libceed-sys 0.12.0

Low-level bindings for libCEED library.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138

// Copyright (c) 2017-2022, Lawrence Livermore National Security, LLC and other CEED contributors.
// All Rights Reserved. See the top-level LICENSE and NOTICE files for details.
//
// SPDX-License-Identifier: BSD-2-Clause
//
// This file is part of CEED:  http://github.com/ceed

#include "../kernel-defines.hpp"

const char *occa_simplex_basis_gpu_source = STRINGIFY_SOURCE(

    @directive("#if TRANSPOSE") typedef CeedScalar * dofArray @dim(Q, elementCount, BASIS_COMPONENT_COUNT, DIM);
    typedef CeedScalar * quadArray @dim(P, BASIS_COMPONENT_COUNT, elementCount, DIM);
    @directive("#else") typedef CeedScalar * dofArray @dim(P, BASIS_COMPONENT_COUNT, elementCount, DIM);
    typedef CeedScalar * quadArray @dim(Q, elementCount, BASIS_COMPONENT_COUNT, DIM); @directive("#endif")

                                                                                          typedef CeedScalar *
                                                                                      quadToDof @dim(P, Q);
    typedef CeedScalar * dQuadToDof @dim(P, Q, DIM); typedef CeedScalar * elementWeightArray @dim(Q, elementCount);

    @kernel void interp(const CeedInt elementCount, const quadToDof B, const dofArray U, quadArray V) {
      for (int elementOffset = 0; elementOffset < elementCount; elementOffset += ELEMENTS_PER_BLOCK; @outer) {
        @shared CeedScalar s_B[P * Q] @dim(P, Q);

        // Store weights in shared memory
        for (int i = 0; i < MAX_PQ; ++i; @inner) {
          for (int j = i; j < (P * Q); j += MAX_PQ) {
            s_B[j] = B[j];
          }
        }

        for (int localElement = 0; localElement < ELEMENTS_PER_BLOCK; ++localElement) {
          for (int i = 0; i < MAX_PQ; ++i; @inner) {
            const int element = elementOffset + localElement;
            if (element < elementCount) {
              // Element operation
              for (int component = 0; component < BASIS_COMPONENT_COUNT; ++component) {
                if (!TRANSPOSE) {
                  const int q = i;
                  if (q < Q) {
                    CeedScalar v = 0;
                    for (int p = 0; p < P; ++p) {
                      v += s_B(p, q) * U(p, component, element, 0);
                    }
                    V(q, element, component, 0) = v;
                  }
                } else {
                  const int p = i;
                  if (p < P) {
                    CeedScalar v = 0;
                    for (int q = 0; q < Q; ++q) {
                      v += s_B(p, q) * U(q, element, component, 0);
                    }
                    V(p, component, element, 0) = v;
                  }
                }
              }
            }
          }
        }
      }
    }

    @kernel void grad(const CeedInt elementCount, const dQuadToDof Bx, const dofArray U, quadArray V) {
      for (int elementOffset = 0; elementOffset < elementCount; elementOffset += ELEMENTS_PER_BLOCK; @outer) {
        @shared CeedScalar s_Bx[Q * P * DIM] @dim(P, Q, DIM);

        // Store weights in shared memory
        for (int i = 0; i < MAX_PQ; ++i; @inner) {
          for (int j = i; j < (P * Q * DIM); j += MAX_PQ) {
            s_Bx[j] = Bx[j];
          }
        }

        for (int localElement = 0; localElement < ELEMENTS_PER_BLOCK; ++localElement) {
          for (int i = 0; i < MAX_PQ; ++i; @inner) {
            const int element = elementOffset + localElement;
            if (element < elementCount) {
              // Element operation
              for (int component = 0; component < BASIS_COMPONENT_COUNT; ++component) {
                if (!TRANSPOSE) {
                  const int q = i;
                  if (q < Q) {
                    CeedScalar v[DIM];
                    for (int dim = 0; dim < DIM; ++dim) {
                      v[dim] = 0;
                    }

                    for (int p = 0; p < P; ++p) {
                      const CeedScalar u = U(p, component, element, 0);
                      for (int dim = 0; dim < DIM; ++dim) {
                        v[dim] += s_Bx(p, q, dim) * u;
                      }
                    }

                    for (int dim = 0; dim < DIM; ++dim) {
                      V(q, element, component, dim) = v[dim];
                    }
                  }
                } else {
                  const int p = i;
                  if (p < P) {
                    CeedScalar v = 0;
                    for (int dim = 0; dim < DIM; ++dim) {
                      for (int q = 0; q < Q; ++q) {
                        v += s_Bx(p, q, dim) * U(q, element, component, dim);
                      }
                    }
                    V(p, component, element, 0) = v;
                  }
                }
              }
            }
          }
        }
      }
    }

    @kernel void weight(const CeedInt elementCount, const CeedScalar *qWeights, elementWeightArray W) {
      for (int elementOffset = 0; elementOffset < elementCount; elementOffset += ELEMENTS_PER_BLOCK; @outer) {
        @shared CeedScalar s_qWeights[Q];

        for (int q = 0; q < Q; ++q; @inner) {
          s_qWeights[q] = qWeights[q];
        }

        for (int localElement = 0; localElement < ELEMENTS_PER_BLOCK; ++localElement) {
          const int element = elementOffset + localElement;
          if (element < elementCount) {
            for (int q = 0; q < Q; ++q; @inner) {
              W(q, element) = s_qWeights[q];
            }
          }
        }
      }
    }

);