-
Notifications
You must be signed in to change notification settings - Fork 14
/
range-avx2.c
277 lines (241 loc) · 10.1 KB
/
range-avx2.c
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
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
#ifdef __AVX2__
#include <stdio.h>
#include <stdint.h>
#include <x86intrin.h>
int utf8_naive(const unsigned char *data, int len);
#if 0
static void print256(const char *s, const __m256i v256)
{
const unsigned char *v8 = (const unsigned char *)&v256;
if (s)
printf("%s:\t", s);
for (int i = 0; i < 32; i++)
printf("%02x ", v8[i]);
printf("\n");
}
#endif
/*
* Map high nibble of "First Byte" to legal character length minus 1
* 0x00 ~ 0xBF --> 0
* 0xC0 ~ 0xDF --> 1
* 0xE0 ~ 0xEF --> 2
* 0xF0 ~ 0xFF --> 3
*/
static const int8_t _first_len_tbl[] = {
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 2, 3,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 2, 3,
};
/* Map "First Byte" to 8-th item of range table (0xC2 ~ 0xF4) */
static const int8_t _first_range_tbl[] = {
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 8, 8, 8, 8,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 8, 8, 8, 8,
};
/*
* Range table, map range index to min and max values
* Index 0 : 00 ~ 7F (First Byte, ascii)
* Index 1,2,3: 80 ~ BF (Second, Third, Fourth Byte)
* Index 4 : A0 ~ BF (Second Byte after E0)
* Index 5 : 80 ~ 9F (Second Byte after ED)
* Index 6 : 90 ~ BF (Second Byte after F0)
* Index 7 : 80 ~ 8F (Second Byte after F4)
* Index 8 : C2 ~ F4 (First Byte, non ascii)
* Index 9~15 : illegal: i >= 127 && i <= -128
*/
static const int8_t _range_min_tbl[] = {
0x00, 0x80, 0x80, 0x80, 0xA0, 0x80, 0x90, 0x80,
0xC2, 0x7F, 0x7F, 0x7F, 0x7F, 0x7F, 0x7F, 0x7F,
0x00, 0x80, 0x80, 0x80, 0xA0, 0x80, 0x90, 0x80,
0xC2, 0x7F, 0x7F, 0x7F, 0x7F, 0x7F, 0x7F, 0x7F,
};
static const int8_t _range_max_tbl[] = {
0x7F, 0xBF, 0xBF, 0xBF, 0xBF, 0x9F, 0xBF, 0x8F,
0xF4, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
0x7F, 0xBF, 0xBF, 0xBF, 0xBF, 0x9F, 0xBF, 0x8F,
0xF4, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
};
/*
* Tables for fast handling of four special First Bytes(E0,ED,F0,F4), after
* which the Second Byte are not 80~BF. It contains "range index adjustment".
