1 /*****************************************************************************
2 * i420_yuy2.h : YUV to YUV conversion module for vlc
3 *****************************************************************************
4 * Copyright (C) 2000, 2001 the VideoLAN team
7 * Authors: Samuel Hocevar <sam@zoy.org>
8 * Damien Fouilleul <damien@videolan.org>
10 * This program is free software; you can redistribute it and/or modify
11 * it under the terms of the GNU General Public License as published by
12 * the Free Software Foundation; either version 2 of the License, or
13 * (at your option) any later version.
15 * This program is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 * GNU General Public License for more details.
20 * You should have received a copy of the GNU General Public License
21 * along with this program; if not, write to the Free Software
22 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
23 *****************************************************************************/
25 #ifdef MODULE_NAME_IS_i420_yuy2_mmx
27 #define MMX_CALL(MMX_INSTRUCTIONS) \
29 __asm__ __volatile__( \
33 : "r" (p_line1), "r" (p_line2), "r" (p_y1), "r" (p_y2), \
34 "r" (p_u), "r" (p_v) ); \
35 p_line1 += 16; p_line2 += 16; p_y1 += 8; p_y2 += 8; p_u += 4; p_v += 4; \
38 #define MMX_YUV420_YUYV " \n\
39 movd (%4), %%mm1 # Load 4 Cb 00 00 00 00 u3 u2 u1 u0 \n\
40 movd (%5), %%mm2 # Load 4 Cr 00 00 00 00 v3 v2 v1 v0 \n\
41 movq (%2), %%mm0 # Load 8 Y y7 y6 y5 y4 y3 y2 y1 y0 \n\
42 movq (%3), %%mm3 # Load 8 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
43 punpcklbw %%mm2, %%mm1 # v3 u3 v2 u2 v1 u1 v0 u0 \n\
44 movq %%mm0, %%mm2 # y7 y6 y5 y4 y3 y2 y1 y0 \n\
45 punpcklbw %%mm1, %%mm2 # v1 y3 u1 y2 v0 y1 u0 y0 \n\
46 movq %%mm2, (%0) # Store low YUYV \n\
47 punpckhbw %%mm1, %%mm0 # v3 y7 u3 y6 v2 y5 u2 y4 \n\
48 movq %%mm0, 8(%0) # Store high YUYV \n\
49 movq %%mm3, %%mm4 # Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
50 punpcklbw %%mm1, %%mm4 # v1 Y3 u1 Y2 v0 Y1 u0 Y0 \n\
51 movq %%mm4, (%1) # Store low YUYV \n\
52 punpckhbw %%mm1, %%mm3 # v3 Y7 u3 Y6 v2 Y5 u2 Y4 \n\
53 movq %%mm3, 8(%1) # Store high YUYV \n\
56 #define MMX_YUV420_YVYU " \n\
57 movd (%4), %%mm2 # Load 4 Cb 00 00 00 00 u3 u2 u1 u0 \n\
58 movd (%5), %%mm1 # Load 4 Cr 00 00 00 00 v3 v2 v1 v0 \n\
59 movq (%2), %%mm0 # Load 8 Y y7 y6 y5 y4 y3 y2 y1 y0 \n\
60 movq (%3), %%mm3 # Load 8 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
61 punpcklbw %%mm2, %%mm1 # u3 v3 u2 v2 u1 v1 u0 v0 \n\
62 movq %%mm0, %%mm2 # y7 y6 y5 y4 y3 y2 y1 y0 \n\
63 punpcklbw %%mm1, %%mm2 # u1 y3 v1 y2 u0 y1 v0 y0 \n\
64 movq %%mm2, (%0) # Store low YUYV \n\
