gradfun.c
Go to the documentation of this file.
1 /*
2  * Copyright (C) 2009 Loren Merritt <lorenm@u.washignton.edu>
3  *
4  * This file is part of Libav.
5  *
6  * Libav is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU Lesser General Public
8  * License as published by the Free Software Foundation; either
9  * version 2.1 of the License, or (at your option) any later version.
10  *
11  * Libav is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14  * Lesser General Public License for more details.
15  *
16  * You should have received a copy of the GNU Lesser General Public
17  * License along with Libav; if not, write to the Free Software
18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19  */
20 
21 #include "libavutil/cpu.h"
22 #include "libavutil/x86_cpu.h"
23 #include "libavfilter/gradfun.h"
24 
25 DECLARE_ALIGNED(16, static const uint16_t, pw_7f)[8] = {0x7F,0x7F,0x7F,0x7F,0x7F,0x7F,0x7F,0x7F};
26 DECLARE_ALIGNED(16, static const uint16_t, pw_ff)[8] = {0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF};
27 
28 void ff_gradfun_filter_line_mmx2(uint8_t *dst, uint8_t *src, uint16_t *dc, int width, int thresh, const uint16_t *dithers)
29 {
30 #if HAVE_MMX
31  intptr_t x;
32  if (width & 3) {
33  x = width & ~3;
34  ff_gradfun_filter_line_c(dst + x, src + x, dc + x / 2, width - x, thresh, dithers);
35  width = x;
36  }
37  x = -width;
38  __asm__ volatile(
39  "movd %4, %%mm5 \n"
40  "pxor %%mm7, %%mm7 \n"
41  "pshufw $0, %%mm5, %%mm5 \n"
42  "movq %6, %%mm6 \n"
43  "movq %5, %%mm4 \n"
44  "1: \n"
45  "movd (%2,%0), %%mm0 \n"
46  "movd (%3,%0), %%mm1 \n"
47  "punpcklbw %%mm7, %%mm0 \n"
48  "punpcklwd %%mm1, %%mm1 \n"
49  "psllw $7, %%mm0 \n"
50  "pxor %%mm2, %%mm2 \n"
51  "psubw %%mm0, %%mm1 \n" // delta = dc - pix
52  "psubw %%mm1, %%mm2 \n"
53  "pmaxsw %%mm1, %%mm2 \n"
54  "pmulhuw %%mm5, %%mm2 \n" // m = abs(delta) * thresh >> 16
55  "psubw %%mm6, %%mm2 \n"
56  "pminsw %%mm7, %%mm2 \n" // m = -max(0, 127-m)
57  "pmullw %%mm2, %%mm2 \n"
58  "paddw %%mm4, %%mm0 \n" // pix += dither
59  "pmulhw %%mm2, %%mm1 \n"
60  "psllw $2, %%mm1 \n" // m = m*m*delta >> 14
61  "paddw %%mm1, %%mm0 \n" // pix += m
62  "psraw $7, %%mm0 \n"
63  "packuswb %%mm0, %%mm0 \n"
64  "movd %%mm0, (%1,%0) \n" // dst = clip(pix>>7)
65  "add $4, %0 \n"
66  "jl 1b \n"
67  "emms \n"
68  :"+r"(x)
69  :"r"(dst+width), "r"(src+width), "r"(dc+width/2),
70  "rm"(thresh), "m"(*dithers), "m"(*pw_7f)
71  :"memory"
72  );
73 #endif
74 }
75 
76 void ff_gradfun_filter_line_ssse3(uint8_t *dst, uint8_t *src, uint16_t *dc, int width, int thresh, const uint16_t *dithers)
