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 <stdbool.h> 11 #include <stdlib.h> 12 #include <string.h> 13 14 #include "ceed-ref.h" 15 16 //------------------------------------------------------------------------------ 17 // Core ElemRestriction Apply Code 18 //------------------------------------------------------------------------------ 19 static inline int CeedElemRestrictionApplyStridedNoTranspose_Ref_Core(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, 20 CeedInt start, CeedInt stop, CeedInt num_elem, CeedInt elem_size, 21 CeedSize v_offset, const CeedScalar *__restrict__ uu, 22 CeedScalar *__restrict__ vv) { 23 // No offsets provided, identity restriction 24 bool has_backend_strides; 25 26 CeedCallBackend(CeedElemRestrictionHasBackendStrides(rstr, &has_backend_strides)); 27 if (has_backend_strides) { 28 // CPU backend strides are {1, elem_size, elem_size*num_comp} 29 // This if branch is left separate to allow better inlining 30 for (CeedSize e = start * block_size; e < stop * block_size; e += block_size) { 31 CeedPragmaSIMD for (CeedSize k = 0; k < num_comp; k++) { 32 CeedPragmaSIMD for (CeedSize n = 0; n < elem_size; n++) { 33 CeedPragmaSIMD for (CeedSize j = 0; j < block_size; j++) { 34 vv[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] = 35 uu[n + k * elem_size + CeedIntMin(e + j, num_elem - 1) * elem_size * (CeedSize)num_comp]; 36 } 37 } 38 } 39 } 40 } else { 41 // User provided strides 42 CeedInt strides[3]; 43 44 CeedCallBackend(CeedElemRestrictionGetStrides(rstr, strides)); 45 for (CeedSize e = start * block_size; e < stop * block_size; e += block_size) { 46 CeedPragmaSIMD for (CeedSize k = 0; k < num_comp; k++) { 47 CeedPragmaSIMD for (CeedSize n = 0; n < elem_size; n++) { 48 CeedPragmaSIMD for (CeedSize j = 0; j < block_size; j++) { 49 vv[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] = 50 uu[n * strides[0] + k * strides[1] + CeedIntMin(e + j, num_elem - 1) * (CeedSize)strides[2]]; 51 } 52 } 53 } 54 } 55 } 56 return CEED_ERROR_SUCCESS; 57 } 58 59 static inline int CeedElemRestrictionApplyOffsetNoTranspose_Ref_Core(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, 60 const CeedInt comp_stride, CeedInt start, CeedInt stop, CeedInt num_elem, 61 CeedInt elem_size, CeedSize v_offset, const CeedScalar *__restrict__ uu, 62 CeedScalar *__restrict__ vv) { 63 // Default restriction with offsets 64 CeedElemRestriction_Ref *impl; 65 66 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 67 for (CeedSize e = start * block_size; e < stop * block_size; e += block_size) { 68 CeedPragmaSIMD for (CeedSize k = 0; k < num_comp; k++) { 69 CeedPragmaSIMD for (CeedSize i = 0; i < elem_size * block_size; i++) { 70 vv[elem_size * (k * block_size + e * num_comp) + i - v_offset] = uu[impl->offsets[i + e * elem_size] + k * comp_stride]; 71 } 72 } 73 } 74 return CEED_ERROR_SUCCESS; 75 } 76 77 static inline int CeedElemRestrictionApplyOrientedNoTranspose_Ref_Core(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, 78 const CeedInt comp_stride, CeedInt start, CeedInt stop, CeedInt num_elem, 79 CeedInt elem_size, CeedSize v_offset, const CeedScalar *__restrict__ uu, 80 CeedScalar *__restrict__ vv) { 81 // Restriction with orientations 82 CeedElemRestriction_Ref *impl; 83 84 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 85 for (CeedSize e = start * block_size; e < stop * block_size; e += block_size) { 86 CeedPragmaSIMD for (CeedSize k = 0; k < num_comp; k++) { 87 CeedPragmaSIMD for (CeedSize i = 0; i < elem_size * block_size; i++) { 88 vv[elem_size * (k * block_size + e * num_comp) + i - v_offset] = 89 uu[impl->offsets[i + e * elem_size] + k * comp_stride] * (impl->orients[i + e * elem_size] ? -1.0 : 1.0); 90 } 91 } 92 } 93 return CEED_ERROR_SUCCESS; 94 } 95 96 static inline int CeedElemRestrictionApplyCurlOrientedNoTranspose_Ref_Core(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, 97 const CeedInt comp_stride, CeedInt start, CeedInt stop, CeedInt num_elem, 98 CeedInt elem_size, CeedSize v_offset, const CeedScalar *__restrict__ uu, 99 CeedScalar *__restrict__ vv) { 100 // Restriction with tridiagonal transformation 101 CeedElemRestriction_Ref *impl; 102 103 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 104 for (CeedSize e = start * block_size; e < stop * block_size; e += block_size) { 105 CeedPragmaSIMD for (CeedSize k = 0; k < num_comp; k++) { 106 CeedSize n = 0; 107 108 CeedPragmaSIMD for (CeedSize j = 0; j < block_size; j++) { 109 vv[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] = 110 uu[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] * 111 impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size] + 112 uu[impl->offsets[j + (n + 1) * block_size + e * elem_size] + k * comp_stride] * 113 impl->curl_orients[j + (3 * n + 2) * block_size + e * 3 * elem_size]; 114 } 115 CeedPragmaSIMD for (n = 1; n < elem_size - 1; n++) { 116 CeedPragmaSIMD for (CeedSize j = 0; j < block_size; j++) { 117 vv[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] = 118 uu[impl->offsets[j + (n - 1) * block_size + e * elem_size] + k * comp_stride] * 119 impl->curl_orients[j + (3 * n + 0) * block_size + e * 3 * elem_size] + 120 uu[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] * 121 impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size] + 122 uu[impl->offsets[j + (n + 1) * block_size + e * elem_size] + k * comp_stride] * 123 impl->curl_orients[j + (3 * n + 2) * block_size + e * 3 * elem_size]; 124 } 125 } 126 CeedPragmaSIMD for (CeedSize j = 0; j < block_size; j++) { 127 vv[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] = 128 uu[impl->offsets[j + (n - 1) * block_size + e * elem_size] + k * comp_stride] * 129 impl->curl_orients[j + (3 * n + 0) * block_size + e * 3 * elem_size] + 130 uu[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] * 131 impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size]; 132 } 133 } 134 } 135 return CEED_ERROR_SUCCESS; 136 } 137 138 static inline int CeedElemRestrictionApplyCurlOrientedUnsignedNoTranspose_Ref_Core(CeedElemRestriction rstr, const CeedInt num_comp, 139 const CeedInt block_size, const CeedInt comp_stride, CeedInt start, 140 CeedInt stop, CeedInt num_elem, CeedInt elem_size, 141 CeedSize v_offset, const CeedScalar *__restrict__ uu, 142 CeedScalar *__restrict__ vv) { 143 // Restriction with (unsigned) tridiagonal transformation 144 CeedElemRestriction_Ref *impl; 145 146 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 147 for (CeedSize e = start * block_size; e < stop * block_size; e += block_size) { 148 CeedPragmaSIMD for (CeedSize k = 0; k < num_comp; k++) { 149 CeedSize n = 0; 150 151 CeedPragmaSIMD for (CeedSize j = 0; j < block_size; j++) { 152 vv[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] = 153 uu[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] * 154 abs(impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size]) + 155 uu[impl->offsets[j + (n + 1) * block_size + e * elem_size] + k * comp_stride] * 156 abs(impl->curl_orients[j + (3 * n + 2) * block_size + e * 3 * elem_size]); 157 } 158 CeedPragmaSIMD for (n = 1; n < elem_size - 1; n++) { 159 CeedPragmaSIMD for (CeedSize j = 0; j < block_size; j++) { 160 vv[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] = 161 uu[impl->offsets[j + (n - 1) * block_size + e * elem_size] + k * comp_stride] * 162 abs(impl->curl_orients[j + (3 * n + 0) * block_size + e * 3 * elem_size]) + 163 uu[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] * 164 abs(impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size]) + 165 uu[impl->offsets[j + (n + 1) * block_size + e * elem_size] + k * comp_stride] * 166 abs(impl->curl_orients[j + (3 * n + 2) * block_size + e * 3 * elem_size]); 167 } 168 } 169 CeedPragmaSIMD for (CeedSize j = 0; j < block_size; j++) { 170 vv[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] = 171 uu[impl->offsets[j + (n - 1) * block_size + e * elem_size] + k * comp_stride] * 172 abs(impl->curl_orients[j + (3 * n + 0) * block_size + e * 3 * elem_size]) + 173 uu[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] * 174 abs(impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size]); 175 } 176 } 177 } 178 return CEED_ERROR_SUCCESS; 179 } 180 181 static inline int CeedElemRestrictionApplyStridedTranspose_Ref_Core(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, 182 CeedInt start, CeedInt stop, CeedInt num_elem, CeedInt elem_size, 183 CeedSize v_offset, const CeedScalar *__restrict__ uu, 184 CeedScalar *__restrict__ vv) { 185 // No offsets provided, identity restriction 186 bool has_backend_strides; 187 188 CeedCallBackend(CeedElemRestrictionHasBackendStrides(rstr, &has_backend_strides)); 189 if (has_backend_strides) { 190 // CPU backend strides are {1, elem_size, elem_size*num_comp} 191 // This if brach is left separate to allow better inlining 192 for (CeedSize e = start * block_size; e < stop * block_size; e += block_size) { 193 CeedPragmaSIMD for (CeedSize k = 0; k < num_comp; k++) { 194 CeedPragmaSIMD for (CeedSize n = 0; n < elem_size; n++) { 195 CeedPragmaSIMD for (CeedSize j = 0; j < CeedIntMin(block_size, num_elem - e); j++) { 196 vv[n + k * elem_size + (e + j) * elem_size * num_comp] += uu[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset]; 197 } 198 } 199 } 200 } 201 } else { 202 // User provided strides 203 CeedInt strides[3]; 204 205 CeedCallBackend(CeedElemRestrictionGetStrides(rstr, strides)); 206 for (CeedInt e = start * block_size; e < stop * block_size; e += block_size) { 207 CeedPragmaSIMD for (CeedSize k = 0; k < num_comp; k++) { 208 CeedPragmaSIMD for (CeedSize n = 0; n < elem_size; n++) { 209 CeedPragmaSIMD for (CeedSize j = 0; j < CeedIntMin(block_size, num_elem - e); j++) { 210 vv[n * strides[0] + k * strides[1] + (e + j) * strides[2]] += 211 uu[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset]; 212 } 213 } 214 } 215 } 216 } 217 return CEED_ERROR_SUCCESS; 218 } 219 220 static inline int CeedElemRestrictionApplyOffsetTranspose_Ref_Core(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, 221 const CeedInt comp_stride, CeedInt start, CeedInt stop, CeedInt num_elem, 222 CeedInt elem_size, CeedSize v_offset, const CeedScalar *__restrict__ uu, 223 CeedScalar *__restrict__ vv) { 224 // Default restriction with offsets 225 CeedElemRestriction_Ref *impl; 226 227 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 228 for (CeedSize e = start * block_size; e < stop * block_size; e += block_size) { 229 for (CeedSize k = 0; k < num_comp; k++) { 230 for (CeedSize i = 0; i < elem_size * block_size; i += block_size) { 231 // Iteration bound set to discard padding elements 232 for (CeedSize j = i; j < i + CeedIntMin(block_size, num_elem - e); j++) { 233 CeedScalar vv_loc; 234 235 vv_loc = uu[elem_size * (k * block_size + e * num_comp) + j - v_offset]; 236 CeedPragmaAtomic vv[impl->offsets[j + e * elem_size] + k * comp_stride] += vv_loc; 237 } 238 } 239 } 240 } 241 return CEED_ERROR_SUCCESS; 242 } 243 244 static inline int CeedElemRestrictionApplyOrientedTranspose_Ref_Core(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, 245 const CeedInt comp_stride, CeedInt start, CeedInt stop, CeedInt num_elem, 246 CeedInt elem_size, CeedSize v_offset, const CeedScalar *__restrict__ uu, 247 CeedScalar *__restrict__ vv) { 248 // Restriction with orientations 249 CeedElemRestriction_Ref *impl; 250 251 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 252 for (CeedSize e = start * block_size; e < stop * block_size; e += block_size) { 253 for (CeedSize k = 0; k < num_comp; k++) { 254 for (CeedSize i = 0; i < elem_size * block_size; i += block_size) { 255 // Iteration bound set to discard padding elements 256 for (CeedSize j = i; j < i + CeedIntMin(block_size, num_elem - e); j++) { 257 CeedScalar vv_loc; 258 259 vv_loc = uu[elem_size * (k * block_size + e * num_comp) + j - v_offset] * (impl->orients[j + e * elem_size] ? -1.0 : 1.0); 260 CeedPragmaAtomic vv[impl->offsets[j + e * elem_size] + k * comp_stride] += vv_loc; 261 } 262 } 263 } 264 } 265 return CEED_ERROR_SUCCESS; 266 } 267 268 static inline int CeedElemRestrictionApplyCurlOrientedTranspose_Ref_Core(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, 269 const CeedInt comp_stride, CeedInt start, CeedInt stop, CeedInt num_elem, 270 CeedInt elem_size, CeedSize v_offset, const CeedScalar *__restrict__ uu, 271 CeedScalar *__restrict__ vv) { 272 // Restriction with tridiagonal transformation 273 CeedElemRestriction_Ref *impl; 274 CeedScalar vv_loc[block_size]; 275 276 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 277 for (CeedSize e = start * block_size; e < stop * block_size; e += block_size) { 278 for (CeedSize k = 0; k < num_comp; k++) { 279 // Iteration bound set to discard padding elements 280 const CeedSize block_end = CeedIntMin(block_size, num_elem - e); 281 CeedSize n = 0; 282 283 CeedPragmaSIMD for (CeedSize j = 0; j < block_end; j++) { 284 vv_loc[j] = uu[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] * 285 impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size] + 286 uu[e * elem_size * num_comp + (k * elem_size + n + 1) * block_size + j - v_offset] * 287 impl->curl_orients[j + (3 * n + 3) * block_size + e * 3 * elem_size]; 288 } 289 for (CeedSize j = 0; j < block_end; j++) { 290 CeedPragmaAtomic vv[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] += vv_loc[j]; 291 } 292 for (n = 1; n < elem_size - 1; n++) { 293 CeedPragmaSIMD for (CeedInt j = 0; j < block_end; j++) { 294 vv_loc[j] = uu[e * elem_size * num_comp + (k * elem_size + n - 1) * block_size + j - v_offset] * 295 impl->curl_orients[j + (3 * n - 1) * block_size + e * 3 * elem_size] + 296 uu[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] * 297 impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size] + 298 uu[e * elem_size * num_comp + (k * elem_size + n + 1) * block_size + j - v_offset] * 299 impl->curl_orients[j + (3 * n + 3) * block_size + e * 3 * elem_size]; 300 } 301 for (CeedSize j = 0; j < block_end; j++) { 302 CeedPragmaAtomic vv[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] += vv_loc[j]; 303 } 304 } 305 CeedPragmaSIMD for (CeedSize j = 0; j < block_end; j++) { 306 vv_loc[j] = uu[e * elem_size * num_comp + (k * elem_size + n - 1) * block_size + j - v_offset] * 307 impl->curl_orients[j + (3 * n - 1) * block_size + e * 3 * elem_size] + 308 uu[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] * 309 impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size]; 310 } 311 for (CeedSize j = 0; j < block_end; j++) { 312 CeedPragmaAtomic vv[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] += vv_loc[j]; 313 } 314 } 315 } 316 return CEED_ERROR_SUCCESS; 317 } 318 319 static inline int CeedElemRestrictionApplyCurlOrientedUnsignedTranspose_Ref_Core(CeedElemRestriction rstr, const CeedInt num_comp, 320 const CeedInt block_size, const CeedInt comp_stride, CeedInt start, 321 CeedInt stop, CeedInt num_elem, CeedInt elem_size, CeedSize v_offset, 322 const CeedScalar *__restrict__ uu, CeedScalar *__restrict__ vv) { 323 // Restriction with (unsigned) tridiagonal transformation 324 CeedElemRestriction_Ref *impl; 325 CeedScalar vv_loc[block_size]; 326 327 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 328 for (CeedSize e = start * block_size; e < stop * block_size; e += block_size) { 329 for (CeedSize k = 0; k < num_comp; k++) { 330 // Iteration bound set to discard padding elements 331 const CeedSize block_end = CeedIntMin(block_size, num_elem - e); 332 CeedSize n = 0; 333 334 CeedPragmaSIMD for (CeedSize j = 0; j < block_end; j++) { 335 vv_loc[j] = uu[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] * 336 abs(impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size]) + 337 uu[e * elem_size * num_comp + (k * elem_size + n + 1) * block_size + j - v_offset] * 338 abs(impl->curl_orients[j + (3 * n + 3) * block_size + e * 3 * elem_size]); 339 } 340 for (CeedSize j = 0; j < block_end; j++) { 341 CeedPragmaAtomic vv[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] += vv_loc[j]; 342 } 343 for (n = 1; n < elem_size - 1; n++) { 344 CeedPragmaSIMD for (CeedSize j = 0; j < block_end; j++) { 345 vv_loc[j] = uu[e * elem_size * num_comp + (k * elem_size + n - 1) * block_size + j - v_offset] * 346 abs(impl->curl_orients[j + (3 * n - 1) * block_size + e * 3 * elem_size]) + 347 uu[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] * 348 abs(impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size]) + 349 uu[e * elem_size * num_comp + (k * elem_size + n + 1) * block_size + j - v_offset] * 350 abs(impl->curl_orients[j + (3 * n + 3) * block_size + e * 3 * elem_size]); 351 } 352 for (CeedSize j = 0; j < block_end; j++) { 353 CeedPragmaAtomic vv[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] += vv_loc[j]; 354 } 355 } 356 CeedPragmaSIMD for (CeedSize j = 0; j < block_end; j++) { 357 vv_loc[j] = uu[e * elem_size * num_comp + (k * elem_size + n - 1) * block_size + j - v_offset] * 358 abs(impl->curl_orients[j + (3 * n - 1) * block_size + e * 3 * elem_size]) + 359 uu[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] * 360 abs(impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size]); 361 } 362 for (CeedSize j = 0; j < block_end; j++) { 363 CeedPragmaAtomic vv[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] += vv_loc[j]; 364 } 365 } 366 } 367 return CEED_ERROR_SUCCESS; 368 } 369 370 static inline int CeedElemRestrictionApplyAtPointsInElement_Ref_Core(CeedElemRestriction rstr, const CeedInt num_comp, CeedInt start, CeedInt stop, 371 CeedTransposeMode t_mode, const CeedScalar *__restrict__ uu, 372 CeedScalar *__restrict__ vv) { 373 CeedInt num_points, l_vec_offset; 374 CeedSize e_vec_offset = 0; 375 CeedElemRestriction_Ref *impl; 376 377 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 378 for (CeedInt e = start; e < stop; e++) { 379 l_vec_offset = impl->offsets[e]; 380 CeedCallBackend(CeedElemRestrictionGetNumPointsInElement(rstr, e, &num_points)); 381 if (t_mode == CEED_NOTRANSPOSE) { 382 for (CeedSize i = 0; i < num_points; i++) { 383 for (CeedSize j = 0; j < num_comp; j++) vv[j * num_points + i + e_vec_offset] = uu[impl->offsets[i + l_vec_offset] * num_comp + j]; 384 } 385 } else { 386 for (CeedSize i = 0; i < num_points; i++) { 387 for (CeedSize j = 0; j < num_comp; j++) vv[impl->offsets[i + l_vec_offset] * num_comp + j] += uu[j * num_points + i + e_vec_offset]; 388 } 389 } 390 e_vec_offset += num_points * (CeedSize)num_comp; 391 } 392 return CEED_ERROR_SUCCESS; 393 } 394 395 static inline int CeedElemRestrictionApply_Ref_Core(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, 396 const CeedInt comp_stride, CeedInt start, CeedInt stop, CeedTransposeMode t_mode, bool use_signs, 397 bool use_orients, CeedVector u, CeedVector v, CeedRequest *request) { 398 CeedInt num_elem, elem_size; 399 CeedSize v_offset = 0; 400 CeedRestrictionType rstr_type; 401 const CeedScalar *uu; 402 CeedScalar *vv; 403 404 CeedCallBackend(CeedElemRestrictionGetNumElements(rstr, &num_elem)); 405 CeedCallBackend(CeedElemRestrictionGetElementSize(rstr, &elem_size)); 406 v_offset = start * block_size * elem_size * (CeedSize)num_comp; 407 CeedCallBackend(CeedElemRestrictionGetType(rstr, &rstr_type)); 408 CeedCallBackend(CeedVectorGetArrayRead(u, CEED_MEM_HOST, &uu)); 409 410 if (t_mode == CEED_TRANSPOSE) { 411 // Sum into for transpose mode, E-vector to L-vector 412 CeedCallBackend(CeedVectorGetArray(v, CEED_MEM_HOST, &vv)); 413 } else { 414 // Overwrite for notranspose mode, L-vector to E-vector 415 CeedCallBackend(CeedVectorGetArrayWrite(v, CEED_MEM_HOST, &vv)); 416 } 417 if (t_mode == CEED_TRANSPOSE) { 418 // Restriction from E-vector to L-vector 419 // Performing v += r^T * u 420 // uu has shape [elem_size, num_comp, num_elem], row-major 421 // vv has shape [nnodes, num_comp] 422 // Sum into for transpose mode 423 switch (rstr_type) { 424 case CEED_RESTRICTION_STRIDED: 425 CeedCallBackend( 426 CeedElemRestrictionApplyStridedTranspose_Ref_Core(rstr, num_comp, block_size, start, stop, num_elem, elem_size, v_offset, uu, vv)); 427 break; 428 case CEED_RESTRICTION_STANDARD: 429 CeedCallBackend(CeedElemRestrictionApplyOffsetTranspose_Ref_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, elem_size, 430 v_offset, uu, vv)); 431 break; 432 case CEED_RESTRICTION_ORIENTED: 433 if (use_signs) { 434 CeedCallBackend(CeedElemRestrictionApplyOrientedTranspose_Ref_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, 435 elem_size, v_offset, uu, vv)); 436 } else { 437 CeedCallBackend(CeedElemRestrictionApplyOffsetTranspose_Ref_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, elem_size, 438 v_offset, uu, vv)); 439 } 440 break; 441 case CEED_RESTRICTION_CURL_ORIENTED: 442 if (use_signs && use_orients) { 443 CeedCallBackend(CeedElemRestrictionApplyCurlOrientedTranspose_Ref_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, 444 elem_size, v_offset, uu, vv)); 445 } else if (use_orients) { 446 CeedCallBackend(CeedElemRestrictionApplyCurlOrientedUnsignedTranspose_Ref_Core(rstr, num_comp, block_size, comp_stride, start, stop, 447 num_elem, elem_size, v_offset, uu, vv)); 448 } else { 449 CeedCallBackend(CeedElemRestrictionApplyOffsetTranspose_Ref_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, elem_size, 450 v_offset, uu, vv)); 451 } 452 break; 453 case CEED_RESTRICTION_POINTS: 454 CeedCallBackend(CeedElemRestrictionApplyAtPointsInElement_Ref_Core(rstr, num_comp, start, stop, t_mode, uu, vv)); 455 break; 456 } 457 } else { 458 // Restriction