1 // Copyright (c) 2017-2024, 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 #include <ceed.h> 9 #include <ceed/backend.h> 10 #include <ceed/jit-tools.h> 11 #include <stdbool.h> 12 #include <stddef.h> 13 #include <string.h> 14 #include <hip/hip_runtime.h> 15 16 #include "../hip/ceed-hip-common.h" 17 #include "../hip/ceed-hip-compile.h" 18 #include "ceed-hip-shared.h" 19 20 //------------------------------------------------------------------------------ 21 // Compute a block size based on required minimum threads 22 //------------------------------------------------------------------------------ 23 static CeedInt ComputeBlockSizeFromRequirement(const CeedInt required) { 24 CeedInt maxSize = 1024; // Max total threads per block 25 CeedInt currentSize = 64; // Start with one group 26 27 while (currentSize < maxSize) { 28 if (currentSize > required) break; 29 else currentSize = currentSize * 2; 30 } 31 return currentSize; 32 } 33 34 //------------------------------------------------------------------------------ 35 // Compute required thread block sizes for basis kernels given P, Q, dim, and 36 // num_comp (num_comp not currently used, but may be again in other basis 37 // parallelization options) 38 //------------------------------------------------------------------------------ 39 static int ComputeBasisThreadBlockSizes(const CeedInt dim, const CeedInt P_1d, const CeedInt Q_1d, const CeedInt num_comp, CeedInt *block_sizes) { 40 // Note that this will use the same block sizes for all dimensions when compiling, 41 // but as each basis object is defined for a particular dimension, we will never 42 // call any kernels except the ones for the dimension for which we have computed the 43 // block sizes. 44 const CeedInt thread_1d = CeedIntMax(P_1d, Q_1d); 45 46 switch (dim) { 47 case 1: { 48 // Interp kernels: 49 block_sizes[0] = 256; 50 51 // Grad kernels: 52 block_sizes[1] = 256; 53 54 // Weight kernels: 55 block_sizes[2] = 256; 56 } break; 57 case 2: { 58 // Interp kernels: 59 CeedInt required = thread_1d * thread_1d; 60 61 block_sizes[0] = CeedIntMax(256, ComputeBlockSizeFromRequirement(required)); 62 63 // Grad kernels: currently use same required minimum threads 64 block_sizes[1] = CeedIntMax(256, ComputeBlockSizeFromRequirement(required)); 65 66 // Weight kernels: 67 required = CeedIntMax(64, Q_1d * Q_1d); 68 block_sizes[2] = CeedIntMax(256, ComputeBlockSizeFromRequirement(required)); 69 70 } break; 71 case 3: { 72 // Interp kernels: 73 CeedInt required = thread_1d * thread_1d; 74 75 block_sizes[0] = CeedIntMax(256, ComputeBlockSizeFromRequirement(required)); 76 77 // Grad kernels: currently use same required minimum threads 78 block_sizes[1] = CeedIntMax(256, ComputeBlockSizeFromRequirement(required)); 79 80 // Weight kernels: 81 required = Q_1d * Q_1d * Q_1d; 82 block_sizes[2] = CeedIntMax(256, ComputeBlockSizeFromRequirement(required)); 83 } 84 } 85 return CEED_ERROR_SUCCESS; 86 } 87 88 //------------------------------------------------------------------------------ 89 // Apply basis 90 //------------------------------------------------------------------------------ 91 static int CeedBasisApplyTensorCore_Hip_shared(CeedBasis