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) 2002 Zbigniew Leyk (zbigniew.leyk@anu.edu.au)
// and David E. Stewart (david.stewart@anu.edu.au)
// and Ronan Collobert (collober@iro.umontreal.ca)
//
//
// This file is part of Torch. Release II.
// [The Ultimate Machine Learning Library]
//
// Torch is free software; you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation; either version 2 of the License, or
// (at your option) any later version.
//
// Torch is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
// along with Torch; if not, write to the Free Software
// Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
#include "mx_low_level.h"
namespace Torch {
/* __ip__ -- inner product */
real mxIp__(real * dp1, real * dp2, int len)
{
#ifdef VUNROLL
int len4;
real sum1, sum2, sum3;
#endif
real sum;
sum = 0.0;
#ifdef VUNROLL
sum1 = sum2 = sum3 = 0.0;
len4 = len / 4;
len = len % 4;
for (int i = 0; i < len4; i++)
{
sum += dp1[4 * i] * dp2[4 * i];
sum1 += dp1[4 * i + 1] * dp2[4 * i + 1];
sum2 += dp1[4 * i + 2] * dp2[4 * i + 2];
sum3 += dp1[4 * i + 3] * dp2[4 * i + 3];
}
sum += sum1 + sum2 + sum3;
dp1 += 4 * len4;
dp2 += 4 * len4;
#endif
for (int i = 0; i < len; i++)
sum += dp1[i] * dp2[i];
return sum;
}
/* __mltadd__ -- scalar multiply and add c.f. v_mltadd() */
void mxRealMulAdd__(real * dp1, real * dp2, real s, int len)
{
#ifdef VUNROLL
int len4;
len4 = len / 4;
len = len % 4;
for (int i = 0; i < len4; i++)
{
dp1[4 * i] += s * dp2[4 * i];
dp1[4 * i + 1] += s * dp2[4 * i + 1];
dp1[4 * i + 2] += s * dp2[4 * i + 2];
dp1[4 * i + 3] += s * dp2[4 * i + 3];
}
dp1 += 4 * len4;
dp2 += 4 * len4;
#endif
for (int i = 0; i < len; i++)
dp1[i] += s * dp2[i];
}
/* __smlt__ scalar multiply array c.f. sv_mlt() */
void mxRealMul__(real * dp, real s, real * out, int len)
{
for (int i = 0; i < len; i++)
out[i] = s * dp[i];
}
/* __add__ -- add arrays c.f. v_add() */
void mxAdd__(real * dp1, real * dp2, real * out, int len)
{
for (int i = 0; i < len; i++)
out[i] = dp1[i] + dp2[i];
}
/* __sub__ -- subtract arrays c.f. v_sub() */
void mxSub__(real * dp1, real * dp2, real * out, int len)
{
for (int i = 0; i < len; i++)
out[i] = dp1[i] - dp2[i];
}
/* __zero__ -- zeros an array of floating point numbers */
void mxZero__(real * dp, int len)
{
/* else, need to zero the array entry by entry */
for (int i = 0; i < len; i++)
dp[i] = 0.0;
}
}
|
