1 // Copyright (c) 2017-2022, 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-memcheck.h" 15 16 //------------------------------------------------------------------------------ 17 // Set backend strides 18 //------------------------------------------------------------------------------ 19 static inline int CeedElemRestrictionGetBackendStrides_Memcheck(CeedElemRestriction rstr, CeedInt strides[3]) { 20 CeedInt elem_size, num_comp, num_elem; 21 22 CeedCallBackend(CeedElemRestrictionGetElementSize(rstr, &elem_size)); 23 CeedCallBackend(CeedElemRestrictionGetNumComponents(rstr, &num_comp)); 24 CeedCallBackend(CeedElemRestrictionGetNumElements(rstr, &num_elem)); 25 // Memcheck default, contiguous by component, then node 26 strides[0] = num_comp; 27 strides[1] = 1; 28 strides[2] = num_comp * elem_size; 29 /** 30 // CPU default, contiguous by node, then component 31 strides[0] = 1; 32 strides[1] = elem_size; 33 strides[2] = elem_size * num_comp; 34 35 // GPU default, contiguous by node, then element 36 strides[0] = 1; 37 strides[1] = num_elem * elem_size; 38 strides[2] = elem_size; 39 **/ 40 return CEED_ERROR_SUCCESS; 41 } 42 43 //------------------------------------------------------------------------------ 44 // Core ElemRestriction Apply Code 45 //------------------------------------------------------------------------------ 46 static inline int CeedElemRestrictionApplyStridedNoTranspose_Memcheck_Core(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, 47 CeedInt start, CeedInt stop, CeedInt num_elem, CeedInt elem_size, 48 CeedInt v_offset, const CeedScalar *__restrict__ uu, 49 CeedScalar *__restrict__ vv) { 50 // Get strides 51 bool has_backend_strides; 52 CeedInt strides[3] = {0}; 53 54 CeedCallBackend(CeedElemRestrictionHasBackendStrides(rstr, &has_backend_strides)); 55 if (has_backend_strides) CeedCallBackend(CeedElemRestrictionGetBackendStrides_Memcheck(rstr, strides)); 56 else CeedCallBackend(CeedElemRestrictionGetStrides(rstr, strides)); 57 58 // Apply restriction 59 for (CeedInt e = start * block_size; e < stop * block_size; e += block_size) { 60 CeedPragmaSIMD for (CeedInt k = 0; k < num_comp; k++) { 61 CeedPragmaSIMD for (CeedInt n = 0; n < elem_size; n++) { 62 CeedPragmaSIMD for (CeedInt j = 0; j < block_size; j++) { 63 vv[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] = 64 uu[n * strides[0] + k * strides[1] + CeedIntMin(e + j, num_elem - 1) * strides[2]]; 65 } 66 } 67 } 68 } 69 return CEED_ERROR_SUCCESS; 70 } 71 72 static inline int CeedElemRestrictionApplyOffsetNoTranspose_Memcheck_Core(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, 73 const CeedInt comp_stride, CeedInt start, CeedInt stop, CeedInt num_elem, 74 CeedInt elem_size, CeedInt v_offset, const CeedScalar *__restrict__ uu, 75 CeedScalar *__restrict__ vv) { 76 // Default restriction with offsets 77 CeedElemRestriction_Memcheck *impl; 78 79 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 80 for (CeedInt e = start * block_size; e < stop * block_size; e += block_size) { 81 CeedPragmaSIMD for (CeedInt k = 0; k < num_comp; k++) { 82 CeedPragmaSIMD for (CeedInt i = 0; i < elem_size * block_size; i++) { 83 vv[elem_size * (k * block_size + e * num_comp) + i - v_offset] = uu[impl->offsets[i + e * elem_size] + k * comp_stride]; 84 } 85 } 86 } 87 return CEED_ERROR_SUCCESS; 88 } 89 90 static inline int CeedElemRestrictionApplyOrientedNoTranspose_Memcheck_Core(CeedElemRestriction rstr, const CeedInt num_comp, 91 const CeedInt block_size, const CeedInt comp_stride, CeedInt start, 92 CeedInt stop, CeedInt num_elem, CeedInt elem_size, CeedInt v_offset, 93 const CeedScalar *__restrict__ uu, CeedScalar *__restrict__ vv) { 94 // Restriction with orientations 95 CeedElemRestriction_Memcheck *impl; 96 97 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 98 for (CeedInt e = start * block_size; e < stop * block_size; e += block_size) { 99 CeedPragmaSIMD for (CeedInt k = 0; k < num_comp; k++) { 100 CeedPragmaSIMD for (CeedInt i = 0; i < elem_size * block_size; i++) { 101 vv[elem_size * (k * block_size + e * num_comp) + i - v_offset] = 102 uu[impl->offsets[i + e * elem_size] + k * comp_stride] * (impl->orients[i + e * elem_size] ? -1.0 : 1.0); 103 } 104 } 105 } 106 return CEED_ERROR_SUCCESS; 107 } 108 109 static inline int CeedElemRestrictionApplyCurlOrientedNoTranspose_Memcheck_Core(CeedElemRestriction rstr, const CeedInt num_comp, 110 const CeedInt block_size, const CeedInt comp_stride, CeedInt start, 111 CeedInt stop, CeedInt num_elem, CeedInt elem_size, CeedInt v_offset, 112 const CeedScalar *__restrict__ uu, CeedScalar *__restrict__ vv) { 113 // Restriction with tridiagonal transformation 