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
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
|
/* This file, swapproc.c, contains general utility routines that are */
/* used by other FITSIO routines to swap bytes. */
/* The FITSIO software was written by William Pence at the High Energy */
/* Astrophysic Science Archive Research Center (HEASARC) at the NASA */
/* Goddard Space Flight Center. */
/* The fast SSE2 and SSSE3 functions were provided by Julian Taylor, ESO */
#include <string.h>
#include <stdlib.h>
#include "fitsio2.h"
/* bswap builtin is available since GCC 4.3 */
#if __GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 3)
#define HAVE_BSWAP
#endif
#ifdef __SSSE3__
#include <tmmintrin.h>
/* swap 16 bytes according to mask, values must be 16 byte aligned */
static inline void swap_ssse3(char * values, __m128i mask)
{
__m128i v = _mm_load_si128((__m128i *)values);
__m128i s = _mm_shuffle_epi8(v, mask);
_mm_store_si128((__m128i*)values, s);
}
#endif
#ifdef __SSE2__
#include <emmintrin.h>
/* swap 8 shorts, values must be 16 byte aligned
* faster than ssse3 variant for shorts */
static inline void swap2_sse2(char * values)
{
__m128i r1 = _mm_load_si128((__m128i *)values);
__m128i r2 = r1;
r1 = _mm_srli_epi16(r1, 8);
r2 = _mm_slli_epi16(r2, 8);
r1 = _mm_or_si128(r1, r2);
_mm_store_si128((__m128i*)values, r1);
}
/* the three shuffles required for 4 and 8 byte variants make
* SSE2 slower than bswap */
/* get number of elements to peel to reach alignment */
static inline size_t get_peel(void * addr, size_t esize, size_t nvals,
size_t alignment)
{
const size_t offset = (size_t)addr % alignment;
size_t peel = offset ? (alignment - offset) / esize : 0;
peel = nvals < peel ? nvals : peel;
return peel;
}
#endif
/*--------------------------------------------------------------------------*/
static void ffswap2_slow(short *svalues, long nvals)
{
register long ii;
unsigned short * usvalues;
usvalues = (unsigned short *) svalues;
for (ii = 0; ii < nvals; ii++)
{
usvalues[ii] = (usvalues[ii]>>8) | (usvalues[ii]<<8);
}
}
/*--------------------------------------------------------------------------*/
#if __SSE2__
void ffswap2(short *svalues, /* IO - pointer to shorts to be swapped */
long nvals) /* I - number of shorts to be swapped */
/*
swap the bytes in the input short integers: ( 0 1 -> 1 0 )
*/
{
if ((long)svalues % 2 != 0) { /* should not happen */
ffswap2_slow(svalues, nvals);
return;
}
long ii;
size_t peel = get_peel((void*)&svalues[0], sizeof(svalues[0]), nvals, 16);
ffswap2_slow(svalues, peel);
for (ii = peel; ii < (nvals - peel - (nvals - peel) % 8); ii+=8) {
swap2_sse2((char*)&svalues[ii]);
}
ffswap2_slow(&svalues[ii], nvals - ii);
}
#else
void ffswap2(short *svalues, /* IO - pointer to shorts to be swapped */
long nvals) /* I - number of shorts to be swapped */
/*
swap the bytes in the input 4-byte integer: ( 0 1 2 3 -> 3 2 1 0 )
*/
{
ffswap2_slow(svalues, nvals);
}
#endif
/*--------------------------------------------------------------------------*/
static void ffswap4_slow(INT32BIT *ivalues, long nvals)
{
register long ii;
#if defined(HAVE_BSWAP)
for (ii = 0; ii < nvals; ii++)
{
ivalues[ii] = __builtin_bswap32(ivalues[ii]);
}
#elif defined(_MSC_VER) && (_MSC_VER >= 1400)
/* intrinsic byte swapping function in Microsoft Visual C++ 8.0 and later */
unsigned int* uivalues = (unsigned int *) ivalues;
