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
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
|
//------------------------------------------------------------------------------
// GB_emult_02_phase1: C = A.*B where A is sparse/hyper and B is bitmap/full
//------------------------------------------------------------------------------
// SuiteSparse:GraphBLAS, Timothy A. Davis, (c) 2017-2025, All Rights Reserved.
// SPDX-License-Identifier: Apache-2.0
//------------------------------------------------------------------------------
// JIT: not needed: factory cases: mask types, M bitmap/full, B bitmap/full,
// A sparse/hyper
// Symbolic analysis phase for GB_emult_02 and GB_emult_03.
#define GB_FREE_ALL ;
#include "ewise/GB_ewise.h"
#include "emult/GB_emult.h"
#include "binaryop/GB_binop.h"
#include "jitifyer/GB_stringify.h"
#include "slice/factory/GB_ek_slice_merge.h"
GrB_Info GB_emult_02_phase1 // symbolic analysis for GB_emult_02 and GB_emult_03
(
// input/output:
GrB_Matrix C,
// input:
const bool C_iso,
const GrB_Matrix M,
const bool Mask_struct,
const bool Mask_comp,
const GrB_Matrix A,
const GrB_Matrix B,
const int64_t *restrict A_ek_slicing,
const int A_ntasks,
const int A_nthreads,
// workspace:
uint64_t *restrict Wfirst,
uint64_t *restrict Wlast,
// output:
uint64_t *Cp_kfirst,
GB_Werk Werk
)
{
//--------------------------------------------------------------------------
// get C, M, A, and B
//--------------------------------------------------------------------------
ASSERT (GB_IS_SPARSE (A) || GB_IS_HYPERSPARSE (A)) ;
ASSERT (GB_IS_BITMAP (B) || GB_IS_FULL (B)) ;
ASSERT ((M == NULL) || GB_IS_BITMAP (M) || GB_IS_FULL (M)) ;
GrB_Info info ;
const int8_t *restrict Mb = (M == NULL) ? NULL : M->b ;
const GB_M_TYPE *restrict Mx = (M == NULL || Mask_struct) ? NULL :
(const GB_M_TYPE *) M->x ;
const size_t msize = (M == NULL) ? 0 : M->type->size ;
GB_Ap_DECLARE (Ap, const) ; GB_Ap_PTR (Ap, A) ;
GB_Ah_DECLARE (Ah, const) ; GB_Ah_PTR (Ah, A) ;
GB_Ai_DECLARE (Ai, const) ; GB_Ai_PTR (Ai, A) ;
const int64_t vlen = A->vlen ;
const int64_t nvec = A->nvec ;
const int64_t anz = GB_nnz (A) ;
const int8_t *restrict Bb = B->b ;
const bool B_is_bitmap = GB_IS_BITMAP (B) ;
GB_Cp_DECLARE (Cp, ) ; GB_Cp_PTR (Cp, C) ;
const bool Cp_is_32 = C->p_is_32 ;
const bool Cj_is_32 = C->j_is_32 ;
const bool Ci_is_32 = C->i_is_32 ;
ASSERT (C->p_is_32 == A->p_is_32) ;
ASSERT (C->j_is_32 == A->j_is_32) ;
ASSERT (C->i_is_32 == A->i_is_32) ;
const int64_t *restrict kfirst_Aslice = A_ek_slicing ;
const int64_t *restrict klast_Aslice = A_ek_slicing + A_ntasks ;
const int64_t *restrict pstart_Aslice = A_ek_slicing + A_ntasks * 2 ;
//--------------------------------------------------------------------------
// count entries in C
//--------------------------------------------------------------------------
// C->nvec_nonempty = A->nvec_nonempty ;
GB_nvec_nonempty_set (C, GB_nvec_nonempty_get (A)) ;
C->nvec = nvec ;
const bool C_has_pattern_of_A = !B_is_bitmap && (M == NULL) ;
if (!C_has_pattern_of_A)
{
// This phase is very similar to GB_select_entry_phase1_template.c.