114 CeedElemRestriction_Memcheck *impl; 115 116 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 117 for (CeedInt e = start * block_size; e < stop * block_size; e += block_size) { 118 CeedPragmaSIMD for (CeedInt k = 0; k < num_comp; k++) { 119 CeedInt n = 0; 120 121 CeedPragmaSIMD for (CeedInt j = 0; j < block_size; j++) { 122 vv[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] = 123 uu[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] * 124 impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size] + 125 uu[impl->offsets[j + (n + 1) * block_size + e * elem_size] + k * comp_stride] * 126 impl->curl_orients[j + (3 * n + 2) * block_size + e * 3 * elem_size]; 127 } 128 CeedPragmaSIMD for (n = 1; n < elem_size - 1; n++) { 129 CeedPragmaSIMD for (CeedInt j = 0; j < block_size; j++) { 130 vv[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] = 131 uu[impl->offsets[j + (n - 1) * block_size + e * elem_size] + k * comp_stride] * 132 impl->curl_orients[j + (3 * n + 0) * block_size + e * 3 * elem_size] + 133 uu[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] * 134 impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size] + 135 uu[impl->offsets[j + (n + 1) * block_size + e * elem_size] + k * comp_stride] * 136 impl->curl_orients[j + (3 * n + 2) * block_size + e * 3 * elem_size]; 137 } 138 } 139 CeedPragmaSIMD for (CeedInt j = 0; j < block_size; j++) { 140 vv[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] = 141 uu[impl->offsets[j + (n - 1) * block_size + e * elem_size] + k * comp_stride] * 142 impl->curl_orients[j + (3 * n + 0) * block_size + e * 3 * elem_size] + 143 uu[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] * 144 impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size]; 145 } 146 } 147 } 148 return CEED_ERROR_SUCCESS; 149 } 150 151 static inline int CeedElemRestrictionApplyCurlOrientedUnsignedNoTranspose_Memcheck_Core( 152 CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, const CeedInt comp_stride, CeedInt start, CeedInt stop, 153 CeedInt num_elem, CeedInt elem_size, CeedInt v_offset, const CeedScalar *__restrict__ uu, CeedScalar *__restrict__ vv) { 154 // Restriction with (unsigned) tridiagonal transformation 155 CeedElemRestriction_Memcheck *impl; 156 157 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 158 for (CeedInt e = start * block_size; e < stop * block_size; e += block_size) { 159 CeedPragmaSIMD for (CeedInt k = 0; k < num_comp; k++) { 160 CeedInt n = 0; 161 162 CeedPragmaSIMD for (CeedInt j = 0; j < block_size; j++) { 163 vv[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] = 164 uu[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] * 165 abs(impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size]) + 166 uu[impl->offsets[j + (n + 1) * block_size + e * elem_size] + k * comp_stride] * 167 abs(impl->curl_orients[j + (3 * n + 2) * block_size + e * 3 * elem_size]); 168 } 169 CeedPragmaSIMD for (n = 1; n < elem_size - 1; n++) { 170 CeedPragmaSIMD for (CeedInt j = 0; j < block_size; j++) { 171 vv[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] = 172 uu[impl->offsets[j + (n - 1) * block_size + e * elem_size] + k * comp_stride] * 173 abs(impl->curl_orients[j + (3 * n + 0) * block_size + e * 3 * elem_size]) + 174 uu[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] * 175 abs(impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size]) + 176 uu[impl->offsets[j + (n + 1) * block_size + e * elem_size] + k * comp_stride] * 177 abs(impl->curl_orients[j + (3 * n + 2) * block_size + e * 3 * elem_size]); 178 } 179 } 180 CeedPragmaSIMD for (CeedInt j = 0; j < block_size; j++) { 181 vv[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] = 182 uu[impl->offsets[j + (n - 1) * block_size + e * elem_size] + k * comp_stride] * 183 abs(impl->curl_orients[j + (3 * n + 0) * block_size + e * 3 * elem_size]) + 184 uu[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] * 185 abs(impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size]); 186 } 187 } 188 } 189 return CEED_ERROR_SUCCESS; 190 } 191 192 static inline int CeedElemRestrictionApplyStridedTranspose_Memcheck_Core(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, 193 CeedInt start, CeedInt stop, CeedInt num_elem, CeedInt elem_size, 194 CeedInt v_offset, const CeedScalar *__restrict__ uu, 195 CeedScalar *__restrict__ vv) { 196 // Get strides 197 bool has_backend_strides; 198 CeedInt strides[3] = {0}; 199 200 CeedCallBackend(CeedElemRestrictionHasBackendStrides(rstr, &has_backend_strides)); 201 if (has_backend_strides) CeedCallBackend(CeedElemRestrictionGetBackendStrides_Memcheck(rstr, strides)); 202 else