basis, bool apply_add, const CeedInt num_elem, CeedTransposeMode t_mode, 92 CeedEvalMode eval_mode, CeedVector u, CeedVector v) { 93 Ceed ceed; 94 Ceed_Hip *ceed_Hip; 95 CeedInt dim, num_comp; 96 const CeedScalar *d_u; 97 CeedScalar *d_v; 98 CeedBasis_Hip_shared *data; 99 100 CeedCallBackend(CeedBasisGetCeed(basis, &ceed)); 101 CeedCallBackend(CeedGetData(ceed, &ceed_Hip)); 102 CeedCallBackend(CeedBasisGetData(basis, &data)); 103 CeedCallBackend(CeedBasisGetDimension(basis, &dim)); 104 CeedCallBackend(CeedBasisGetNumComponents(basis, &num_comp)); 105 106 // Get read/write access to u, v 107 if (u != CEED_VECTOR_NONE) CeedCallBackend(CeedVectorGetArrayRead(u, CEED_MEM_DEVICE, &d_u)); 108 else CeedCheck(eval_mode == CEED_EVAL_WEIGHT, ceed, CEED_ERROR_BACKEND, "An input vector is required for this CeedEvalMode"); 109 if (apply_add) CeedCallBackend(CeedVectorGetArray(v, CEED_MEM_DEVICE, &d_v)); 110 else CeedCallBackend(CeedVectorGetArrayWrite(v, CEED_MEM_DEVICE, &d_v)); 111 112 // Apply basis operation 113 switch (eval_mode) { 114 case CEED_EVAL_INTERP: { 115 CeedInt P_1d, Q_1d; 116 CeedInt block_size = data->block_sizes[0]; 117 118 CeedCallBackend(CeedBasisGetNumNodes1D(basis, &P_1d)); 119 CeedCallBackend(CeedBasisGetNumQuadraturePoints1D(basis, &Q_1d)); 120 CeedInt thread_1d = CeedIntMax(Q_1d, P_1d); 121 void *interp_args[] = {(void *)&num_elem, &data->d_interp_1d, &d_u, &d_v}; 122 123 if (dim == 1) { 124 CeedInt elems_per_block = 64 * thread_1d > 256 ? 256 / thread_1d : 64; 125 elems_per_block = elems_per_block > 0 ? elems_per_block : 1; 126 CeedInt grid = num_elem / elems_per_block + ((num_elem / elems_per_block * elems_per_block < num_elem) ? 1 : 0); 127 CeedInt shared_mem = elems_per_block * thread_1d * sizeof(CeedScalar); 128 129 if (t_mode == CEED_TRANSPOSE) { 130 CeedCallBackend(CeedRunKernelDimShared_Hip(ceed, apply_add ? data->InterpTransposeAdd : data->InterpTranspose, grid, thread_1d, 1, 131 elems_per_block, shared_mem, interp_args)); 132 } else { 133 CeedCallBackend(CeedRunKernelDimShared_Hip(ceed, data->Interp, grid, thread_1d, 1, elems_per_block, shared_mem, interp_args)); 134 } 135 } else if (dim == 2) { 136 // Check if required threads is small enough to do multiple elems 137 const CeedInt elems_per_block = CeedIntMax(block_size / (thread_1d * thread_1d), 1); 138 CeedInt grid = num_elem / elems_per_block + ((num_elem / elems_per_block * elems_per_block < num_elem) ? 1 : 0); 139 CeedInt shared_mem = elems_per_block * thread_1d * thread_1d * sizeof(CeedScalar); 140 141 if (t_mode == CEED_TRANSPOSE) { 142 CeedCallBackend(CeedRunKernelDimShared_Hip(ceed, apply_add ? data->InterpTransposeAdd : data->InterpTranspose, grid, thread_1d, thread_1d, 143 elems_per_block, shared_mem, interp_args)); 144 } else { 145 CeedCallBackend(CeedRunKernelDimShared_Hip(ceed, data->Interp, grid, thread_1d, thread_1d, elems_per_block, shared_mem, interp_args)); 146 } 147 } else if (dim == 3) { 148 const CeedInt elems_per_block = CeedIntMax(block_size / (thread_1d * thread_1d), 1); 149 CeedInt grid = num_elem / elems_per_block + ((num_elem / elems_per_block * elems_per_block < num_elem) ? 