if (M == NULL)
{
//------------------------------------------------------------------
// Method2/3(a): C = A.*B where A is sparse/hyper and B is bitmap
//------------------------------------------------------------------
ASSERT (GB_IS_BITMAP (B)) ;
int tid ;
#pragma omp parallel for num_threads(A_nthreads) schedule(dynamic,1)
for (tid = 0 ; tid < A_ntasks ; tid++)
{
int64_t kfirst = kfirst_Aslice [tid] ;
int64_t klast = klast_Aslice [tid] ;
Wfirst [tid] = 0 ;
Wlast [tid] = 0 ;
for (int64_t k = kfirst ; k <= klast ; k++)
{
// count the entries in C(:,j)
int64_t j = GBh_A (Ah, k) ;
int64_t pB_start = j * vlen ;
GB_GET_PA (pA, pA_end, tid, k, kfirst, klast, pstart_Aslice,
GB_IGET (Ap, k), GB_IGET (Ap, k+1)) ;
int64_t cjnz = 0 ;
for ( ; pA < pA_end ; pA++)
{
cjnz += Bb [pB_start + GB_IGET (Ai, pA)] ;
}
if (k == kfirst)
{
Wfirst [tid] = cjnz ;
}
else if (k == klast)
{
Wlast [tid] = cjnz ;
}
else
{
GB_ISET (Cp, k, cjnz) ; // Cp [k] = cjnz ;
}
}
}
}
else
{
//------------------------------------------------------------------
// Method2/3(c): C<#M> = A.*B; A is sparse/hyper; M, B bitmap/full
//------------------------------------------------------------------
ASSERT (M != NULL) ;
ASSERT (GB_IS_BITMAP (M) || GB_IS_FULL (M)) ;
ASSERT (GB_IS_BITMAP (B) || GB_IS_FULL (B)) ;
int tid ;
#pragma omp parallel for num_threads(A_nthreads) schedule(dynamic,1)
for (tid = 0 ; tid < A_ntasks ; tid++)
{
int64_t kfirst = kfirst_Aslice [tid] ;
int64_t klast = klast_Aslice [tid] ;
Wfirst [tid] = 0 ;
Wlast [tid] = 0 ;
for (int64_t k = kfirst ; k <= klast ; k++)
{
// count the entries in C(:,j)
int64_t j = GBh_A (Ah, k) ;
int64_t pB_start = j * vlen ;
GB_GET_PA (pA, pA_end, tid, k, kfirst, klast, pstart_Aslice,
GB_IGET (Ap, k), GB_IGET (Ap, k+1)) ;
int64_t cjnz = 0 ;
for ( ; pA < pA_end ; pA++)
{
int64_t i = GB_IGET (Ai, pA) ;
int64_t pB = pB_start + i ;
bool mij = GBb_M (Mb, pB) && GB_MCAST (Mx, pB, msize) ;
mij = mij ^ Mask_comp ;
cjnz += (mij && GBb_M (Bb, pB)) ;
}
if (k == kfirst)
{
Wfirst [tid] = cjnz ;
}
else if (k == klast)
{
Wlast [tid] = cjnz ;
}
else
{
GB_ISET (Cp, k, cjnz) ; // Cp [k] = cjnz ;
}
}
}
}
//----------------------------------------------------------------------
// finalize Cp, cumulative sum of Cp and compute Cp_kfirst
//----------------------------------------------------------------------
GB_ek_slice_merge1 (Cp, Cp_is_32,
Wfirst, Wlast, A_ek_slicing, A_ntasks) ;
int64_t nvec_nonempty ;
GB_cumsum (Cp, Cp_is_32, nvec, &nvec_nonempty, A_nthreads, Werk) ;
GB_nvec_nonempty_set (C, nvec_nonempty) ;
GB_ek_slice_merge2 (Cp_kfirst, Cp, Cp_is_32,
Wfirst, Wlast, A_ek_slicing, A_ntasks) ;
}
//--------------------------------------------------------------------------
// allocate C->i and C->x
//--------------------------------------------------------------------------
int64_t cnz = (C_has_pattern_of_A) ? anz : GB_IGET (Cp, nvec) ;
GB_OK (GB_bix_alloc (C, cnz, GxB_SPARSE, false, true, C_iso)) ;
//--------------------------------------------------------------------------
// copy pattern into C
//--------------------------------------------------------------------------
// FUTURE: could make these components of C shallow instead of memcpy
size_t cpsize = Cp_is_32 ? sizeof (uint32_t) : sizeof (uint64_t) ;
size_t cjsize = Cj_is_32 ? sizeof (uint32_t) : sizeof (uint64_t) ;
size_t cisize = Ci_is_32 ? sizeof (uint32_t) : sizeof (uint64_t) ;
if (GB_IS_HYPERSPARSE (A))
{
// copy A->h into C->h
GB_memcpy (C->h, Ah, nvec * cjsize, A_nthreads) ;
}
if (C_has_pattern_of_A)
{
// Method2/3(b): B is full and no mask present, so the pattern of C is
// the same as the pattern of A
GB_memcpy (Cp, Ap, (nvec+1) * cpsize, A_nthreads) ;
GB_memcpy (C->i, Ai, cnz * cisize, A_nthreads) ;
}
C->nvals = cnz ;
C->jumbled = A->jumbled ;
C->magic = GB_MAGIC ;
return (GrB_SUCCESS) ;
}
|
