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
|
/*
-- MAGMA (version 2.9.0) --
Univ. of Tennessee, Knoxville
Univ. of California, Berkeley
Univ. of Colorado, Denver
@date January 2025
@generated from magmablas/zsetmatrix_transpose.cpp, normal z -> d, Wed Jan 22 14:41:49 2025
*/
#include "magma_internal.h"
/***************************************************************************//**
Copy and transpose matrix hA on CPU host to dAT on GPU device.
@param[in] m Number of rows of input matrix hA. m >= 0.
@param[in] n Number of columns of input matrix hA. n >= 0.
@param[in] nb Block size. nb >= 0.
@param[in] hA The m-by-n matrix A on the CPU, of dimension (lda,n).
@param[in] lda Leading dimension of matrix hA. lda >= m.
@param[out] dAT The n-by-m matrix A^T on the GPU, of dimension (ldda,m).
@param[in] ldda Leading dimension of matrix dAT. ldda >= n.
@param[out] dwork Workspace on the GPU, of dimension (2*lddw*nb).
@param[in] lddw Leading dimension of dwork. lddw >= m.
@param[in] queues Array of two queues, to pipeline operation.
@ingroup magma_setmatrix_transpose
*******************************************************************************/
extern "C" void
magmablas_dsetmatrix_transpose(
magma_int_t m, magma_int_t n, magma_int_t nb,
const double *hA, magma_int_t lda,
magmaDouble_ptr dAT, magma_int_t ldda,
magmaDouble_ptr dwork, magma_int_t lddw,
magma_queue_t queues[2] )
{
#define hA(i_, j_) (hA + (i_) + (j_)*lda)
#define dAT(i_, j_) (dAT + (i_) + (j_)*ldda)
#define dwork(i_, j_) (dwork + (i_) + (j_)*lddw)
magma_int_t i = 0, j = 0, ib;
/* Quick return */
if ( (m == 0) || (n == 0) )
return;
// TODO standard check arguments
if (lda < m || ldda < n || lddw < m) {
fprintf( stderr, "%s: wrong arguments.\n", __func__ );
return;
}
/* Move data from CPU to GPU in the first panel in the dwork buffer */
ib = min(n-i, nb);
magma_dsetmatrix_async( m, ib,
hA(0,i), lda,
dwork(0,(j%2)*nb), lddw, queues[j%2] );
j++;
for (i=nb; i < n; i += nb) {
/* Move data from CPU to GPU in the second panel in the dwork buffer */
ib = min(n-i, nb);
magma_dsetmatrix_async( m, ib,
hA(0,i), lda,
dwork(0,(j%2)*nb), lddw, queues[j%2] );
j++;
/* Note that the previous panel (i.e., j%2) comes through the queue
for the kernel so there is no need to synchronize. */
// TODO should this be ib not nb?
magmablas_dtranspose( m, nb, dwork(0,(j%2)*nb), lddw, dAT(i-nb,0), ldda, queues[j%2] );
}
/* Transpose the last part of the matrix. */
j++;
magmablas_dtranspose( m, ib, dwork(0,(j%2)*nb), lddw, dAT(i-nb,0), ldda, queues[j%2] );
}
|