65 punpckhbw %%mm1, %%mm0 # u3 y7 v3 y6 u2 y5 v2 y4 \n\
66 movq %%mm0, 8(%0) # Store high YUYV \n\
67 movq %%mm3, %%mm4 # Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
68 punpcklbw %%mm1, %%mm4 # u1 Y3 v1 Y2 u0 Y1 v0 Y0 \n\
69 movq %%mm4, (%1) # Store low YUYV \n\
70 punpckhbw %%mm1, %%mm3 # u3 Y7 v3 Y6 u2 Y5 v2 Y4 \n\
71 movq %%mm3, 8(%1) # Store high YUYV \n\
74 #define MMX_YUV420_UYVY " \n\
75 movd (%4), %%mm1 # Load 4 Cb 00 00 00 00 u3 u2 u1 u0 \n\
76 movd (%5), %%mm2 # Load 4 Cr 00 00 00 00 v3 v2 v1 v0 \n\
77 movq (%2), %%mm0 # Load 8 Y y7 y6 y5 y4 y3 y2 y1 y0 \n\
78 movq (%3), %%mm3 # Load 8 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
79 punpcklbw %%mm2, %%mm1 # v3 u3 v2 u2 v1 u1 v0 u0 \n\
80 movq %%mm1, %%mm2 # v3 u3 v2 u2 v1 u1 v0 u0 \n\
81 punpcklbw %%mm0, %%mm2 # y3 v1 y2 u1 y1 v0 y0 u0 \n\
82 movq %%mm2, (%0) # Store low UYVY \n\
83 movq %%mm1, %%mm2 # u3 v3 u2 v2 u1 v1 u0 v0 \n\
84 punpckhbw %%mm0, %%mm2 # y3 v1 y2 u1 y1 v0 y0 u0 \n\
85 movq %%mm2, 8(%0) # Store high UYVY \n\
86 movq %%mm1, %%mm4 # u3 v3 u2 v2 u1 v1 u0 v0 \n\
87 punpcklbw %%mm3, %%mm4 # Y3 v1 Y2 u1 Y1 v0 Y0 u0 \n\
88 movq %%mm4, (%1) # Store low UYVY \n\
89 punpckhbw %%mm3, %%mm1 # Y7 v3 Y6 u3 Y5 v2 Y4 u2 \n\
90 movq %%mm1, 8(%1) # Store high UYVY \n\
93 /* FIXME: this code does not work ! Chroma seems to be wrong. */
94 #define MMX_YUV420_Y211 " \n\
95 movq (%2), %%mm0 # Load 8 Y y7 y6 y5 y4 y3 y2 y1 y0 \n\
96 movq (%3), %%mm1 # Load 8 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
97 movd (%4), %%mm2 # Load 4 Cb 00 00 00 00 u3 u2 u1 u0 \n\
98 movd (%5), %%mm3 # Load 4 Cr 00 00 00 00 v3 v2 v1 v0 \n\
99 pand i_00ffw, %%mm0 # get Y even 00 Y6 00 Y4 00 Y2 00 Y0 \n\
100 packuswb %%mm0, %%mm0 # pack Y y6 y4 y2 y0 y6 y4 y2 y0 \n\
101 pand i_00ffw, %%mm2 # get U even 00 u6 00 u4 00 u2 00 u0 \n\
102 packuswb %%mm2, %%mm2 # pack U 00 00 u2 u0 00 00 u2 u0 \n\
103 pand i_00ffw, %%mm3 # get V even 00 v6 00 v4 00 v2 00 v0 \n\
104 packuswb %%mm3, %%mm3 # pack V 00 00 v2 v0 00 00 v2 v0 \n\
105 punpcklbw %%mm3, %%mm2 # 00 00 00 00 v2 u2 v0 u0 \n\
106 psubsw i_80w, %%mm2 # U,V -= 128 \n\
107 punpcklbw %%mm2, %%mm0 # v2 y6 u2 y4 v0 y2 u0 y0 \n\
108 movq %%mm0, (%0) # Store YUYV \n\
109 pand i_00ffw, %%mm1 # get Y even 00 Y6 00 Y4 00 Y2 00 Y0 \n\
110 packuswb %%mm1, %%mm1 # pack Y Y6 Y4 Y2 Y0 Y6 Y4 Y2 Y0 \n\
111 punpcklbw %%mm2, %%mm1 # v2 Y6 u2 Y4 v0 Y2 u0 Y0 \n\
112 movq %%mm1, (%1) # Store YUYV \n\
115 #elif defined( MODULE_NAME_IS_i420_yuy2_sse2 )
119 #define SSE2_CALL(SSE2_INSTRUCTIONS) \
121 __asm__ __volatile__( \
125 : "r" (p_line1), "r" (p_line2), "r" (p_y1), "r" (p_y2), \
126 "r" (p_u), "r" (p_v) ); \
