr_type) { 472 case CEED_RESTRICTION_STANDARD: { 473 CeedInt comp_stride; 474 475 CeedCallBackend(CeedElemRestrictionGetLVectorSize(elem_rstr, &l_size)); 476 code << tab << "const CeedInt l_size" << var_suffix << " = " << l_size << ";\n"; 477 CeedCallBackend(CeedElemRestrictionGetCompStride(elem_rstr, &comp_stride)); 478 code << tab << "const CeedInt comp_stride" << var_suffix << " = " << comp_stride << ";\n"; 479 data->indices.outputs[i] = (CeedInt *)rstr_data->d_offsets; 480 code << tab << "WriteLVecStandard" << (is_all_tensor ? max_dim : 1) << "d<num_comp" << var_suffix << ", comp_stride" << var_suffix << ", " 481 << P_name << ">(data, l_size" << var_suffix << ", elem, indices.outputs[" << i << "], r_e" << var_suffix << ", d" << var_suffix 482 << ");\n"; 483 break; 484 } 485 case CEED_RESTRICTION_STRIDED: { 486 bool has_backend_strides; 487 CeedInt num_elem; 488 489 CeedCallBackend(CeedElemRestrictionHasBackendStrides(elem_rstr, &has_backend_strides)); 490 CeedCallBackend(CeedElemRestrictionGetNumElements(elem_rstr, &num_elem)); 491 CeedInt strides[3] = {1, elem_size * num_elem, elem_size}; 492 493 if (!has_backend_strides) { 494 CeedCallBackend(CeedElemRestrictionGetStrides(elem_rstr, strides)); 495 } 496 code << tab << "const CeedInt strides" << var_suffix << "_0 = " << strides[0] << ", strides" << var_suffix << "_1 = " << strides[1] 497 << ", strides" << var_suffix << "_2 = " << strides[2] << ";\n"; 498 code << tab << "WriteLVecStrided" << (is_all_tensor ? max_dim : 1) << "d<num_comp" << var_suffix << ", " << P_name << ", strides" 499 << var_suffix << "_0, strides" << var_suffix << "_1, strides" << var_suffix << "_2>(data, elem, r_e" << var_suffix << ", d" << var_suffix 500 << ");\n"; 501 break; 502 } 503 case CEED_RESTRICTION_POINTS: 504 data->indices.outputs[i] = (CeedInt *)rstr_data->d_offsets; 505 break; 506 // LCOV_EXCL_START 507 case CEED_RESTRICTION_ORIENTED: 508 case CEED_RESTRICTION_CURL_ORIENTED: 509 break; // TODO: Not implemented 510 // LCOV_EXCL_STOP 511 } 512 } 513 CeedCallBackend(CeedElemRestrictionDestroy(&elem_rstr)); 514 return CEED_ERROR_SUCCESS; 515 } 516 517 //------------------------------------------------------------------------------ 518 // Basis 519 //------------------------------------------------------------------------------ 520 static int CeedOperatorBuildKernelBasis_Cuda_gen(std::ostringstream &code, CeedOperator_Cuda_gen *data, Tab &tab, CeedInt i, 521 CeedOperatorField op_field, CeedQFunctionField qf_field, CeedInt max_dim, CeedInt Q_1d, 522 bool is_input, bool is_all_tensor, bool is_at_points, bool use_3d_slices) { 523 bool is_tensor = true, is_collocated = true; 524 CeedBasis basis; 525 CeedCallBackend(CeedOperatorFieldGetBasis(op_field, &basis)); 526 CeedCallBackend(CeedBasisIsTensor(basis, &is_tensor)); 527 CeedCallBackend(CeedBasisIsCollocated(basis, &is_collocated)); 528 529 std::string var_suffix = (is_input ? "_in_" : "_out_") + std::to_string(i); 530 std::string P_name = (is_tensor ? "P_1d" : "P") + var_suffix, Q_name = is_tensor ? "Q_1d" : "Q"; 531 CeedEvalMode eval_mode = CEED_EVAL_NONE; 532 CeedInt dim = max_dim, elem_size = 0, num_comp = 0, P_1d = 0; 533 CeedElemRestriction elem_rstr; 534 535 // Get field data 536 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_field, &elem_rstr)); 537 if (elem_rstr != CEED_ELEMRESTRICTION_NONE) { 538 CeedCallBackend(CeedElemRestrictionGetElementSize(elem_rstr, &elem_size)); 539 CeedCallBackend(CeedElemRestrictionGetNumComponents(elem_rstr, &num_comp)); 540 } 541 CeedCallBackend(CeedElemRestrictionDestroy(&elem_rstr)); 542 if (basis != CEED_BASIS_NONE) { 543 CeedCallBackend(CeedBasisGetDimension(basis, &dim)); 544 if (is_tensor) CeedCallBackend(CeedBasisGetNumNodes1D(basis, &P_1d)); 545 else CeedCallBackend(CeedBasisGetNumNodes(basis, &P_1d)); 546 } 547 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_field, &eval_mode)); 548 549 // Basis 550 code << tab << "// EvalMode: " << CeedEvalModes[eval_mode] << "\n"; 551 if (is_input) { 552 switch (eval_mode) { 553 case CEED_EVAL_NONE: 554 if (!use_3d_slices && !is_at_points) { 555 code << tab << "CeedScalar *r_q" << var_suffix << " = r_e" << var_suffix << ";\n"; 556 } 557 break; 558 case CEED_EVAL_INTERP: 559 if (is_at_points) { 560 std::string function_name = (dim == 1 ? "Interp" : "InterpTensor") + std::to_string(dim) + "d"; 561 562 code << tab << "CeedScalar r_c" << var_suffix << "[num_comp" << var_suffix << "*" << (dim >= 3 ? Q_name : "1") << "];\n"; 563 code << tab << function_name << "<num_comp" << var_suffix << ", " << P_name << ", " << Q_name << ", OP_T_1D>(data, r_e" << var_suffix 564 << ", s_B" << var_suffix << ", r_c" << var_suffix << ");\n"; 565 } else { 566 std::string function_name = is_tensor ? ((dim == 1 ? "Interp" : "InterpTensor") + std::string(is_collocated ? "CollocatedNodes" : "") + 567 std::to_string(dim) + "d" + (is_all_tensor ? "" : "Flattened")) 568 : "InterpNonTensor"; 569 std::string op_t_1d_name = (is_all_tensor || !is_tensor) ? "OP_T_1D" : (P_1d > Q_1d ? P_name : Q_name); 570 571 code << tab << "CeedScalar r_q" << var_suffix << "[num_comp" << var_suffix << "*" << (is_all_tensor && (dim >= 3) ? Q_name : "1") << "];\n"; 572 code << tab << function_name << "<num_comp" << var_suffix << ", " << P_name << ", " << Q_name << ", " << op_t_1d_name << ">(data, r_e" 573 << var_suffix << ", s_B" << var_suffix << ", r_q" << var_suffix << ");\n"; 574 } 575 break; 576 case CEED_EVAL_GRAD: 577 if (is_at_points) { 578 std::string function_name = (dim == 1 ? "Interp" : "InterpTensor") + std::to_string(dim) + "d"; 579 580 code << tab << "CeedScalar r_c" << var_suffix << "[num_comp" << var_suffix << "*" << (dim >= 3 ? Q_name : "1") << "];\n"; 581 code << tab << function_name << "<num_comp" << var_suffix << ", " << P_name << ", " << Q_name << ", OP_T_1D>(data, r_e" << var_suffix 582 << ", s_B" << var_suffix << ", r_c" << var_suffix << ");\n"; 583 } else if (use_3d_slices) { 584 std::string function_name = 585 (dim > 1 ? "InterpTensor" : "Interp") + std::string(is_collocated ? "CollocatedNodes" : "") + std::to_string(dim) + "d"; 586 587 code << tab << "CeedScalar r_q" << var_suffix << "[num_comp" << var_suffix << "*" << Q_name << "];\n"; 588 code << tab << function_name << "<num_comp" << var_suffix << ", " << P_name << ", " << Q_name << ", OP_T_1D>(data, r_e" << var_suffix 589 << ", s_B" << var_suffix << ", r_q" << var_suffix << ");\n"; 590 } else if (is_tensor) { 591 bool is_collocated_grad = dim == 3 && Q_1d >= P_1d; 592 std::string function_name = 593 (dim == 1 ? "Grad" : ("GradTensor" + std::string(is_collocated ? "CollocatedNodes" : (is_collocated_grad ? "Collocated" : "")))) + 594 std::to_string(dim) + "d" + (is_all_tensor ? "" : "Flattened"); 595 std::string op_t_1d_name = is_all_tensor ? "OP_T_1D" : (P_1d > Q_1d ? P_name : Q_name); 596 597 code << tab << "CeedScalar r_q" << var_suffix << "[num_comp" << var_suffix << "*dim" << var_suffix << "*" 598 << (is_all_tensor && dim >= 3 ? Q_name : "1") << "];\n"; 599 code << tab << function_name << "<num_comp" << var_suffix << ", " << P_name << ", " << Q_name << ", " << op_t_1d_name << ">(data, r_e" 600 << var_suffix << ", s_B" << var_suffix << ", s_G" << var_suffix << ", r_q" << var_suffix << ");\n"; 601 } else { 602 std::string function_name = "GradNonTensor"; 603 604 code << tab << "CeedScalar r_q" << var_suffix << "[num_comp" << var_suffix << "*dim" << var_suffix << "];\n"; 605 code << tab << function_name << "<num_comp" << var_suffix << ", dim" << var_suffix << ", " << P_name << ", " << Q_name 606 << ", OP_T_1D>(data, r_e" << var_suffix << ", s_G" << var_suffix << ", r_q" << var_suffix << ");\n"; 607 } 608 break; 609 case CEED_EVAL_WEIGHT: { 610 if (is_at_points) { 611 code << tab << "// Nothing to do AtPoints\n"; 612 } else { 613 CeedBasis_Cuda_shared *basis_data; 614 std::string function_name = is_tensor 615 ? ((dim == 1 ? "Weight" : "WeightTensor") + std::to_string(dim) + "d" + (is_all_tensor ? "" : "Flattened")) 616 : "WeightNonTensor"; 617 618 code << tab << "CeedScalar r_q" << var_suffix << "[" << (is_all_tensor && (dim >= 3) ? Q_name : "1") << "];\n"; 619 CeedCallBackend(CeedBasisGetData(basis, &basis_data)); 620 data->W = basis_data->d_q_weight_1d; 621 code << tab << function_name << "<" << P_name << ", " << Q_name << ">(data, W, r_q" << var_suffix << ");\n"; 622 } 623 break; 624 } 625 // LCOV_EXCL_START 626 case CEED_EVAL_DIV: 627 case CEED_EVAL_CURL: 628 break; // TODO: Not implemented 629 // LCOV_EXCL_STOP 630 } 631 } else { 632 switch (eval_mode) { 633 case CEED_EVAL_NONE: 634 code << tab << "CeedScalar *r_e" << var_suffix << " = r_q" << var_suffix << ";\n"; 635 break; // No action 636 case CEED_EVAL_INTERP: 637 code << tab << "CeedScalar *r_e" << var_suffix << " = r_e_scratch;\n"; 638 if (is_at_points) { 639 std::string function_name = (dim == 1 ? "InterpTranspose" : "InterpTransposeTensor") + std::to_string(dim) + "d"; 640 641 code << tab << function_name << "<num_comp" << var_suffix << ", " << P_name << ", " << Q_name << ", OP_T_1D>(data, r_c" << var_suffix 642 << ", s_B" << var_suffix << ", r_e" << var_suffix << ");\n"; 643 } else { 644 std::string function_name = 645 is_tensor ? ((dim == 1 ? "InterpTranspose" : "InterpTransposeTensor") + std::string(is_collocated ? "CollocatedNodes" : "") + 646 std::to_string(dim) + "d" + (is_all_tensor ? "" : "Flattened")) 647 : "InterpTransposeNonTensor"; 648 std::string op_t_1d_name = (is_all_tensor || !is_tensor) ? "OP_T_1D" : (P_1d > Q_1d ? P_name : Q_name); 649 650 code << tab << function_name << "<num_comp" << var_suffix << ", " << P_name << ", " << Q_name << ", " << op_t_1d_name << ">(data, r_q" 651 << var_suffix << ", s_B" << var_suffix << ", r_e" << var_suffix << ");\n"; 652 } 653 break; 654 case CEED_EVAL_GRAD: 655 code << tab << "CeedScalar *r_e" << var_suffix << " = r_e_scratch;\n"; 656 if (is_at_points) { 657 std::string function_name = (dim == 1 ? "InterpTranspose" : "InterpTransposeTensor") + std::to_string(dim) + "d"; 658 659 code << tab << function_name << "<num_comp" << var_suffix << ", " << P_name << ", " << Q_name << ", OP_T_1D>(data, r_c" << var_suffix 660 << ", s_B" << var_suffix << ", r_e" << var_suffix << ");\n"; 661 } else if (use_3d_slices) { 662 std::string function_name = (dim == 1 ? "InterpTranspose" : "InterpTransposeTensor") + std::string(is_collocated ? "CollocatedNodes" : "") + 663 std::to_string(dim) + "d"; 664 665 code << tab << function_name << "<num_comp" << var_suffix << ", " << P_name << ", " << Q_name << ", OP_T_1D>(data, r_q" << var_suffix 666 << ", s_B" << var_suffix << ", r_e" << var_suffix << ");\n"; 667 } else if (is_tensor) { 668 bool is_collocated_grad = dim == 3 && Q_1d >= P_1d; 669 std::string function_name = 670 (dim == 1 ? "GradTranspose" 671 : ("GradTransposeTensor" + std::string(is_collocated ? "CollocatedNodes" : (is_collocated_grad ? "Collocated" : "")))) + 672 std::to_string(dim) + "d" + (is_all_tensor ? "" : "Flattened"); 673 std::string op_t_1d_name = is_all_tensor ? "OP_T_1D" : (P_1d > Q_1d ? P_name : Q_name); 674 675 code << tab << function_name << "<num_comp" << var_suffix << ", " << P_name << ", " << Q_name << ", " << op_t_1d_name << ">(data, r_q" 676 << var_suffix << ", s_B" << var_suffix << ", s_G" << var_suffix << ", r_e" << var_suffix << ");\n"; 677 } else { 678 std::string function_name = "GradTransposeNonTensor"; 679 680 code << tab << function_name << "<num_comp" << var_suffix << ", dim" << var_suffix << ", " << P_name << ", " << Q_name 681 << ", OP_T_1D>(data, r_q" << var_suffix << ", s_G" << var_suffix << ", r_e" << var_suffix << ");\n"; 682 } 683 break; 684 // LCOV_EXCL_START 685 case CEED_EVAL_WEIGHT: 686 break; // Should not occur 687 case CEED_EVAL_DIV: 688 case CEED_EVAL_CURL: 689 break; // TODO: Not implemented 690 // LCOV_EXCL_STOP 691 } 692 } 693 CeedCallBackend(CeedBasisDestroy(&basis)); 694 return CEED_ERROR_SUCCESS; 695 } 696 697 //------------------------------------------------------------------------------ 698 // QFunction 699 //------------------------------------------------------------------------------ 700 static int CeedOperatorBuildKernelQFunction_Cuda_gen(std::ostringstream &code, CeedOperator_Cuda_gen *data, Tab &tab, CeedInt max_dim, 701 CeedInt max_num_points, CeedInt num_input_fields, CeedOperatorField *op_input_fields, 702 