77 {
78 #if HAVE_SSSE3
79  intptr_t x;
80  if (width & 7) {
81  // could be 10% faster if I somehow eliminated this
82  x = width & ~7;
83  ff_gradfun_filter_line_c(dst + x, src + x, dc + x / 2, width - x, thresh, dithers);
84  width = x;
85  }
86  x = -width;
87  __asm__ volatile(
88  "movd %4, %%xmm5 \n"
89  "pxor %%xmm7, %%xmm7 \n"
90  "pshuflw $0,%%xmm5, %%xmm5 \n"
91  "movdqa %6, %%xmm6 \n"
92  "punpcklqdq %%xmm5, %%xmm5 \n"
93  "movdqa %5, %%xmm4 \n"
94  "1: \n"
95  "movq (%2,%0), %%xmm0 \n"
96  "movq (%3,%0), %%xmm1 \n"
97  "punpcklbw %%xmm7, %%xmm0 \n"
98  "punpcklwd %%xmm1, %%xmm1 \n"
99  "psllw $7, %%xmm0 \n"
100  "psubw %%xmm0, %%xmm1 \n" // delta = dc - pix
101  "pabsw %%xmm1, %%xmm2 \n"
102  "pmulhuw %%xmm5, %%xmm2 \n" // m = abs(delta) * thresh >> 16
103  "psubw %%xmm6, %%xmm2 \n"
104  "pminsw %%xmm7, %%xmm2 \n" // m = -max(0, 127-m)
105  "pmullw %%xmm2, %%xmm2 \n"
106  "psllw $1, %%xmm2 \n"
107  "paddw %%xmm4, %%xmm0 \n" // pix += dither
108  "pmulhrsw %%xmm2, %%xmm1 \n" // m = m*m*delta >> 14
109  "paddw %%xmm1, %%xmm0 \n" // pix += m
110  "psraw $7, %%xmm0 \n"
111  "packuswb %%xmm0, %%xmm0 \n"
112  "movq %%xmm0, (%1,%0) \n" // dst = clip(pix>>7)
113  "add $8, %0 \n"
114  "jl 1b \n"
115  :"+&r"(x)
116  :"r"(dst+width), "r"(src+width), "r"(dc+width/2),
117  "rm"(thresh), "m"(*dithers), "m"(*pw_7f)
118  :"memory"
119  );
120 #endif // HAVE_SSSE3
121 }
122 
123 void ff_gradfun_blur_line_sse2(uint16_t *dc, uint16_t *buf, uint16_t *buf1, uint8_t *src, int src_linesize, int width)
124 {
125 #if HAVE_SSE
126 #define BLURV(load)\
127  intptr_t x = -2*width;\
128  __asm__ volatile(\
129  "movdqa %6, %%xmm7 \n"\
130  "1: \n"\
131  load" (%4,%0), %%xmm0 \n"\
132  load" (%5,%0), %%xmm1 \n"\
133  "movdqa %%xmm0, %%xmm2 \n"\
134  "movdqa %%xmm1, %%xmm3 \n"\
135  "psrlw $8, %%xmm0 \n"\
136  "psrlw $8, %%xmm1 \n"\
137  "pand %%xmm7, %%xmm2 \n"\
138  "pand %%xmm7, %%xmm3 \n"\
139  "paddw %%xmm1, %%xmm0 \n"\
140  "paddw %%xmm3, %%xmm2 \n"\
141  "paddw %%xmm2, %%xmm0 \n"\
142  "paddw (%2,%0), %%xmm0 \n"\
143  "movdqa (%1,%0), %%xmm1 \n"\
144  "movdqa %%xmm0, (%1,%0) \n"\
145  "psubw %%xmm1, %%xmm0 \n"\
146  "movdqa %%xmm0, (%3,%0) \n"\
147  "add $16, %0 \n"\
148  "jl 1b \n"\
149  :"+&r"(x)\
150  :"r"(buf+width),\
151  "r"(buf1+width),\
152  "r"(dc+width),\
153  "r"(src+width*2),\
154  "r"(src+width*2+src_linesize),\
155  "m"(*pw_ff)\
156  :"memory"\
157  );
158  if (((intptr_t) src | src_linesize) & 15) {
159  BLURV("movdqu");
160  } else {
161  BLURV("movdqa");
162  }
163 #endif // HAVE_SSE
164 }