127 p_line1 += 32; p_line2 += 32; p_y1 += 16; p_y2 += 16; \
128 p_u += 8; p_v += 8; \
131 #define SSE2_YUV420_YUYV_ALIGNED " \n\
132 movq (%4), %%xmm1 # Load 8 Cb u7 u6 u5 u4 u3 u2 u1 u0 \n\
133 movq (%5), %%xmm2 # Load 8 Cr v7 06 v5 v4 v3 v2 v1 v0 \n\
134 movdqa (%2), %%xmm0 # Load 16 Y y15 y14 y13 .. y2 y1 y0 \n\
135 movdqa (%3), %%xmm3 # Load 16 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
136 punpcklbw %%xmm2, %%xmm1 # v7 u7 v6 u6 .. u1 v0 u0 \n\
137 movdqa %%xmm0, %%xmm2 # y15 y14 y13 .. y2 y1 y0 \n\
138 punpcklbw %%xmm1, %%xmm2 # v3 y7 u3 .. v0 y1 u0 y0 \n\
139 movntdq %%xmm2, (%0) # Store low YUYV \n\
140 punpckhbw %%xmm1, %%xmm0 # v3 y7 u3 y6 v2 y5 u2 y4 \n\
141 movntdq %%xmm0, 16(%0) # Store high YUYV \n\
142 movdqa %%xmm3, %%xmm4 # Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
143 punpcklbw %%xmm1, %%xmm4 # v1 Y3 u1 Y2 v0 Y1 u0 Y0 \n\
144 movntdq %%xmm4, (%1) # Store low YUYV \n\
145 punpckhbw %%xmm1, %%xmm3 # v3 Y7 u3 Y6 v2 Y5 u2 Y4 \n\
146 movntdq %%xmm3, 16(%1) # Store high YUYV \n\
149 #define SSE2_YUV420_YUYV_UNALIGNED " \n\
150 movq (%4), %%xmm1 # Load 8 Cb 00 00 00 00 u3 u2 u1 u0 \n\
151 movq (%5), %%xmm2 # Load 8 Cr 00 00 00 00 v3 v2 v1 v0 \n\
152 movdqu (%2), %%xmm0 # Load 16 Y y7 y6 y5 y4 y3 y2 y1 y0 \n\
153 movdqu (%3), %%xmm3 # Load 16 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
154 punpcklbw %%xmm2, %%xmm1 # v3 u3 v2 u2 v1 u1 v0 u0 \n\
155 movdqa %%xmm0, %%xmm2 # y7 y6 y5 y4 y3 y2 y1 y0 \n\
156 punpcklbw %%xmm1, %%xmm2 # v1 y3 u1 y2 v0 y1 u0 y0 \n\
157 movdqu %%xmm2, (%0) # Store low YUYV \n\
158 punpckhbw %%xmm1, %%xmm0 # v3 y7 u3 y6 v2 y5 u2 y4 \n\
159 movdqu %%xmm0, 16(%0) # Store high YUYV \n\
160 movdqa %%xmm3, %%xmm4 # Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
161 punpcklbw %%xmm1, %%xmm4 # v1 Y3 u1 Y2 v0 Y1 u0 Y0 \n\
162 movdqu %%xmm4, (%1) # Store low YUYV \n\
163 punpckhbw %%xmm1, %%xmm3 # v3 Y7 u3 Y6 v2 Y5 u2 Y4 \n\
164 movdqu %%xmm3, 16(%1) # Store high YUYV \n\
167 #define SSE2_YUV420_YVYU_ALIGNED " \n\
168 movq (%4), %%xmm2 # Load 8 Cb 00 00 00 00 u3 u2 u1 u0 \n\
169 movq (%5), %%xmm1 # Load 8 Cr 00 00 00 00 v3 v2 v1 v0 \n\
170 movdqa (%2), %%xmm0 # Load 16 Y y7 y6 y5 y4 y3 y2 y1 y0 \n\
171 movdqa (%3), %%xmm3 # Load 16 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
172 punpcklbw %%xmm2, %%xmm1 # u3 v3 u2 v2 u1 v1 u0 v0 \n\
173 movdqa %%xmm0, %%xmm2 # y7 y6 y5 y4 y3 y2 y1 y0 \n\
174 punpcklbw %%xmm1, %%xmm2 # u1 y3 v1 y2 u0 y1 v0 y0 \n\
175 movntdq %%xmm2, (%0) # Store low YUYV \n\
176 punpckhbw %%xmm1, %%xmm0 # u3 y7 v3 y6 u2 y5 v2 y4 \n\
177 movntdq %%xmm0, 16(%0) # Store high YUYV \n\
