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
|
//------------------------------------------------------------------------------
// GB_split_sparse_template: split a single tile from a sparse matrix
//------------------------------------------------------------------------------
// SuiteSparse:GraphBLAS, Timothy A. Davis, (c) 2017-2025, All Rights Reserved.
// SPDX-License-Identifier: Apache-2.0
//------------------------------------------------------------------------------
// A is sparse or hypersparse. C has the same sparsity format as A.
{
//--------------------------------------------------------------------------
// get A and C, and the slicing of C
//--------------------------------------------------------------------------
#ifndef GB_ISO_SPLIT
const GB_A_TYPE *restrict Ax = (GB_A_TYPE *) A->x ;
GB_C_TYPE *restrict Cx = (GB_C_TYPE *) C->x ;
#endif
#ifdef GB_JIT_KERNEL
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) ;
GB_Cp_DECLARE (Cp, ) ; GB_Cp_PTR (Cp, C) ;
GB_Ci_DECLARE (Ci, ) ; GB_Ci_PTR (Ci, C) ;
const GB_Ap_TYPE *restrict Wp = (const GB_Ap_TYPE *) Wp_workspace ;
const int64_t *restrict kfirst_Cslice = C_ek_slicing ;
const int64_t *restrict klast_Cslice = C_ek_slicing + C_ntasks ;
const int64_t *restrict pstart_Cslice = C_ek_slicing + C_ntasks * 2 ;
#endif
//--------------------------------------------------------------------------
// copy the tile from A to C
//--------------------------------------------------------------------------
int tid ;
#pragma omp parallel for num_threads(C_nthreads) schedule(dynamic,1)
for (tid = 0 ; tid < C_ntasks ; tid++)
{
int64_t kfirst = kfirst_Cslice [tid] ;
int64_t klast = klast_Cslice [tid] ;
for (int64_t k = kfirst ; k <= klast ; k++)
{
// int64_t jA = GBh_A (Ah, k+akstart) ; not needed
int64_t p0 = GB_IGET (Cp, k) ;
GB_GET_PA (pC_start, pC_end, tid, k,
kfirst, klast, pstart_Cslice, p0, GB_IGET (Cp, k+1)) ;
int64_t pA_offset = GB_IGET (Wp, k + akstart) ;
// copy the vector from A to C
for (int64_t pC = pC_start ; pC < pC_end ; pC++)
{
// get the index of A(iA,jA)
int64_t pA = pA_offset + pC - p0 ;
int64_t iA = GB_IGET (Ai, pA) ;
// shift the index and copy into C(i,j)
int64_t iC = iA - aistart ;
GB_ISET (Ci, pC, iC) ; // Ci [pC] = iC ;
GB_COPY (pC, pA) ;
}
}
}
}
#undef GB_C_TYPE
#undef GB_A_TYPE
#undef GB_ISO_SPLIT
|