CeedCallBackend(CeedElemRestrictionGetStrides(rstr, strides)); 203 204 // Apply restriction 205 for (CeedInt e = start * block_size; e < stop * block_size; e += block_size) { 206 CeedPragmaSIMD for (CeedInt k = 0; k < num_comp; k++) { 207 CeedPragmaSIMD for (CeedInt n = 0; n < elem_size; n++) { 208 CeedPragmaSIMD for (CeedInt j = 0; j < CeedIntMin(block_size, num_elem - e); j++) { 209 vv[n * strides[0] + k * strides[1] + (e + j) * strides[2]] += 210 uu[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset]; 211 } 212 } 213 } 214 } 215 return CEED_ERROR_SUCCESS; 216 } 217 218 static inline int CeedElemRestrictionApplyOffsetTranspose_Memcheck_Core(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, 219 const CeedInt comp_stride, CeedInt start, CeedInt stop, CeedInt num_elem, 220 CeedInt elem_size, CeedInt v_offset, const CeedScalar *__restrict__ uu, 221 CeedScalar *__restrict__ vv) { 222 // Default restriction with offsets 223 CeedElemRestriction_Memcheck *impl; 224 225 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 226 for (CeedInt e = start * block_size; e < stop * block_size; e += block_size) { 227 for (CeedInt k = 0; k < num_comp; k++) { 228 for (CeedInt i = 0; i < elem_size * block_size; i += block_size) { 229 // Iteration bound set to discard padding elements 230 for (CeedInt j = i; j < i + CeedIntMin(block_size, num_elem - e); j++) { 231 CeedScalar vv_loc; 232 233 vv_loc = uu[elem_size * (k * block_size + e * num_comp) + j - v_offset]; 234 CeedPragmaAtomic vv[impl->offsets[j + e * elem_size] + k * comp_stride] += vv_loc; 235 } 236 } 237 } 238 } 239 return CEED_ERROR_SUCCESS; 240 } 241 242 static inline int CeedElemRestrictionApplyOrientedTranspose_Memcheck_Core(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, 243 const CeedInt comp_stride, CeedInt start, CeedInt stop, CeedInt num_elem, 244 CeedInt elem_size, CeedInt v_offset, const CeedScalar *__restrict__ uu, 245 CeedScalar *__restrict__ vv) { 246 // Restriction with orientations 247 CeedElemRestriction_Memcheck *impl; 248 249 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 250 for (CeedInt e = start * block_size; e < stop * block_size; e += block_size) { 251 for (CeedInt k = 0; k < num_comp; k++) { 252 for (CeedInt i = 0; i < elem_size * block_size; i += block_size) { 253 // Iteration bound set to discard padding elements 254 for (CeedInt j = i; j < i + CeedIntMin(block_size, num_elem - e); j++) { 255 CeedScalar vv_loc; 256 257 vv_loc = uu[elem_size * (k * block_size + e * num_comp) + j - v_offset] * (impl->orients[j + e * elem_size] ? -1.0 : 1.0); 258 CeedPragmaAtomic vv[impl->offsets[j + e * elem_size] + k * comp_stride] += vv_loc; 259 } 260 } 261 } 262 } 263 return CEED_ERROR_SUCCESS; 264 } 265 266 static inline int CeedElemRestrictionApplyCurlOrientedTranspose_Memcheck_Core(CeedElemRestriction rstr, const CeedInt num_comp, 267 const CeedInt block_size, const CeedInt comp_stride, CeedInt start, 268 CeedInt stop, CeedInt num_elem, CeedInt elem_size, CeedInt v_offset, 269 const CeedScalar *__restrict__ uu, CeedScalar *__restrict__ vv) { 270 // Restriction with tridiagonal transformation 271 CeedElemRestriction_Memcheck *impl; 272 CeedScalar vv_loc[block_size]; 273 274 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 275 for (CeedInt e = start * block_size; e < stop * block_size; e += block_size) { 276 for (CeedInt k = 0; k < num_comp; k++) { 277 // Iteration bound set to discard padding elements 278 const CeedInt block_end = CeedIntMin(block_size, num_elem - e); 279 CeedInt n = 0; 280 281 CeedPragmaSIMD for (CeedInt j = 0; j < block_end; j++) { 282 vv_loc[j] = uu[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] * 283 impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size] + 284 uu[e * elem_size * num_comp + (k * elem_size + n + 1) * block_size + j - v_offset] * 285 impl->curl_orients[j + (3 * n + 3) * block_size + e * 3 * elem_size]; 286 } 287 for (CeedInt j = 0; j < block_end; j++) { 288 CeedPragmaAtomic vv[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] += vv_loc[j]; 289 } 290 for (n = 1; n < elem_size - 1; n++) { 291 CeedPragmaSIMD for (CeedInt j = 0; j < block_end; j++) { 292 vv_loc[j] = uu[e * elem_size * num_comp + (k * elem_size + n - 1) * block_size + j - v_offset] * 293 impl->curl_orients[j + (3 * n - 1) * block_size + e * 3 * elem_size] + 294 uu[e * elem_size * num_comp + (k * elem_size + n) * 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 + 1) * block_size + j - v_offset] * 297 impl->curl_orients[j + (3 * n + 3) * block_size + e * 3 * elem_size]; 298 } 299 for (CeedInt j = 0; j < block_end; j++) { 300 CeedPragmaAtomic vv[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] += vv_loc[j]; 