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
|
/*
* Copyright © 2019 Manuel Stoeckl
*
* Permission is hereby granted, free of charge, to any person obtaining
* a copy of this software and associated documentation files (the
* "Software"), to deal in the Software without restriction, including
* without limitation the rights to use, copy, modify, merge, publish,
* distribute, sublicense, and/or sell copies of the Software, and to
* permit persons to whom the Software is furnished to do so, subject to
* the following conditions:
*
* The above copyright notice and this permission notice (including the
* next paragraph) shall be included in all copies or substantial
* portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
* BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
* ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#include <stdbool.h>
#include <stdint.h>
#include <stdio.h>
#include <x86intrin.h>
#ifdef __x86_64__
static inline int tzcnt(uint64_t v) { return (int)_tzcnt_u64(v); }
#else
static inline int tzcnt(uint64_t v) { return v ? __builtin_ctzll(v) : 64; }
#endif
#ifdef __x86_64__
static inline int lzcnt(uint64_t v) { return (int)_lzcnt_u64(v); }
#else
static inline int lzcnt(uint64_t v) { return v ? __builtin_clzll(v) : 64; }
#endif
size_t run_interval_diff_avx2(const int diff_window_size,
const void *__restrict__ imod, void *__restrict__ ibase,
uint32_t *__restrict__ diff, size_t i, const size_t i_end)
{
const __m256i *__restrict__ mod = imod;
__m256i *__restrict__ base = ibase;
size_t dc = 0;
while (1) {
/* Loop: no changes */
uint32_t *ctrl_blocks = &diff[dc];
dc += 2;
int trailing_unchanged = 0;
for (; i < i_end; i++) {
__m256i m0 = _mm256_load_si256(&mod[2 * i]);
__m256i m1 = _mm256_load_si256(&mod[2 * i + 1]);
__m256i b0 = _mm256_load_si256(&base[2 * i]);
__m256i b1 = _mm256_load_si256(&base[2 * i + 1]);
__m256i eq0 = _mm256_cmpeq_epi32(m0, b0);
__m256i eq1 = _mm256_cmpeq_epi32(m1, b1);
/* It's very hard to tell which loop exit method is
* better, since the routine is typically bandwidth
* limited */
#if 1
uint32_t mask0 = (uint32_t)_mm256_movemask_epi8(eq0);
uint32_t mask1 = (uint32_t)_mm256_movemask_epi8(eq1);
uint64_t mask = mask0 + mask1 * 0x100000000uLL;
if (~mask) {
#else
__m256i andv = _mm256_and_si256(eq0, eq1);
if (_mm256_testz_si256(andv, _mm256_set1_epi8(-1))) {
uint32_t mask0 = (uint32_t)_mm256_movemask_epi8(
eq0);
uint32_t mask1 = (uint32_t)_mm256_movemask_epi8(
eq1);
uint64_t mask = mask0 + mask1 * 0x100000000uLL;
#endif
_mm256_store_si256(&base[2 * i], m0);
_mm256_store_si256(&base[2 * i + 1], m1);
/* Write the changed bytes, starting at the
* first modified term,
* and set the n_unchanged counter */
size_t ncom = (size_t)tzcnt(~mask) >> 2;
size_t block_shift = (ncom & 7);
uint64_t esmask = 0xffffffffuLL
<< (block_shift * 4);
__m128i halfsize = _mm_set_epi64x(
0uLL, (long long)esmask);
__m256i estoremask =
_mm256_cvtepi8_epi64(halfsize);
_mm256_maskstore_epi32(
(int *)&diff[dc - block_shift],
estoremask, ncom < 8 ? m0 : m1);
if (ncom < 8) {
_mm256_storeu_si256(
(__m256i *)&diff[dc +
8 -
block_shift],
m1);
}
dc += 16 - ncom;
trailing_unchanged = lzcnt(~mask) >> 2;
ctrl_blocks[0] = (uint32_t)(16 * i + ncom);
i++;
if (i >= i_end) {
/* Last block, hence will not enter copy
* loop */
ctrl_blocks[1] = (uint32_t)(16 * i);
dc += 2;
}
break;
}
}
if (i >= i_end) {
dc -= 2;
break;
}
/* Loop: until no changes for DIFF_WINDOW +/- 4 spaces */
for (; i < i_end; i++) {
__m256i m0 = _mm256_load_si256(&mod[2 * i]);
__m256i m1 = _mm256_load_si256(&mod[2 * i + 1]);
__m256i b0 = _mm256_load_si256(&base[2 * i]);
__m256i b1 = _mm256_load_si256(&base[2 * i + 1]);
__m256i eq0 = _mm256_cmpeq_epi32(m0, b0);
__m256i eq1 = _mm256_cmpeq_epi32(m1, b1);
uint32_t mask0 = (uint32_t)_mm256_movemask_epi8(eq0);
uint32_t mask1 = (uint32_t)_mm256_movemask_epi8(eq1);
uint64_t mask = mask0 + mask1 * 0x100000000uLL;
/* Reset trailing counter if anything changed */
bool clear = ~mask == 0;
trailing_unchanged = clear * trailing_unchanged +
(lzcnt(~mask) >> 2);
_mm256_storeu_si256((__m256i *)&diff[dc], m0);
_mm256_storeu_si256((__m256i *)&diff[dc + 8], m1);
dc += 16;
if (trailing_unchanged > diff_window_size) {
i++;
break;
}
_mm256_store_si256(&base[2 * i], m0);
_mm256_store_si256(&base[2 * i + 1], m1);
}
/* Write coda */
dc -= (size_t)trailing_unchanged;
ctrl_blocks[1] =
(uint32_t)(16 * i - (size_t)trailing_unchanged);
if (i >= i_end) {
break;
}
}
return dc;
}
|