CeedQFunctionField *qf_input_fields, CeedInt num_output_fields, 703 CeedOperatorField *op_output_fields, CeedQFunctionField *qf_output_fields, 704 std::string qfunction_name, CeedInt Q_1d, bool is_all_tensor, bool is_at_points, 705 bool use_3d_slices) { 706 std::string Q_name = is_all_tensor ? "Q_1d" : "Q"; 707 CeedEvalMode eval_mode = CEED_EVAL_NONE; 708 CeedElemRestriction elem_rstr; 709 710 // Setup output arrays 711 code << "\n"; 712 code << tab << "// -- Output field setup\n"; 713 for (CeedInt i = 0; i < num_output_fields; i++) { 714 const char *field_name; 715 std::string var_suffix = "_out_" + std::to_string(i); 716 717 CeedCallBackend(CeedOperatorFieldGetName(op_output_fields[i], &field_name)); 718 code << tab << "// ---- Output field " << i << ": " << field_name << "\n"; 719 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_output_fields[i], &eval_mode)); 720 switch (eval_mode) { 721 case CEED_EVAL_NONE: 722 if (is_at_points) { 723 code << tab << "CeedScalar r_q" << var_suffix << "[num_comp" << var_suffix << "];\n"; 724 } else { 725 code << tab << "CeedScalar r_q" << var_suffix << "[num_comp" << var_suffix << "*" << (is_all_tensor && (max_dim >= 3) ? Q_name : "1") 726 << "];\n"; 727 } 728 break; 729 case CEED_EVAL_INTERP: 730 if (is_at_points) { 731 // Accumulator for point data 732 code << tab << "CeedScalar r_c" << var_suffix << "[num_comp" << var_suffix << "*" << (max_dim >= 3 ? Q_name : "1") << "];\n"; 733 code << tab << "for (CeedInt i = 0; i < num_comp" << var_suffix << "*" << (max_dim >= 3 ? Q_name : "1") << "; i++) r_c" << var_suffix 734 << "[i] = 0.0;\n"; 735 } else { 736 code << tab << "CeedScalar r_q" << var_suffix << "[num_comp" << var_suffix << "*" << (is_all_tensor && (max_dim >= 3) ? Q_name : "1") 737 << "];\n"; 738 } 739 break; 740 case CEED_EVAL_GRAD: 741 if (is_at_points) { 742 // Accumulator for point data 743 code << tab << "CeedScalar r_c" << var_suffix << "[num_comp" << var_suffix << "*" << (max_dim >= 3 ? Q_name : "1") << "];\n"; 744 code << tab << "for (CeedInt i = 0; i < num_comp" << var_suffix << "*" << (max_dim >= 3 ? Q_name : "1") << "; i++) r_c" << var_suffix 745 << "[i] = 0.0;\n"; 746 } else if (use_3d_slices) { 747 // Accumulator for gradient slices 748 code << tab << "CeedScalar r_q" << var_suffix << "[num_comp" << var_suffix << "*" << Q_name << "];\n"; 749 code << tab << "for (CeedInt i = 0; i < num_comp" << var_suffix << "*" << Q_name << "; i++) r_q" << var_suffix << "[i] = 0.0;\n"; 750 } else { 751 code << tab << "CeedScalar r_q" << var_suffix << "[num_comp" << var_suffix << "*dim" << var_suffix << "*" 752 << (is_all_tensor && (max_dim >= 3) ? Q_name : "1") << "];\n"; 753 } 754 break; 755 case CEED_EVAL_WEIGHT: 756 break; 757 // LCOV_EXCL_START 758 case CEED_EVAL_DIV: 759 case CEED_EVAL_CURL: 760 break; // TODO: Not implemented 761 // LCOV_EXCL_STOP 762 } 763 } 764 765 if (is_at_points) { 766 // We need to handle batches of points 767 code << "\n"; 768 code << tab << "// Note: Using batches of points\n"; 769 code << tab << "const CeedInt point_loop_bound = (blockDim.x*blockDim.y) * ceil((1.0*max_num_points) / (blockDim.x*blockDim.y));\n\n"; 770 code << tab << "#pragma unroll\n"; 771 code << tab << "for (CeedInt i = threadIdx.x + threadIdx.y*blockDim.x; i < point_loop_bound; i += blockDim.x*blockDim.y) {\n"; 772 tab.push(); 773 code << tab << "const CeedInt p = i % max_num_points;\n\n"; 774 775 code << tab << "// -- Coordinates\n"; 776 code << tab << "CeedScalar r_x[max_dim];\n"; 777 code << tab << "ReadPoint<max_dim, coords_comp_stride, max_num_points>(data, elem, p, max_num_points, points.indices, points.coords, r_x);\n\n"; 778 779 code << tab << "// -- Input fields\n"; 780 for (CeedInt i = 0; i < num_input_fields; i++) { 781 const char *field_name; 782 std::string var_suffix = "_in_" + std::to_string(i); 783 std::string P_name = "P_1d" + var_suffix; 784 785 CeedCallBackend(CeedOperatorFieldGetName(op_input_fields[i], &field_name)); 786 code << tab << "// ---- Input field " << i << ": " << field_name << "\n"; 787 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[i], &eval_mode)); 788 // Basis action 789 code << tab << "// EvalMode: " << CeedEvalModes[eval_mode] << "\n"; 790 switch (eval_mode) { 791 case CEED_EVAL_NONE: 792 code << tab << "CeedScalar r_s" << var_suffix << "[num_comp" << var_suffix << "];\n"; 793 code << tab << "ReadPoint<num_comp" << var_suffix << ", comp_stride" << var_suffix 794 << ", max_num_points>(data, elem, p, max_num_points, indices.inputs[" << i << "], d" << var_suffix << ", r_s" << var_suffix << ");\n"; 795 break; 796 case CEED_EVAL_INTERP: 797 code << tab << "CeedScalar r_s" << var_suffix << "[num_comp" << var_suffix << "];\n"; 798 code << tab << "InterpAtPoints" << max_dim << "d<num_comp" << var_suffix << ", max_num_points, " << P_name << ", " << Q_name 799 << ">(data, i, r_c" << var_suffix << ", r_x, r_s" << var_suffix << ");\n"; 800 break; 801 case CEED_EVAL_GRAD: 802 code << tab << "CeedScalar r_s" << var_suffix << "[num_comp" << var_suffix << "*dim" << var_suffix << "];\n"; 803 code << tab << "GradAtPoints" << max_dim << "d<num_comp" << var_suffix << ", max_num_points, " << P_name << ", " << Q_name 804 << ">(data, i, r_c" << var_suffix << ", r_x, r_s" << var_suffix << ");\n"; 805 break; 806 case CEED_EVAL_WEIGHT: 807 code << tab << "CeedScalar r_s" << var_suffix << "[1];\n"; 808 code << tab << "r_s" << var_suffix << "[0] = 1.0;\n"; 809 break; 810 // LCOV_EXCL_START 811 case CEED_EVAL_DIV: 812 case CEED_EVAL_CURL: 813 break; // TODO: Not implemented 814 // LCOV_EXCL_STOP 815 } 816 } 817 code << "\n"; 818 code << tab << "// -- Output fields\n"; 819 for (CeedInt i = 0; i < num_output_fields; i++) { 820 const char *field_name; 821 std::string var_suffix = "_out_" + std::to_string(i); 822 823 CeedCallBackend(CeedOperatorFieldGetName(op_output_fields[i], &field_name)); 824 code << tab << "// ---- Output field " << i << ": " << field_name << "\n"; 825 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_output_fields[i], &eval_mode)); 826 // Basis action 827 switch (eval_mode) { 828 case CEED_EVAL_NONE: 829 code << tab << "CeedScalar r_s" << var_suffix << "[num_comp" << var_suffix << "];\n"; 830 break; 831 case CEED_EVAL_INTERP: 832 code << tab << "CeedScalar r_s" << var_suffix << "[num_comp" << var_suffix << "];\n"; 833 break; 834 case CEED_EVAL_GRAD: 835 code << tab << "CeedScalar r_s" << var_suffix << "[num_comp" << var_suffix << "*dim" << var_suffix << "];\n"; 836 break; 837 // LCOV_EXCL_START 838 case CEED_EVAL_WEIGHT: 839 break; // Should not occur 840 case CEED_EVAL_DIV: 841 case CEED_EVAL_CURL: 842 break; // TODO: Not implemented 843 // LCOV_EXCL_STOP 844 } 845 } 846 847 } else if (use_3d_slices) { 848 // We treat quadrature points per slice in 3d to save registers 849 code << "\n"; 850 code << tab << "// Note: Using planes of 3D elements\n"; 851 code << tab << "#pragma unroll\n"; 852 code << tab << "for (CeedInt q = 0; q < " << Q_name << "; q++) {\n"; 853 tab.push(); 854 code << tab << "// -- Input fields\n"; 855 for (CeedInt i = 0; i < num_input_fields; i++) { 856 const char *field_name; 857 std::string var_suffix = "_in_" + std::to_string(i); 858 859 CeedCallBackend(CeedOperatorFieldGetName(op_input_fields[i], &field_name)); 860 code << tab << "// ---- Input field " << i << ": " << field_name << "\n"; 861 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[i], &eval_mode)); 862 // Basis action 863 code << tab << "// EvalMode: " << CeedEvalModes[eval_mode] << "\n"; 864 switch (eval_mode) { 865 case CEED_EVAL_NONE: 866 bool is_strided; 867 868 code << tab << "CeedScalar r_s" << var_suffix << "[num_comp" << var_suffix << "];\n"; 869 870 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_input_fields[i], &elem_rstr)); 871 CeedCallBackend(CeedElemRestrictionIsStrided(elem_rstr, &is_strided)); 872 if (is_strided) { 873 bool has_backend_strides; 874 CeedInt num_elem, elem_size; 875 876 CeedCallBackend(CeedElemRestrictionGetElementSize(elem_rstr, &elem_size)); 877 CeedCallBackend(CeedElemRestrictionHasBackendStrides(elem_rstr, &has_backend_strides)); 878 CeedCallBackend(CeedElemRestrictionGetNumElements(elem_rstr, &num_elem)); 879 CeedInt strides[3] = {1, elem_size * num_elem, elem_size}; 880 881 if (!has_backend_strides) { 882 CeedCallBackend(CeedElemRestrictionGetStrides(elem_rstr, strides)); 883 } 884 code << tab << "const CeedInt strides" << var_suffix << "_0 = " << strides[0] << ", strides" << var_suffix << "_1 = " << strides[1] 885 << ", strides" << var_suffix << "_2 = " << strides[2] << ";\n"; 886 code << tab << "ReadEVecSliceStrided3d<num_comp" << var_suffix << ", " << Q_name << ", strides" << var_suffix << "_0, strides" 887 << var_suffix << "_1, strides" << var_suffix << "_2>(data, elem, q, d" << var_suffix << ", r_s" << var_suffix << ");\n"; 888 } else { 889 CeedSize l_size = 0; 890 CeedInt comp_stride; 891 CeedElemRestriction_Cuda *rstr_data; 892 893 CeedCallBackend(CeedElemRestrictionGetLVectorSize(elem_rstr, &l_size)); 894 code << tab << "const CeedInt l_size" << var_suffix << " = " << l_size << ";\n"; 895 CeedCallBackend(CeedElemRestrictionGetCompStride(elem_rstr, &comp_stride)); 896 code << tab << "const CeedInt comp_stride" << var_suffix << " = " << comp_stride << ";\n"; 897 CeedCallBackend(CeedElemRestrictionGetData(elem_rstr, &rstr_data)); 898 data->indices.inputs[i] = (CeedInt *)rstr_data->d_offsets; 899 code << tab << "ReadEVecSliceStandard3d<num_comp" << var_suffix << ", comp_stride" << var_suffix << ", " << Q_name << ">(data, l_size" 900 << var_suffix << ", elem, q, indices.inputs[" << i << "], d" << var_suffix << ", r_s" << var_suffix << ");\n"; 901 } 902 CeedCallBackend(CeedElemRestrictionDestroy(&elem_rstr)); 903 break; 904 case CEED_EVAL_INTERP: 905 code << tab << "CeedScalar r_s" << var_suffix << "[num_comp" << var_suffix << "];\n"; 906 code << tab << "for (CeedInt j = 0; j < num_comp" << var_suffix << "; j++) {\n"; 907 tab.push(); 908 code << tab << "r_s" << var_suffix << "[j] = r_q" << var_suffix << "[q + j*" << Q_name << "];\n"; 909 tab.pop(); 910 code << tab << "}\n"; 911 break; 912 case CEED_EVAL_GRAD: 913 code << tab << "CeedScalar r_s" << var_suffix << "[num_comp" << var_suffix << "*dim" << var_suffix << "];\n"; 914 code << tab << "GradColloSlice3d<num_comp" << var_suffix << ", " << Q_name << ", OP_T_1D>(data, q, r_q" << var_suffix << ", s_G" 915 << var_suffix << ", r_s" << var_suffix << ");\n"; 916 break; 917 case CEED_EVAL_WEIGHT: 918 code << tab << "CeedScalar r_s" << var_suffix << "[1];\n"; 919 code << tab << "r_s" << var_suffix << "[0] = r_q" << var_suffix << "[q];\n"; 920 break; 921 // LCOV_EXCL_START 922 case CEED_EVAL_DIV: 923 case CEED_EVAL_CURL: 924 break; // TODO: Not implemented 925 // LCOV_EXCL_STOP 926 } 927 } 928 code << "\n"; 929 code << tab << "// -- Output fields\n"; 930 for (CeedInt i = 0; i < num_output_fields; i++) { 931 const char *field_name; 932 std::string var_suffix = "_out_" + std::to_string(i); 933 934 CeedCallBackend(CeedOperatorFieldGetName(op_output_fields[i], &field_name)); 935 code << tab << "// ---- Output field " << i << ": " << field_name << "\n"; 936 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_output_fields[i], &eval_mode)); 937 // Basis action 938 switch (eval_mode) { 939 case CEED_EVAL_NONE: 940 code << tab << "CeedScalar r_s" << var_suffix << "[num_comp" << var_suffix << "];\n"; 941 break; 942 case CEED_EVAL_INTERP: 943 code << tab << "CeedScalar r_s" << var_suffix << "[num_comp" << var_suffix << "];\n"; 944 break; 945 case CEED_EVAL_GRAD: 946 code << tab << "CeedScalar r_s" << var_suffix << "[num_comp" << var_suffix << "*dim" << var_suffix << "];\n"; 947 break; 948 // LCOV_EXCL_START 949 case CEED_EVAL_WEIGHT: 950 break; // Should not occur 951 case CEED_EVAL_DIV: 952 case CEED_EVAL_CURL: 953 break; // TODO: Not implemented 954 // LCOV_EXCL_STOP 955 } 956 } 957 } else { 958 code << "\n"; 959 code << tab << "// Note: Using full elements\n"; 960 code << tab << "{\n"; 961 tab.push(); 962 code << tab << "// -- Input fields\n"; 963 for (CeedInt i = 0; i < num_input_fields; i++) { 964 const char *field_name; 965 966 CeedCallBackend(CeedOperatorFieldGetName(op_input_fields[i], &field_name)); 967 code << tab << "// ---- Input field " << i << ": " << field_name << "\n"; 968 code << tab << "CeedScalar *r_s_in_" << i << " = r_q_in_" << i << ";\n"; 969 } 970 code << tab << "// -- Output fields\n"; 971 for (CeedInt i = 0; i < num_output_fields; i++) { 972 const char *field_name; 973 974 CeedCallBackend(CeedOperatorFieldGetName(op_output_fields[i], &field_name)); 975 code << tab << "// ---- Output field " << i << ": " << field_name << "\n"; 976 code << tab << "CeedScalar *r_s_out_" << i << " = r_q_out_" << i << ";\n"; 977 } 978 } 979 980 // Input and output buffers 981 code << "\n"; 982 code << tab << "// -- QFunction inputs and outputs\n"; 983 code << tab << "// ---- Inputs\n"; 984 code << tab << "CeedScalar *inputs[" << CeedIntMax(num_input_fields, 1) << "];\n"; 985 for (CeedInt i = 0; i < num_input_fields; i++) { 986 const char *field_name; 987 988 CeedCallBackend(CeedOperatorFieldGetName(op_input_fields[i], &field_name)); 989 code << tab << "// ------ Input field " << i << ": " << field_name << "\n"; 990 code << tab << "inputs[" << i << "] = r_s_in_" << i << ";\n"; 991 } 992 code << tab << "// ---- Outputs\n"; 993 code << tab << "CeedScalar *outputs[" << CeedIntMax(num_output_fields, 1) << "];\n"; 994 for (CeedInt i = 0; i < num_output_fields; i++) { 995 const char *field_name; 996 997 CeedCallBackend(CeedOperatorFieldGetName(op_output_fields[i], &field_name)); 998 code << tab << "// ------ Output field " << i << ": " << field_name << "\n"; 999 code << tab << "outputs[" << i << "] = r_s_out_" << i << ";\n"; 1000 } 1001 1002 // Apply QFunction 1003 code << "\n"; 1004 code << tab << "// -- Apply QFunction\n"; 1005 code << tab << "" << qfunction_name << "(ctx, "; 1006 if (max_dim != 3 || is_at_points || use_3d_slices || !is_all_tensor) { 1007 code << "1"; 1008 } else { 1009 code << Q_name; 1010 } 1011 code << ", inputs, outputs);\n"; 1012 1013 if (is_at_points) { 1014 // Map back to coefficients 1015 code << "\n"; 1016 code << tab << "// -- Output fields\n"; 1017 for (CeedInt i = 0; i < num_output_fields; i++) { 1018 const char *field_name; 1019 std::string var_suffix = "_out_" + std::to_string(i); 1020 std::string P_name = "P_1d" + var_suffix; 1021 1022 CeedCallBackend(CeedOperatorFieldGetName(op_output_fields[i], &field_name)); 1023 code << tab << "// ---- Output field " << i << ": " << field_name << "\n"; 1024 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_output_fields[i], &eval_mode)); 1025 // Basis action 1026 code << tab << "// EvalMode: " << CeedEvalModes[eval_mode] << "\n"; 1027 switch (eval_mode) { 1028 case CEED_EVAL_NONE: { 1029 CeedInt comp_stride; 1030 CeedElemRestriction elem_rstr; 1031 1032 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_output_fields[i], &elem_rstr)); 1033 CeedCallBackend(CeedElemRestrictionGetCompStride(elem_rstr, &comp_stride)); 1034 CeedCallBackend(CeedElemRestrictionDestroy(&elem_rstr)); 1035 code << tab << "const CeedInt comp_stride" << var_suffix << " = " << comp_stride << ";\n"; 1036 code << tab << "WritePoint<num_comp" << var_suffix << ", comp_stride" << var_suffix 1037 << ", max_num_points>(data, elem, i, points.num_per_elem[elem], indices.outputs[" << i << "]" 1038 << ", r_s" << var_suffix << ", d" << var_suffix << ");\n"; 1039 break; 1040 } 1041 case CEED_EVAL_INTERP: 1042 code << tab << "if (i >= points.num_per_elem[elem]) {\n"; 1043 tab.push(); 1044 code << tab << "for (CeedInt j = 0; j < num_comp" << var_suffix << "; j++) r_s" << var_suffix << "[j] = 0.0;\n"; 1045 tab.pop(); 1046 code << tab << "}\n"; 1047 code << tab << "InterpTransposeAtPoints" << max_dim << "d<num_comp" << var_suffix << ", max_num_points, " << P_name << ", " << Q_name 1048 << ">(data, i, r_s" << var_suffix << ", r_x, r_c" << var_suffix << ");\n"; 1049 break; 1050 case CEED_EVAL_GRAD: 1051 code << tab << "if (i >= points.num_per_elem[elem]) {\n"; 1052 tab.push(); 1053 code << tab << "for (CeedInt j = 0; j < num_comp" << var_suffix << "*dim" << var_suffix << "; j++) r_s" << var_suffix << "[j] = 0.0;\n"; 1054 tab.pop(); 1055 code << tab << "}\n"; 1056 code << tab << "GradTransposeAtPoints" << max_dim << "d<num_comp" << var_suffix << ", max_num_points, " << P_name << ", " << Q_name 1057 << ">(data, i, r_s" << var_suffix << ", r_x, r_c" << var_suffix << ");\n"; 1058 break; 1059 // LCOV_EXCL_START 1060 case CEED_EVAL_WEIGHT: 1061 break; // Should not occur 1062 case CEED_EVAL_DIV: 1063 case CEED_EVAL_CURL: 1064 break; // TODO: Not implemented 1065 // LCOV_EXCL_STOP 1066 } 1067 } 1068 } else if (use_3d_slices) { 1069 // Copy or apply transpose grad, if needed 1070 code << "\n"; 1071 code << tab << "// -- Output fields\n"; 1072 for (CeedInt i = 0; i < num_output_fields; i++) { 1073 const char *field_name; 1074 std::string var_suffix = "_out_" + std::to_string(i); 1075 std::string P_name = "P_1d" + var_suffix; 1076 1077 CeedCallBackend(CeedOperatorFieldGetName(op_output_fields[i], &field_name)); 1078 code << tab << "// ---- Output field " << i << ": " << field_name << "\n"; 1079 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_output_fields[i], &eval_mode)); 1080 // Basis action 1081 code << tab << "// EvalMode: " << CeedEvalModes[eval_mode] << "\n"; 1082 switch (eval_mode) { 1083 case CEED_EVAL_NONE: 1084 code << tab << "for (CeedInt j = 0; j < num_comp" << var_suffix << " ; j++) {\n"; 1085 tab.push(); 1086 code << tab << "r_q" << var_suffix << "[q + j*" << Q_name << "] = r_s" << var_suffix << "[j];\n"; 1087 tab.pop(); 1088 code << tab << "}\n"; 1089 break; 1090 case CEED_EVAL_INTERP: 1091 code << tab << "for (CeedInt j = 0; j < num_comp" << var_suffix << " ; j++) {\n"; 1092 tab.push(); 1093 code << tab << "r_q" << var_suffix << "[q + j*" << Q_name << "] = r_s" << var_suffix << "[j];\n"; 1094 tab.pop(); 1095 code << tab << "}\n"; 1096 break; 1097 case CEED_EVAL_GRAD: 1098 code << tab << "GradColloSliceTranspose3d<num_comp" << var_suffix << ", " << Q_name << ", OP_T_1D>(data, q, r_s" << var_suffix << ", s_G" 1099 << var_suffix << ", r_q" << var_suffix << ");\n"; 1100 break; 1101 // LCOV_EXCL_START 1102 case CEED_EVAL_WEIGHT: 1103 break; // Should not occur 1104 case CEED_EVAL_DIV: 1105 case CEED_EVAL_CURL: 1106 break; // TODO: Not implemented 1107 // LCOV_EXCL_STOP 1108 } 1109 } 1110 } 1111 tab.pop(); 1112 code << tab << "}\n"; 1113 return CEED_ERROR_SUCCESS; 1114 } 1115 1116 //------------------------------------------------------------------------------ 1117 // Build single operator kernel 1118 //------------------------------------------------------------------------------ 1119 extern "C" int CeedOperatorBuildKernel_Cuda_gen(CeedOperator op, bool *is_good_build) { 1120 bool is_all_tensor = true, is_all_nontensor = true, is_at_points = false, use_3d_slices = false; 1121 Ceed ceed; 1122 CeedInt Q = 0, Q_1d = 0, num_input_fields, num_output_fields, max_dim = 1, max_num_points = 0, coords_comp_stride = 0; 1123 CeedQFunctionField *qf_input_fields, *qf_output_fields; 1124 CeedQFunction_Cuda_gen *qf_data; 1125 CeedQFunction qf; 1126 CeedOperatorField *op_input_fields, *op_output_fields; 1127 CeedOperator_Cuda_gen *data; 1128 std::ostringstream code; 1129 Tab tab; 1130 1131 CeedCallBackend(CeedOperatorGetData(op, &data)); 1132 { 1133 bool is_setup_done; 1134 1135 CeedCallBackend(CeedOperatorIsSetupDone(op, &is_setup_done)); 1136 if (is_setup_done) { 1137 *is_good_build = !data->use_fallback; 1138 return CEED_ERROR_SUCCESS; 1139 } 1140 } 1141 1142 // Check field compatibility 1143 CeedCallBackend(CeedOperatorGetFields(op, &num_input_fields, &op_input_fields, &num_output_fields, &op_output_fields)); 1144 { 1145 bool has_shared_bases = true; 1146 1147 for (CeedInt i = 0; i < num_input_fields; i++) { 1148 CeedBasis basis; 1149 1150 CeedCallBackend(CeedOperatorFieldGetBasis(op_input_fields[i], &basis)); 1151 if (basis != CEED_BASIS_NONE) { 1152 bool is_tensor = true; 1153 const char *resource; 1154 char *resource_root; 1155 Ceed basis_ceed; 1156 1157 CeedCallBackend(CeedBasisIsTensor(basis, &is_tensor)); 1158 is_all_tensor = is_all_tensor && is_tensor; 1159 is_all_nontensor = is_all_nontensor && !is_tensor; 1160 CeedCallBackend(CeedBasisGetCeed(basis, &basis_ceed)); 1161 CeedCallBackend(CeedGetResource(basis_ceed, &resource)); 1162 CeedCallBackend(CeedGetResourceRoot(basis_ceed, resource, ":", &resource_root)); 1163 has_shared_bases = has_shared_bases && !strcmp(resource_root, "/gpu/cuda/shared"); 1164 CeedCallBackend(CeedFree(&resource_root)); 1165 CeedCallBackend(CeedDestroy(&basis_ceed)); 1166 } 1167 CeedCallBackend(CeedBasisDestroy(&basis)); 1168 } 1169 1170 for (CeedInt i = 0; i < num_output_fields; i++) { 1171 CeedBasis basis; 1172 1173 CeedCallBackend(CeedOperatorFieldGetBasis(op_output_fields[i], &basis)); 1174 if (basis != CEED_BASIS_NONE) { 1175 bool is_tensor = true; 1176 const char *resource; 1177 char *resource_root; 1178 Ceed basis_ceed; 1179 1180 CeedCallBackend(CeedBasisIsTensor(basis, &is_tensor)); 1181 is_all_tensor = is_all_tensor && is_tensor; 1182 is_all_nontensor = is_all_nontensor && !is_tensor; 1183 1184 CeedCallBackend(CeedBasisGetCeed(basis, &basis_ceed)); 1185 CeedCallBackend(CeedGetResource(basis_ceed, &resource)); 1186 CeedCallBackend(CeedGetResourceRoot(basis_ceed, resource, ":", &resource_root)); 1187 has_shared_bases = has_shared_bases && !strcmp(resource_root, "/gpu/cuda/shared"); 1188 CeedCallBackend(CeedFree(&resource_root)); 1189 CeedCallBackend(CeedDestroy(&basis_ceed)); 1190 } 1191 CeedCallBackend(CeedBasisDestroy(&basis)); 1192 } 1193 // -- Fallback to ref if not all bases are shared 1194 if (!has_shared_bases) { 1195 *is_good_build = false; 1196 return CEED_ERROR_SUCCESS; 1197 } 1198 } 1199 CeedCallBackend(CeedOperatorGetCeed(op, &ceed)); 1200 CeedCallBackend(CeedOperatorGetQFunction(op, &qf)); 1201 CeedCallBackend(CeedQFunctionGetData(qf, &qf_data)); 1202 CeedCallBackend(CeedQFunctionGetFields(qf, NULL, &qf_input_fields, NULL, &qf_output_fields)); 1203 1204 // Get operator data 1205 CeedCallBackend(CeedOperatorIsAtPoints(op, &is_at_points)); 1206 { 1207 CeedInt max_P = 0, max_P_1d = 0; 1208 1209 CeedCallBackend(CeedOperatorBuildKernelData_Cuda_gen(ceed, num_input_fields, op_input_fields, qf_input_fields, num_output_fields, 1210 op_output_fields, qf_output_fields, &max_P, &max_P_1d, &Q, &Q_1d, &max_dim, &is_all_tensor, 1211 &use_3d_slices)); 1212 data->max_P_1d = is_all_tensor ? max_P_1d : max_P; 1213 } 1214 if (max_dim == 0) max_dim = 1; 1215 data->dim = max_dim; 1216 if (is_at_points) { 1217 CeedElemRestriction_Cuda *rstr_data; 1218 CeedElemRestriction rstr_points = NULL; 1219 1220 CeedCallBackend(CeedOperatorAtPointsGetPoints(op, &rstr_points, NULL)); 1221 CeedCallBackend(CeedElemRestrictionGetMaxPointsInElement(rstr_points, &max_num_points)); 1222 CeedCallBackend(CeedElemRestrictionGetCompStride(rstr_points, &coords_comp_stride)); 1223 CeedCallBackend(CeedElemRestrictionGetData(rstr_points, &rstr_data)); 1224 data->points.indices = (CeedInt *)rstr_data->d_offsets; 1225 CeedCallBackend(CeedElemRestrictionDestroy(&rstr_points)); 1226 } 1227 if (is_at_points) use_3d_slices = false; 1228 if (Q_1d == 0) { 1229 if (is_at_points) Q_1d = max_num_points; 1230 else CeedCallBackend(CeedOperatorGetNumQuadraturePoints(op, &Q_1d)); 1231 } 1232 if (Q == 0) Q = Q_1d; 1233 data->Q = Q; 1234 data->Q_1d = Q_1d; 1235 1236 // Check for restriction only identity operator 1237 { 1238 bool is_identity_qf; 1239 1240 CeedCallBackend(CeedQFunctionIsIdentity(qf, &is_identity_qf)); 1241 if (is_identity_qf) { 1242 CeedEvalMode eval_mode_in, eval_mode_out; 1243 1244 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[0], &eval_mode_in)); 1245 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_output_fields[0], &eval_mode_out)); 1246 CeedCheck(eval_mode_in != CEED_EVAL_NONE || eval_mode_out != CEED_EVAL_NONE, ceed, CEED_ERROR_BACKEND, 1247 "Backend does not implement restriction only identity operators"); 1248 } 1249 } 1250 1251 // Add atomicAdd function for old NVidia architectures 1252 { 1253 Ceed_Cuda *ceed_data; 1254 struct cudaDeviceProp prop; 1255 1256 CeedCallBackend(CeedGetData(ceed, &ceed_data)); 1257 CeedCallBackend(cudaGetDeviceProperties(&prop, ceed_data->device_id)); 1258 if ((prop.major < 6) && (CEED_SCALAR_TYPE != CEED_SCALAR_FP32)) { 1259 code << tab << "// AtomicAdd fallback source\n"; 1260 code << tab << "#include <ceed/jit-source/cuda/cuda-atomic-add-fallback.h>\n\n"; 1261 } 1262 } 1263 1264 // Load basis source files 