301 } 302 } 303 CeedPragmaSIMD for (CeedInt j = 0; j < block_end; j++) { 304 vv_loc[j] = uu[e * elem_size * num_comp + (k * elem_size + n - 1) * block_size + j - v_offset] * 305 impl->curl_orients[j + (3 * n - 1) * block_size + e * 3 * elem_size] + 306 uu[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] * 307 impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size]; 308 } 309 for (CeedInt j = 0; j < block_end; j++) { 310 CeedPragmaAtomic vv[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] += vv_loc[j]; 311 } 312 } 313 } 314 return CEED_ERROR_SUCCESS; 315 } 316 317 static inline int CeedElemRestrictionApplyCurlOrientedUnsignedTranspose_Memcheck_Core( 318 CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, const CeedInt comp_stride, CeedInt start, CeedInt stop, 319 CeedInt num_elem, CeedInt elem_size, CeedInt v_offset, const CeedScalar *__restrict__ uu, CeedScalar *__restrict__ vv) { 320 // Restriction with (unsigned) tridiagonal transformation 321 CeedElemRestriction_Memcheck *impl; 322 CeedScalar vv_loc[block_size]; 323 324 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 325 for (CeedInt e = start * block_size; e < stop * block_size; e += block_size) { 326 for (CeedInt k = 0; k < num_comp; k++) { 327 // Iteration bound set to discard padding elements 328 const CeedInt block_end = CeedIntMin(block_size, num_elem - e); 329 CeedInt n = 0; 330 331 CeedPragmaSIMD for (CeedInt j = 0; j < block_end; j++) { 332 vv_loc[j] = uu[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] * 333 abs(impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size]) + 334 uu[e * elem_size * num_comp + (k * elem_size + n + 1) * block_size + j - v_offset] * 335 abs(impl->curl_orients[j + (3 * n + 3) * block_size + e * 3 * elem_size]); 336 } 337 for (CeedInt j = 0; j < block_end; j++) { 338 CeedPragmaAtomic vv[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] += vv_loc[j]; 339 } 340 for (n = 1; n < elem_size - 1; n++) { 341 CeedPragmaSIMD for (CeedInt j = 0; j < block_end; j++) { 342 vv_loc[j] = uu[e * elem_size * num_comp + (k * elem_size + n - 1) * block_size + j - v_offset] * 343 abs(impl->curl_orients[j + (3 * n - 1) * block_size + e * 3 * elem_size]) + 344 uu[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] * 345 abs(impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size]) + 346 uu[e * elem_size * num_comp + (k * elem_size + n + 1) * block_size + j - v_offset] * 347 abs(impl->curl_orients[j + (3 * n + 3) * block_size + e * 3 * elem_size]); 348 } 349 for (CeedInt j = 0; j < block_end; j++) { 350 CeedPragmaAtomic vv[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] += vv_loc[j]; 351 } 352 } 353 CeedPragmaSIMD for (CeedInt j = 0; j < block_end; j++) { 354 vv_loc[j] = uu[e * elem_size * num_comp + (k * elem_size + n - 1) * block_size + j - v_offset] * 355 abs(impl->curl_orients[j + (3 * n - 1) * block_size + e * 3 * elem_size]) + 356 uu[e * elem_size * num_comp + (k * elem_size + n) * block_size + j - v_offset] * 357 abs(impl->curl_orients[j + (3 * n + 1) * block_size + e * 3 * elem_size]); 358 } 359 for (CeedInt j = 0; j < block_end; j++) { 360 CeedPragmaAtomic vv[impl->offsets[j + n * block_size + e * elem_size] + k * comp_stride] += vv_loc[j]; 361 } 362 } 363 } 364 return CEED_ERROR_SUCCESS; 365 } 366 367 static inline int CeedElemRestrictionApplyAtPointsInElement_Memcheck_Core(CeedElemRestriction rstr, const CeedInt num_comp, CeedInt start, 368 CeedInt stop, CeedTransposeMode t_mode, const CeedScalar *__restrict__ uu, 369 CeedScalar *__restrict__ vv) { 370 CeedInt num_points, l_vec_offset, e_vec_offset = 0; 371 CeedElemRestriction_Memcheck *impl; 372 373 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 374 for (CeedInt e = start; e < stop; e++) { 375 l_vec_offset = impl->offsets[e]; 376 CeedCallBackend(CeedElemRestrictionGetNumPointsInElement(rstr, e, &num_points)); 377 if (t_mode == CEED_NOTRANSPOSE) { 378 for (CeedInt i = 0; i < num_points; i++) { 379 for (CeedInt j = 0; j < num_comp; j++) vv[j * num_points + i + e_vec_offset] = uu[impl->offsets[i + l_vec_offset] * num_comp + j]; 380 } 381 } else { 382 for (CeedInt i = 0; i < num_points; i++) { 383 for (CeedInt j = 0; j < num_comp; j++) vv[impl->offsets[i + l_vec_offset] * num_comp + j] = uu[j * num_points + i + e_vec_offset]; 384 } 385 } 386 e_vec_offset += num_points * num_comp; 387 } 388 return CEED_ERROR_SUCCESS; 389 } 390 391 static inline int CeedElemRestrictionApply_Memcheck_Core(CeedElemRestriction rstr, const CeedInt num_comp, const CeedInt block_size, 392 const CeedInt comp_stride, CeedInt start, CeedInt stop, CeedTransposeMode t_mode, 393 bool use_signs, bool