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
|
/*
* Loongson MMI optimizations for libjpeg-turbo
*
* Copyright (C) 2016-2017, Loongson Technology Corporation Limited, BeiJing.
* All Rights Reserved.
* Authors: ZhuChen <zhuchen@loongson.cn>
* CaiWanwei <caiwanwei@loongson.cn>
* SunZhangzhi <sunzhangzhi-cq@loongson.cn>
* Copyright (C) 2018-2019, D. R. Commander. All Rights Reserved.
*
* Based on the x86 SIMD extension for IJG JPEG library
* Copyright (C) 1999-2006, MIYASAKA Masaru.
*
* This software is provided 'as-is', without any express or implied
* warranty. In no event will the authors be held liable for any damages
* arising from the use of this software.
*
* Permission is granted to anyone to use this software for any purpose,
* including commercial applications, and to alter it and redistribute it
* freely, subject to the following restrictions:
*
* 1. The origin of this software must not be misrepresented; you must not
* claim that you wrote the original software. If you use this software
* in a product, an acknowledgment in the product documentation would be
* appreciated but is not required.
* 2. Altered source versions must be plainly marked as such, and must not be
* misrepresented as being the original software.
* 3. This notice may not be removed or altered from any source distribution.
*/
/* INTEGER QUANTIZATION AND SAMPLE CONVERSION */
#include "jsimd_mmi.h"
#define DO_QUANT() { \
__m64 rowl, rowh, rowls, rowhs, rowlsave, rowhsave; \
__m64 corrl, corrh, recipl, reciph, scalel, scaleh; \
\
rowl = _mm_load_si64((__m64 *)&workspace[0]); \
rowh = _mm_load_si64((__m64 *)&workspace[4]); \
\
/* Branch-less absolute value */ \
rowls = _mm_srai_pi16(rowl, (WORD_BIT - 1)); /* -1 if value < 0, */ \
/* 0 otherwise */ \
rowhs = _mm_srai_pi16(rowh, (WORD_BIT - 1)); \
\
rowl = _mm_xor_si64(rowl, rowls); /* val = -val */ \
rowh = _mm_xor_si64(rowh, rowhs); \
rowl = _mm_sub_pi16(rowl, rowls); \
rowh = _mm_sub_pi16(rowh, rowhs); \
\
corrl = _mm_load_si64((__m64 *)&divisors[DCTSIZE2 * 1]); /* correction */ \
corrh = _mm_load_si64((__m64 *)&divisors[DCTSIZE2 * 1 + 4]); \
\
rowlsave = rowl = _mm_add_pi16(rowl, corrl); /* correction + roundfactor */ \
rowhsave = rowh = _mm_add_pi16(rowh, corrh); \
\
recipl = _mm_load_si64((__m64 *)&divisors[DCTSIZE2 * 0]); /* reciprocal */ \
reciph = _mm_load_si64((__m64 *)&divisors[DCTSIZE2 * 0 + 4]); \
\
rowl = _mm_mulhi_pi16(rowl, recipl); \
rowh = _mm_mulhi_pi16(rowh, reciph); \
\
/* reciprocal is always negative (MSB=1), so we always need to add the */ \
/* initial value (input value is never negative as we inverted it at the */ \
/* start of this routine) */ \
rowlsave = rowl = _mm_add_pi16(rowl, rowlsave); \
rowhsave = rowh = _mm_add_pi16(rowh, rowhsave); \
\
scalel = _mm_load_si64((__m64 *)&divisors[DCTSIZE2 * 2]); /* scale */ \
scaleh = _mm_load_si64((__m64 *)&divisors[DCTSIZE2 * 2 + 4]); \
\
rowl = _mm_mulhi_pi16(rowl, scalel); \
rowh = _mm_mulhi_pi16(rowh, scaleh); \
\
/* determine if scale is negative */ \
scalel = _mm_srai_pi16(scalel, (WORD_BIT - 1)); \
scaleh = _mm_srai_pi16(scaleh, (WORD_BIT - 1)); \
\
/* and add input if it is */ \
scalel = _mm_and_si64(scalel, rowlsave); \
scaleh = _mm_and_si64(scaleh, rowhsave); \
rowl = _mm_add_pi16(rowl, scalel); \
rowh = _mm_add_pi16(rowh, scaleh); \
\
/* then check if negative input */ \
rowlsave = _mm_srai_pi16(rowlsave, (WORD_BIT - 1)); \
rowhsave = _mm_srai_pi16(rowhsave, (WORD_BIT - 1)); \
\
/* and add scale if it is */ \
rowlsave = _mm_and_si64(rowlsave, scalel); \
rowhsave = _mm_and_si64(rowhsave, scaleh); \
rowl = _mm_add_pi16(rowl, rowlsave); \
rowh = _mm_add_pi16(rowh, rowhsave); \
\
rowl = _mm_xor_si64(rowl, rowls); /* val = -val */ \
rowh = _mm_xor_si64(rowh, rowhs); \
rowl = _mm_sub_pi16(rowl, rowls); \
rowh = _mm_sub_pi16(rowh, rowhs); \
\
_mm_store_si64((__m64 *)&output_ptr[0], rowl); \
_mm_store_si64((__m64 *)&output_ptr[4], rowh); \
\
workspace += DCTSIZE; \
divisors += DCTSIZE; \
output_ptr += DCTSIZE; \
}
void jsimd_quantize_mmi(JCOEFPTR coef_block, DCTELEM *divisors,
DCTELEM *workspace)
{
JCOEFPTR output_ptr = coef_block;
DO_QUANT()
DO_QUANT()
DO_QUANT()
DO_QUANT()
DO_QUANT()
DO_QUANT()
DO_QUANT()
DO_QUANT()
}
|