* +------------+---------------+------------------+----------------+
* | First Byte | original range| range adjustment | adjusted range |
* +------------+---------------+------------------+----------------+
* | E0 | 2 | 2 | 4 |
* +------------+---------------+------------------+----------------+
* | ED | 2 | 3 | 5 |
* +------------+---------------+------------------+----------------+
* | F0 | 3 | 3 | 6 |
* +------------+---------------+------------------+----------------+
* | F4 | 4 | 4 | 8 |
* +------------+---------------+------------------+----------------+
*/
/* index1 -> E0, index14 -> ED */
static const int8_t _df_ee_tbl[] = {
0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 0,
0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 0,
};
/* index1 -> F0, index5 -> F4 */
static const int8_t _ef_fe_tbl[] = {
0, 3, 0, 0, 0, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 3, 0, 0, 0, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
};
#define RET_ERR_IDX 0 /* Define 1 to return index of first error char */
static inline __m256i push_last_byte_of_a_to_b(__m256i a, __m256i b) {
return _mm256_alignr_epi8(b, _mm256_permute2x128_si256(a, b, 0x21), 15);
}
static inline __m256i push_last_2bytes_of_a_to_b(__m256i a, __m256i b) {
return _mm256_alignr_epi8(b, _mm256_permute2x128_si256(a, b, 0x21), 14);
}
static inline __m256i push_last_3bytes_of_a_to_b(__m256i a, __m256i b) {
return _mm256_alignr_epi8(b, _mm256_permute2x128_si256(a, b, 0x21), 13);
}
/* 5x faster than naive method */
/* Return 0 - success, -1 - error, >0 - first error char(if RET_ERR_IDX = 1) */
int utf8_range_avx2(const unsigned char *data, int len)
{
#if RET_ERR_IDX
int err_pos = 1;
#endif
if (len >= 32) {
__m256i prev_input = _mm256_set1_epi8(0);
__m256i prev_first_len = _mm256_set1_epi8(0);
/* Cached tables */
const __m256i first_len_tbl =
_mm256_loadu_si256((const __m256i *)_first_len_tbl);
const __m256i first_range_tbl =
_mm256_loadu_si256((const __m256i *)_first_range_tbl);
const __m256i range_min_tbl =
_mm256_loadu_si256((const __m256i *)_range_min_tbl);
const __m256i range_max_tbl =
_mm256_loadu_si256((const __m256i *)_range_max_tbl);
const __m256i df_ee_tbl =
_mm256_loadu_si256((const __m256i *)_df_ee_tbl);
const __m256i ef_fe_tbl =
_mm256_loadu_si256((const __m256i *)_ef_fe_tbl);
#if !RET_ERR_IDX
__m256i error1 = _mm256_set1_epi8(0);
__m256i error2 = _mm256_set1_epi8(0);
#endif
while (len >= 32) {
const __m256i input = _mm256_loadu_si256((const __m256i *)data);
/* high_nibbles = input >> 4 */
const __m256i high_nibbles =
_mm256_and_si256(_mm256_srli_epi16(input, 4), _mm256_set1_epi8(0x0F));
/* first_len = legal character length minus 1 */
/* 0 for 00~7F, 1 for C0~DF, 2 for E0~EF, 3 for F0~FF */
/* first_len = first_len_tbl[high_nibbles] */
__m256i first_len = _mm256_shuffle_epi8(first_len_tbl, high_nibbles);
/* First Byte: set range index to 8 for bytes within 0xC0 ~ 0xFF */
/* range = first_range_tbl[high_nibbles] */
__m256i range = _mm256_shuffle_epi8(first_range_tbl, high_nibbles);
/* Second Byte: set range index to first_len */
/* 0 for 00~7F, 1 for C0~DF, 2 for E0~EF, 3 for F0~FF */
/* range |= (first_len, prev_first_len) << 1 byte */
range = _mm256_or_si256(
range, push_last_byte_of_a_to_b(prev_first_len, first_len));
/* Third Byte: set range index to saturate_sub(first_len, 1) */