use_orients, CeedVector u, CeedVector v, CeedRequest *request) { 394 CeedInt num_elem, elem_size, v_offset; 395 CeedRestrictionType rstr_type; 396 const CeedScalar *uu; 397 CeedScalar *vv; 398 399 CeedCallBackend(CeedElemRestrictionGetNumElements(rstr, &num_elem)); 400 CeedCallBackend(CeedElemRestrictionGetElementSize(rstr, &elem_size)); 401 v_offset = start * block_size * elem_size * num_comp; 402 CeedCallBackend(CeedElemRestrictionGetType(rstr, &rstr_type)); 403 CeedCallBackend(CeedVectorGetArrayRead(u, CEED_MEM_HOST, &uu)); 404 405 if (t_mode == CEED_TRANSPOSE) { 406 // Sum into for transpose mode, E-vector to L-vector 407 CeedCallBackend(CeedVectorGetArray(v, CEED_MEM_HOST, &vv)); 408 } else { 409 // Overwrite for notranspose mode, L-vector to E-vector 410 CeedCallBackend(CeedVectorGetArrayWrite(v, CEED_MEM_HOST, &vv)); 411 } 412 413 if (t_mode == CEED_TRANSPOSE) { 414 // Restriction from E-vector to L-vector 415 // Performing v += r^T * u 416 // uu has shape [elem_size, num_comp, num_elem], row-major 417 // vv has shape [nnodes, num_comp] 418 // Sum into for transpose mode 419 switch (rstr_type) { 420 case CEED_RESTRICTION_STRIDED: 421 CeedCallBackend( 422 CeedElemRestrictionApplyStridedTranspose_Memcheck_Core(rstr, num_comp, block_size, start, stop, num_elem, elem_size, v_offset, uu, vv)); 423 break; 424 case CEED_RESTRICTION_STANDARD: 425 CeedCallBackend(CeedElemRestrictionApplyOffsetTranspose_Memcheck_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, 426 elem_size, v_offset, uu, vv)); 427 break; 428 case CEED_RESTRICTION_ORIENTED: 429 if (use_signs) { 430 CeedCallBackend(CeedElemRestrictionApplyOrientedTranspose_Memcheck_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, 431 elem_size, v_offset, uu, vv)); 432 } else { 433 CeedCallBackend(CeedElemRestrictionApplyOffsetTranspose_Memcheck_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, 434 elem_size, v_offset, uu, vv)); 435 } 436 break; 437 case CEED_RESTRICTION_CURL_ORIENTED: 438 if (use_signs && use_orients) { 439 CeedCallBackend(CeedElemRestrictionApplyCurlOrientedTranspose_Memcheck_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, 440 elem_size, v_offset, uu, vv)); 441 } else if (use_orients) { 442 CeedCallBackend(CeedElemRestrictionApplyCurlOrientedUnsignedTranspose_Memcheck_Core(rstr, num_comp, block_size, comp_stride, start, stop, 443 num_elem, elem_size, v_offset, uu, vv)); 444 } else { 445 CeedCallBackend(CeedElemRestrictionApplyOffsetTranspose_Memcheck_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, 446 elem_size, v_offset, uu, vv)); 447 } 448 break; 449 case CEED_RESTRICTION_POINTS: 450 CeedCallBackend(CeedElemRestrictionApplyAtPointsInElement_Memcheck_Core(rstr, num_comp, start, stop, t_mode, uu, vv)); 451 break; 452 } 453 } else { 454 // Restriction from L-vector to E-vector 455 // Perform: v = r * u 456 // vv has shape [elem_size, num_comp, num_elem], row-major 457 // uu has shape [nnodes, num_comp] 458 // Overwrite for notranspose mode 459 switch (rstr_type) { 460 case CEED_RESTRICTION_STRIDED: 461 CeedCallBackend( 462 CeedElemRestrictionApplyStridedNoTranspose_Memcheck_Core(rstr, num_comp, block_size, start, stop, num_elem, elem_size, v_offset, uu, vv)); 463 break; 464 case CEED_RESTRICTION_STANDARD: 465 CeedCallBackend(CeedElemRestrictionApplyOffsetNoTranspose_Memcheck_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, 466 elem_size, v_offset, uu, vv)); 467 break; 468 case CEED_RESTRICTION_ORIENTED: 469 if (use_signs) { 470 CeedCallBackend(CeedElemRestrictionApplyOrientedNoTranspose_Memcheck_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, 471 elem_size, v_offset, uu, vv)); 472 } else { 473 CeedCallBackend(CeedElemRestrictionApplyOffsetNoTranspose_Memcheck_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, 474 elem_size, v_offset, uu, vv)); 475 } 476 break; 477 case CEED_RESTRICTION_CURL_ORIENTED: 478 if (use_signs && use_orients) { 479 CeedCallBackend(CeedElemRestrictionApplyCurlOrientedNoTranspose_Memcheck_Core(rstr, num_comp, block_size, comp_stride, start, stop, 480 num_elem, elem_size, v_offset, uu, vv)); 481 } else if (use_orients) { 482 CeedCallBackend(CeedElemRestrictionApplyCurlOrientedUnsignedNoTranspose_Memcheck_Core(rstr, num_comp, block_size, comp_stride, start, stop, 483 num_elem, elem_size, v_offset, uu, vv)); 484 } else { 485 CeedCallBackend(CeedElemRestrictionApplyOffsetNoTranspose_Memcheck_Core(rstr, num_comp, block_size, comp_stride, start, stop, num_elem, 486 elem_size, v_offset, uu, vv)); 487 } 488 break; 489 case CEED_RESTRICTION_POINTS: 490 CeedCallBackend(CeedElemRestrictionApplyAtPointsInElement_Memcheck_Core(rstr, num_comp, start, stop, t_mode, uu, vv)); 