1 : 0); 150 CeedInt shared_mem = elems_per_block * thread_1d * thread_1d * sizeof(CeedScalar); 151 152 if (t_mode == CEED_TRANSPOSE) { 153 CeedCallBackend(CeedRunKernelDimShared_Hip(ceed, apply_add ? data->InterpTransposeAdd : data->InterpTranspose, grid, thread_1d, thread_1d, 154 elems_per_block, shared_mem, interp_args)); 155 } else { 156 CeedCallBackend(CeedRunKernelDimShared_Hip(ceed, data->Interp, grid, thread_1d, thread_1d, elems_per_block, shared_mem, interp_args)); 157 } 158 } 159 } break; 160 case CEED_EVAL_GRAD: { 161 CeedInt P_1d, Q_1d; 162 CeedInt block_size = data->block_sizes[1]; 163 164 CeedCallBackend(CeedBasisGetNumNodes1D(basis, &P_1d)); 165 CeedCallBackend(CeedBasisGetNumQuadraturePoints1D(basis, &Q_1d)); 166 CeedInt thread_1d = CeedIntMax(Q_1d, P_1d); 167 CeedScalar *d_grad_1d = data->d_grad_1d; 168 169 if (data->d_collo_grad_1d) { 170 d_grad_1d = data->d_collo_grad_1d; 171 } 172 void *grad_args[] = {(void *)&num_elem, &data->d_interp_1d, &d_grad_1d, &d_u, &d_v}; 173 if (dim == 1) { 174 CeedInt elems_per_block = 64 * thread_1d > 256 ? 256 / thread_1d : 64; 175 elems_per_block = elems_per_block > 0 ? elems_per_block : 1; 176 CeedInt grid = num_elem / elems_per_block + ((num_elem / elems_per_block * elems_per_block < num_elem) ? 1 : 0); 177 CeedInt shared_mem = elems_per_block * thread_1d * sizeof(CeedScalar); 178 179 if (t_mode == CEED_TRANSPOSE) { 180 CeedCallBackend(CeedRunKernelDimShared_Hip(ceed, apply_add ? data->GradTransposeAdd : data->GradTranspose, grid, thread_1d, 1, 181 elems_per_block, shared_mem, grad_args)); 182 } else { 183 CeedCallBackend(CeedRunKernelDimShared_Hip(ceed, data->Grad, grid, thread_1d, 1, elems_per_block, shared_mem, grad_args)); 184 } 185 } else if (dim == 2) { 186 // Check if required threads is small enough to do multiple elems 187 const CeedInt elems_per_block = CeedIntMax(block_size / (thread_1d * thread_1d), 1); 188 CeedInt grid = num_elem / elems_per_block + ((num_elem / elems_per_block * elems_per_block < num_elem) ? 1 : 0); 189 CeedInt shared_mem = elems_per_block * thread_1d * thread_1d * sizeof(CeedScalar); 190 191 if (t_mode == CEED_TRANSPOSE) { 192 CeedCallBackend(CeedRunKernelDimShared_Hip(ceed, apply_add ? data->GradTransposeAdd : data->GradTranspose, grid, thread_1d, thread_1d, 193 elems_per_block, shared_mem, grad_args)); 194 } else { 195 CeedCallBackend(CeedRunKernelDimShared_Hip(ceed, data->Grad, grid, thread_1d, thread_1d, elems_per_block, shared_mem, grad_args)); 196 } 197 } else if (dim == 3) { 198 const CeedInt elems_per_block = CeedIntMax(block_size / (thread_1d * thread_1d), 1); 199 CeedInt grid = num_elem / elems_per_block + ((num_elem / elems_per_block * elems_per_block < num_elem) ? 1 : 0); 200 CeedInt shared_mem = elems_per_block * thread_1d * thread_1d * sizeof(CeedScalar); 201 202 if (t_mode == CEED_TRANSPOSE) { 203 CeedCallBackend(CeedRunKernelDimShared_Hip(ceed, apply_add ? data->GradTransposeAdd : data->GradTranspose, grid, thread_1d, thread_1d, 204 elems_per_block, shared_mem, grad_args)); 205 } else { 206 CeedCallBackend(CeedRunKernelDimShared_Hip(ceed, data->Grad, grid, thread_1d, thread_1d, elems_per_block, shared_mem, grad_args)); 207 } 208 } 209 } break; 210 case CEED_EVAL_WEIGHT: { 211 CeedInt Q_1d; 212 CeedInt