/* intrinsic byte swapping function in Microsoft Visual C++ */
for (ii = 0; ii < nvals; ii++)
{
uivalues[ii] = _byteswap_ulong(uivalues[ii]);
}
#else
char *cvalues, tmp;
for (ii = 0; ii < nvals; ii++)
{
cvalues = (char *)&ivalues[ii];
tmp = cvalues[0];
cvalues[0] = cvalues[3];
cvalues[3] = tmp;
tmp = cvalues[1];
cvalues[1] = cvalues[2];
cvalues[2] = tmp;
}
#endif
}
/*--------------------------------------------------------------------------*/
#ifdef __SSSE3__
void ffswap4(INT32BIT *ivalues, /* IO - pointer to INT*4 to be swapped */
long nvals) /* I - number of floats to be swapped */
/*
swap the bytes in the input 4-byte integer: ( 0 1 2 3 -> 3 2 1 0 )
*/
{
if ((long)ivalues % 4 != 0) { /* should not happen */
ffswap4_slow(ivalues, nvals);
return;
}
long ii;
const __m128i cmask4 = _mm_set_epi8(12, 13, 14, 15,
8, 9, 10, 11,
4, 5, 6, 7,
0, 1, 2 ,3);
size_t peel = get_peel((void*)&ivalues[0], sizeof(ivalues[0]), nvals, 16);
ffswap4_slow(ivalues, peel);
for (ii = peel; ii < (nvals - peel - (nvals - peel) % 4); ii+=4) {
swap_ssse3((char*)&ivalues[ii], cmask4);
}
ffswap4_slow(&ivalues[ii], nvals - ii);
}
#else
void ffswap4(INT32BIT *ivalues, /* IO - pointer to INT*4 to be swapped */
long nvals) /* I - number of floats to be swapped */
/*
swap the bytes in the input 4-byte integer: ( 0 1 2 3 -> 3 2 1 0 )
*/
{
ffswap4_slow(ivalues, nvals);
}
#endif
/*--------------------------------------------------------------------------*/
static void ffswap8_slow(double *dvalues, long nvals)
{
register long ii;
#ifdef HAVE_BSWAP
LONGLONG * llvalues = (LONGLONG*)dvalues;
for (ii = 0; ii < nvals; ii++) {
llvalues[ii] = __builtin_bswap64(llvalues[ii]);
}
#elif defined(_MSC_VER) && (_MSC_VER >= 1400)
/* intrinsic byte swapping function in Microsoft Visual C++ 8.0 and later */
unsigned __int64 * llvalues = (unsigned __int64 *) dvalues;
for (ii = 0; ii < nvals; ii++)
{
llvalues[ii] = _byteswap_uint64(llvalues[ii]);
}
#else
register char *cvalues;
register char temp;
cvalues = (char *) dvalues; /* copy the pointer value */
for (ii = 0; ii < nvals*8; ii += 8)
{
temp = cvalues[ii];
cvalues[ii] = cvalues[ii+7];
cvalues[ii+7] = temp;
temp = cvalues[ii+1];
cvalues[ii+1] = cvalues[ii+6];
cvalues[ii+6] = temp;
temp = cvalues[ii+2];
cvalues[ii+2] = cvalues[ii+5];
cvalues[ii+5] = temp;
temp = cvalues[ii+3];
cvalues[ii+3] = cvalues[ii+4];
cvalues[ii+4] = temp;
}
#endif
}
/*--------------------------------------------------------------------------*/
#ifdef __SSSE3__
void ffswap8(double *dvalues, /* IO - pointer to doubles to be swapped */
long nvals) /* I - number of doubles to be swapped */
/*
swap the bytes in the input doubles: ( 01234567 -> 76543210 )
*/
{
if ((long)dvalues % 8 != 0) { /* should not happen on amd64 */
ffswap8_slow(dvalues, nvals);
return;
}
long ii;
const __m128i cmask8 = _mm_set_epi8(8, 9, 10, 11, 12, 13, 14, 15,
0, 1, 2 ,3, 4, 5, 6, 7);
size_t peel = get_peel((void*)&dvalues[0], sizeof(dvalues[0]), nvals, 16);
ffswap8_slow(dvalues, peel);
for (ii = peel; ii < (nvals - peel - (nvals - peel) % 2); ii+=2) {
swap_ssse3((char*)&dvalues[ii], cmask8);
}
ffswap8_slow(&dvalues[ii], nvals - ii);
}
#else
void ffswap8(double *dvalues, /* IO - pointer to doubles to be swapped */
long nvals) /* I - number of doubles to be swapped */
/*
swap the bytes in the input doubles: ( 01234567 -> 76543210 )
*/
{
ffswap8_slow(dvalues, nvals);
}
#endif
|
