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
|
//------------------------------------------------------------------------------
// GB_dense_subassign_05d_template: C<M> = x where C is as-if-full
//------------------------------------------------------------------------------
// SuiteSparse:GraphBLAS, Timothy A. Davis, (c) 2017-2022, All Rights Reserved.
// SPDX-License-Identifier: Apache-2.0
//------------------------------------------------------------------------------
{
//--------------------------------------------------------------------------
// get C and M
//--------------------------------------------------------------------------
ASSERT (GB_JUMBLED_OK (M)) ;
ASSERT (!C->iso) ;
const int64_t *restrict Mp = M->p ;
const int8_t *restrict Mb = M->b ;
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 size_t mvlen = M->vlen ;
GB_CTYPE *restrict Cx = (GB_CTYPE *) C->x ;
const int64_t cvlen = C->vlen ;
const int64_t *restrict kfirst_Mslice = M_ek_slicing ;
const int64_t *restrict klast_Mslice = M_ek_slicing + M_ntasks ;
const int64_t *restrict pstart_Mslice = M_ek_slicing + M_ntasks * 2 ;
//--------------------------------------------------------------------------
// C<M> = x
//--------------------------------------------------------------------------
int taskid ;
#pragma omp parallel for num_threads(M_nthreads) schedule(dynamic,1)
for (taskid = 0 ; taskid < M_ntasks ; taskid++)
{
// if kfirst > klast then taskid does no work at all
int64_t kfirst = kfirst_Mslice [taskid] ;
int64_t klast = klast_Mslice [taskid] ;
//----------------------------------------------------------------------
// C<M(:,kfirst:klast)> = x
//----------------------------------------------------------------------
for (int64_t k = kfirst ; k <= klast ; k++)
{
//------------------------------------------------------------------
// find the part of M(:,k) to be operated on by this task
//------------------------------------------------------------------
int64_t j = GBH (Mh, k) ;
int64_t pM_start, pM_end ;
GB_get_pA (&pM_start, &pM_end, taskid, k,
kfirst, klast, pstart_Mslice, Mp, mvlen) ;
// pC points to the start of C(:,j) if C is dense
int64_t pC = j * cvlen ;
//------------------------------------------------------------------
// C<M(:,j)> = x
//------------------------------------------------------------------
if (Mx == NULL && Mb == NULL)
{
GB_PRAGMA_SIMD_VECTORIZE
for (int64_t pM = pM_start ; pM < pM_end ; pM++)
{
int64_t p = pC + GBI (Mi, pM, mvlen) ;
GB_COPY_SCALAR_TO_C (p, cwork) ; // Cx [p] = scalar
}
}
else
{
GB_PRAGMA_SIMD_VECTORIZE
for (int64_t pM = pM_start ; pM < pM_end ; pM++)
{
if (GBB (Mb, pM) && GB_mcast (Mx, pM, msize))
{
int64_t p = pC + GBI (Mi, pM, mvlen) ;
GB_COPY_SCALAR_TO_C (p, cwork) ; // Cx [p] = scalar
}
}
}
}
}
}
|