block_size = data->block_sizes[2]; 213 214 CeedCheck(data->d_q_weight_1d, ceed, CEED_ERROR_BACKEND, "%s not supported; q_weights_1d not set", CeedEvalModes[eval_mode]); 215 CeedCallBackend(CeedBasisGetNumQuadraturePoints1D(basis, &Q_1d)); 216 void *weight_args[] = {(void *)&num_elem, (void *)&data->d_q_weight_1d, &d_v}; 217 218 if (dim == 1) { 219 const CeedInt opt_elems = block_size / Q_1d; 220 const CeedInt elems_per_block = opt_elems > 0 ? opt_elems : 1; 221 const CeedInt grid_size = num_elem / elems_per_block + ((num_elem / elems_per_block * elems_per_block < num_elem) ? 1 : 0); 222 223 CeedCallBackend(CeedRunKernelDim_Hip(ceed, data->Weight, grid_size, Q_1d, elems_per_block, 1, weight_args)); 224 } else if (dim == 2) { 225 const CeedInt opt_elems = block_size / (Q_1d * Q_1d); 226 const CeedInt elems_per_block = opt_elems > 0 ? opt_elems : 1; 227 const CeedInt grid_size = num_elem / elems_per_block + ((num_elem / elems_per_block * elems_per_block < num_elem) ? 1 : 0); 228 229 CeedCallBackend(CeedRunKernelDim_Hip(ceed, data->Weight, grid_size, Q_1d, Q_1d, elems_per_block, weight_args)); 230 } else if (dim == 3) { 231 const CeedInt opt_elems = block_size / (Q_1d * Q_1d); 232 const CeedInt elems_per_block = opt_elems > 0 ? opt_elems : 1; 233 const CeedInt grid_size = num_elem / elems_per_block + ((num_elem / elems_per_block * elems_per_block < num_elem) ? 1 : 0); 234 235 CeedCallBackend(CeedRunKernelDim_Hip(ceed, data->Weight, grid_size, Q_1d, Q_1d, elems_per_block, weight_args)); 236 } 237 } break; 238 case CEED_EVAL_NONE: /* handled separately below */ 239 break; 240 // LCOV_EXCL_START 241 case CEED_EVAL_DIV: 242 case CEED_EVAL_CURL: 243 return CeedError(ceed, CEED_ERROR_BACKEND, "%s not supported", CeedEvalModes[eval_mode]); 244 // LCOV_EXCL_STOP 245 } 246 247 // Restore vectors, cover CEED_EVAL_NONE 248 CeedCallBackend(CeedVectorRestoreArray(v, &d_v)); 249 if (eval_mode == CEED_EVAL_NONE) CeedCallBackend(CeedVectorSetArray(v, CEED_MEM_DEVICE, CEED_COPY_VALUES, (CeedScalar *)d_u)); 250 if (eval_mode != CEED_EVAL_WEIGHT) CeedCallBackend(CeedVectorRestoreArrayRead(u, &d_u)); 251 return CEED_ERROR_SUCCESS; 252 } 253 254 int CeedBasisApplyTensor_Hip_shared(CeedBasis basis, const CeedInt num_elem, CeedTransposeMode t_mode, CeedEvalMode eval_mode, CeedVector u, 255 CeedVector v) { 256 CeedCallBackend(CeedBasisApplyTensorCore_Hip_shared(basis, false, num_elem, t_mode, eval_mode, u, v)); 257 return CEED_ERROR_SUCCESS; 258 } 259 260 int CeedBasisApplyAddTensor_Hip_shared(CeedBasis basis, const CeedInt num_elem, CeedTransposeMode t_mode, CeedEvalMode eval_mode, CeedVector u, 261 CeedVector v) { 262 CeedCallBackend(CeedBasisApplyTensorCore_Hip_shared(basis, true, num_elem, t_mode, eval_mode, u, v)); 263 return CEED_ERROR_SUCCESS; 264 } 265 266 //------------------------------------------------------------------------------ 267 // Basis apply - tensor AtPoints 268 //------------------------------------------------------------------------------ 269 static int CeedBasisApplyAtPointsCore_Hip_shared(CeedBasis basis, bool apply_add, const CeedInt num_elem, const CeedInt *num_points, 270 CeedTransposeMode t_mode, CeedEvalMode eval_mode, CeedVector