from L-vector to E-vector 459 // Perform: v = r * u 460 // vv has shape [elem_size, num_comp, num_elem], row-major 461 // uu has shape [nnodes, num_comp] 462 // Overwrite for notranspose mode 463 switch (rstr_type) { 464 case CEED_RESTRICTION_STRIDED: 465 CeedCallBackend( 466 CeedElemRestrictionApplyStridedNoTranspose_Ref_Core(rstr, num_comp, block_size, start, stop, num_elem, elem_size, v_offset, uu, vv)); 467 break; 468 case CEED_RESTRICTION_STANDARD: 469 CeedCallBackend(CeedElemRestrictionApplyOffsetNoTranspose_Ref_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, elem_size, 470 v_offset, uu, vv)); 471 break; 472 case CEED_RESTRICTION_ORIENTED: 473 if (use_signs) { 474 CeedCallBackend(CeedElemRestrictionApplyOrientedNoTranspose_Ref_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, 475 elem_size, v_offset, uu, vv)); 476 } else { 477 CeedCallBackend(CeedElemRestrictionApplyOffsetNoTranspose_Ref_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, 478 elem_size, v_offset, uu, vv)); 479 } 480 break; 481 case CEED_RESTRICTION_CURL_ORIENTED: 482 if (use_signs && use_orients) { 483 CeedCallBackend(CeedElemRestrictionApplyCurlOrientedNoTranspose_Ref_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, 484 elem_size, v_offset, uu, vv)); 485 } else if (use_orients) { 486 CeedCallBackend(CeedElemRestrictionApplyCurlOrientedUnsignedNoTranspose_Ref_Core(rstr, num_comp, block_size, comp_stride, start, stop, 487 num_elem, elem_size, v_offset, uu, vv)); 488 } else { 489 CeedCallBackend(CeedElemRestrictionApplyOffsetNoTranspose_Ref_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, 490 elem_size, v_offset, uu, vv)); 491 } 492 break; 493 case CEED_RESTRICTION_POINTS: 494 CeedCallBackend(CeedElemRestrictionApplyAtPointsInElement_Ref_Core(rstr, num_comp, start, stop, t_mode, uu, vv)); 495 break; 496 } 497 } 498 CeedCallBackend(CeedVectorRestoreArrayRead(u, &uu)); 499 CeedCallBackend(CeedVectorRestoreArray(v, &vv)); 500 if (request != CEED_REQUEST_IMMEDIATE && request != CEED_REQUEST_ORDERED) *request = NULL; 501 return CEED_ERROR_SUCCESS; 502 } 503 504 //------------------------------------------------------------------------------ 505 // ElemRestriction Apply - Common Sizes 506 //------------------------------------------------------------------------------ 507 static int CeedElemRestrictionApply_Ref_110(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, const CeedInt comp_stride, 508 CeedInt start, CeedInt stop, CeedTransposeMode t_mode, bool use_signs, bool use_orients, CeedVector u, 509 CeedVector v, CeedRequest *request) { 510 return CeedElemRestrictionApply_Ref_Core(rstr, 1, 1, comp_stride, start, stop, t_mode, use_signs, use_orients, u, v, request); 511 } 512 513 static int CeedElemRestrictionApply_Ref_111(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, const CeedInt comp_stride, 514 CeedInt start, CeedInt stop, CeedTransposeMode t_mode, bool use_signs, bool use_orients, CeedVector u, 515 CeedVector v, CeedRequest *request) { 516 return CeedElemRestrictionApply_Ref_Core(rstr, 1, 1, 1, start, stop, t_mode, use_signs, use_orients, u, v, request); 517 } 518 519 static int CeedElemRestrictionApply_Ref_180(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, const CeedInt comp_stride, 520 CeedInt start, CeedInt stop, CeedTransposeMode t_mode, bool use_signs, bool use_orients, CeedVector u, 521 CeedVector v, CeedRequest *request) { 522 return CeedElemRestrictionApply_Ref_Core(rstr, 1, 8, comp_stride, start, stop, t_mode, use_signs, use_orients, u, v, request); 523 } 524 525 static int CeedElemRestrictionApply_Ref_181(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, const CeedInt comp_stride, 526 CeedInt start, CeedInt stop, CeedTransposeMode t_mode, bool use_signs, bool use_orients, CeedVector u, 527 CeedVector v, CeedRequest *request) { 528 return CeedElemRestrictionApply_Ref_Core(rstr, 1, 8, 1, start, stop, t_mode, use_signs, use_orients, u, v, request); 529 } 530 531 static int CeedElemRestrictionApply_Ref_310(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, const CeedInt comp_stride, 532 CeedInt start, CeedInt stop, CeedTransposeMode t_mode, bool use_signs, bool use_orients, CeedVector u, 533 CeedVector v, CeedRequest *request) { 534 return CeedElemRestrictionApply_Ref_Core(rstr, 3, 1, comp_stride, start, stop, t_mode, use_signs, use_orients, u, v, request); 535 } 536 537 static int CeedElemRestrictionApply_Ref_311(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, const CeedInt comp_stride, 538 CeedInt start, CeedInt stop, CeedTransposeMode t_mode, bool use_signs, bool use_orients, CeedVector u, 539 CeedVector v, CeedRequest *request) { 540 return CeedElemRestrictionApply_Ref_Core(rstr, 3, 1, 1, start, stop, t_mode, use_signs, use_orients, u, v, request); 541 } 542 543 static int CeedElemRestrictionApply_Ref_380(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, const CeedInt comp_stride, 544 CeedInt start, CeedInt stop, CeedTransposeMode t_mode, bool use_signs, bool use_orients, CeedVector u, 545 CeedVector v, CeedRequest *request) { 546 return CeedElemRestrictionApply_Ref_Core(rstr, 3, 8, comp_stride, start, stop, t_mode, use_signs, use_orients, u, v, request); 547 } 548 549 static int CeedElemRestrictionApply_Ref_381(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, const CeedInt comp_stride, 550 CeedInt start, CeedInt