forked from arrayfire/arrayfire
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathtranspose.cpp
More file actions
114 lines (101 loc) · 4.08 KB
/
transpose.cpp
File metadata and controls
114 lines (101 loc) · 4.08 KB
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
/*******************************************************
* Copyright (c) 2014, ArrayFire
* All rights reserved.
*
* This file is distributed under 3-clause BSD license.
* The complete license agreement can be obtained at:
* http://arrayfire.com/licenses/BSD-3-Clause
********************************************************/
#include <af/dim4.hpp>
#include <af/defines.h>
#include <af/blas.h>
#include <af/data.h>
#include <af/arith.h>
#include <err_common.hpp>
#include <handle.hpp>
#include <backend.hpp>
#include <transpose.hpp>
using af::dim4;
using namespace detail;
template<typename T>
static inline af_array trs(const af_array in, const bool conjugate)
{
return getHandle<T>(detail::transpose<T>(getArray<T>(in), conjugate));
}
af_err af_transpose(af_array *out, af_array in, const bool conjugate)
{
try {
ArrayInfo info = getInfo(in);
af_dtype type = info.getType();
af::dim4 dims = info.dims();
if (dims[0]==1 || dims[1]==1) {
af::dim4 outDims(dims[1],dims[0],dims[2],dims[3]);
if(conjugate) {
af_array temp = 0;
AF_CHECK(af_conjg(&temp, in));
AF_CHECK(af_moddims(out, temp, outDims.ndims(), outDims.get()));
AF_CHECK(af_release_array(temp));
return AF_SUCCESS;
} else {
// for a vector OR a batch of vectors
// we can use modDims to transpose
AF_CHECK(af_moddims(out, in, outDims.ndims(), outDims.get()));
return AF_SUCCESS;
}
}
af_array output;
switch(type) {
case f32: output = trs<float> (in, conjugate); break;
case c32: output = trs<cfloat> (in, conjugate); break;
case f64: output = trs<double> (in, conjugate); break;
case c64: output = trs<cdouble>(in, conjugate); break;
case b8 : output = trs<char> (in, conjugate); break;
case s32: output = trs<int> (in, conjugate); break;
case u32: output = trs<uint> (in, conjugate); break;
case u8 : output = trs<uchar> (in, conjugate); break;
case s64: output = trs<intl> (in, conjugate); break;
case u64: output = trs<uintl> (in, conjugate); break;
case s16: output = trs<short> (in, conjugate); break;
case u16: output = trs<ushort> (in, conjugate); break;
default : TYPE_ERROR(1, type);
}
std::swap(*out,output);
}
CATCHALL;
return AF_SUCCESS;
}
template<typename T>
static inline void transpose_inplace(af_array in, const bool conjugate)
{
return detail::transpose_inplace<T>(getWritableArray<T>(in), conjugate);
}
af_err af_transpose_inplace(af_array in, const bool conjugate)
{
try {
ArrayInfo info = getInfo(in);
af_dtype type = info.getType();
af::dim4 dims = info.dims();
// InPlace only works on square matrices
DIM_ASSERT(0, dims[0] == dims[1]);
// If singleton element
if(dims[0] == 1)
return AF_SUCCESS;
switch(type) {
case f32: transpose_inplace<float> (in, conjugate); break;
case c32: transpose_inplace<cfloat> (in, conjugate); break;
case f64: transpose_inplace<double> (in, conjugate); break;
case c64: transpose_inplace<cdouble>(in, conjugate); break;
case b8 : transpose_inplace<char> (in, conjugate); break;
case s32: transpose_inplace<int> (in, conjugate); break;
case u32: transpose_inplace<uint> (in, conjugate); break;
case u8 : transpose_inplace<uchar> (in, conjugate); break;
case s64: transpose_inplace<intl> (in, conjugate); break;
case u64: transpose_inplace<uintl> (in, conjugate); break;
case s16: transpose_inplace<short> (in, conjugate); break;
case u16: transpose_inplace<ushort> (in, conjugate); break;
default : TYPE_ERROR(1, type);
}
}
CATCHALL;
return AF_SUCCESS;
}