x_ref, CeedVector u, CeedVector v) { 271 Ceed ceed; 272 CeedInt Q_1d, dim, max_num_points = num_points[0]; 273 const CeedInt is_transpose = t_mode == CEED_TRANSPOSE; 274 const int max_block_size = 32; 275 const CeedScalar *d_x, *d_u; 276 CeedScalar *d_v; 277 CeedBasis_Hip_shared *data; 278 279 CeedCallBackend(CeedBasisGetCeed(basis, &ceed)); 280 CeedCallBackend(CeedBasisGetData(basis, &data)); 281 CeedCallBackend(CeedBasisGetNumQuadraturePoints1D(basis, &Q_1d)); 282 CeedCallBackend(CeedBasisGetDimension(basis, &dim)); 283 284 // Weight handled separately 285 if (eval_mode == CEED_EVAL_WEIGHT) { 286 CeedCallBackend(CeedVectorSetValue(v, 1.0)); 287 return CEED_ERROR_SUCCESS; 288 } 289 290 // Check padded to uniform number of points per elem 291 for (CeedInt i = 1; i < num_elem; i++) max_num_points = CeedIntMax(max_num_points, num_points[i]); 292 { 293 CeedInt num_comp, q_comp; 294 CeedSize len, len_required; 295 296 CeedCallBackend(CeedBasisGetNumComponents(basis, &num_comp)); 297 CeedCallBackend(CeedBasisGetNumQuadratureComponents(basis, eval_mode, &q_comp)); 298 CeedCallBackend(CeedVectorGetLength(is_transpose ? u : v, &len)); 299 len_required = (CeedSize)num_comp * (CeedSize)q_comp * (CeedSize)num_elem * (CeedSize)max_num_points; 300 CeedCheck(len >= len_required, ceed, CEED_ERROR_BACKEND, 301 "Vector at points must be padded to the same number of points in each element for BasisApplyAtPoints on GPU backends." 302 " Found %" CeedSize_FMT ", Required %" CeedSize_FMT, 303 len, len_required); 304 } 305 306 // Move num_points array to device 307 if (is_transpose) { 308 const CeedInt num_bytes = num_elem * sizeof(CeedInt); 309 310 if (num_elem != data->num_elem_at_points) { 311 data->num_elem_at_points = num_elem; 312 313 if (data->d_points_per_elem) CeedCallHip(ceed, hipFree(data->d_points_per_elem)); 314 CeedCallHip(ceed, hipMalloc((void **)&data->d_points_per_elem, num_bytes)); 315 CeedCallBackend(CeedFree(&data->h_points_per_elem)); 316 CeedCallBackend(CeedCalloc(num_elem, &data->h_points_per_elem)); 317 } 318 if (memcmp(data->h_points_per_elem, num_points, num_bytes)) { 319 memcpy(data->h_points_per_elem, num_points, num_bytes); 320 CeedCallHip(ceed, hipMemcpy(data->d_points_per_elem, num_points, num_bytes, hipMemcpyHostToDevice)); 321 } 322 } 323 324 // Build kernels if needed 325 if (data->num_points != max_num_points) { 326 CeedInt P_1d; 327 328 CeedCallBackend(CeedBasisGetNumNodes1D(basis, &P_1d)); 329 data->num_points = max_num_points; 330 331 // -- Create interp matrix to Chebyshev coefficients 332 if (!data->d_chebyshev_interp_1d) { 333 CeedSize interp_bytes; 334 CeedScalar *chebyshev_interp_1d; 335 336 interp_bytes = P_1d * Q_1d * sizeof(CeedScalar); 337 CeedCallBackend(CeedCalloc(P_1d * Q_1d, &chebyshev_interp_1d)); 338 CeedCallBackend(CeedBasisGetChebyshevInterp1D(basis, chebyshev_interp_1d)); 339 CeedCallHip(ceed, hipMalloc((void **)&data->d_chebyshev_interp_1d, interp_bytes)); 340 CeedCallHip(ceed, hipMemcpy(data->d_chebyshev_interp_1d, chebyshev_interp_1d, interp_bytes, hipMemcpyHostToDevice)); 341 CeedCallBackend(CeedFree(&chebyshev_interp_1d)); 342 } 343 344 // -- Compile kernels 345 const char basis_kernel_source[] = "// AtPoints