stop, CeedTransposeMode t_mode, bool use_signs, bool use_orients, CeedVector u, 551 CeedVector v, CeedRequest *request) { 552 return CeedElemRestrictionApply_Ref_Core(rstr, 3, 8, 1, start, stop, t_mode, use_signs, use_orients, u, v, request); 553 } 554 555 // LCOV_EXCL_START 556 static int CeedElemRestrictionApply_Ref_510(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, const CeedInt comp_stride, 557 CeedInt start, CeedInt stop, CeedTransposeMode t_mode, bool use_signs, bool use_orients, CeedVector u, 558 CeedVector v, CeedRequest *request) { 559 return CeedElemRestrictionApply_Ref_Core(rstr, 5, 1, comp_stride, start, stop, t_mode, use_signs, use_orients, u, v, request); 560 } 561 // LCOV_EXCL_STOP 562 563 static int CeedElemRestrictionApply_Ref_511(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, const CeedInt comp_stride, 564 CeedInt start, CeedInt stop, CeedTransposeMode t_mode, bool use_signs, bool use_orients, CeedVector u, 565 CeedVector v, CeedRequest *request) { 566 return CeedElemRestrictionApply_Ref_Core(rstr, 5, 1, 1, start, stop, t_mode, use_signs, use_orients, u, v, request); 567 } 568 569 // LCOV_EXCL_START 570 static int CeedElemRestrictionApply_Ref_580(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, const CeedInt comp_stride, 571 CeedInt start, CeedInt stop, CeedTransposeMode t_mode, bool use_signs, bool use_orients, CeedVector u, 572 CeedVector v, CeedRequest *request) { 573 return CeedElemRestrictionApply_Ref_Core(rstr, 5, 8, comp_stride, start, stop, t_mode, use_signs, use_orients, u, v, request); 574 } 575 // LCOV_EXCL_STOP 576 577 static int CeedElemRestrictionApply_Ref_581(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, const CeedInt comp_stride, 578 CeedInt start, CeedInt stop, CeedTransposeMode t_mode, bool use_signs, bool use_orients, CeedVector u, 579 CeedVector v, CeedRequest *request) { 580 return CeedElemRestrictionApply_Ref_Core(rstr, 5, 8, 1, start, stop, t_mode, use_signs, use_orients, u, v, request); 581 } 582 583 //------------------------------------------------------------------------------ 584 // ElemRestriction Apply 585 //------------------------------------------------------------------------------ 586 static int CeedElemRestrictionApply_Ref(CeedElemRestriction rstr, CeedTransposeMode t_mode, CeedVector u, CeedVector v, CeedRequest *request) { 587 CeedInt num_block, block_size, num_comp, comp_stride; 588 CeedElemRestriction_Ref *impl; 589 590 CeedCallBackend(CeedElemRestrictionGetNumBlocks(rstr, &num_block)); 591 CeedCallBackend(CeedElemRestrictionGetBlockSize(rstr, &block_size)); 592 CeedCallBackend(CeedElemRestrictionGetNumComponents(rstr, &num_comp)); 593 CeedCallBackend(CeedElemRestrictionGetCompStride(rstr, &comp_stride)); 594 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 595 CeedCallBackend(impl->Apply(rstr, num_comp, block_size, comp_stride, 0, num_block, t_mode, true, true, u, v, request)); 596 return CEED_ERROR_SUCCESS; 597 } 598 599 //------------------------------------------------------------------------------ 600 // ElemRestriction Apply Unsigned 601 //------------------------------------------------------------------------------ 602 static int CeedElemRestrictionApplyUnsigned_Ref(CeedElemRestriction rstr, CeedTransposeMode t_mode, CeedVector u, CeedVector v, 603 CeedRequest *request) { 604 CeedInt num_block, block_size, num_comp, comp_stride; 605 CeedElemRestriction_Ref *impl; 606 607 CeedCallBackend(CeedElemRestrictionGetNumBlocks(rstr, &num_block)); 608 CeedCallBackend(CeedElemRestrictionGetBlockSize(rstr, &block_size)); 609 CeedCallBackend(CeedElemRestrictionGetNumComponents(rstr, &num_comp)); 610 CeedCallBackend(CeedElemRestrictionGetCompStride(rstr, &comp_stride)); 611 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 612 CeedCallBackend(impl->Apply(rstr, num_comp, block_size, comp_stride, 0, num_block, t_mode, false, true, u, v, request)); 613 return CEED_ERROR_SUCCESS; 614 } 615 616 //------------------------------------------------------------------------------ 617 // ElemRestriction Apply Unoriented 618 //------------------------------------------------------------------------------ 619 static int CeedElemRestrictionApplyUnoriented_Ref(CeedElemRestriction rstr, CeedTransposeMode t_mode, CeedVector u, CeedVector v, 620 CeedRequest *request) { 621 CeedInt num_block, block_size, num_comp, comp_stride; 622 CeedElemRestriction_Ref *impl; 623 624 CeedCallBackend(CeedElemRestrictionGetNumBlocks(rstr, &num_block)); 625 CeedCallBackend(CeedElemRestrictionGetBlockSize(rstr, &block_size)); 626 CeedCallBackend(CeedElemRestrictionGetNumComponents(rstr, &num_comp)); 627 CeedCallBackend(CeedElemRestrictionGetCompStride(rstr, &comp_stride)); 628 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 629 CeedCallBackend(impl->Apply(rstr, num_comp, block_size, comp_stride, 0, num_block, t_mode, false, false, u, v, request)); 630 return CEED_ERROR_SUCCESS; 631 } 632 633 //------------------------------------------------------------------------------ 634 // ElemRestriction Apply Points 635 //------------------------------------------------------------------------------ 636 static int CeedElemRestrictionApplyAtPointsInElement_Ref(CeedElemRestriction rstr, CeedInt elem, CeedTransposeMode t_mode, CeedVector u, CeedVector v, 637 CeedRequest *request) { 638 CeedInt num_comp; 639 