1265 if (!is_all_nontensor) { 1266 code << tab << "// Tensor basis source\n"; 1267 code << tab << "#include <ceed/jit-source/cuda/cuda-shared-basis-tensor-templates.h>\n\n"; 1268 } 1269 if (!is_all_tensor) { 1270 code << tab << "// Non-tensor basis source\n"; 1271 code << tab << "#include <ceed/jit-source/cuda/cuda-shared-basis-nontensor-templates.h>\n\n"; 1272 } 1273 if (!is_all_tensor && !is_all_nontensor) { 1274 code << "// Tensor basis source\n"; 1275 code << "#include <ceed/jit-source/cuda/cuda-shared-basis-tensor-flattened-templates.h>\n\n"; 1276 } 1277 if (is_at_points) { 1278 code << "// AtPoints basis source\n"; 1279 code << "#include <ceed/jit-source/cuda/cuda-shared-basis-tensor-at-points-templates.h>\n\n"; 1280 } 1281 code << "// CodeGen operator source\n"; 1282 code << "#include <ceed/jit-source/cuda/cuda-gen-templates.h>\n\n"; 1283 1284 // Get QFunction name 1285 std::string qfunction_name(qf_data->qfunction_name); 1286 std::string operator_name; 1287 1288 operator_name = "CeedKernelCudaGenOperator_" + qfunction_name; 1289 1290 // Define CEED_Q_VLA 1291 code << "\n" << tab << "#undef CEED_Q_VLA\n"; 1292 if (max_dim != 3 || is_at_points || use_3d_slices || !is_all_tensor) { 1293 code << tab << "#define CEED_Q_VLA 1\n\n"; 1294 } else { 1295 code << tab << "#define CEED_Q_VLA " << Q_1d << "\n\n"; 1296 } 1297 1298 // Add user QFunction source 1299 { 1300 const char *source_path; 1301 1302 CeedCallBackend(CeedQFunctionGetSourcePath(qf, &source_path)); 1303 CeedCheck(source_path, ceed, CEED_ERROR_UNSUPPORTED, "/gpu/cuda/gen backend requires QFunction source code file"); 1304 1305 code << tab << "// User QFunction source\n"; 1306 code << tab << "#include \"" << source_path << "\"\n\n"; 1307 } 1308 1309 // Setup 1310 code << "\n" << tab << "// -----------------------------------------------------------------------------\n"; 1311 code << tab << "// Operator Kernel\n"; 1312 code << tab << "// \n"; 1313 code << tab << "// d_[in,out]_i: CeedVector device array\n"; 1314 code << tab << "// r_[in,out]_e_i: Element vector register\n"; 1315 code << tab << "// r_[in,out]_q_i: Quadrature space vector register\n"; 1316 code << tab << "// r_[in,out]_c_i: AtPoints Chebyshev coefficients register\n"; 1317 code << tab << "// r_[in,out]_s_i: Quadrature space slice vector register\n"; 1318 code << tab << "// \n"; 1319 code << tab << "// s_B_[in,out]_i: Interpolation matrix, shared memory\n"; 1320 code << tab << "// s_G_[in,out]_i: Gradient matrix, shared memory\n"; 1321 code << tab << "// -----------------------------------------------------------------------------\n"; 1322 code << tab << "extern \"C\" __global__ void " << operator_name 1323 << "(CeedInt num_elem, void* ctx, FieldsInt_Cuda indices, Fields_Cuda fields, Fields_Cuda B, Fields_Cuda G, CeedScalar *W, Points_Cuda " 1324 "points) {\n"; 1325 tab.push(); 1326 1327 // Scratch buffers 1328 for (CeedInt i = 0; i < num_input_fields; i++) { 1329 CeedEvalMode eval_mode; 1330 1331 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[i], &eval_mode)); 1332 if (eval_mode != CEED_EVAL_WEIGHT) { // Skip CEED_EVAL_WEIGHT 1333 code << tab << "const CeedScalar *__restrict__ d_in_" << i << " = fields.inputs[" << i << "];\n"; 1334 } 1335 } 1336 for (CeedInt i = 0; i < num_output_fields; i++) { 1337 code << tab << "CeedScalar *__restrict__ d_out_" << i << " = fields.outputs[" << i << "];\n"; 1338 } 1339 1340 code << tab << "const CeedInt max_dim = " << max_dim << ";\n"; 1341 if (!is_all_tensor) { 1342 code << tab << "const CeedInt Q = " << Q << ";\n"; 1343 } 1344 if (!is_all_nontensor) { 1345 code << tab << "const CeedInt Q_1d = " << Q_1d << ";\n"; 1346 } 1347 if (is_at_points) { 1348 code << tab << "const CeedInt max_num_points = " << max_num_points << ";\n"; 1349 code << tab << "const CeedInt coords_comp_stride = " << coords_comp_stride << ";\n"; 1350 } 1351 1352 // Shared data 1353 code << tab << "extern __shared__ CeedScalar slice[];\n"; 1354 code << tab << "SharedData_Cuda data;\n"; 1355 code << tab << "data.t_id_x = threadIdx.x;\n"; 1356 code << tab << "data.t_id_y = threadIdx.y;\n"; 1357 code << tab << "data.t_id_z = threadIdx.z;\n"; 1358 code << tab << "data.t_id = threadIdx.x + threadIdx.y*blockDim.x + threadIdx.z*blockDim.y*blockDim.x;\n"; 1359 code << tab << "data.slice = slice + data.t_id_z*OP_T_1D" << ((!is_all_tensor || max_dim == 1) ? "" : "*OP_T_1D") << ";\n"; 1360 1361 // -- Determine input mat reuse 1362 FieldReuse_Cuda input_matrix_reuse[CEED_FIELD_MAX]; 1363 1364 for (CeedInt i = 0; i < num_input_fields; i++) { 1365 input_matrix_reuse[i].index = -1; 1366 } 1367 for (CeedInt i = 0; i < num_input_fields; i++) { 1368 bool is_tensor = true; 1369 CeedEvalMode eval_mode_i; 1370 CeedBasis basis_i; 1371 1372 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[i], &eval_mode_i)); 1373 if (eval_mode_i == CEED_EVAL_WEIGHT) continue; 1374 CeedCallBackend(CeedOperatorFieldGetBasis(op_input_fields[i], &basis_i)); 1375 CeedCallBackend(CeedBasisIsTensor(basis_i, &is_tensor)); 1376 for (CeedInt j = 0; (input_matrix_reuse[i].index == -1) && (j < i); j++) { 1377 CeedEvalMode eval_mode_j; 1378 CeedBasis basis_j; 1379 1380 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[j], &eval_mode_j)); 1381 if (eval_mode_j == CEED_EVAL_WEIGHT) continue; 1382 CeedCallBackend(CeedOperatorFieldGetBasis(op_input_fields[j], &basis_j)); 1383 if (basis_i == basis_j) { 1384 if (is_tensor) { 1385 input_matrix_reuse[i].index = j; 1386 input_matrix_reuse[i].is_input = true; 1387 input_matrix_reuse[i].eval_mode = eval_mode_j; 1388 } else { 1389 // For non-tensor can only re-use with the same eval mode 1390 if (eval_mode_i == eval_mode_j) { 1391 input_matrix_reuse[i].index = j; 1392 input_matrix_reuse[i].is_input = true; 1393 input_matrix_reuse[i].eval_mode = eval_mode_j; 1394 } 1395 } 1396 } 1397 CeedCallBackend(CeedBasisDestroy(&basis_j)); 1398 } 1399 CeedCallBackend(CeedBasisDestroy(&basis_i)); 1400 } 1401 1402 // -- Determine output mat reuse 1403 FieldReuse_Cuda output_matrix_reuse[CEED_FIELD_MAX]; 1404 1405 for (CeedInt i = 0; i < num_output_fields; i++) { 1406 output_matrix_reuse[i].index = -1; 1407 } 1408 for (CeedInt i = 0; i < num_output_fields; i++) { 1409 bool is_tensor = true; 1410 CeedEvalMode eval_mode_i; 1411 CeedBasis basis_i; 1412 1413 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_output_fields[i], &eval_mode_i)); 1414 CeedCallBackend(CeedOperatorFieldGetBasis(op_output_fields[i], &basis_i)); 1415 CeedCallBackend(CeedBasisIsTensor(basis_i, &is_tensor)); 1416 for (CeedInt j = 0; (output_matrix_reuse[i].index == -1) && (j < num_input_fields); j++) { 1417 CeedEvalMode eval_mode_j; 1418 CeedBasis basis_j; 1419 1420 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[j], &eval_mode_j)); 1421 if (eval_mode_j == CEED_EVAL_WEIGHT) continue; 1422 CeedCallBackend(CeedOperatorFieldGetBasis(op_input_fields[j], &basis_j)); 1423 if (basis_i == basis_j) { 1424 if (is_tensor) { 1425 output_matrix_reuse[i].index = j; 1426 output_matrix_reuse[i].is_input = true; 1427 output_matrix_reuse[i].eval_mode = eval_mode_j; 1428 } else { 1429 // For non-tensor can only re-use with the same eval mode 1430 if (eval_mode_i == eval_mode_j) { 1431 output_matrix_reuse[i].index = j; 1432 output_matrix_reuse[i].is_input = true; 1433 output_matrix_reuse[i].eval_mode = eval_mode_j; 1434 } 1435 } 1436 } 1437 CeedCallBackend(CeedBasisDestroy(&basis_j)); 1438 } 1439 for (CeedInt j = 0; (output_matrix_reuse[i].index == -1) && (j < i); j++) { 1440 CeedEvalMode eval_mode_j; 1441 CeedBasis basis_j; 1442 1443 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_output_fields[j], &eval_mode_j)); 1444 if (eval_mode_j == CEED_EVAL_WEIGHT) continue; 1445 CeedCallBackend(CeedOperatorFieldGetBasis(op_output_fields[j], &basis_j)); 1446 if (basis_i == basis_j) { 1447 if (is_tensor) { 1448 output_matrix_reuse[i].index = j; 1449 output_matrix_reuse[i].is_input = false; 1450 output_matrix_reuse[i].eval_mode = eval_mode_j; 1451 } else { 1452 // For non-tensor can only re-use with the same eval mode 1453 if (eval_mode_i == eval_mode_j) { 1454 output_matrix_reuse[i].index = j; 1455 output_matrix_reuse[i].is_input = false; 1456 output_matrix_reuse[i].eval_mode = eval_mode_j; 1457 } 1458 } 1459 } 1460 CeedCallBackend(CeedBasisDestroy(&basis_j)); 1461 } 1462 CeedCallBackend(CeedBasisDestroy(&basis_i)); 1463 } 1464 1465 // Initialize constants, and matrices B and G 1466 code << "\n" << tab << "// Input field constants and basis data\n"; 1467 for (CeedInt i = 0; i < num_input_fields; i++) { 1468 CeedCallBackend(CeedOperatorBuildKernelFieldData_Cuda_gen(code, data, tab, i, op_input_fields[i], qf_input_fields[i], input_matrix_reuse[i], 1469 max_dim, Q, Q_1d, true, is_all_tensor, is_at_points, use_3d_slices, false)); 1470 } 1471 code << "\n" << tab << "// Output field constants and basis data\n"; 1472 for (CeedInt i = 0; i < num_output_fields; i++) { 1473 CeedCallBackend(CeedOperatorBuildKernelFieldData_Cuda_gen(code, data, tab, i, op_output_fields[i], qf_output_fields[i], output_matrix_reuse[i], 1474 max_dim, Q, Q_1d, false, is_all_tensor, is_at_points, use_3d_slices, false)); 1475 } 1476 1477 // Loop over all elements 1478 code << "\n" << tab << "// Element loop\n"; 1479 code << tab << "__syncthreads();\n"; 1480 code << tab << "for (CeedInt elem = blockIdx.x*blockDim.z + threadIdx.z; elem < num_elem; elem += gridDim.x*blockDim.z) {\n"; 1481 tab.push(); 1482 1483 // -- Compute minimum buffer space needed 1484 CeedInt max_rstr_buffer_size = 1; 1485 1486 for (CeedInt i = 0; i < num_input_fields; i++) { 1487 CeedEvalMode eval_mode; 1488 1489 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[i], &eval_mode)); 1490 if (eval_mode != CEED_EVAL_NONE && eval_mode != CEED_EVAL_WEIGHT) { 1491 CeedInt num_comp; 1492 CeedElemRestriction elem_rstr; 1493 1494 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_input_fields[i], &elem_rstr)); 1495 CeedCallBackend(CeedElemRestrictionGetNumComponents(elem_rstr, &num_comp)); 1496 max_rstr_buffer_size = CeedIntMax(max_rstr_buffer_size, num_comp * (is_all_tensor && (max_dim >= 3) ? Q_1d : 1)); 1497 CeedCallBackend(CeedElemRestrictionDestroy(&elem_rstr)); 1498 } 1499 } 1500 for (CeedInt i = 0; i < num_output_fields; i++) { 1501 CeedEvalMode eval_mode; 1502 1503 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_output_fields[i], &eval_mode)); 1504 if (eval_mode != CEED_EVAL_NONE) { 1505 CeedInt num_comp; 1506 CeedElemRestriction elem_rstr; 1507 1508 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_output_fields[i], &elem_rstr)); 1509 CeedCallBackend(CeedElemRestrictionGetNumComponents(elem_rstr, &num_comp)); 1510 max_rstr_buffer_size = CeedIntMax(max_rstr_buffer_size, num_comp * (is_all_tensor && (max_dim >= 3) ? Q_1d : 1)); 1511 CeedCallBackend(CeedElemRestrictionDestroy(&elem_rstr)); 1512 } 1513 } 1514 code << tab << "// Scratch restriction buffer space\n"; 1515 code << tab << "CeedScalar r_e_scratch[" << max_rstr_buffer_size << "];\n"; 1516 1517 // -- Determine best input field processing order 1518 CeedInt field_rstr_in_buffer[CEED_FIELD_MAX], input_field_order[CEED_FIELD_MAX]; 1519 1520 for (CeedInt i = 0; i < num_input_fields; i++) { 1521 field_rstr_in_buffer[i] = -1; 1522 input_field_order[i] = -1; 1523 } 1524 { 1525 bool is_ordered[CEED_FIELD_MAX]; 1526 CeedInt curr_index = 0; 1527 1528 for (CeedInt i = 0; i < num_input_fields; i++) is_ordered[i] = false; 1529 for (CeedInt i = 0; i < num_input_fields; i++) { 1530 CeedVector vec_i; 1531 CeedElemRestriction rstr_i; 1532 1533 if (is_ordered[i]) continue; 1534 field_rstr_in_buffer[i] = i; 1535 is_ordered[i] = true; 1536 input_field_order[curr_index] = i; 1537 curr_index++; 1538 CeedCallBackend(CeedOperatorFieldGetVector(op_input_fields[i], &vec_i)); 1539 if (vec_i == CEED_VECTOR_NONE) continue; // CEED_EVAL_WEIGHT 1540 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_input_fields[i], &rstr_i)); 1541 for (CeedInt j = i + 1; j < num_input_fields; j++) { 1542 CeedVector vec_j; 1543 CeedElemRestriction rstr_j; 1544 1545 CeedCallBackend(CeedOperatorFieldGetVector(op_input_fields[j], &vec_j)); 1546 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_input_fields[j], &rstr_j)); 1547 if (rstr_i == rstr_j && vec_i == vec_j) { 1548 field_rstr_in_buffer[j] = i; 1549 is_ordered[j] = true; 1550 