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
|
//------------------------------------------------------------------------------
// GB_dense_subassign_06d: C(:,:)<A> = A; C is full/bitmap, M and A are aliased
//------------------------------------------------------------------------------
// SuiteSparse:GraphBLAS, Timothy A. Davis, (c) 2017-2022, All Rights Reserved.
// SPDX-License-Identifier: Apache-2.0
//------------------------------------------------------------------------------
// Method 06d: C(:,:)<A> = A ; no S, C is dense, M and A are aliased
// M: present
// Mask_comp: false
// Mask_struct: true or false (both cases handled)
// C_replace: false
// accum: NULL
// A: matrix, and aliased to M
// S: none
// C must be bitmap or as-if-full. No entries are deleted and thus no zombies
// are introduced into C. C can be hypersparse, sparse, bitmap, or full, and
// its sparsity structure does not change. If C is hypersparse, sparse, or
// full, then the pattern does not change (all entries are present, and this
// does not change), and these cases can all be treated the same (as if full).
// If C is bitmap, new entries can be inserted into the bitmap C->b.
// C and A can have any sparsity structure.
#include "GB_subassign_methods.h"
#include "GB_dense.h"
#ifndef GBCUDA_DEV
#include "GB_type__include.h"
#endif
#undef GB_FREE_ALL
#define GB_FREE_ALL \
{ \
GB_WERK_POP (A_ek_slicing, int64_t) ; \
}
GrB_Info GB_dense_subassign_06d
(
GrB_Matrix C,
// input:
const GrB_Matrix A,
bool Mask_struct,
GB_Context Context
)
{
//--------------------------------------------------------------------------
// get inputs
//--------------------------------------------------------------------------
GrB_Info info ;
GB_WERK_DECLARE (A_ek_slicing, int64_t) ;
ASSERT_MATRIX_OK (C, "C for subassign method_06d", GB0) ;
ASSERT (!GB_ZOMBIES (C)) ;
ASSERT (!GB_JUMBLED (C)) ;
ASSERT (!GB_PENDING (C)) ;
ASSERT (GB_IS_BITMAP (C) || GB_as_if_full (C)) ;
ASSERT (!GB_aliased (C, A)) ; // NO ALIAS of C==A
ASSERT_MATRIX_OK (A, "A for subassign method_06d", GB0) ;
ASSERT (!GB_ZOMBIES (A)) ;
ASSERT (GB_JUMBLED_OK (A)) ;
ASSERT (!GB_PENDING (A)) ;
const GB_Type_code ccode = C->type->code ;
const bool C_is_bitmap = GB_IS_BITMAP (C) ;
const bool A_is_bitmap = GB_IS_BITMAP (A) ;
const bool A_is_dense = GB_as_if_full (A) ;
//--------------------------------------------------------------------------
// Method 06d: C(:,:)<A> = A ; no S; C is dense, M and A are aliased
//--------------------------------------------------------------------------
// Time: Optimal: the method must iterate over all entries in A,
// and the time is O(nnz(A)).
//--------------------------------------------------------------------------
// Parallel: slice A into equal-sized chunks
//--------------------------------------------------------------------------
GB_GET_NTHREADS_MAX (nthreads_max, chunk, Context) ;
//--------------------------------------------------------------------------
// slice the entries for each task
//--------------------------------------------------------------------------
int A_ntasks, A_nthreads ;
if (A_is_bitmap || A_is_dense)
{
// no need to construct tasks
int64_t anz = GB_nnz_held (A) ;
A_nthreads = GB_nthreads ((anz + A->nvec), 32*chunk, nthreads_max) ;
A_ntasks = (A_nthreads == 1) ? 1 : (8 * A_nthreads) ;
}
else
{
GB_SLICE_MATRIX (A, 8, 32*chunk) ;
}
//--------------------------------------------------------------------------
// C<A> = A for built-in types
//--------------------------------------------------------------------------
if (C->iso)
{
//----------------------------------------------------------------------
// C is iso
//----------------------------------------------------------------------
// Since C is iso, A must be iso (or effectively iso), which is also
// the mask M. An iso mask matrix M is converted into a structural
// mask by GB_get_mask, and thus Mask_struct must be true if C is iso.
ASSERT (Mask_struct) ;
#define GB_ISO_ASSIGN
#include "GB_dense_subassign_06d_template.c"
}
else
{
//----------------------------------------------------------------------
// C is non iso
//----------------------------------------------------------------------
bool done = false ;
#ifndef GBCUDA_DEV
//------------------------------------------------------------------
// define the worker for the switch factory
//------------------------------------------------------------------
#define GB_Cdense_06d(cname) GB (_Cdense_06d_ ## cname)
#define GB_WORKER(cname) \
{ \
info = GB_Cdense_06d(cname) (C, A, Mask_struct, \
A_ek_slicing, A_ntasks, A_nthreads) ; \
done = (info != GrB_NO_VALUE) ; \
} \
break ;
//------------------------------------------------------------------
// launch the switch factory
//------------------------------------------------------------------
if (C->type == A->type && ccode < GB_UDT_code)
{
// C<A> = A
switch (ccode)
{
case GB_BOOL_code : GB_WORKER (_bool )
case GB_INT8_code : GB_WORKER (_int8 )
case GB_INT16_code : GB_WORKER (_int16 )
case GB_INT32_code : GB_WORKER (_int32 )
case GB_INT64_code : GB_WORKER (_int64 )
case GB_UINT8_code : GB_WORKER (_uint8 )
case GB_UINT16_code : GB_WORKER (_uint16)
case GB_UINT32_code : GB_WORKER (_uint32)
case GB_UINT64_code : GB_WORKER (_uint64)
case GB_FP32_code : GB_WORKER (_fp32 )
case GB_FP64_code : GB_WORKER (_fp64 )
case GB_FC32_code : GB_WORKER (_fc32 )
case GB_FC64_code : GB_WORKER (_fc64 )
default: ;
}
}
#endif
//----------------------------------------------------------------------
// C<A> = A for user-defined types, and typecasting
//----------------------------------------------------------------------
if (!done)
{
//------------------------------------------------------------------
// get operators, functions, workspace, contents of A and C
//------------------------------------------------------------------
GB_BURBLE_MATRIX (A, "(generic C(:,:)<Z>=Z assign) ") ;
const size_t csize = C->type->size ;
const size_t asize = A->type->size ;
const GB_Type_code acode = A->type->code ;
GB_cast_function cast_A_to_C = GB_cast_factory (ccode, acode) ;
// Cx [p] = (ctype) Ax [pA]
#define GB_COPY_A_TO_C(Cx,p,Ax,pA,A_iso) \
cast_A_to_C (Cx + ((p)*csize), \
Ax + (A_iso ? 0:(pA)*asize), asize)
#define GB_AX_MASK(Ax,pA,asize) \
GB_mcast (Ax, pA, asize)
#define GB_CTYPE GB_void
#define GB_ATYPE GB_void
// no vectorization
#define GB_PRAGMA_SIMD_VECTORIZE ;
#undef GB_PRAGMA_SIMD_REDUCTION
#define GB_PRAGMA_SIMD_REDUCTION(op,s) ;
#include "GB_dense_subassign_06d_template.c"
}
}
//--------------------------------------------------------------------------
// free workspace and return result
//--------------------------------------------------------------------------
GB_FREE_ALL ;
ASSERT_MATRIX_OK (C, "C output for subassign method_06d", GB0) ;
return (GrB_SUCCESS) ;
}
|