CeedElemRestriction_Ref *impl; 640 641 CeedCallBackend(CeedElemRestrictionGetNumComponents(rstr, &num_comp)); 642 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 643 return impl->Apply(rstr, num_comp, 0, 1, elem, elem + 1, t_mode, false, false, u, v, request); 644 } 645 646 //------------------------------------------------------------------------------ 647 // ElemRestriction Apply Block 648 //------------------------------------------------------------------------------ 649 static int CeedElemRestrictionApplyBlock_Ref(CeedElemRestriction rstr, CeedInt block, CeedTransposeMode t_mode, CeedVector u, CeedVector v, 650 CeedRequest *request) { 651 CeedInt block_size, num_comp, comp_stride; 652 CeedElemRestriction_Ref *impl; 653 654 CeedCallBackend(CeedElemRestrictionGetBlockSize(rstr, &block_size)); 655 CeedCallBackend(CeedElemRestrictionGetNumComponents(rstr, &num_comp)); 656 CeedCallBackend(CeedElemRestrictionGetCompStride(rstr, &comp_stride)); 657 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 658 CeedCallBackend(impl->Apply(rstr, num_comp, block_size, comp_stride, block, block + 1, t_mode, true, true, u, v, request)); 659 return CEED_ERROR_SUCCESS; 660 } 661 662 //------------------------------------------------------------------------------ 663 // ElemRestriction Get Offsets 664 //------------------------------------------------------------------------------ 665 static int CeedElemRestrictionGetOffsets_Ref(CeedElemRestriction rstr, CeedMemType mem_type, const CeedInt **offsets) { 666 CeedElemRestriction_Ref *impl; 667 668 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 669 670 CeedCheck(mem_type == CEED_MEM_HOST, CeedElemRestrictionReturnCeed(rstr), CEED_ERROR_BACKEND, "Can only provide to HOST memory"); 671 672 *offsets = impl->offsets; 673 return CEED_ERROR_SUCCESS; 674 } 675 676 //------------------------------------------------------------------------------ 677 // ElemRestriction Get Orientations 678 //------------------------------------------------------------------------------ 679 static int CeedElemRestrictionGetOrientations_Ref(CeedElemRestriction rstr, CeedMemType mem_type, const bool **orients) { 680 CeedElemRestriction_Ref *impl; 681 682 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 683 684 CeedCheck(mem_type == CEED_MEM_HOST, CeedElemRestrictionReturnCeed(rstr), CEED_ERROR_BACKEND, "Can only provide to HOST memory"); 685 686 *orients = impl->orients; 687 return CEED_ERROR_SUCCESS; 688 } 689 690 //------------------------------------------------------------------------------ 691 // ElemRestriction Get Curl-Conforming Orientations 692 //------------------------------------------------------------------------------ 693 static int CeedElemRestrictionGetCurlOrientations_Ref(CeedElemRestriction rstr, CeedMemType mem_type, const CeedInt8 **curl_orients) { 694 CeedElemRestriction_Ref *impl; 695 696 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 697 698 CeedCheck(mem_type == CEED_MEM_HOST, CeedElemRestrictionReturnCeed(rstr), CEED_ERROR_BACKEND, "Can only provide to HOST memory"); 699 700 *curl_orients = impl->curl_orients; 701 return CEED_ERROR_SUCCESS; 702 } 703 704 //------------------------------------------------------------------------------ 705 // ElemRestriction Destroy 706 //------------------------------------------------------------------------------ 707 static int CeedElemRestrictionDestroy_Ref(CeedElemRestriction rstr) { 708 CeedElemRestriction_Ref *impl; 709 710 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 711 CeedCallBackend(CeedFree(&impl->offsets_owned)); 712 CeedCallBackend(CeedFree(&impl->orients_owned)); 713 CeedCallBackend(CeedFree(&impl->curl_orients_owned)); 714 CeedCallBackend(CeedFree(&impl)); 715 return CEED_ERROR_SUCCESS; 716 } 717 718 //------------------------------------------------------------------------------ 719 // ElemRestriction Create 720 //------------------------------------------------------------------------------ 721 int CeedElemRestrictionCreate_Ref(CeedMemType mem_type, CeedCopyMode copy_mode, const CeedInt *offsets, const bool *orients, 722 const CeedInt8 *curl_orients, CeedElemRestriction rstr) { 723 Ceed ceed; 724 CeedInt num_elem, elem_size, num_block, block_size, num_comp, comp_stride, num_points = 0, num_offsets; 725 CeedRestrictionType rstr_type; 726 CeedElemRestriction_Ref *impl; 727 728 CeedCallBackend(CeedElemRestrictionGetCeed(rstr, &ceed)); 729 CeedCallBackend(CeedElemRestrictionGetNumElements(rstr, &num_elem)); 730 CeedCallBackend(CeedElemRestrictionGetElementSize(rstr, &elem_size)); 731 CeedCallBackend(CeedElemRestrictionGetNumBlocks(rstr, &num_block)); 732 CeedCallBackend(CeedElemRestrictionGetBlockSize(rstr, &block_size)); 733 CeedCallBackend(CeedElemRestrictionGetNumComponents(rstr, &num_comp)); 734 CeedCallBackend(CeedElemRestrictionGetCompStride(rstr, &comp_stride)); 735 CeedCallBackend(CeedElemRestrictionGetType(rstr, &rstr_type)); 736 737 CeedCheck(mem_type == CEED_MEM_HOST, ceed, CEED_ERROR_BACKEND, "Only MemType = HOST supported"); 738 739 CeedCallBackend(CeedCalloc(1, &impl)); 740 CeedCallBackend(CeedElemRestrictionSetData(rstr, impl)); 741 742 // Set layouts 743 { 744 bool has_backend_strides; 745 CeedInt layout[3] = {1, elem_size, elem_size * num_comp}; 746 747 CeedCallBackend(CeedElemRestrictionSetELayout(rstr, layout)); 748 if (rstr_type == CEED_RESTRICTION_STRIDED) { 