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
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
|
//------------------------------------------------------------------------------
// GB_AxB_dot3: compute C<M> = A'*B in parallel
//------------------------------------------------------------------------------
// SuiteSparse:GraphBLAS, Timothy A. Davis, (c) 2017-2022, All Rights Reserved.
// SPDX-License-Identifier: Apache-2.0
//------------------------------------------------------------------------------
// This function only computes C<M>=A'*B. The mask must be present, and not
// complemented, and can be either valued or structural. The mask is always
// applied. C and M are both sparse or hypersparse, and have the same sparsity
// structure.
#include "GB_mxm.h"
#include "GB_binop.h"
#include "GB_AxB__include1.h"
#ifndef GBCUDA_DEV
#include "GB_AxB__include2.h"
#endif
#include "GB_unused.h"
#define GB_FREE_WORKSPACE \
{ \
GB_FREE_WORK (&TaskList, TaskList_size) ; \
}
#define GB_FREE_ALL \
{ \
GB_FREE_WORKSPACE ; \
GB_phybix_free (C) ; \
}
GB_PUBLIC
GrB_Info GB_AxB_dot3 // C<M> = A'*B using dot product method
(
GrB_Matrix C, // output matrix, static header
const bool C_iso, // true if C is iso
const GB_void *cscalar, // iso value of C
const GrB_Matrix M, // mask matrix
const bool Mask_struct, // if true, use the only structure of M
const GrB_Matrix A, // input matrix
const GrB_Matrix B, // input matrix
const GrB_Semiring semiring, // semiring that defines C=A*B
const bool flipxy, // if true, do z=fmult(b,a) vs fmult(a,b)
GB_Context Context
)
{
//--------------------------------------------------------------------------
// check inputs
//--------------------------------------------------------------------------
GrB_Info info ;
ASSERT (C != NULL && (C->static_header || GBNSTATIC)) ;
ASSERT_MATRIX_OK (M, "M for dot3 A'*B", GB0) ;
ASSERT_MATRIX_OK (A, "A for dot3 A'*B", GB0) ;
ASSERT_MATRIX_OK (B, "B for dot3 A'*B", GB0) ;
ASSERT (!GB_ZOMBIES (M)) ;
ASSERT (GB_JUMBLED_OK (M)) ; // C is jumbled if M is jumbled
ASSERT (!GB_PENDING (M)) ;
ASSERT (!GB_ZOMBIES (A)) ;
ASSERT (!GB_JUMBLED (A)) ;
ASSERT (!GB_PENDING (A)) ;
ASSERT (!GB_ZOMBIES (B)) ;
ASSERT (!GB_JUMBLED (B)) ;
ASSERT (!GB_PENDING (B)) ;
ASSERT (!GB_IS_BITMAP (M)) ;
ASSERT (!GB_IS_FULL (M)) ;
ASSERT_SEMIRING_OK (semiring, "semiring for numeric A'*B", GB0) ;
int ntasks, nthreads ;
GB_task_struct *TaskList = NULL ; size_t TaskList_size = 0 ;
//--------------------------------------------------------------------------
// get the semiring operators
//--------------------------------------------------------------------------
GrB_BinaryOp mult = semiring->multiply ;
GrB_Monoid add = semiring->add ;
ASSERT (mult->ztype == add->op->ztype) ;
bool op_is_first = mult->opcode == GB_FIRST_binop_code ;
bool op_is_second = mult->opcode == GB_SECOND_binop_code ;
bool op_is_pair = mult->opcode == GB_PAIR_binop_code ;
bool A_is_pattern = false ;
bool B_is_pattern = false ;
if (flipxy)
{
// z = fmult (b,a) will be computed
A_is_pattern = op_is_first || op_is_pair ;
B_is_pattern = op_is_second || op_is_pair ;
ASSERT (GB_IMPLIES (!A_is_pattern,
GB_Type_compatible (A->type, mult->ytype))) ;
ASSERT (GB_IMPLIES (!B_is_pattern,
GB_Type_compatible (B->type, mult->xtype))) ;
}
else
{
// z = fmult (a,b) will be computed
A_is_pattern = op_is_second || op_is_pair ;
B_is_pattern = op_is_first || op_is_pair ;
ASSERT (GB_IMPLIES (!A_is_pattern,
GB_Type_compatible (A->type, mult->xtype))) ;
ASSERT (GB_IMPLIES (!B_is_pattern,
GB_Type_compatible (B->type, mult->ytype))) ;
}
//--------------------------------------------------------------------------
// get M, A, and B
//--------------------------------------------------------------------------
const int64_t *restrict Mp = M->p ;
const int64_t *restrict Mh = M->h ;
