File: GB_subassign_01.c

package info (click to toggle)
suitesparse-graphblas 7.4.0%2Bdfsg-2
  • links: PTS, VCS
  • area: main
  • in suites: forky, sid, trixie
  • size: 67,112 kB
  • sloc: ansic: 1,072,243; cpp: 8,081; sh: 512; makefile: 506; asm: 369; python: 125; awk: 10
file content (220 lines) | stat: -rw-r--r-- 8,053 bytes parent folder | download | duplicates (2)
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
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
//------------------------------------------------------------------------------
// GB_subassign_01: C(I,J) = scalar ; using S
//------------------------------------------------------------------------------

// SuiteSparse:GraphBLAS, Timothy A. Davis, (c) 2017-2022, All Rights Reserved.
// SPDX-License-Identifier: Apache-2.0

//------------------------------------------------------------------------------

// Method 01: C(I,J) = scalar ; using S

// M:           NULL
// Mask_comp:   false
// C_replace:   false
// accum:       NULL
// A:           scalar
// S:           constructed

// C: not bitmap

#include "GB_subassign_methods.h"

GrB_Info GB_subassign_01
(
    GrB_Matrix C,
    // input:
    const GrB_Index *I,
    const int64_t ni,
    const int64_t nI,
    const int Ikind,
    const int64_t Icolon [3],
    const GrB_Index *J,
    const int64_t nj,
    const int64_t nJ,
    const int Jkind,
    const int64_t Jcolon [3],
    const void *scalar,
    const GrB_Type atype,
    GB_Context Context
)
{

    //--------------------------------------------------------------------------
    // check inputs
    //--------------------------------------------------------------------------

    ASSERT (!GB_IS_BITMAP (C)) ;

    //--------------------------------------------------------------------------
    // S = C(I,J)
    //--------------------------------------------------------------------------

    GB_EMPTY_TASKLIST ;
    GB_CLEAR_STATIC_HEADER (S, &S_header) ;
    GB_OK (GB_subassign_symbolic (S, C, I, ni, J, nj, true, Context)) ;

    //--------------------------------------------------------------------------
    // get inputs
    //--------------------------------------------------------------------------

    GB_GET_C ;      // C must not be bitmap
    const int64_t *restrict Ch = C->h ;
    const int64_t *restrict Cp = C->p ;
    const bool C_is_hyper = (Ch != NULL) ;
    const int64_t Cnvec = C->nvec ;
    GB_GET_SCALAR ;
    GB_GET_S ;
    GrB_BinaryOp accum = NULL ;

    //--------------------------------------------------------------------------
    // Method 01: C(I,J) = scalar ; using S
    //--------------------------------------------------------------------------

    // Time: Optimal; must visit all IxJ, so Omega(|I|*|J|) is required.

    // Entries in S are found and the corresponding entry in C replaced with
    // the scalar.  The traversal of S is identical to the traversal of M in
    // Method 4.

    // Method 01 and Method 03 are very similar.

    //--------------------------------------------------------------------------
    // Parallel: all IxJ (Methods 01, 03, 13, 15, 17, 19)
    //--------------------------------------------------------------------------

    GB_SUBASSIGN_IXJ_SLICE ;

    //--------------------------------------------------------------------------
    // phase 1: create zombies, update entries, and count pending tuples
    //--------------------------------------------------------------------------

    #pragma omp parallel for num_threads(nthreads) schedule(dynamic,1) \
        reduction(+:nzombies)
    for (taskid = 0 ; taskid < ntasks ; taskid++)
    {

        //----------------------------------------------------------------------
        // get the task descriptor
        //----------------------------------------------------------------------

        GB_GET_IXJ_TASK_DESCRIPTOR_PHASE1 (iA_start, iA_end) ;

        //----------------------------------------------------------------------
        // compute all vectors in this task
        //----------------------------------------------------------------------

        for (int64_t j = kfirst ; j <= klast ; j++)
        {

            //------------------------------------------------------------------
            // get jC, the corresponding vector of C
            //------------------------------------------------------------------

            int64_t jC = GB_ijlist (J, j, Jkind, Jcolon) ;

            //------------------------------------------------------------------
            // get S(iA_start:end,j)
            //------------------------------------------------------------------

            GB_LOOKUP_VECTOR_FOR_IXJ (S, iA_start) ;

            //------------------------------------------------------------------
            // C(I(iA_start,iA_end-1),jC) = scalar
            //------------------------------------------------------------------

            for (int64_t iA = iA_start ; iA < iA_end ; iA++)
            {
                bool found = (pS < pS_end) && (GBI (Si, pS, Svlen) == iA) ;
                if (!found)
                { 
                    // ----[. A 1]----------------------------------------------
                    // S (i,j) is not present, the scalar is present
                    // [. A 1]: action: ( insert )
                    task_pending++ ;
                }
                else
                { 
                    // ----[C A 1] or [X A 1]-----------------------------------
                    // both S (i,j) and A (i,j) present
                    // [C A 1]: action: ( =A ): scalar to C, no accum
                    // [X A 1]: action: ( undelete ): zombie lives
                    GB_C_S_LOOKUP ;
                    GB_noaccum_C_A_1_scalar ;
                    GB_NEXT (S) ;
                }
            }
        }

        GB_PHASE1_TASK_WRAPUP ;
    }

    //--------------------------------------------------------------------------
    // phase 2: insert pending tuples
    //--------------------------------------------------------------------------

    GB_PENDING_CUMSUM ;

    #pragma omp parallel for num_threads(nthreads) schedule(dynamic,1) \
        reduction(&&:pending_sorted)
    for (taskid = 0 ; taskid < ntasks ; taskid++)
    {

        //----------------------------------------------------------------------
        // get the task descriptor
        //----------------------------------------------------------------------

        GB_GET_IXJ_TASK_DESCRIPTOR_PHASE2 (iA_start, iA_end) ;

        //----------------------------------------------------------------------
        // compute all vectors in this task
        //----------------------------------------------------------------------

        for (int64_t j = kfirst ; j <= klast ; j++)
        {

            //------------------------------------------------------------------
            // get jC, the corresponding vector of C
            //------------------------------------------------------------------

            int64_t jC = GB_ijlist (J, j, Jkind, Jcolon) ;

            //------------------------------------------------------------------
            // get S(iA_start:end,j)
            //------------------------------------------------------------------

            GB_LOOKUP_VECTOR_FOR_IXJ (S, iA_start) ;

            //------------------------------------------------------------------
            // C(I(iA_start,iA_end-1),jC) = scalar
            //------------------------------------------------------------------

            for (int64_t iA = iA_start ; iA < iA_end ; iA++)
            {
                bool found = (pS < pS_end) && (GBI (Si, pS, Svlen) == iA) ;
                if (!found)
                { 
                    // ----[. A 1]----------------------------------------------
                    // S (i,j) is not present, the scalar is present
                    // [. A 1]: action: ( insert )
                    int64_t iC = GB_ijlist (I, iA, Ikind, Icolon) ;
                    GB_PENDING_INSERT (scalar) ;
                }
                else
                { 
                    // both S (i,j) and A (i,j) present
                    GB_NEXT (S) ;
                }
            }
        }

        GB_PHASE2_TASK_WRAPUP ;
    }

    //--------------------------------------------------------------------------
    // finalize the matrix and return result
    //--------------------------------------------------------------------------

    GB_SUBASSIGN_WRAPUP ;
}