749 CeedCallBackend(CeedElemRestrictionHasBackendStrides(rstr, &has_backend_strides)); 750 if (has_backend_strides) { 751 CeedCallBackend(CeedElemRestrictionSetLLayout(rstr, layout)); 752 } 753 } 754 } 755 756 // Expand E-vector size for AtPoints 757 if (rstr_type == CEED_RESTRICTION_POINTS) { 758 CeedSize max_points = 0, num_points_total = 0; 759 760 for (CeedInt i = 0; i < num_elem; i++) { 761 CeedInt num_points = offsets[i + 1] - offsets[i]; 762 763 max_points = CeedIntMax(max_points, num_points); 764 num_points_total += num_points; 765 } 766 // -- Increase size for last element 767 num_points_total += (max_points - (offsets[num_elem] - offsets[num_elem - 1])); 768 CeedCallBackend(CeedElemRestrictionSetAtPointsEVectorSize(rstr, num_points_total * num_comp)); 769 } 770 771 // Offsets data 772 if (rstr_type != CEED_RESTRICTION_STRIDED) { 773 const char *resource; 774 775 // Check indices for ref or memcheck backends 776 { 777 Ceed current = ceed, parent = NULL; 778 779 CeedCallBackend(CeedGetParent(current, &parent)); 780 while (current != parent) { 781 current = parent; 782 CeedCallBackend(CeedGetParent(current, &parent)); 783 } 784 CeedCallBackend(CeedGetResource(parent, &resource)); 785 } 786 if (!strcmp(resource, "/cpu/self/ref/serial") || !strcmp(resource, "/cpu/self/ref/blocked")) { 787 CeedSize l_size; 788 789 CeedCallBackend(CeedElemRestrictionGetLVectorSize(rstr, &l_size)); 790 for (CeedInt i = 0; i < num_elem * elem_size; i++) { 791 CeedCheck(offsets[i] >= 0 && offsets[i] + (num_comp - 1) * comp_stride < l_size, ceed, CEED_ERROR_BACKEND, 792 "Restriction offset %" CeedInt_FMT " (%" CeedInt_FMT ") out of range [0, %" CeedInt_FMT "]", i, offsets[i], l_size); 793 } 794 } 795 796 // Copy data 797 if (rstr_type == CEED_RESTRICTION_POINTS) CeedCallBackend(CeedElemRestrictionGetNumPoints(rstr, &num_points)); 798 num_offsets = rstr_type == CEED_RESTRICTION_POINTS ? (num_elem + 1 + num_points) : (num_elem * elem_size); 799 CeedCallBackend(CeedSetHostCeedIntArray(offsets, copy_mode, num_offsets, &impl->offsets_owned, &impl->offsets_borrowed, &impl->offsets)); 800 801 // Orientation data 802 if (rstr_type == CEED_RESTRICTION_ORIENTED) { 803 CeedCheck(orients != NULL, ceed, CEED_ERROR_BACKEND, "No orients array provided for oriented restriction"); 804 CeedCallBackend(CeedSetHostBoolArray(orients, copy_mode, num_offsets, &impl->orients_owned, &impl->orients_borrowed, &impl->orients)); 805 } else if (rstr_type == CEED_RESTRICTION_CURL_ORIENTED) { 806 CeedCheck(curl_orients != NULL, ceed, CEED_ERROR_BACKEND, "No curl_orients array provided for oriented restriction"); 807 CeedCallBackend(CeedSetHostCeedInt8Array(curl_orients, copy_mode, 3 * num_offsets, &impl->curl_orients_owned, &impl->curl_orients_borrowed, 808 &impl->curl_orients)); 809 } 810 } 811 812 // Set apply function based upon num_comp, block_size, and comp_stride 813 CeedInt index = -1; 814 815 if (block_size < 10) index = 100 * num_comp + 10 * block_size + (comp_stride == 1); 816 switch (index) { 817 case 110: 818 impl->Apply = CeedElemRestrictionApply_Ref_110; 819 break; 820 case 111: 821 impl->Apply = CeedElemRestrictionApply_Ref_111; 822 break; 823 case 180: 824 impl->Apply = CeedElemRestrictionApply_Ref_180; 825 break; 826 case 181: 827 impl->Apply = CeedElemRestrictionApply_Ref_181; 828 break; 829 case 310: 830 impl->Apply = CeedElemRestrictionApply_Ref_310; 831 break; 832 case 311: 833 impl->Apply = CeedElemRestrictionApply_Ref_311; 834 break; 835 case 380: 836 impl->Apply = CeedElemRestrictionApply_Ref_380; 837 break; 838 case 381: 839 impl->Apply = CeedElemRestrictionApply_Ref_381; 840 break; 841 // LCOV_EXCL_START 842 case 510: 843 impl->Apply = CeedElemRestrictionApply_Ref_510; 844 break; 845 // LCOV_EXCL_STOP 846 case 511: 847 impl->Apply = CeedElemRestrictionApply_Ref_511; 848 break; 849 // LCOV_EXCL_START 850 case 580: 851 impl->Apply = CeedElemRestrictionApply_Ref_580; 852 break; 853 // LCOV_EXCL_STOP 854 case 581: 855 impl->Apply = CeedElemRestrictionApply_Ref_581; 856 break; 857 default: 858 impl->Apply = CeedElemRestrictionApply_Ref_Core; 859 break; 860 } 861 862 // Register backend functions 863 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "Apply", CeedElemRestrictionApply_Ref)); 864 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "ApplyUnsigned", CeedElemRestrictionApplyUnsigned_Ref)); 865 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "ApplyUnoriented", CeedElemRestrictionApplyUnoriented_Ref)); 866 if (rstr_type == CEED_RESTRICTION_POINTS) { 867 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "ApplyAtPointsInElement", CeedElemRestrictionApplyAtPointsInElement_Ref)); 868 } 869 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "ApplyBlock", CeedElemRestrictionApplyBlock_Ref)); 870 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "GetOffsets", CeedElemRestrictionGetOffsets_Ref)); 871 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "GetOrientations", CeedElemRestrictionGetOrientations_Ref)); 872 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "GetCurlOrientations", CeedElemRestrictionGetCurlOrientations_Ref)); 873 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "Destroy", CeedElemRestrictionDestroy_Ref)); 874 return CEED_ERROR_SUCCESS; 875 } 876 877 //------------------------------------------------------------------------------ 878