491 break; 492 } 493 } 494 CeedCallBackend(CeedVectorRestoreArrayRead(u, &uu)); 495 CeedCallBackend(CeedVectorRestoreArray(v, &vv)); 496 if (request != CEED_REQUEST_IMMEDIATE && request != CEED_REQUEST_ORDERED) *request = NULL; 497 return CEED_ERROR_SUCCESS; 498 } 499 500 //------------------------------------------------------------------------------ 501 // ElemRestriction Apply 502 //------------------------------------------------------------------------------ 503 static int CeedElemRestrictionApply_Memcheck(CeedElemRestriction rstr, CeedTransposeMode t_mode, CeedVector u, CeedVector v, CeedRequest *request) { 504 CeedInt num_block, block_size, num_comp, comp_stride; 505 CeedElemRestriction_Memcheck *impl; 506 507 CeedCallBackend(CeedElemRestrictionGetNumBlocks(rstr, &num_block)); 508 CeedCallBackend(CeedElemRestrictionGetBlockSize(rstr, &block_size)); 509 CeedCallBackend(CeedElemRestrictionGetNumComponents(rstr, &num_comp)); 510 CeedCallBackend(CeedElemRestrictionGetCompStride(rstr, &comp_stride)); 511 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 512 CeedCallBackend(impl->Apply(rstr, num_comp, block_size, comp_stride, 0, num_block, t_mode, true, true, u, v, request)); 513 return CEED_ERROR_SUCCESS; 514 } 515 516 //------------------------------------------------------------------------------ 517 // ElemRestriction Apply Unsigned 518 //------------------------------------------------------------------------------ 519 static int CeedElemRestrictionApplyUnsigned_Memcheck(CeedElemRestriction rstr, CeedTransposeMode t_mode, CeedVector u, CeedVector v, 520 CeedRequest *request) { 521 CeedInt num_block, block_size, num_comp, comp_stride; 522 CeedElemRestriction_Memcheck *impl; 523 524 CeedCallBackend(CeedElemRestrictionGetNumBlocks(rstr, &num_block)); 525 CeedCallBackend(CeedElemRestrictionGetBlockSize(rstr, &block_size)); 526 CeedCallBackend(CeedElemRestrictionGetNumComponents(rstr, &num_comp)); 527 CeedCallBackend(CeedElemRestrictionGetCompStride(rstr, &comp_stride)); 528 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 529 CeedCallBackend(impl->Apply(rstr, num_comp, block_size, comp_stride, 0, num_block, t_mode, false, true, u, v, request)); 530 return CEED_ERROR_SUCCESS; 531 } 532 533 //------------------------------------------------------------------------------ 534 // ElemRestriction Apply Unoriented 535 //------------------------------------------------------------------------------ 536 static int CeedElemRestrictionApplyUnoriented_Memcheck(CeedElemRestriction rstr, CeedTransposeMode t_mode, CeedVector u, CeedVector v, 537 CeedRequest *request) { 538 CeedInt num_block, block_size, num_comp, comp_stride; 539 CeedElemRestriction_Memcheck *impl; 540 541 CeedCallBackend(CeedElemRestrictionGetNumBlocks(rstr, &num_block)); 542 CeedCallBackend(CeedElemRestrictionGetBlockSize(rstr, &block_size)); 543 CeedCallBackend(CeedElemRestrictionGetNumComponents(rstr, &num_comp)); 544 CeedCallBackend(CeedElemRestrictionGetCompStride(rstr, &comp_stride)); 545 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 546 CeedCallBackend(impl->Apply(rstr, num_comp, block_size, comp_stride, 0, num_block, t_mode, false, false, u, v, request)); 547 return CEED_ERROR_SUCCESS; 548 } 549 550 //------------------------------------------------------------------------------ 551 // ElemRestriction Apply Points 552 //------------------------------------------------------------------------------ 553 static int CeedElemRestrictionApplyAtPointsInElement_Memcheck(CeedElemRestriction rstr, CeedInt elem, CeedTransposeMode t_mode, CeedVector u, 554 CeedVector v, CeedRequest *request) { 555 CeedInt num_comp; 556 CeedElemRestriction_Memcheck *impl; 557 558 CeedCallBackend(CeedElemRestrictionGetNumComponents(rstr, &num_comp)); 559 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 560 return impl->Apply(rstr, num_comp, 0, 1, elem, elem + 1, t_mode, false, false, u, v, request); 561 } 562 563 //------------------------------------------------------------------------------ 564 // ElemRestriction Apply Block 565 //------------------------------------------------------------------------------ 566 static int CeedElemRestrictionApplyBlock_Memcheck(CeedElemRestriction rstr, CeedInt block, CeedTransposeMode t_mode, CeedVector u, CeedVector v, 567 CeedRequest *request) { 568 CeedInt block_size, num_comp, comp_stride; 569 CeedElemRestriction_Memcheck *impl; 570 571 CeedCallBackend(CeedElemRestrictionGetBlockSize(rstr, &block_size)); 572 CeedCallBackend(CeedElemRestrictionGetNumComponents(rstr, &num_comp)); 573 CeedCallBackend(CeedElemRestrictionGetCompStride(rstr, &comp_stride)); 574 