input_field_order[curr_index] = j; 1551 curr_index++; 1552 } 1553 CeedCallBackend(CeedVectorDestroy(&vec_j)); 1554 CeedCallBackend(CeedElemRestrictionDestroy(&rstr_j)); 1555 } 1556 CeedCallBackend(CeedVectorDestroy(&vec_i)); 1557 CeedCallBackend(CeedElemRestrictionDestroy(&rstr_i)); 1558 } 1559 } 1560 1561 // -- Input restriction and basis 1562 code << "\n" << tab << "// -- Input field restrictions and basis actions\n"; 1563 for (CeedInt i = 0; i < num_input_fields; i++) { 1564 const char *field_name; 1565 const CeedInt f = input_field_order[i]; 1566 1567 CeedCallBackend(CeedOperatorFieldGetName(op_input_fields[f], &field_name)); 1568 code << tab << "// ---- Input field " << f << ": " << field_name << "\n"; 1569 1570 // ---- Restriction 1571 CeedCallBackend(CeedOperatorBuildKernelRestriction_Cuda_gen(code, data, tab, f, field_rstr_in_buffer, op_input_fields[f], qf_input_fields[f], 1572 max_dim, Q_1d, true, is_all_tensor, is_at_points, use_3d_slices)); 1573 1574 // ---- Basis action 1575 CeedCallBackend(CeedOperatorBuildKernelBasis_Cuda_gen(code, data, tab, f, op_input_fields[f], qf_input_fields[f], max_dim, Q_1d, true, 1576 is_all_tensor, is_at_points, use_3d_slices)); 1577 } 1578 1579 // -- Q function 1580 CeedCallBackend(CeedOperatorBuildKernelQFunction_Cuda_gen(code, data, tab, max_dim, max_num_points, num_input_fields, op_input_fields, 1581 qf_input_fields, num_output_fields, op_output_fields, qf_output_fields, qfunction_name, 1582 Q_1d, is_all_tensor, is_at_points, use_3d_slices)); 1583 1584 // -- Output basis and restriction 1585 code << "\n" << tab << "// -- Output field basis action and restrictions\n"; 1586 for (CeedInt i = 0; i < num_output_fields; i++) { 1587 const char *field_name; 1588 1589 CeedCallBackend(CeedOperatorFieldGetName(op_output_fields[i], &field_name)); 1590 code << tab << "// ---- Output field " << i << ": " << field_name << "\n"; 1591 1592 // ---- Basis action 1593 CeedCallBackend(CeedOperatorBuildKernelBasis_Cuda_gen(code, data, tab, i, op_output_fields[i], qf_output_fields[i], max_dim, Q_1d, false, 1594 is_all_tensor, is_at_points, use_3d_slices)); 1595 1596 // ---- Restriction 1597 CeedCallBackend(CeedOperatorBuildKernelRestriction_Cuda_gen(code, data, tab, i, NULL, op_output_fields[i], qf_output_fields[i], max_dim, Q_1d, 1598 false, is_all_tensor, is_at_points, use_3d_slices)); 1599 } 1600 1601 // Close loop and function 1602 tab.pop(); 1603 code << tab << "}\n"; 1604 tab.pop(); 1605 code << tab << "}\n"; 1606 code << tab << "// -----------------------------------------------------------------------------\n\n"; 1607 1608 // Compile 1609 { 1610 bool is_compile_good = false; 1611 const CeedInt T_1d = CeedIntMax(is_all_tensor ? Q_1d : Q, data->max_P_1d); 1612 1613 data->thread_1d = T_1d; 1614 CeedCallBackend(CeedTryCompile_Cuda(ceed, code.str().c_str(), &is_compile_good, &data->module, 1, "OP_T_1D", T_1d)); 1615 if (is_compile_good) { 1616 *is_good_build = true; 1617 CeedCallBackend(CeedGetKernel_Cuda(ceed, data->module, operator_name.c_str(), &data->op)); 1618 } else { 1619 *is_good_build = false; 1620 data->use_fallback = true; 1621 } 1622 } 1623 CeedCallBackend(CeedOperatorSetSetupDone(op)); 1624 CeedCallBackend(CeedDestroy(&ceed)); 1625 CeedCallBackend(CeedQFunctionDestroy(&qf)); 1626 return CEED_ERROR_SUCCESS; 1627 } 1628 1629 //------------------------------------------------------------------------------ 1630 // Build AtPoints assembly operator kernel 1631 //------------------------------------------------------------------------------ 1632 static int CeedOperatorBuildKernelAssemblyAtPoints_Cuda_gen(CeedOperator op, bool is_full, bool *is_good_build) { 1633 bool is_all_tensor = true, is_at_points = false, use_3d_slices = false; 1634 Ceed ceed; 1635 CeedInt Q, Q_1d, num_input_fields, num_output_fields, max_dim = 1, max_num_points = 0, coords_comp_stride = 0; 1636 CeedQFunctionField *qf_input_fields, *qf_output_fields; 1637 CeedQFunction_Cuda_gen *qf_data; 1638 CeedQFunction qf; 1639 CeedOperatorField *op_input_fields, *op_output_fields; 1640 CeedOperator_Cuda_gen *data; 1641 std::ostringstream code; 1642 Tab tab; 1643 1644 // Check compatibility 1645 CeedCallBackend(CeedOperatorGetCeed(op, &ceed)); 1646 CeedCallBackend(CeedOperatorIsAtPoints(op, &is_at_points)); 1647 CeedCheck(is_at_points, ceed, CEED_ERROR_BACKEND, "Only AtPoints operator assembly supported"); 1648 1649 // Retrieve operator data 1650 CeedCallBackend(CeedOperatorGetData(op, &data)); 1651 Q = data->Q; 1652 Q_1d = data->Q_1d; 1653 max_dim = data->dim; 1654 { 1655 CeedElemRestriction rstr_points = NULL; 1656 1657 CeedCallBackend(CeedOperatorAtPointsGetPoints(op, &rstr_points, NULL)); 1658 CeedCallBackend(CeedElemRestrictionGetMaxPointsInElement(rstr_points, &max_num_points)); 1659 CeedCallBackend(CeedElemRestrictionGetCompStride(rstr_points, &coords_comp_stride)); 1660 CeedCallBackend(CeedElemRestrictionDestroy(&rstr_points)); 1661 } 1662 CeedCallBackend(CeedOperatorGetQFunction(op, &qf)); 1663 CeedCallBackend(CeedQFunctionGetData(qf, &qf_data)); 1664 CeedCallBackend(CeedQFunctionGetFields(qf, NULL, &qf_input_fields, NULL, &qf_output_fields)); 1665 CeedCallBackend(CeedOperatorGetFields(op, &num_input_fields, &op_input_fields, &num_output_fields, &op_output_fields)); 1666 1667 // Add atomicAdd function for old NVidia architectures 1668 { 1669 Ceed_Cuda *ceed_data; 1670 struct cudaDeviceProp prop; 1671 1672 CeedCallBackend(CeedGetData(ceed, &ceed_data)); 1673 CeedCallBackend(cudaGetDeviceProperties(&prop, ceed_data->device_id)); 1674 if ((prop.major < 6) && (CEED_SCALAR_TYPE != CEED_SCALAR_FP32)) { 1675 code << tab << "// AtomicAdd fallback source\n"; 1676 code << tab << "#include <ceed/jit-source/cuda/cuda-atomic-add-fallback.h>\n\n"; 1677 } 1678 } 1679 1680 // Load basis source files 1681 code << tab << "// Tensor basis source\n"; 1682 code << tab << "#include <ceed/jit-source/cuda/cuda-shared-basis-tensor-templates.h>\n\n"; 1683 code << tab << "// AtPoints basis source\n"; 1684 code << tab << "#include <ceed/jit-source/cuda/cuda-shared-basis-tensor-at-points-templates.h>\n\n"; 1685 code << tab << "// CodeGen operator source\n"; 1686 code << tab << "#include <ceed/jit-source/cuda/cuda-gen-templates.h>\n\n"; 1687 1688 // Get QFunction name 1689 std::string qfunction_name(qf_data->qfunction_name); 1690 std::string operator_name; 1691 1692 if (is_full) { 1693 operator_name = "CeedKernelCudaGenOperatorFullAssembly_" + qfunction_name; 1694 } else { 1695 operator_name = "CeedKernelCudaGenOperatorDiagonalAssembly_" + qfunction_name; 1696 } 1697 1698 // Define CEED_Q_VLA 1699 code << "\n" << tab << "#undef CEED_Q_VLA\n"; 1700 code << tab << "#define CEED_Q_VLA 1\n\n"; 1701 1702 // Add user QFunction source 1703 { 1704 const char *source_path; 1705 1706 CeedCallBackend(CeedQFunctionGetSourcePath(qf, &source_path)); 1707 CeedCheck(source_path, ceed, CEED_ERROR_UNSUPPORTED, "/gpu/cuda/gen backend requires QFunction source code file"); 1708 1709 code << tab << "// User QFunction source\n"; 1710 code << tab << "#include \"" << source_path << "\"\n\n"; 1711 } 1712 1713 // Setup 1714 code << "\n" << tab << "// -----------------------------------------------------------------------------\n"; 1715 code << tab << "// Operator Assembly Kernel\n"; 1716 code << tab << "// \n"; 1717 code << tab << "// d_[in,out]_i: CeedVector device array\n"; 1718 code << tab << "// r_[in,out]_e_i: Element vector register\n"; 1719 code << tab << "// r_[in,out]_q_i: Quadrature space vector register\n"; 1720 code << tab << "// r_[in,out]_c_i: AtPoints Chebyshev coefficients register\n"; 1721 code << tab << "// r_[in,out]_s_i: Quadrature space slice vector register\n"; 1722 code << tab << "// \n"; 1723 code << tab << "// s_B_[in,out]_i: Interpolation matrix, shared memory\n"; 1724 code << tab << "// s_G_[in,out]_i: Gradient matrix, shared memory\n"; 1725 code << tab << "// -----------------------------------------------------------------------------\n"; 1726 code << tab << "extern \"C\" __global__ void " << operator_name 1727 << "(CeedInt num_elem, void* ctx, FieldsInt_Cuda indices, Fields_Cuda fields, Fields_Cuda B, Fields_Cuda G, CeedScalar *W, Points_Cuda " 1728 "points, CeedScalar *__restrict__ values_array) {\n"; 1729 tab.push(); 1730 1731 // Scratch buffers 1732 for (CeedInt i = 0; i < num_input_fields; i++) { 1733 CeedEvalMode eval_mode; 1734 1735 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[i], &eval_mode)); 1736 if (eval_mode != CEED_EVAL_WEIGHT) { // Skip CEED_EVAL_WEIGHT 1737 code << tab << "const CeedScalar *__restrict__ d_in_" << i << " = fields.inputs[" << i << "];\n"; 1738 } 1739 } 1740 for (CeedInt i = 0; i < num_output_fields; i++) { 1741 code << tab << "CeedScalar *__restrict__ d_out_" << i << " = fields.outputs[" << i << "];\n"; 1742 } 1743 1744 code << tab << "const CeedInt max_dim = " << max_dim << ";\n"; 1745 code << tab << "const CeedInt Q_1d = " << Q_1d << ";\n"; 1746 code << tab << "const CeedInt max_num_points = " << max_num_points << ";\n"; 1747 code << tab << "const CeedInt coords_comp_stride = " << coords_comp_stride << ";\n"; 1748 1749 // Shared data 1750 code << tab << "extern __shared__ CeedScalar slice[];\n"; 1751 code << tab << "SharedData_Cuda data;\n"; 1752 code << tab << "data.t_id_x = threadIdx.x;\n"; 1753 code << tab << "data.t_id_y = threadIdx.y;\n"; 1754 code << tab << "data.t_id_z = threadIdx.z;\n"; 1755 code << tab << "data.t_id = threadIdx.x + threadIdx.y*blockDim.x + threadIdx.z*blockDim.y*blockDim.x;\n"; 1756 code << tab << "data.slice = slice + data.t_id_z*OP_T_1D" << ((!is_all_tensor || max_dim == 1) ? "" : "*OP_T_1D") << ";\n"; 1757 1758 // -- Determine input mat reuse 1759 FieldReuse_Cuda input_matrix_reuse[CEED_FIELD_MAX]; 1760 1761 for (CeedInt i = 0; i < num_input_fields; i++) { 1762 input_matrix_reuse[i].index = -1; 1763 } 1764 for (CeedInt i = 0; i < num_input_fields; i++) { 1765 CeedEvalMode eval_mode_i; 1766 CeedBasis basis_i; 1767 1768 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[i], &eval_mode_i)); 1769 if (eval_mode_i == CEED_EVAL_WEIGHT) continue; 1770 CeedCallBackend(CeedOperatorFieldGetBasis(op_input_fields[i], &basis_i)); 1771 for (CeedInt j = 0; (input_matrix_reuse[i].index == -1) && (j < i); j++) { 1772 CeedEvalMode eval_mode_j; 1773 CeedBasis basis_j; 1774 1775 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[j], &eval_mode_j)); 1776 if (eval_mode_j == CEED_EVAL_WEIGHT) continue; 1777 CeedCallBackend(CeedOperatorFieldGetBasis(op_input_fields[j], &basis_j)); 1778 if (basis_i == basis_j) { 1779 input_matrix_reuse[i].index = j; 1780 input_matrix_reuse[i].is_input = true; 1781 input_matrix_reuse[i].eval_mode = eval_mode_j; 1782 } 1783 CeedCallBackend(CeedBasisDestroy(&basis_j)); 1784 } 1785 CeedCallBackend(CeedBasisDestroy(&basis_i)); 1786 } 1787 1788 // -- Determine output mat reuse 1789 FieldReuse_Cuda output_matrix_reuse[CEED_FIELD_MAX]; 1790 1791 for (CeedInt i = 0; i < num_output_fields; i++) { 1792 output_matrix_reuse[i].index = -1; 1793 } 1794 for (CeedInt i = 0; i < num_output_fields; i++) { 1795 CeedEvalMode eval_mode_i; 1796 CeedBasis basis_i; 1797 1798 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_output_fields[i], &eval_mode_i)); 1799 CeedCallBackend(CeedOperatorFieldGetBasis(op_output_fields[i], &basis_i)); 1800 for (CeedInt j = 0; (output_matrix_reuse[i].index == -1) && (j < num_input_fields); j++) { 1801 CeedEvalMode eval_mode_j; 1802 CeedBasis basis_j; 1803 1804 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[j], &eval_mode_j)); 1805 if (eval_mode_j == CEED_EVAL_WEIGHT) continue; 1806 CeedCallBackend(CeedOperatorFieldGetBasis(op_input_fields[j], &basis_j)); 1807 if (basis_i == basis_j) { 1808 output_matrix_reuse[i].index = j; 1809 output_matrix_reuse[i].is_input = true; 1810 output_matrix_reuse[i].eval_mode = eval_mode_j; 1811 } 1812 CeedCallBackend(CeedBasisDestroy(&basis_j)); 1813 } 1814 for (CeedInt j = 0; (output_matrix_reuse[i].index == -1) && (j < i); j++) { 1815 CeedEvalMode eval_mode_j; 1816 CeedBasis basis_j; 1817 1818 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_output_fields[j], &eval_mode_j)); 1819 if (eval_mode_j == CEED_EVAL_WEIGHT) continue; 1820 CeedCallBackend(CeedOperatorFieldGetBasis(op_output_fields[j], &basis_j)); 1821 if (basis_i == basis_j) { 1822 output_matrix_reuse[i].index = j; 1823 output_matrix_reuse[i].is_input = false; 1824 output_matrix_reuse[i].eval_mode = eval_mode_j; 1825 } 1826 CeedCallBackend(CeedBasisDestroy(&basis_j)); 1827 } 1828 CeedCallBackend(CeedBasisDestroy(&basis_i)); 1829 } 1830 1831 // Initialize constants, and matrices B and G 1832 code << "\n" << tab << "// Input field constants and basis data\n"; 1833 for (CeedInt i = 0; i < num_input_fields; i++) { 1834 CeedCallBackend(CeedOperatorBuildKernelFieldData_Cuda_gen(code, data, tab, i, op_input_fields[i], qf_input_fields[i], input_matrix_reuse[i], 1835 max_dim, Q, Q_1d, true, is_all_tensor, is_at_points, use_3d_slices, false)); 1836 } 1837 code << "\n" << tab << "// Output field constants and basis data\n"; 1838 for (CeedInt i = 0; i < num_output_fields; i++) { 1839 CeedCallBackend(CeedOperatorBuildKernelFieldData_Cuda_gen(code, data, tab, i, op_output_fields[i], qf_output_fields[i], output_matrix_reuse[i], 1840 max_dim, Q, Q_1d, false, is_all_tensor, is_at_points, use_3d_slices, false)); 1841 } 1842 1843 // Loop over all elements 1844 code << "\n" << tab << "// Element loop\n"; 1845 code << tab << "__syncthreads();\n"; 1846 code << tab << "for (CeedInt elem = blockIdx.x*blockDim.z + threadIdx.z; elem < num_elem; elem += gridDim.x*blockDim.z) {\n"; 1847 tab.push(); 1848 1849 // -- Compute minimum buffer space needed 1850 CeedInt max_rstr_buffer_size = 1; 1851 1852 for (CeedInt i = 0; i < num_input_fields; i++) { 1853 CeedEvalMode eval_mode; 1854 1855 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[i], &eval_mode)); 1856 if (eval_mode != CEED_EVAL_NONE && eval_mode != CEED_EVAL_WEIGHT) { 1857 CeedInt num_comp; 1858 CeedElemRestriction elem_rstr; 1859 1860 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_input_fields[i], &elem_rstr)); 1861 CeedCallBackend(CeedElemRestrictionGetNumComponents(elem_rstr, &num_comp)); 1862 max_rstr_buffer_size = CeedIntMax(max_rstr_buffer_size, num_comp * (is_all_tensor && (max_dim >= 3) ? Q_1d : 1)); 1863 CeedCallBackend(CeedElemRestrictionDestroy(&elem_rstr)); 1864 } 1865 } 1866 for (CeedInt i = 0; i < num_output_fields; i++) { 1867 CeedEvalMode eval_mode; 1868 1869 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_output_fields[i], &eval_mode)); 1870 if (eval_mode != CEED_EVAL_NONE) { 1871 CeedInt num_comp; 1872 CeedElemRestriction elem_rstr; 1873 1874 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_output_fields[i], &elem_rstr)); 1875 CeedCallBackend(CeedElemRestrictionGetNumComponents(elem_rstr, &num_comp)); 1876 max_rstr_buffer_size = CeedIntMax(max_rstr_buffer_size, num_comp * (is_all_tensor && (max_dim >= 3) ? Q_1d : 1)); 1877 CeedCallBackend(CeedElemRestrictionDestroy(&elem_rstr)); 1878 } 1879 } 1880 code << tab << "// Scratch restriction buffer space\n"; 1881 code << tab << "CeedScalar r_e_scratch[" << max_rstr_buffer_size << "];\n"; 1882 1883 // -- Determine best input field processing order 1884 CeedInt field_rstr_in_buffer[CEED_FIELD_MAX], input_field_order[CEED_FIELD_MAX]; 1885 1886 for (CeedInt i = 0; i < num_input_fields; i++) { 1887 field_rstr_in_buffer[i] = -1; 1888 input_field_order[i] = -1; 1889 } 1890 { 1891 bool is_ordered[CEED_FIELD_MAX]; 1892 CeedInt curr_index = 0; 1893 1894 for (CeedInt i = 0; i < num_input_fields; i++) is_ordered[i] = false; 1895 for (CeedInt i = 0; i < num_input_fields; i++) { 1896 CeedVector vec_i; 1897 CeedElemRestriction rstr_i; 1898 1899 if (is_ordered[i]) continue; 1900 field_rstr_in_buffer[i] = i; 1901 is_ordered[i] = true; 1902 input_field_order[curr_index] = i; 1903 curr_index++; 1904 CeedCallBackend(CeedOperatorFieldGetVector(op_input_fields[i], &vec_i)); 1905 if (vec_i == CEED_VECTOR_NONE) continue; // CEED_EVAL_WEIGHT 1906 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_input_fields[i], &rstr_i)); 1907 for (CeedInt j = i + 1; j < num_input_fields; j++) { 1908 CeedVector vec_j; 1909 CeedElemRestriction rstr_j; 1910 1911 CeedCallBackend(CeedOperatorFieldGetVector(op_input_fields[j], &vec_j)); 1912 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_input_fields[j], &rstr_j)); 1913 if (rstr_i == rstr_j && vec_i == vec_j) { 1914 field_rstr_in_buffer[j] = i; 1915 is_ordered[j] = true; 1916 input_field_order[curr_index] = j; 1917 curr_index++; 1918 } 1919 CeedCallBackend(CeedVectorDestroy(&vec_j)); 1920 CeedCallBackend(CeedElemRestrictionDestroy(&rstr_j)); 1921 } 1922 CeedCallBackend(CeedVectorDestroy(&vec_i)); 1923 CeedCallBackend(CeedElemRestrictionDestroy(&rstr_i)); 1924 } 1925 } 1926 1927 // -- Input restriction and basis 1928 code << "\n" << tab << "// -- Input field restrictions and basis actions\n"; 1929 CeedInt active_field_index = -1; 1930 1931 for (CeedInt i = 0; i < num_input_fields; i++) { 1932 bool is_active = false; 1933 const char *field_name; 1934 const CeedInt f = input_field_order[i]; 1935 1936 { 1937 CeedVector vec; 1938 1939 CeedCallBackend(CeedOperatorFieldGetVector(op_input_fields[f], &vec)); 1940 is_active = vec == CEED_VECTOR_ACTIVE; 1941 CeedCallBackend(CeedVectorDestroy(&vec)); 1942 } 1943 1944 CeedCallBackend(CeedOperatorFieldGetName(op_input_fields[f], &field_name)); 1945 code << tab << "// ---- Input field " << f << ": " << field_name << "\n"; 1946 1947 if (is_active) { 1948 std::string var_suffix = "_in_" + std::to_string(f); 1949 1950 code << tab << "// Active field - no restriction or basis action here\n"; 1951 if (active_field_index == -1) { 1952 active_field_index = f; 1953 code << tab << "CeedScalar r_e" << var_suffix << "[num_comp" << var_suffix << "*" << (max_dim >= 3 ? "P_1d" + var_suffix : "1") 1954 << "] = {0.0};\n"; 1955 } else { 1956 code << tab << "CeedScalar *r_e" << var_suffix << " = r_e_in_" << active_field_index << ";\n"; 1957 } 1958 } else { 1959 // ---- Restriction 1960 CeedCallBackend(CeedOperatorBuildKernelRestriction_Cuda_gen(code, data, tab, f, field_rstr_in_buffer, op_input_fields[f], qf_input_fields[f], 1961 max_dim, Q_1d, true, is_all_tensor, is_at_points, use_3d_slices)); 1962 1963 // ---- Basis action 1964 CeedCallBackend(CeedOperatorBuildKernelBasis_Cuda_gen(code, data, tab, f, op_input_fields[f], qf_input_fields[f], max_dim, Q_1d, true, 1965 is_all_tensor, is_at_points, use_3d_slices)); 1966 } 1967 } 1968 1969 // -- Loop over active field 1970 std::string active_var_suffix = "_in_" + std::to_string(active_field_index); 1971 1972 code << "\n" << tab << "// Loop over nodes in active field\n"; 1973 code << tab << "for (CeedInt n = 0; n < num_comp" << active_var_suffix << "*P_1d" << active_var_suffix 1974 << (max_dim > 1 ? "*P_1d" + active_var_suffix : "") << (max_dim > 2 ? "*P_1d" + active_var_suffix : "") << "; n++) {\n"; 1975 tab.push(); 1976 1977 // -- Set current active node and component to 1 1978 code << tab << "// Set current active node and component to 1.0\n"; 1979 code << tab << "SetEVecStandard" << max_dim << "d_Single<num_comp" << active_var_suffix << ", P_1d" << active_var_suffix << ">(data, n, 1.0, r_e" 1980 << active_var_suffix << ");\n\n"; 1981 1982 for (CeedInt i = 0; i < num_input_fields; i++) { 1983 bool is_active = false; 1984 const char *field_name; 1985 const CeedInt f = input_field_order[i]; 1986 1987 { 1988 CeedVector vec; 1989 1990 CeedCallBackend(CeedOperatorFieldGetVector(op_input_fields[f], &vec)); 1991 is_active = vec == CEED_VECTOR_ACTIVE; 1992 CeedCallBackend(CeedVectorDestroy(&vec)); 1993 } 1994 if (!is_active) continue; 1995 1996 CeedCallBackend(CeedOperatorFieldGetName(op_input_fields[f], &field_name)); 1997 code << tab << "// ---- Input field " << f << ": " << field_name << "\n"; 1998 1999 // ---- Basis action 2000 CeedCallBackend(CeedOperatorBuildKernelBasis_Cuda_gen(code, data, tab, f, op_input_fields[f], qf_input_fields[f], max_dim, Q_1d, true, 2001 is_all_tensor, is_at_points, use_3d_slices)); 2002 } 2003 2004 // -- Q function 2005 CeedCallBackend(CeedOperatorBuildKernelQFunction_Cuda_gen(code, data, tab, max_dim, max_num_points, num_input_fields, op_input_fields, 2006 qf_input_fields, num_output_fields, op_output_fields, qf_output_fields, qfunction_name, 2007 Q_1d, is_all_tensor, is_at_points, use_3d_slices)); 2008 2009 // -- Output basis and restriction 2010 code << "\n" << tab << "// -- Output field basis action and restrictions\n"; 2011 for (CeedInt i = 0; i < num_output_fields; i++) { 2012 bool is_active = false; 2013 const char *field_name; 2014 2015 { 2016 CeedVector vec; 2017 2018 CeedCallBackend(CeedOperatorFieldGetVector(op_output_fields[i], &vec)); 2019 is_active = vec == CEED_VECTOR_ACTIVE; 2020 CeedCallBackend(CeedVectorDestroy(&vec)); 2021 } 2022 if (!is_active) continue; 2023 2024 CeedCallBackend(CeedOperatorFieldGetName(op_output_fields[i], &field_name)); 2025 code << tab << "// ---- Output field " << i << ": " << field_name << "\n"; 2026 2027 // ---- Basis action 2028 CeedCallBackend(CeedOperatorBuildKernelBasis_Cuda_gen(code, data, tab, i, op_output_fields[i], qf_output_fields[i], max_dim, Q_1d, false, 2029 is_all_tensor, is_at_points, use_3d_slices)); 2030 2031 // ---- Restriction 2032 if (is_full) { 2033 std::string var_suffix = "_out_" + std::to_string(i); 2034 CeedInt comp_stride; 2035 CeedSize l_size; 2036 CeedElemRestriction elem_rstr; 2037 2038 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_output_fields[i], &elem_rstr)); 2039 CeedCallBackend(CeedElemRestrictionGetLVectorSize(elem_rstr, &l_size)); 2040 code << tab << "const CeedInt l_size" << var_suffix << " = " << l_size << ";\n"; 2041 CeedCallBackend(CeedElemRestrictionGetCompStride(elem_rstr, &comp_stride)); 2042 code << tab << "const CeedInt comp_stride" << var_suffix << " = " << comp_stride << ";\n"; 2043 code << tab << "WriteLVecStandard" << max_dim << "d_Assembly<num_comp" << var_suffix << ", comp_stride" << var_suffix << ", P_1d" + var_suffix 2044 << ">(data, l_size" << var_suffix << ", elem, n, r_e" << var_suffix << ", values_array);\n"; 2045 CeedCallBackend(CeedElemRestrictionDestroy(&elem_rstr)); 2046 } else { 2047 std::string var_suffix = "_out_" + std::to_string(i); 2048 CeedInt comp_stride; 2049 CeedSize l_size; 2050 CeedElemRestriction elem_rstr; 2051 2052 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_output_fields[i], &elem_rstr)); 2053 CeedCallBackend(CeedElemRestrictionGetLVectorSize(elem_rstr, &l_size)); 2054 code << tab << "const CeedInt l_size" << var_suffix << " = " << l_size << ";\n"; 2055 CeedCallBackend(CeedElemRestrictionGetCompStride(elem_rstr, &comp_stride)); 2056 code << tab << "const CeedInt comp_stride" << var_suffix << " = " << comp_stride << ";\n"; 2057 code << tab << "WriteLVecStandard" << max_dim << "d_Single<num_comp" << var_suffix << ", comp_stride" << var_suffix << ", P_1d" + var_suffix 2058 << ">(data, l_size" << var_suffix << ", elem, n, indices.outputs[" << i << "], r_e" << var_suffix << ", values_array);\n"; 2059 CeedCallBackend(CeedElemRestrictionDestroy(&elem_rstr)); 2060 } 2061 } 2062 2063 // -- Reset current active node and component 2064 code << "\n" << tab << "// Reset current active node and component to 0.0\n"; 2065 code << tab << "SetEVecStandard" << max_dim << "d_Single<num_comp" << active_var_suffix << ", P_1d" << active_var_suffix << ">(data, n, 0.0, r_e" 2066 << active_var_suffix << ");\n"; 2067 2068 // -- End of loop over active field 2069 tab.pop(); 2070 code << tab << "}\n"; 2071 2072 // Close loop and function 2073 tab.pop(); 2074 code << tab << "}\n"; 2075 tab.pop(); 2076 code << tab << "}\n"; 2077 code << tab << "// -----------------------------------------------------------------------------\n\n"; 2078 2079 // Compile 2080 { 2081 bool is_compile_good = false; 2082 const CeedInt T_1d = CeedIntMax(is_all_tensor ? Q_1d : Q, data->max_P_1d); 2083 2084 data->thread_1d = T_1d; 2085 CeedCallBackend(CeedTryCompile_Cuda(ceed, code.str().c_str(), &is_compile_good, 2086 is_full ? &data->module_assemble_full : &data->module_assemble_diagonal, 1, "OP_T_1D", T_1d)); 2087 if (is_compile_good) { 2088 *is_good_build = true; 2089 CeedCallBackend(CeedGetKernel_Cuda(ceed, is_full ? data->module_assemble_full : data->module_assemble_diagonal, operator_name.c_str(), 2090 is_full ? &data->assemble_full : &data->assemble_diagonal)); 2091 } else { 2092 *is_good_build = false; 2093 data->use_assembly_fallback = true; 2094 } 2095 } 2096 CeedCallBackend(CeedDestroy(&ceed)); 2097 CeedCallBackend(CeedQFunctionDestroy(&qf)); 2098 return