basis source\n#include <ceed/jit-source/hip/hip-ref-basis-tensor-at-points.h>\n"; 346 CeedInt num_comp; 347 348 if (data->moduleAtPoints) CeedCallHip(ceed, hipModuleUnload(data->moduleAtPoints)); 349 CeedCallBackend(CeedBasisGetNumComponents(basis, &num_comp)); 350 CeedCallBackend(CeedCompile_Hip(ceed, basis_kernel_source, &data->moduleAtPoints, 9, "BASIS_Q_1D", Q_1d, "BASIS_P_1D", P_1d, "BASIS_BUF_LEN", 351 Q_1d * CeedIntPow(Q_1d > P_1d ? Q_1d : P_1d, dim - 1), "BASIS_DIM", dim, "BASIS_NUM_COMP", num_comp, 352 "BASIS_NUM_NODES", CeedIntPow(P_1d, dim), "BASIS_NUM_QPTS", CeedIntPow(Q_1d, dim), "BASIS_NUM_PTS", 353 max_num_points, "POINTS_BUFF_LEN", CeedIntPow(Q_1d, dim - 1))); 354 CeedCallBackend(CeedGetKernel_Hip(ceed, data->moduleAtPoints, "InterpAtPoints", &data->InterpAtPoints)); 355 CeedCallBackend(CeedGetKernel_Hip(ceed, data->moduleAtPoints, "GradAtPoints", &data->GradAtPoints)); 356 } 357 358 // Get read/write access to u, v 359 CeedCallBackend(CeedVectorGetArrayRead(x_ref, CEED_MEM_DEVICE, &d_x)); 360 if (u != CEED_VECTOR_NONE) CeedCallBackend(CeedVectorGetArrayRead(u, CEED_MEM_DEVICE, &d_u)); 361 else CeedCheck(eval_mode == CEED_EVAL_WEIGHT, ceed, CEED_ERROR_BACKEND, "An input vector is required for this CeedEvalMode"); 362 if (apply_add) CeedCallBackend(CeedVectorGetArray(v, CEED_MEM_DEVICE, &d_v)); 363 else CeedCallBackend(CeedVectorGetArrayWrite(v, CEED_MEM_DEVICE, &d_v)); 364 365 // Clear v for transpose operation 366 if (is_transpose && !apply_add) { 367 CeedInt num_comp, q_comp, num_nodes; 368 CeedSize length; 369 370 CeedCallBackend(CeedBasisGetNumComponents(basis, &num_comp)); 371 CeedCallBackend(CeedBasisGetNumQuadratureComponents(basis, eval_mode, &q_comp)); 372 CeedCallBackend(CeedBasisGetNumNodes(basis, &num_nodes)); 373 length = 374 (CeedSize)num_elem * (CeedSize)num_comp * (t_mode == CEED_TRANSPOSE ? (CeedSize)num_nodes : ((CeedSize)max_num_points * (CeedSize)q_comp)); 375 CeedCallHip(ceed, hipMemset(d_v, 0, length * sizeof(CeedScalar))); 376 } 377 378 // Basis action 379 switch (eval_mode) { 380 case CEED_EVAL_INTERP: { 381 void *interp_args[] = {(void *)&num_elem, (void *)&is_transpose, &data->d_chebyshev_interp_1d, &data->d_points_per_elem, &d_x, &d_u, &d_v}; 382 const CeedInt block_size = CeedIntMin(CeedIntPow(Q_1d, dim), max_block_size); 383 384 CeedCallBackend(CeedRunKernel_Hip(ceed, data->InterpAtPoints, num_elem, block_size, interp_args)); 385 } break; 386 case CEED_EVAL_GRAD: { 387 void *grad_args[] = {(void *)&num_elem, (void *)&is_transpose, &data->d_chebyshev_interp_1d, &data->d_points_per_elem, &d_x, &d_u, &d_v}; 388 const CeedInt block_size = CeedIntMin(CeedIntPow(Q_1d, dim), max_block_size); 389 390 CeedCallBackend(CeedRunKernel_Hip(ceed, data->GradAtPoints, num_elem, block_size, grad_args)); 391 } break; 392 case CEED_EVAL_WEIGHT: 393 case CEED_EVAL_NONE: /* handled separately below */ 394 break; 395 // LCOV_EXCL_START 396 case CEED_EVAL_DIV: 397 case CEED_EVAL_CURL: 398 return CeedError(ceed, CEED_ERROR_BACKEND, "%s not supported", CeedEvalModes[eval_mode]); 399 // LCOV_EXCL_STOP 400 } 401 402 // Restore vectors, cover CEED_EVAL_NONE 403 CeedCallBackend(CeedVectorRestoreArrayRead(x_ref, &d_x)); 