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 575 CeedCallBackend(impl->Apply(rstr, num_comp, block_size, comp_stride, block, block + 1, t_mode, true, true, u, v, request)); 576 return CEED_ERROR_SUCCESS; 577 } 578 579 //------------------------------------------------------------------------------ 580 // ElemRestriction Get Offsets 581 //------------------------------------------------------------------------------ 582 static int CeedElemRestrictionGetOffsets_Memcheck(CeedElemRestriction rstr, CeedMemType mem_type, const CeedInt **offsets) { 583 CeedElemRestriction_Memcheck *impl; 584 585 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 586 587 CeedCheck(mem_type == CEED_MEM_HOST, CeedElemRestrictionReturnCeed(rstr), CEED_ERROR_BACKEND, "Can only provide to HOST memory"); 588 589 *offsets = impl->offsets; 590 return CEED_ERROR_SUCCESS; 591 } 592 593 //------------------------------------------------------------------------------ 594 // ElemRestriction Get Orientations 595 //------------------------------------------------------------------------------ 596 static int CeedElemRestrictionGetOrientations_Memcheck(CeedElemRestriction rstr, CeedMemType mem_type, const bool **orients) { 597 CeedElemRestriction_Memcheck *impl; 598 599 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 600 601 CeedCheck(mem_type == CEED_MEM_HOST, CeedElemRestrictionReturnCeed(rstr), CEED_ERROR_BACKEND, "Can only provide to HOST memory"); 602 603 *orients = impl->orients; 604 return CEED_ERROR_SUCCESS; 605 } 606 607 //------------------------------------------------------------------------------ 608 // ElemRestriction Get Curl-Conforming Orientations 609 //------------------------------------------------------------------------------ 610 static int CeedElemRestrictionGetCurlOrientations_Memcheck(CeedElemRestriction rstr, CeedMemType mem_type, const CeedInt8 **curl_orients) { 611 CeedElemRestriction_Memcheck *impl; 612 613 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 614 615 CeedCheck(mem_type == CEED_MEM_HOST, CeedElemRestrictionReturnCeed(rstr), CEED_ERROR_BACKEND, "Can only provide to HOST memory"); 616 617 *curl_orients = impl->curl_orients; 618 return CEED_ERROR_SUCCESS; 619 } 620 621 //------------------------------------------------------------------------------ 622 // ElemRestriction Destroy 623 //------------------------------------------------------------------------------ 624 static int CeedElemRestrictionDestroy_Memcheck(CeedElemRestriction rstr) { 625 CeedElemRestriction_Memcheck *impl; 626 627 CeedCallBackend(CeedElemRestrictionGetData(rstr, &impl)); 628 CeedCallBackend(CeedFree(&impl->offsets_allocated)); 629 CeedCallBackend(CeedFree(&impl->orients_allocated)); 630 CeedCallBackend(CeedFree(&impl->curl_orients_allocated)); 631 CeedCallBackend(CeedFree(&impl)); 632 return CEED_ERROR_SUCCESS; 633 } 634 635 //------------------------------------------------------------------------------ 636 // ElemRestriction Create 637 //------------------------------------------------------------------------------ 638 int CeedElemRestrictionCreate_Memcheck(CeedMemType mem_type, CeedCopyMode copy_mode, const CeedInt *offsets, const bool *orients, 639 const CeedInt8 *curl_orients, CeedElemRestriction rstr) { 640 Ceed ceed; 641 CeedInt num_elem, elem_size, num_block, block_size, num_comp, comp_stride, num_points = 0, num_offsets; 642 CeedRestrictionType rstr_type; 643 CeedElemRestriction_Memcheck *impl; 644 645 CeedCallBackend(CeedElemRestrictionGetCeed(rstr, &ceed)); 646 CeedCallBackend(CeedElemRestrictionGetNumElements(rstr, &num_elem)); 647 CeedCallBackend(CeedElemRestrictionGetElementSize(rstr, &elem_size)); 648 CeedCallBackend(CeedElemRestrictionGetNumBlocks(rstr, &num_block)); 649 CeedCallBackend(CeedElemRestrictionGetBlockSize(rstr, &block_size)); 650 CeedCallBackend(CeedElemRestrictionGetNumComponents(rstr, &num_comp)); 651 CeedCallBackend(CeedElemRestrictionGetCompStride(rstr, &comp_stride)); 652 CeedCallBackend(CeedElemRestrictionGetType(rstr, &rstr_type)); 653 654 CeedCheck(mem_type == CEED_MEM_HOST, ceed, CEED_ERROR_BACKEND, "Only MemType = HOST supported"); 655 656 CeedCallBackend(CeedCalloc(1, &impl)); 657 CeedCallBackend(CeedElemRestrictionSetData(rstr, impl)); 658 659 // Set layouts 660 { 661 bool has_backend_strides; 662 CeedInt e_layout[3] = {1, elem_size, elem_size * num_comp}, l_layout[3] = {0}; 663 664 CeedCallBackend(CeedElemRestrictionSetELayout(rstr, e_layout)); 665 if (rstr_type == CEED_RESTRICTION_STRIDED) { 666 CeedCallBackend(CeedElemRestrictionHasBackendStrides(rstr, &has_backend_strides)); 667 if (has_backend_strides) { 668 