const int64_t *restrict Mi = M->i ;
const GB_void *restrict Mx = (GB_void *) (Mask_struct ? NULL : (M->x)) ;
const size_t msize = M->type->size ;
const int64_t mvlen = M->vlen ;
const int64_t mvdim = M->vdim ;
const int64_t mnz = GB_nnz (M) ;
const int64_t mnvec = M->nvec ;
const bool M_is_hyper = GB_IS_HYPERSPARSE (M) ;
const bool M_is_sparse = GB_IS_SPARSE (M) ;
const int64_t *restrict Ap = A->p ;
const int64_t *restrict Ah = A->h ;
const int64_t vlen = A->vlen ;
const int64_t anvec = A->nvec ;
const bool A_is_hyper = GB_IS_HYPERSPARSE (A) ;
const bool A_is_sparse = GB_IS_SPARSE (A) ;
const bool A_is_bitmap = GB_IS_BITMAP (A) ;
const int64_t *restrict Bp = B->p ;
const int64_t *restrict Bh = B->h ;
const int64_t bnvec = B->nvec ;
const bool B_is_hyper = GB_IS_HYPERSPARSE (B) ;
const bool B_is_sparse = GB_IS_SPARSE (B) ;
const bool B_is_bitmap = GB_IS_BITMAP (B) ;
ASSERT (A->vlen == B->vlen) ;
ASSERT (vlen > 0) ;
const GrB_Matrix A_Y = A->Y ;
const int64_t *restrict A_Yp = (A_is_hyper) ? A_Y->p : NULL ;
const int64_t *restrict A_Yi = (A_is_hyper) ? A_Y->i : NULL ;
const int64_t *restrict A_Yx = (A_is_hyper) ? A_Y->x : NULL ;
const int64_t A_hash_bits = (A_is_hyper) ? (A_Y->vdim - 1) : 0 ;
const GrB_Matrix B_Y = B->Y ;
const int64_t *restrict B_Yp = (B_is_hyper) ? B_Y->p : NULL ;
const int64_t *restrict B_Yi = (B_is_hyper) ? B_Y->i : NULL ;
const int64_t *restrict B_Yx = (B_is_hyper) ? B_Y->x : NULL ;
const int64_t B_hash_bits = (B_is_hyper) ? (B_Y->vdim - 1) : 0 ;
//--------------------------------------------------------------------------
// allocate C, the same size and # of entries as M
//--------------------------------------------------------------------------
GrB_Type ctype = add->op->ztype ;
int64_t cvlen = mvlen ;
int64_t cvdim = mvdim ;
int64_t cnz = mnz ;
int64_t cnvec = mnvec ;
int C_sparsity = (M_is_hyper) ? GxB_HYPERSPARSE : GxB_SPARSE ;
// C is sparse or hypersparse, not full or bitmap
// set C->iso = C_iso OK
GB_OK (GB_new_bix (&C, // sparse or hyper (from M), existing header
ctype, cvlen, cvdim, GB_Ap_malloc, true,
C_sparsity, true, M->hyper_switch, cnvec,
cnz+1, // add one to cnz for GB_cumsum of Cwork in GB_AxB_dot3_slice
true, C_iso, Context)) ;
int64_t *restrict Cp = C->p ;
int64_t *restrict Ch = C->h ;
int64_t *restrict Cwork = C->i ; // use C->i as workspace
//--------------------------------------------------------------------------
// determine the # of threads to use
//--------------------------------------------------------------------------
GB_GET_NTHREADS_MAX (nthreads_max, chunk, Context) ;
//--------------------------------------------------------------------------
// copy Mp and Mh into C
//--------------------------------------------------------------------------
// M is sparse or hypersparse; C is the same as M
nthreads = GB_nthreads (cnvec, chunk, nthreads_max) ;
// TODO: try this with Cp and Ch shallow
GB_memcpy (Cp, Mp, (cnvec+1) * sizeof (int64_t), nthreads) ;
if (M_is_hyper)
{
GB_memcpy (Ch, Mh, cnvec * sizeof (int64_t), nthreads) ;
}
C->nvec_nonempty = M->nvec_nonempty ;
C->nvec = M->nvec ;
C->nvals = M->nvals ;
C->magic = GB_MAGIC ;
//--------------------------------------------------------------------------
// construct the tasks for the first phase
//--------------------------------------------------------------------------
nthreads = GB_nthreads (cnz, chunk, nthreads_max) ;
GB_OK (GB_AxB_dot3_one_slice (&TaskList, &TaskList_size, &ntasks, &nthreads,
M, Context)) ;
//--------------------------------------------------------------------------
// phase1: estimate the work to compute each entry in C
//--------------------------------------------------------------------------
// The work to compute C(i,j) is held in Cwork [p], if C(i,j) appears in
// as the pth entry in C.