/* 0 for 00~7F, 0 for C0~DF, 1 for E0~EF, 2 for F0~FF */
__m256i tmp1, tmp2;
/* tmp1 = (first_len, prev_first_len) << 2 bytes */
tmp1 = push_last_2bytes_of_a_to_b(prev_first_len, first_len);
/* tmp2 = saturate_sub(tmp1, 1) */
tmp2 = _mm256_subs_epu8(tmp1, _mm256_set1_epi8(1));
/* range |= tmp2 */
range = _mm256_or_si256(range, tmp2);
/* Fourth Byte: set range index to saturate_sub(first_len, 2) */
/* 0 for 00~7F, 0 for C0~DF, 0 for E0~EF, 1 for F0~FF */
/* tmp1 = (first_len, prev_first_len) << 3 bytes */
tmp1 = push_last_3bytes_of_a_to_b(prev_first_len, first_len);
/* tmp2 = saturate_sub(tmp1, 2) */
tmp2 = _mm256_subs_epu8(tmp1, _mm256_set1_epi8(2));
/* range |= tmp2 */
range = _mm256_or_si256(range, tmp2);
/*
* Now we have below range indices caluclated
* Correct cases:
* - 8 for C0~FF
* - 3 for 1st byte after F0~FF
* - 2 for 1st byte after E0~EF or 2nd byte after F0~FF
* - 1 for 1st byte after C0~DF or 2nd byte after E0~EF or
* 3rd byte after F0~FF
* - 0 for others
* Error cases:
* 9,10,11 if non ascii First Byte overlaps
* E.g., F1 80 C2 90 --> 8 3 10 2, where 10 indicates error
*/
/* Adjust Second Byte range for special First Bytes(E0,ED,F0,F4) */
/* Overlaps lead to index 9~15, which are illegal in range table */
__m256i shift1, pos, range2;
/* shift1 = (input, prev_input) << 1 byte */
shift1 = push_last_byte_of_a_to_b(prev_input, input);
pos = _mm256_sub_epi8(shift1, _mm256_set1_epi8(0xEF));
/*
* shift1: | EF F0 ... FE | FF 00 ... ... DE | DF E0 ... EE |
* pos: | 0 1 15 | 16 17 239| 240 241 255|
* pos-240: | 0 0 0 | 0 0 0 | 0 1 15 |
* pos+112: | 112 113 127| >= 128 | >= 128 |
*/
tmp1 = _mm256_subs_epu8(pos, _mm256_set1_epi8(240));
range2 = _mm256_shuffle_epi8(df_ee_tbl, tmp1);
tmp2 = _mm256_adds_epu8(pos, _mm256_set1_epi8(112));
range2 = _mm256_add_epi8(range2, _mm256_shuffle_epi8(ef_fe_tbl, tmp2));
range = _mm256_add_epi8(range, range2);
/* Load min and max values per calculated range index */
__m256i minv = _mm256_shuffle_epi8(range_min_tbl, range);
__m256i maxv = _mm256_shuffle_epi8(range_max_tbl, range);
/* Check value range */
#if RET_ERR_IDX
__m256i error = _mm256_cmpgt_epi8(minv, input);
error = _mm256_or_si256(error, _mm256_cmpgt_epi8(input, maxv));
/* 5% performance drop from this conditional branch */
if (!_mm256_testz_si256(error, error))
break;
#else
error1 = _mm256_or_si256(error1, _mm256_cmpgt_epi8(minv, input));
error2 = _mm256_or_si256(error2, _mm256_cmpgt_epi8(input, maxv));
#endif
prev_input = input;
prev_first_len = first_len;
data += 32;
len -= 32;
#if RET_ERR_IDX
err_pos += 32;
#endif
}
#if RET_ERR_IDX
/* Error in first 16 bytes */
if (err_pos == 1)
goto do_naive;
#else
__m256i error = _mm256_or_si256(error1, error2);
if (!_mm256_testz_si256(error, error))
return -1;
#endif
/* Find previous token (not 80~BF) */
int32_t token4 = _mm256_extract_epi32(prev_input, 7);
const int8_t *token = (const int8_t *)&token4;
int lookahead = 0;
if (token[3] > (int8_t)0xBF)
lookahead = 1;
else if (token[2] > (int8_t)0xBF)
lookahead = 2;
else if (token[1] > (int8_t)0xBF)
lookahead = 3;
data -= lookahead;
len += lookahead;
#if RET_ERR_IDX
err_pos -= lookahead;
#endif
}
/* Check remaining bytes with naive method */
#if RET_ERR_IDX
int err_pos2;
do_naive:
err_pos2 = utf8_naive(data, len);
if (err_pos2)
return err_pos + err_pos2 - 1;
return 0;
#else
return utf8_naive(data, len);
#endif
}
#endif