CeedCallBackend(CeedElemRestrictionGetBackendStrides_Memcheck(rstr, l_layout)); 669 CeedCallBackend(CeedElemRestrictionSetLLayout(rstr, l_layout)); 670 } 671 } 672 } 673 674 // Offsets data 675 if (rstr_type != CEED_RESTRICTION_STRIDED) { 676 const char *resource; 677 678 // Check indices for ref or memcheck backends 679 { 680 Ceed current = ceed, parent = NULL; 681 682 CeedCallBackend(CeedGetParent(current, &parent)); 683 while (current != parent) { 684 current = parent; 685 CeedCallBackend(CeedGetParent(current, &parent)); 686 } 687 CeedCallBackend(CeedGetResource(parent, &resource)); 688 } 689 if (!strcmp(resource, "/cpu/self/ref/serial") || !strcmp(resource, "/cpu/self/ref/blocked") || !strcmp(resource, "/cpu/self/memcheck/serial") || 690 !strcmp(resource, "/cpu/self/memcheck/blocked")) { 691 CeedSize l_size; 692 693 CeedCallBackend(CeedElemRestrictionGetLVectorSize(rstr, &l_size)); 694 for (CeedInt i = 0; i < num_elem * elem_size; i++) { 695 CeedCheck(offsets[i] >= 0 && offsets[i] + (num_comp - 1) * comp_stride < l_size, ceed, CEED_ERROR_BACKEND, 696 "Restriction offset %" CeedInt_FMT " (%" CeedInt_FMT ") out of range [0, %" CeedInt_FMT "]", i, offsets[i], l_size); 697 } 698 } 699 700 // Copy data 701 if (rstr_type == CEED_RESTRICTION_POINTS) CeedCallBackend(CeedElemRestrictionGetNumPoints(rstr, &num_points)); 702 num_offsets = rstr_type == CEED_RESTRICTION_POINTS ? (num_elem + 1 + num_points) : (num_elem * elem_size); 703 switch (copy_mode) { 704 case CEED_COPY_VALUES: 705 CeedCallBackend(CeedMalloc(num_offsets, &impl->offsets_allocated)); 706 memcpy(impl->offsets_allocated, offsets, num_offsets * sizeof(offsets[0])); 707 impl->offsets = impl->offsets_allocated; 708 break; 709 case CEED_OWN_POINTER: 710 impl->offsets_allocated = (CeedInt *)offsets; 711 impl->offsets = impl->offsets_allocated; 712 break; 713 case CEED_USE_POINTER: 714 impl->offsets = offsets; 715 } 716 717 // Orientation data 718 if (rstr_type == CEED_RESTRICTION_ORIENTED) { 719 CeedCheck(orients != NULL, ceed, CEED_ERROR_BACKEND, "No orients array provided for oriented restriction"); 720 switch (copy_mode) { 721 case CEED_COPY_VALUES: 722 CeedCallBackend(CeedMalloc(num_offsets, &impl->orients_allocated)); 723 memcpy(impl->orients_allocated, orients, num_offsets * sizeof(orients[0])); 724 impl->orients = impl->orients_allocated; 725 break; 726 case CEED_OWN_POINTER: 727 impl->orients_allocated = (bool *)orients; 728 impl->orients = impl->orients_allocated; 729 break; 730 case CEED_USE_POINTER: 731 impl->orients = orients; 732 } 733 } else if (rstr_type == CEED_RESTRICTION_CURL_ORIENTED) { 734 CeedCheck(curl_orients != NULL, ceed, CEED_ERROR_BACKEND, "No curl_orients array provided for oriented restriction"); 735 switch (copy_mode) { 736 case CEED_COPY_VALUES: 737 CeedCallBackend(CeedMalloc(3 * num_offsets, &impl->curl_orients_allocated)); 738 memcpy(impl->curl_orients_allocated, curl_orients, 3 * num_offsets * sizeof(curl_orients[0])); 739 impl->curl_orients = impl->curl_orients_allocated; 740 break; 741 case CEED_OWN_POINTER: 742 impl->curl_orients_allocated = (CeedInt8 *)curl_orients; 743 impl->curl_orients = impl->curl_orients_allocated; 744 break; 745 case CEED_USE_POINTER: 746 impl->curl_orients = curl_orients; 747 } 748 } 749 } 750 751 // Set apply function 752 impl->Apply = CeedElemRestrictionApply_Memcheck_Core; 753 754 // Register backend functions 755 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "Apply", CeedElemRestrictionApply_Memcheck)); 756 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "ApplyUnsigned", CeedElemRestrictionApplyUnsigned_Memcheck)); 757 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "ApplyUnoriented", CeedElemRestrictionApplyUnoriented_Memcheck)); 758 if (rstr_type == CEED_RESTRICTION_POINTS) { 759 CeedCallBackend( 760 CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "ApplyAtPointsInElement", CeedElemRestrictionApplyAtPointsInElement_Memcheck)); 761 } 762 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "ApplyBlock", CeedElemRestrictionApplyBlock_Memcheck)); 763 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "GetOffsets", CeedElemRestrictionGetOffsets_Memcheck)); 764 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "GetOrientations", CeedElemRestrictionGetOrientations_Memcheck)); 765 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "GetCurlOrientations", CeedElemRestrictionGetCurlOrientations_Memcheck)); 766 CeedCallBackend(CeedSetBackendFunction(ceed, "ElemRestriction", rstr, "Destroy", CeedElemRestrictionDestroy_Memcheck)); 767 return CEED_ERROR_SUCCESS; 768 } 769 770 //------------------------------------------------------------------------------ 771