CEED_ERROR_SUCCESS; 2099 } 2100 2101 extern "C" int CeedOperatorBuildKernelDiagonalAssemblyAtPoints_Cuda_gen(CeedOperator op, bool *is_good_build) { 2102 return CeedOperatorBuildKernelAssemblyAtPoints_Cuda_gen(op, false, is_good_build); 2103 } 2104 2105 extern "C" int CeedOperatorBuildKernelFullAssemblyAtPoints_Cuda_gen(CeedOperator op, bool *is_good_build) { 2106 return CeedOperatorBuildKernelAssemblyAtPoints_Cuda_gen(op, true, is_good_build); 2107 } 2108 2109 //------------------------------------------------------------------------------ 2110 // Build QFunction assembly operator kernel 2111 //------------------------------------------------------------------------------ 2112 extern "C" int CeedOperatorBuildKernelLinearAssembleQFunction_Cuda_gen(CeedOperator op, bool *is_good_build) { 2113 bool is_all_tensor = true, is_all_nontensor = true, is_at_points = false, use_3d_slices = false; 2114 Ceed ceed; 2115 CeedInt Q, Q_1d, num_input_fields, num_output_fields, max_dim = 1, max_num_points = 0; 2116 CeedQFunctionField *qf_input_fields, *qf_output_fields; 2117 CeedQFunction_Cuda_gen *qf_data; 2118 CeedQFunction qf; 2119 CeedOperatorField *op_input_fields, *op_output_fields; 2120 CeedOperator_Cuda_gen *data; 2121 std::ostringstream code; 2122 Tab tab; 2123 2124 // Check compatibility 2125 CeedCallBackend(CeedOperatorGetCeed(op, &ceed)); 2126 CeedCallBackend(CeedOperatorIsAtPoints(op, &is_at_points)); 2127 CeedCheck(!is_at_points, ceed, CEED_ERROR_BACKEND, "AtPoints QFunction assembly is not supported"); 2128 2129 // Check field compatibility 2130 CeedCallBackend(CeedOperatorGetFields(op, &num_input_fields, &op_input_fields, &num_output_fields, &op_output_fields)); 2131 { 2132 bool has_shared_bases = true; 2133 2134 for (CeedInt i = 0; i < num_input_fields; i++) { 2135 CeedBasis basis; 2136 2137 CeedCallBackend(CeedOperatorFieldGetBasis(op_input_fields[i], &basis)); 2138 if (basis != CEED_BASIS_NONE) { 2139 bool is_tensor = true; 2140 const char *resource; 2141 char *resource_root; 2142 Ceed basis_ceed; 2143 2144 CeedCallBackend(CeedBasisIsTensor(basis, &is_tensor)); 2145 is_all_tensor = is_all_tensor && is_tensor; 2146 is_all_nontensor = is_all_nontensor && !is_tensor; 2147 CeedCallBackend(CeedBasisGetCeed(basis, &basis_ceed)); 2148 CeedCallBackend(CeedGetResource(basis_ceed, &resource)); 2149 CeedCallBackend(CeedGetResourceRoot(basis_ceed, resource, ":", &resource_root)); 2150 has_shared_bases = has_shared_bases && !strcmp(resource_root, "/gpu/cuda/shared"); 2151 CeedCallBackend(CeedFree(&resource_root)); 2152 CeedCallBackend(CeedDestroy(&basis_ceed)); 2153 } 2154 CeedCallBackend(CeedBasisDestroy(&basis)); 2155 } 2156 2157 for (CeedInt i = 0; i < num_output_fields; i++) { 2158 CeedBasis basis; 2159 2160 CeedCallBackend(CeedOperatorFieldGetBasis(op_output_fields[i], &basis)); 2161 if (basis != CEED_BASIS_NONE) { 2162 bool is_tensor = true; 2163 const char *resource; 2164 char *resource_root; 2165 Ceed basis_ceed; 2166 2167 CeedCallBackend(CeedBasisIsTensor(basis, &is_tensor)); 2168 is_all_tensor = is_all_tensor && is_tensor; 2169 is_all_nontensor = is_all_nontensor && !is_tensor; 2170 2171 CeedCallBackend(CeedBasisGetCeed(basis, &basis_ceed)); 2172 CeedCallBackend(CeedGetResource(basis_ceed, &resource)); 2173 CeedCallBackend(CeedGetResourceRoot(basis_ceed, resource, ":", &resource_root)); 2174 has_shared_bases = has_shared_bases && !strcmp(resource_root, "/gpu/cuda/shared"); 2175 CeedCallBackend(CeedFree(&resource_root)); 2176 CeedCallBackend(CeedDestroy(&basis_ceed)); 2177 } 2178 CeedCallBackend(CeedBasisDestroy(&basis)); 2179 } 2180 } 2181 2182 // Retrieve operator data 2183 CeedCallBackend(CeedOperatorGetData(op, &data)); 2184 Q = data->Q; 2185 Q_1d = data->Q_1d; 2186 max_dim = data->dim; 2187 CeedCallBackend(CeedOperatorGetQFunction(op, &qf)); 2188 CeedCallBackend(CeedQFunctionGetData(qf, &qf_data)); 2189 CeedCallBackend(CeedQFunctionGetFields(qf, NULL, &qf_input_fields, NULL, &qf_output_fields)); 2190 2191 // Add atomicAdd function for old NVidia architectures 2192 { 2193 Ceed_Cuda *ceed_data; 2194 struct cudaDeviceProp prop; 2195 2196 CeedCallBackend(CeedGetData(ceed, &ceed_data)); 2197 CeedCallBackend(cudaGetDeviceProperties(&prop, ceed_data->device_id)); 2198 if ((prop.major < 6) && (CEED_SCALAR_TYPE != CEED_SCALAR_FP32)) { 2199 code << tab << "// AtomicAdd fallback source\n"; 2200 code << tab << "#include <ceed/jit-source/cuda/cuda-atomic-add-fallback.h>\n\n"; 2201 } 2202 } 2203 2204 // Load basis source files 2205 if (!is_all_nontensor) { 2206 code << tab << "// Tensor basis source\n"; 2207 code << tab << "#include <ceed/jit-source/cuda/cuda-shared-basis-tensor-templates.h>\n\n"; 2208 } 2209 if (!is_all_tensor) { 2210 code << tab << "// Non-tensor basis source\n"; 2211 code << tab << "#include <ceed/jit-source/cuda/cuda-shared-basis-nontensor-templates.h>\n\n"; 2212 } 2213 if (!is_all_tensor && !is_all_nontensor) { 2214 code << "// Tensor basis source\n"; 2215 code << "#include <ceed/jit-source/cuda/cuda-shared-basis-tensor-flattened-templates.h>\n\n"; 2216 } 2217 code << "// CodeGen operator source\n"; 2218 code << "#include <ceed/jit-source/cuda/cuda-gen-templates.h>\n\n"; 2219 2220 // Get QFunction name 2221 std::string qfunction_name(qf_data->qfunction_name); 2222 std::string operator_name; 2223 2224 operator_name = "CeedKernelCudaGenQFunctionAssembly_" + qfunction_name; 2225 2226 // Define CEED_Q_VLA 2227 code << "\n" << tab << "#undef CEED_Q_VLA\n"; 2228 if (max_dim != 3 || is_at_points || use_3d_slices || !is_all_tensor) { 2229 code << tab << "#define CEED_Q_VLA 1\n\n"; 2230 } else { 2231 code << tab << "#define CEED_Q_VLA " << Q_1d << "\n\n"; 2232 } 2233 2234 // Add user QFunction source 2235 { 2236 const char *source_path; 2237 2238 CeedCallBackend(CeedQFunctionGetSourcePath(qf, &source_path)); 2239 CeedCheck(source_path, ceed, CEED_ERROR_UNSUPPORTED, "/gpu/cuda/gen backend requires QFunction source code file"); 2240 2241 code << tab << "// User QFunction source\n"; 2242 code << tab << "#include \"" << source_path << "\"\n\n"; 2243 } 2244 2245 // Setup 2246 code << "\n" << tab << "// -----------------------------------------------------------------------------\n"; 2247 code << tab << "// Operator Assembly Kernel\n"; 2248 code << tab << "// \n"; 2249 code << tab << "// d_[in,out]_i: CeedVector device array\n"; 2250 code << tab << "// r_[in,out]_e_i: Element vector register\n"; 2251 code << tab << "// r_[in,out]_q_i: Quadrature space vector register\n"; 2252 code << tab << "// r_[in,out]_c_i: AtPoints Chebyshev coefficients register\n"; 2253 code << tab << "// r_[in,out]_s_i: Quadrature space slice vector register\n"; 2254 code << tab << "// \n"; 2255 code << tab << "// s_B_[in,out]_i: Interpolation matrix, shared memory\n"; 2256 code << tab << "// s_G_[in,out]_i: Gradient matrix, shared memory\n"; 2257 code << tab << "// -----------------------------------------------------------------------------\n"; 2258 code << tab << "extern \"C\" __global__ void " << operator_name 2259 << "(CeedInt num_elem, void* ctx, FieldsInt_Cuda indices, Fields_Cuda fields, Fields_Cuda B, Fields_Cuda G, CeedScalar *W, Points_Cuda " 2260 "points, CeedScalar *__restrict__ values_array) {\n"; 2261 tab.push(); 2262 2263 // Scratch buffers 2264 for (CeedInt i = 0; i < num_input_fields; i++) { 2265 CeedEvalMode eval_mode; 2266 2267 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[i], &eval_mode)); 2268 if (eval_mode != CEED_EVAL_WEIGHT) { // Skip CEED_EVAL_WEIGHT 2269 code << tab << "const CeedScalar *__restrict__ d_in_" << i << " = fields.inputs[" << i << "];\n"; 2270 } 2271 } 2272 for (CeedInt i = 0; i < num_output_fields; i++) { 2273 code << tab << "CeedScalar *__restrict__ d_out_" << i << " = fields.outputs[" << i << "];\n"; 2274 } 2275 2276 code << tab << "const CeedInt max_dim = " << max_dim << ";\n"; 2277 if (!is_all_tensor) { 2278 code << tab << "const CeedInt Q = " << Q << ";\n"; 2279 } 2280 if (!is_all_nontensor) { 2281 code << tab << "const CeedInt Q_1d = " << Q_1d << ";\n"; 2282 } 2283 2284 // Shared data 2285 code << tab << "extern __shared__ CeedScalar slice[];\n"; 2286 code << tab << "SharedData_Cuda data;\n"; 2287 code << tab << "data.t_id_x = threadIdx.x;\n"; 2288 code << tab << "data.t_id_y = threadIdx.y;\n"; 2289 code << tab << "data.t_id_z = threadIdx.z;\n"; 2290 code << tab << "data.t_id = threadIdx.x + threadIdx.y*blockDim.x + threadIdx.z*blockDim.y*blockDim.x;\n"; 2291 code << tab << "data.slice = slice + data.t_id_z*OP_T_1D" << ((!is_all_tensor || max_dim == 1) ? "" : "*OP_T_1D") << ";\n"; 2292 2293 // -- Determine input mat reuse 2294 FieldReuse_Cuda input_matrix_reuse[CEED_FIELD_MAX]; 2295 2296 for (CeedInt i = 0; i < num_input_fields; i++) { 2297 input_matrix_reuse[i].index = -1; 2298 } 2299 for (CeedInt i = 0; i < num_input_fields; i++) { 2300 bool is_tensor = true; 2301 CeedEvalMode eval_mode_i; 2302 CeedBasis basis_i; 2303 2304 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[i], &eval_mode_i)); 2305 if (eval_mode_i == CEED_EVAL_WEIGHT) continue; 2306 CeedCallBackend(CeedOperatorFieldGetBasis(op_input_fields[i], &basis_i)); 2307 CeedCallBackend(CeedBasisIsTensor(basis_i, &is_tensor)); 2308 for (CeedInt j = 0; (input_matrix_reuse[i].index == -1) && (j < i); j++) { 2309 CeedEvalMode eval_mode_j; 2310 CeedBasis basis_j; 2311 2312 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[j], &eval_mode_j)); 2313 if (eval_mode_j == CEED_EVAL_WEIGHT) continue; 2314 CeedCallBackend(CeedOperatorFieldGetBasis(op_input_fields[j], &basis_j)); 2315 if (basis_i == basis_j) { 2316 if (is_tensor) { 2317 input_matrix_reuse[i].index = j; 2318 input_matrix_reuse[i].is_input = true; 2319 input_matrix_reuse[i].eval_mode = eval_mode_j; 2320 } else { 2321 // For non-tensor can only re-use with the same eval mode 2322 if (eval_mode_i == eval_mode_j) { 2323 input_matrix_reuse[i].index = j; 2324 input_matrix_reuse[i].is_input = true; 2325 input_matrix_reuse[i].eval_mode = eval_mode_j; 2326 } 2327 } 2328 } 2329 CeedCallBackend(CeedBasisDestroy(&basis_j)); 2330 } 2331 CeedCallBackend(CeedBasisDestroy(&basis_i)); 2332 } 2333 2334 // -- Determine output mat reuse 2335 FieldReuse_Cuda output_matrix_reuse[CEED_FIELD_MAX]; 2336 2337 for (CeedInt i = 0; i < num_output_fields; i++) { 2338 output_matrix_reuse[i].index = -1; 2339 } 2340 for (CeedInt i = 0; i < num_output_fields; i++) { 2341 bool is_tensor = true; 2342 CeedEvalMode eval_mode_i; 2343 CeedBasis basis_i; 2344 2345 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_output_fields[i], &eval_mode_i)); 2346 CeedCallBackend(CeedOperatorFieldGetBasis(op_output_fields[i], &basis_i)); 2347 CeedCallBackend(CeedBasisIsTensor(basis_i, &is_tensor)); 2348 for (CeedInt j = 0; (output_matrix_reuse[i].index == -1) && (j < num_input_fields); j++) { 2349 CeedEvalMode eval_mode_j; 2350 CeedBasis basis_j; 2351 2352 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[j], &eval_mode_j)); 2353 if (eval_mode_j == CEED_EVAL_WEIGHT) continue; 2354 CeedCallBackend(CeedOperatorFieldGetBasis(op_input_fields[j], &basis_j)); 2355 if (basis_i == basis_j) { 2356 if (is_tensor) { 2357 output_matrix_reuse[i].index = j; 2358 output_matrix_reuse[i].is_input = true; 2359 output_matrix_reuse[i].eval_mode = eval_mode_j; 2360 } else { 2361 // For non-tensor can only re-use with the same eval mode 2362 if (eval_mode_i == eval_mode_j) { 2363 output_matrix_reuse[i].index = j; 2364 output_matrix_reuse[i].is_input = true; 2365 output_matrix_reuse[i].eval_mode = eval_mode_j; 2366 } 2367 } 2368 } 2369 CeedCallBackend(CeedBasisDestroy(&basis_j)); 2370 } 2371 for (CeedInt j = 0; (output_matrix_reuse[i].index == -1) && (j < i); j++) { 2372 CeedEvalMode eval_mode_j; 2373 CeedBasis basis_j; 2374 2375 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_output_fields[j], &eval_mode_j)); 2376 if (eval_mode_j == CEED_EVAL_WEIGHT) continue; 2377 CeedCallBackend(CeedOperatorFieldGetBasis(op_output_fields[j], &basis_j)); 2378 if (basis_i == basis_j) { 2379 if (is_tensor) { 2380 output_matrix_reuse[i].index = j; 2381 output_matrix_reuse[i].is_input = false; 2382 output_matrix_reuse[i].eval_mode = eval_mode_j; 2383 } else { 2384 // For non-tensor can only re-use with the same eval mode 2385 if (eval_mode_i == eval_mode_j) { 2386 output_matrix_reuse[i].index = j; 2387 