#define GB_DOT3
#define GB_DOT3_PHASE1
if (M_is_sparse && Mask_struct)
{
// special case: M is sparse and structural
#define GB_MASK_SPARSE_AND_STRUCTURAL
#include "GB_meta16_factory.c"
#undef GB_MASK_SPARSE_AND_STRUCTURAL
// TODO: skip phase1 if A and B are both bitmap/full.
}
else
{
// general case: M sparse/hyper, structural/valued
#include "GB_meta16_factory.c"
}
#undef GB_DOT3
#undef GB_DOT3_PHASE1
//--------------------------------------------------------------------------
// free the current tasks and construct the tasks for the second phase
//--------------------------------------------------------------------------
GB_FREE_WORK (&TaskList, TaskList_size) ;
GB_OK (GB_AxB_dot3_slice (&TaskList, &TaskList_size, &ntasks, &nthreads,
C, Context)) ;
GBURBLE ("nthreads %d ntasks %d ", nthreads, ntasks) ;
//--------------------------------------------------------------------------
// C<M> = A'*B, via masked dot product method and built-in semiring
//--------------------------------------------------------------------------
if (C_iso)
{
//----------------------------------------------------------------------
// C is iso; compute the pattern of C<M>=A'*B with the any_pair semiring
//----------------------------------------------------------------------
memcpy (C->x, cscalar, ctype->size) ;
GB_OK (GB (_Adot3B__any_pair_iso) (C, M, Mask_struct, A, B,
TaskList, ntasks, nthreads)) ;
}
else
{
//----------------------------------------------------------------------
// C is non-iso
//----------------------------------------------------------------------
bool done = false ;
#ifndef GBCUDA_DEV
//------------------------------------------------------------------
// define the worker for the switch factory
//------------------------------------------------------------------
#define GB_Adot3B(add,mult,xname) \
GB (_Adot3B_ ## add ## mult ## xname)
#define GB_AxB_WORKER(add,mult,xname) \
{ \
info = GB_Adot3B (add,mult,xname) (C, M, Mask_struct, A, B, \
TaskList, ntasks, nthreads) ; \
done = (info != GrB_NO_VALUE) ; \
} \
break ;
//------------------------------------------------------------------
// launch the switch factory
//------------------------------------------------------------------
GB_Opcode mult_binop_code, add_binop_code ;
GB_Type_code xcode, ycode, zcode ;
if (GB_AxB_semiring_builtin (A, A_is_pattern, B, B_is_pattern,
semiring, flipxy, &mult_binop_code, &add_binop_code, &xcode,
&ycode, &zcode))
{
#include "GB_AxB_factory.c"
}
#endif
//----------------------------------------------------------------------
// C<M> = A'*B, via masked dot product method and typecasting
//----------------------------------------------------------------------
if (!done)
{
#define GB_DOT3_GENERIC
GB_BURBLE_MATRIX (C, "(generic C<M>=A'*B) ") ;
#include "GB_AxB_dot_generic.c"
}
}
//--------------------------------------------------------------------------
// free workspace and return result
//--------------------------------------------------------------------------
GB_FREE_WORKSPACE ;
C->jumbled = GB_JUMBLED (M) ; // C is jumbled if M is jumbled
ASSERT_MATRIX_OK (C, "dot3: C<M> = A'*B output", GB0) ;
ASSERT (GB_ZOMBIES_OK (C)) ;
ASSERT (GB_JUMBLED_OK (C)) ;
ASSERT (!GB_PENDING (C)) ;
return (GrB_SUCCESS) ;
}
|
