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
|
//------------------------------------------------------------------------------
// GB_positional_op_ijp: C = positional_op (A), depending j
//------------------------------------------------------------------------------
// SuiteSparse:GraphBLAS, Timothy A. Davis, (c) 2017-2022, All Rights Reserved.
// SPDX-License-Identifier: Apache-2.0
//------------------------------------------------------------------------------
// TODO: use this kernel for GrB_extractTuples, to create J array.
// A can be jumbled. If A is jumbled, so is C.
// if A and C are bitmap, not all of Cx need to be written to, but it's faster
// just to write to all of it. C->b is copied from A->b in the caller.
{
//--------------------------------------------------------------------------
// slice the entries for each task
//--------------------------------------------------------------------------
GB_WERK_DECLARE (A_ek_slicing, int64_t) ;
int A_ntasks, A_nthreads ;
GB_SLICE_MATRIX (A, 32, chunk) ;
//--------------------------------------------------------------------------
// Cx = positional_op (A)
//--------------------------------------------------------------------------
int tid ;
#pragma omp parallel for num_threads(A_nthreads) schedule(dynamic,1)
for (tid = 0 ; tid < A_ntasks ; tid++)
{
// if kfirst > klast then task tid does no work at all
int64_t kfirst = kfirst_Aslice [tid] ;
int64_t klast = klast_Aslice [tid] ;
//----------------------------------------------------------------------
// C(:,kfirst:klast) = op (A(:,kfirst:klast))
//----------------------------------------------------------------------
for (int64_t k = kfirst ; k <= klast ; k++)
{
//------------------------------------------------------------------
// find the part of A(:,k) and Cx to be operated on by this task
//------------------------------------------------------------------
int64_t j = GBH (Ah, k) ;
int64_t pA_start, pA_end ;
GB_get_pA (&pA_start, &pA_end, tid, k,
kfirst, klast, pstart_Aslice, Ap, avlen) ;
//------------------------------------------------------------------
// C(:,j) = op (A(:,j))
//------------------------------------------------------------------
for (int64_t p = pA_start ; p < pA_end ; p++)
{
// Cx [p] = op (A (i,j))
GB_APPLY (p) ;
}
}
}
//--------------------------------------------------------------------------
// free workspace
//--------------------------------------------------------------------------
GB_WERK_POP (A_ek_slicing, int64_t) ;
}
#undef GB_APPLY
|