404 CeedCallBackend(CeedVectorRestoreArray(v, &d_v)); 405 if (eval_mode == CEED_EVAL_NONE) CeedCallBackend(CeedVectorSetArray(v, CEED_MEM_DEVICE, CEED_COPY_VALUES, (CeedScalar *)d_u)); 406 if (eval_mode != CEED_EVAL_WEIGHT) CeedCallBackend(CeedVectorRestoreArrayRead(u, &d_u)); 407 return CEED_ERROR_SUCCESS; 408 } 409 410 static int CeedBasisApplyAtPoints_Hip_shared(CeedBasis basis, const CeedInt num_elem, const CeedInt *num_points, CeedTransposeMode t_mode, 411 CeedEvalMode eval_mode, CeedVector x_ref, CeedVector u, CeedVector v) { 412 CeedCallBackend(CeedBasisApplyAtPointsCore_Hip_shared(basis, false, num_elem, num_points, t_mode, eval_mode, x_ref, u, v)); 413 return CEED_ERROR_SUCCESS; 414 } 415 416 static int CeedBasisApplyAddAtPoints_Hip_shared(CeedBasis basis, const CeedInt num_elem, const CeedInt *num_points, CeedTransposeMode t_mode, 417 CeedEvalMode eval_mode, CeedVector x_ref, CeedVector u, CeedVector v) { 418 CeedCallBackend(CeedBasisApplyAtPointsCore_Hip_shared(basis, true, num_elem, num_points, t_mode, eval_mode, x_ref, u, v)); 419 return CEED_ERROR_SUCCESS; 420 } 421 422 //------------------------------------------------------------------------------ 423 // Destroy basis 424 //------------------------------------------------------------------------------ 425 static int CeedBasisDestroy_Hip_shared(CeedBasis basis) { 426 Ceed ceed; 427 CeedBasis_Hip_shared *data; 428 429 CeedCallBackend(CeedBasisGetCeed(basis, &ceed)); 430 CeedCallBackend(CeedBasisGetData(basis, &data)); 431 CeedCallHip(ceed, hipModuleUnload(data->module)); 432 if (data->moduleAtPoints) CeedCallHip(ceed, hipModuleUnload(data->moduleAtPoints)); 433 if (data->d_q_weight_1d) CeedCallHip(ceed, hipFree(data->d_q_weight_1d)); 434 CeedCallBackend(CeedFree(&data->h_points_per_elem)); 435 if (data->d_points_per_elem) CeedCallHip(ceed, hipFree(data->d_points_per_elem)); 436 CeedCallHip(ceed, hipFree(data->d_interp_1d)); 437 CeedCallHip(ceed, hipFree(data->d_grad_1d)); 438 CeedCallHip(ceed, hipFree(data->d_collo_grad_1d)); 439 CeedCallHip(ceed, hipFree(data->d_chebyshev_interp_1d)); 440 CeedCallBackend(CeedFree(&data)); 441 return CEED_ERROR_SUCCESS; 442 } 443 444 //------------------------------------------------------------------------------ 445 // Create tensor basis 446 //------------------------------------------------------------------------------ 447 int CeedBasisCreateTensorH1_Hip_shared(CeedInt dim, CeedInt P_1d, CeedInt Q_1d, const CeedScalar *interp_1d, const CeedScalar *grad_1d, 448 const CeedScalar *q_ref_1d, const CeedScalar *q_weight_1d, CeedBasis basis) { 449 Ceed ceed; 450 CeedInt num_comp; 451 const CeedInt q_bytes = Q_1d * sizeof(CeedScalar); 452 const CeedInt interp_bytes = q_bytes * P_1d; 453 CeedBasis_Hip_shared *data; 454 455 CeedCallBackend(CeedBasisGetCeed(basis, &ceed)); 456 CeedCallBackend(CeedCalloc(1, &data)); 457 458 // Copy basis data to GPU 459 if (q_weight_1d) { 460 CeedCallHip(ceed, hipMalloc((void **)&data->d_q_weight_1d, q_bytes)); 461 CeedCallHip(ceed, hipMemcpy(data->d_q_weight_1d, q_weight_1d, q_bytes, hipMemcpyHostToDevice)); 462 } 463 CeedCallHip(ceed, hipMalloc((void **)&data->d_interp_1d, interp_bytes)); 464 