output_matrix_reuse[i].is_input = false; 2388 output_matrix_reuse[i].eval_mode = eval_mode_j; 2389 } 2390 } 2391 } 2392 CeedCallBackend(CeedBasisDestroy(&basis_j)); 2393 } 2394 CeedCallBackend(CeedBasisDestroy(&basis_i)); 2395 } 2396 2397 // Initialize constants, and matrices B and G 2398 code << "\n" << tab << "// Input field constants and basis data\n"; 2399 for (CeedInt i = 0; i < num_input_fields; i++) { 2400 CeedCallBackend(CeedOperatorBuildKernelFieldData_Cuda_gen(code, data, tab, i, op_input_fields[i], qf_input_fields[i], input_matrix_reuse[i], 2401 max_dim, Q, Q_1d, true, is_all_tensor, is_at_points, use_3d_slices, true)); 2402 } 2403 code << "\n" << tab << "// Output field constants and basis data\n"; 2404 for (CeedInt i = 0; i < num_output_fields; i++) { 2405 CeedCallBackend(CeedOperatorBuildKernelFieldData_Cuda_gen(code, data, tab, i, op_output_fields[i], qf_output_fields[i], output_matrix_reuse[i], 2406 max_dim, Q, Q_1d, false, is_all_tensor, is_at_points, use_3d_slices, true)); 2407 } 2408 2409 // Loop over all elements 2410 code << "\n" << tab << "// Element loop\n"; 2411 code << tab << "__syncthreads();\n"; 2412 code << tab << "for (CeedInt elem = blockIdx.x*blockDim.z + threadIdx.z; elem < num_elem; elem += gridDim.x*blockDim.z) {\n"; 2413 tab.push(); 2414 2415 // -- Compute minimum buffer space needed 2416 CeedInt max_rstr_buffer_size = 1; 2417 2418 for (CeedInt i = 0; i < num_input_fields; i++) { 2419 CeedEvalMode eval_mode; 2420 2421 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[i], &eval_mode)); 2422 if (eval_mode != CEED_EVAL_NONE && eval_mode != CEED_EVAL_WEIGHT) { 2423 CeedInt num_comp; 2424 CeedElemRestriction elem_rstr; 2425 2426 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_input_fields[i], &elem_rstr)); 2427 CeedCallBackend(CeedElemRestrictionGetNumComponents(elem_rstr, &num_comp)); 2428 max_rstr_buffer_size = CeedIntMax(max_rstr_buffer_size, num_comp * (is_all_tensor && (max_dim >= 3) ? Q_1d : 1)); 2429 CeedCallBackend(CeedElemRestrictionDestroy(&elem_rstr)); 2430 } 2431 } 2432 for (CeedInt i = 0; i < num_output_fields; i++) { 2433 CeedEvalMode eval_mode; 2434 2435 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_output_fields[i], &eval_mode)); 2436 if (eval_mode != CEED_EVAL_NONE) { 2437 CeedInt num_comp; 2438 CeedElemRestriction elem_rstr; 2439 2440 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_output_fields[i], &elem_rstr)); 2441 CeedCallBackend(CeedElemRestrictionGetNumComponents(elem_rstr, &num_comp)); 2442 max_rstr_buffer_size = CeedIntMax(max_rstr_buffer_size, num_comp * (is_all_tensor && (max_dim >= 3) ? Q_1d : 1)); 2443 CeedCallBackend(CeedElemRestrictionDestroy(&elem_rstr)); 2444 } 2445 } 2446 code << tab << "// Scratch restriction buffer space\n"; 2447 code << tab << "CeedScalar r_e_scratch[" << max_rstr_buffer_size << "];\n"; 2448 2449 // -- Determine best input field processing order 2450 CeedInt field_rstr_in_buffer[CEED_FIELD_MAX], input_field_order[CEED_FIELD_MAX]; 2451 2452 for (CeedInt i = 0; i < num_input_fields; i++) { 2453 field_rstr_in_buffer[i] = -1; 2454 input_field_order[i] = -1; 2455 } 2456 { 2457 bool is_ordered[CEED_FIELD_MAX]; 2458 CeedInt curr_index = 0; 2459 2460 for (CeedInt i = 0; i < num_input_fields; i++) is_ordered[i] = false; 2461 for (CeedInt i = 0; i < num_input_fields; i++) { 2462 CeedVector vec_i; 2463 CeedElemRestriction rstr_i; 2464 2465 if (is_ordered[i]) continue; 2466 field_rstr_in_buffer[i] = i; 2467 is_ordered[i] = true; 2468 input_field_order[curr_index] = i; 2469 curr_index++; 2470 CeedCallBackend(CeedOperatorFieldGetVector(op_input_fields[i], &vec_i)); 2471 if (vec_i == CEED_VECTOR_NONE) continue; // CEED_EVAL_WEIGHT 2472 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_input_fields[i], &rstr_i)); 2473 for (CeedInt j = i + 1; j < num_input_fields; j++) { 2474 CeedVector vec_j; 2475 CeedElemRestriction rstr_j; 2476 2477 CeedCallBackend(CeedOperatorFieldGetVector(op_input_fields[j], &vec_j)); 2478 CeedCallBackend(CeedOperatorFieldGetElemRestriction(op_input_fields[j], &rstr_j)); 2479 if (rstr_i == rstr_j && vec_i == vec_j) { 2480 field_rstr_in_buffer[j] = i; 2481 is_ordered[j] = true; 2482 input_field_order[curr_index] = j; 2483 curr_index++; 2484 } 2485 CeedCallBackend(CeedVectorDestroy(&vec_j)); 2486 CeedCallBackend(CeedElemRestrictionDestroy(&rstr_j)); 2487 } 2488 CeedCallBackend(CeedVectorDestroy(&vec_i)); 2489 CeedCallBackend(CeedElemRestrictionDestroy(&rstr_i)); 2490 } 2491 } 2492 2493 // -- Input restriction and basis 2494 code << "\n" << tab << "// -- Input field restrictions and basis actions\n"; 2495 CeedInt num_active_in = 0, num_active_out = 0, qf_assembly_size_out = 0; 2496 CeedInt active_fields_in[CEED_FIELD_MAX], active_fields_out[CEED_FIELD_MAX]; 2497 2498 for (CeedInt i = 0; i < num_input_fields; i++) { 2499 bool is_active = false; 2500 const char *field_name; 2501 const CeedInt f = input_field_order[i]; 2502 2503 { 2504 CeedVector vec; 2505 2506 CeedCallBackend(CeedOperatorFieldGetVector(op_input_fields[f], &vec)); 2507 is_active = vec == CEED_VECTOR_ACTIVE; 2508 CeedCallBackend(CeedVectorDestroy(&vec)); 2509 } 2510 2511 CeedCallBackend(CeedOperatorFieldGetName(op_input_fields[f], &field_name)); 2512 code << tab << "// ---- Input field " << f << ": " << field_name << "\n"; 2513 2514 if (is_active) { 2515 CeedEvalMode eval_mode; 2516 CeedInt field_size; 2517 2518 active_fields_in[num_active_in] = f; 2519 num_active_in++; 2520 CeedCallBackend(CeedQFunctionFieldGetSize(qf_input_fields[f], &field_size)); 2521 CeedCallBackend(CeedQFunctionFieldGetEvalMode(qf_input_fields[f], &eval_mode)); 2522 if (eval_mode == CEED_EVAL_GRAD) { 2523 code << tab << "CeedScalar r_q_in_" << f << "[num_comp_in_" << f << "*" << "dim_in_" << f << "*" 2524 << (is_all_tensor && (max_dim >= 3) ? "Q_1d" : "1") << "] = {0.};\n"; 2525 } else { 2526 code << tab << "CeedScalar r_q_in_" << f << "[num_comp_in_" << f << "*" << (is_all_tensor && (max_dim >= 3) ? "Q_1d" : "1") << "] = {0.};\n"; 2527 } 2528 code << tab << "const CeedInt field_size_in_" << f << " = " << field_size << ";\n"; 2529 } else { 2530 // ---- Restriction 2531 CeedCallBackend(CeedOperatorBuildKernelRestriction_Cuda_gen(code, data, tab, f, field_rstr_in_buffer, op_input_fields[f], qf_input_fields[f], 2532 max_dim, Q_1d, true, is_all_tensor, is_at_points, use_3d_slices)); 2533 2534 // ---- Basis action 2535 CeedCallBackend(CeedOperatorBuildKernelBasis_Cuda_gen(code, data, tab, f, op_input_fields[f], qf_input_fields[f], max_dim, Q_1d, true, 2536 is_all_tensor, is_at_points, use_3d_slices)); 2537 } 2538 } 2539 code << tab << "const CeedInt field_sizes_in[" << num_active_in << "] = {"; 2540 for (CeedInt i = 0; i < num_active_in; i++) { 2541 code << "field_size_in_" << active_fields_in[i] << (i < num_active_in - 1 ? ", " : ""); 2542 } 2543 code << "};\n"; 2544 code << tab << "CeedScalar * r_q_in[" << num_active_in << "] = {"; 2545 for (CeedInt i = 0; i < num_active_in; i++) { 2546 code << "r_q_in_" << active_fields_in[i] << (i < num_active_in - 1 ? ", " : ""); 2547 } 2548 code << "};\n"; 2549 2550 for (CeedInt i = 0; i < num_output_fields; i++) { 2551 bool is_active = false; 2552 2553 { 2554 CeedVector vec; 2555 2556 CeedCallBackend(CeedOperatorFieldGetVector(op_output_fields[i], &vec)); 2557 is_active = vec == CEED_VECTOR_ACTIVE; 2558 CeedCallBackend(CeedVectorDestroy(&vec)); 2559 } 2560 if (is_active) { 2561 const char *field_name; 2562 CeedInt field_size; 2563 2564 active_fields_out[num_active_out] = i; 2565 num_active_out++; 2566 CeedCallBackend(CeedQFunctionFieldGetSize(qf_output_fields[i], &field_size)); 2567 qf_assembly_size_out += field_size; 2568 CeedCallBackend(CeedOperatorFieldGetName(op_output_fields[i], &field_name)); 2569 code << tab << "// ---- Output field " << i << ": " << field_name << "\n"; 2570 code << tab << "const CeedInt field_size_out_" << i << " = " << field_size << ";\n"; 2571 } 2572 } 2573 code << tab << "const CeedInt field_sizes_out[" << num_active_out << "] = {"; 2574 for (CeedInt i = 0; i < num_active_out; i++) { 2575 code << "field_size_out_" << active_fields_out[i] << (i < num_active_out - 1 ? ", " : ""); 2576 } 2577 code << "};\n"; 2578 code << tab << "const CeedInt total_size_out = " << qf_assembly_size_out << ";\n"; 2579 2580 // -- Loop over active field 2581 code << "\n" << tab << "CeedInt input_offset = 0;\n"; 2582 code << tab << "// Loop over active QFunction input fields\n"; 2583 code << tab << "const CeedInt num_active_in = " << num_active_in << ";\n"; 2584 code << tab << "for (CeedInt a = 0; a < num_active_in; a++) {\n"; 2585 tab.push(); 2586 2587 // -- Loop over size of active field 2588 code << "\n" << tab << "// Loop over current active input field size\n"; 2589 code << tab << "const CeedInt field_size_in = field_sizes_in[a];\n"; 2590 code << tab << "for (CeedInt s = 0; s < field_size_in; s++) {\n"; 2591 tab.push(); 2592 2593 // -- Set current active point and component to 1 2594 code << tab << "// Set current active point and component to 1.0\n"; 2595 if (is_all_tensor && (max_dim >= 3)) { 2596 code << tab << "for (CeedInt i = 0; i < Q_1d; i++) r_q_in[a][i + s * Q_1d] = 1.0;\n"; 2597 } else { 2598 code << tab << "r_q_in[a][s] = 1.0;\n"; 2599 } 2600 2601 // -- Q function 2602 CeedCallBackend(CeedOperatorBuildKernelQFunction_Cuda_gen(code, data, tab, max_dim, max_num_points, num_input_fields, op_input_fields, 2603 qf_input_fields, num_output_fields, op_output_fields, qf_output_fields, qfunction_name, 2604 Q_1d, is_all_tensor, is_at_points, use_3d_slices)); 2605 2606 // -- Output basis and restriction 2607 code << "\n" << tab << "// -- Output field basis action and restrictions\n"; 2608 CeedScalar offset = 0; 2609 2610 for (CeedInt i = 0; i < num_output_fields; i++) { 2611 bool is_active = false; 2612 const char *field_name; 2613 2614 { 2615 CeedVector vec; 2616 2617 CeedCallBackend(CeedOperatorFieldGetVector(op_output_fields[i], &vec)); 2618 is_active = vec == CEED_VECTOR_ACTIVE; 2619 CeedCallBackend(CeedVectorDestroy(&vec)); 2620 } 2621 if (!is_active) continue; 2622 2623 CeedCallBackend(CeedOperatorFieldGetName(op_output_fields[i], &field_name)); 2624 code << tab << "// ---- Output field " << i << ": " << field_name << "\n"; 2625 2626 // ---- Restriction 2627 CeedInt field_size; 2628 2629 code << tab << "WriteLVecStandard" << (is_all_tensor ? max_dim : 1) << "d_QFAssembly<total_size_out, field_size_out_" << i << ", " 2630 << (is_all_tensor ? "Q_1d" : "Q") << ">(data, num_elem, elem, input_offset + s, " << offset << ", r_q_out_" << i << ", values_array);\n"; 2631 CeedCallBackend(CeedQFunctionFieldGetSize(qf_output_fields[i], &field_size)); 2632 offset += field_size; 2633 } 2634 2635 // -- Reset current active node and component 2636 code << "\n" << tab << "// Reset current active node and component to 0.0\n"; 2637 if (is_all_tensor && (max_dim >= 3)) { 2638 code << tab << "for (CeedInt i = 0; i < Q_1d; i++) r_q_in[a][i + s * Q_1d] = 0.0;\n"; 2639 } else { 2640 code << tab << "r_q_in[a][s] = 0.0;\n"; 2641 } 2642 2643 // -- End of loop over size of active field 2644 tab.pop(); 2645 code << tab << "}\n"; 2646 code << tab << "input_offset += field_size_in;\n"; 2647 2648 // -- End of loop over active field 2649 tab.pop(); 2650 code << tab << "}\n"; 2651 2652 // Close loop and function 2653 tab.pop(); 2654 code << tab << "}\n"; 2655 tab.pop(); 2656 code << tab << "}\n"; 2657 code << tab << "// -----------------------------------------------------------------------------\n\n"; 2658 2659 // Compile 2660 { 2661 bool is_compile_good = false; 2662 const CeedInt T_1d = CeedIntMax(is_all_tensor ? Q_1d : Q, data->max_P_1d); 2663 2664 data->thread_1d = T_1d; 2665 CeedCallBackend(CeedTryCompile_Cuda(ceed, code.str().c_str(), &is_compile_good, &data->module_assemble_qfunction, 1, "OP_T_1D", T_1d)); 2666 if (is_compile_good) { 2667 *is_good_build = true; 2668 CeedCallBackend(CeedGetKernel_Cuda(ceed, data->module_assemble_qfunction, operator_name.c_str(), &data->assemble_qfunction)); 2669 } else { 2670 *is_good_build = false; 2671 data->use_assembly_fallback = true; 2672 } 2673 } 2674 CeedCallBackend(CeedDestroy(&ceed)); 2675 CeedCallBackend(CeedQFunctionDestroy(&qf)); 2676 return CEED_ERROR_SUCCESS; 2677 } 2678 2679 //------------------------------------------------------------------------------ 2680