CeedCallHip(ceed, hipMemcpy(data->d_interp_1d, interp_1d, interp_bytes, hipMemcpyHostToDevice)); 465 CeedCallHip(ceed, hipMalloc((void **)&data->d_grad_1d, interp_bytes)); 466 CeedCallHip(ceed, hipMemcpy(data->d_grad_1d, grad_1d, interp_bytes, hipMemcpyHostToDevice)); 467 468 // Compute collocated gradient and copy to GPU 469 data->d_collo_grad_1d = NULL; 470 bool has_collocated_grad = dim == 3 && Q_1d >= P_1d; 471 472 if (has_collocated_grad) { 473 CeedScalar *collo_grad_1d; 474 475 CeedCallBackend(CeedMalloc(Q_1d * Q_1d, &collo_grad_1d)); 476 CeedCallBackend(CeedBasisGetCollocatedGrad(basis, collo_grad_1d)); 477 CeedCallHip(ceed, hipMalloc((void **)&data->d_collo_grad_1d, q_bytes * Q_1d)); 478 CeedCallHip(ceed, hipMemcpy(data->d_collo_grad_1d, collo_grad_1d, q_bytes * Q_1d, hipMemcpyHostToDevice)); 479 CeedCallBackend(CeedFree(&collo_grad_1d)); 480 } 481 482 // Set number of threads per block for basis kernels 483 CeedCallBackend(CeedBasisGetNumComponents(basis, &num_comp)); 484 CeedCallBackend(ComputeBasisThreadBlockSizes(dim, P_1d, Q_1d, num_comp, data->block_sizes)); 485 486 // Compile basis kernels 487 const char basis_kernel_source[] = "// Tensor basis source\n#include <ceed/jit-source/hip/hip-shared-basis-tensor.h>\n"; 488 489 CeedCallBackend(CeedCompile_Hip(ceed, basis_kernel_source, &data->module, 11, "BASIS_Q_1D", Q_1d, "BASIS_P_1D", P_1d, "T_1D", 490 CeedIntMax(Q_1d, P_1d), "BASIS_DIM", dim, "BASIS_NUM_COMP", num_comp, "BASIS_NUM_NODES", CeedIntPow(P_1d, dim), 491 "BASIS_NUM_QPTS", CeedIntPow(Q_1d, dim), "BASIS_INTERP_BLOCK_SIZE", data->block_sizes[0], "BASIS_GRAD_BLOCK_SIZE", 492 data->block_sizes[1], "BASIS_WEIGHT_BLOCK_SIZE", data->block_sizes[2], "BASIS_HAS_COLLOCATED_GRAD", 493 has_collocated_grad)); 494 CeedCallBackend(CeedGetKernel_Hip(ceed, data->module, "Interp", &data->Interp)); 495 CeedCallBackend(CeedGetKernel_Hip(ceed, data->module, "InterpTranspose", &data->InterpTranspose)); 496 CeedCallBackend(CeedGetKernel_Hip(ceed, data->module, "InterpTransposeAdd", &data->InterpTransposeAdd)); 497 CeedCallBackend(CeedGetKernel_Hip(ceed, data->module, "Grad", &data->Grad)); 498 CeedCallBackend(CeedGetKernel_Hip(ceed, data->module, "GradTranspose", &data->GradTranspose)); 499 CeedCallBackend(CeedGetKernel_Hip(ceed, data->module, "GradTransposeAdd", &data->GradTransposeAdd)); 500 CeedCallBackend(CeedGetKernel_Hip(ceed, data->module, "Weight", &data->Weight)); 501 502 CeedCallBackend(CeedBasisSetData(basis, data)); 503 504 // Register backend functions 505 CeedCallBackend(CeedSetBackendFunction(ceed, "Basis", basis, "Apply", CeedBasisApplyTensor_Hip_shared)); 506 CeedCallBackend(CeedSetBackendFunction(ceed, "Basis", basis, "ApplyAdd", CeedBasisApplyAddTensor_Hip_shared)); 507 CeedCallBackend(CeedSetBackendFunction(ceed, "Basis", basis, "ApplyAtPoints", CeedBasisApplyAtPoints_Hip_shared)); 508 CeedCallBackend(CeedSetBackendFunction(ceed, "Basis", basis, "ApplyAddAtPoints", CeedBasisApplyAddAtPoints_Hip_shared)); 509 CeedCallBackend(CeedSetBackendFunction(ceed, "Basis", basis, "Destroy", CeedBasisDestroy_Hip_shared)); 510 return CEED_ERROR_SUCCESS; 511 } 512 513 //------------------------------------------------------------------------------ 514