178 movdqa %%xmm3, %%xmm4 # Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
179 punpcklbw %%xmm1, %%xmm4 # u1 Y3 v1 Y2 u0 Y1 v0 Y0 \n\
180 movntdq %%xmm4, (%1) # Store low YUYV \n\
181 punpckhbw %%xmm1, %%xmm3 # u3 Y7 v3 Y6 u2 Y5 v2 Y4 \n\
182 movntdq %%xmm3, 16(%1) # Store high YUYV \n\
185 #define SSE2_YUV420_YVYU_UNALIGNED " \n\
186 movq (%4), %%xmm2 # Load 8 Cb 00 00 00 00 u3 u2 u1 u0 \n\
187 movq (%5), %%xmm1 # Load 8 Cr 00 00 00 00 v3 v2 v1 v0 \n\
188 movdqu (%2), %%xmm0 # Load 16 Y y7 y6 y5 y4 y3 y2 y1 y0 \n\
189 movdqu (%3), %%xmm3 # Load 16 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
190 punpcklbw %%xmm2, %%xmm1 # u3 v3 u2 v2 u1 v1 u0 v0 \n\
191 movdqu %%xmm0, %%xmm2 # y7 y6 y5 y4 y3 y2 y1 y0 \n\
192 punpcklbw %%xmm1, %%xmm2 # u1 y3 v1 y2 u0 y1 v0 y0 \n\
193 movdqu %%xmm2, (%0) # Store low YUYV \n\
194 punpckhbw %%xmm1, %%xmm0 # u3 y7 v3 y6 u2 y5 v2 y4 \n\
195 movdqu %%xmm0, 16(%0) # Store high YUYV \n\
196 movdqu %%xmm3, %%xmm4 # Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
197 punpcklbw %%xmm1, %%xmm4 # u1 Y3 v1 Y2 u0 Y1 v0 Y0 \n\
198 movdqu %%xmm4, (%1) # Store low YUYV \n\
199 punpckhbw %%xmm1, %%xmm3 # u3 Y7 v3 Y6 u2 Y5 v2 Y4 \n\
200 movdqu %%xmm3, 16(%1) # Store high YUYV \n\
203 #define SSE2_YUV420_UYVY_ALIGNED " \n\
204 movq (%4), %%xmm1 # Load 8 Cb 00 00 00 00 u3 u2 u1 u0 \n\
205 movq (%5), %%xmm2 # Load 8 Cr 00 00 00 00 v3 v2 v1 v0 \n\
206 movdqa (%2), %%xmm0 # Load 16 Y y7 y6 y5 y4 y3 y2 y1 y0 \n\
207 movdqa (%3), %%xmm3 # Load 16 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
208 punpcklbw %%xmm2, %%xmm1 # v3 u3 v2 u2 v1 u1 v0 u0 \n\
209 movdqa %%xmm1, %%xmm2 # v3 u3 v2 u2 v1 u1 v0 u0 \n\
210 punpcklbw %%xmm0, %%xmm2 # y3 v1 y2 u1 y1 v0 y0 u0 \n\
211 movntdq %%xmm2, (%0) # Store low UYVY \n\
212 movdqa %%xmm1, %%xmm2 # u3 v3 u2 v2 u1 v1 u0 v0 \n\
213 punpckhbw %%xmm0, %%xmm2 # y3 v1 y2 u1 y1 v0 y0 u0 \n\
214 movntdq %%xmm2, 16(%0) # Store high UYVY \n\
215 movdqa %%xmm1, %%xmm4 # u3 v3 u2 v2 u1 v1 u0 v0 \n\
216 punpcklbw %%xmm3, %%xmm4 # Y3 v1 Y2 u1 Y1 v0 Y0 u0 \n\
217 movntdq %%xmm4, (%1) # Store low UYVY \n\
218 punpckhbw %%xmm3, %%xmm1 # Y7 v3 Y6 u3 Y5 v2 Y4 u2 \n\
219 movntdq %%xmm1, 16(%1) # Store high UYVY \n\
222 #define SSE2_YUV420_UYVY_UNALIGNED " \n\
223 movq (%4), %%xmm1 # Load 8 Cb 00 00 00 00 u3 u2 u1 u0 \n\
224 movq (%5), %%xmm2 # Load 8 Cr 00 00 00 00 v3 v2 v1 v0 \n\
225 movdqu (%2), %%xmm0 # Load 16 Y y7 y6 y5 y4 y3 y2 y1 y0 \n\
226 movdqu (%3), %%xmm3 # Load 16 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
227 punpcklbw %%xmm2, %%xmm1 # v3 u3 v2 u2 v1 u1 v0 u0 \n\
228 movdqu %%xmm1, %%xmm2 # v3 u3 v2 u2 v1 u1 v0 u0 \n\
229 punpcklbw %%xmm0, %%xmm2 # y3 v1 y2 u1 y1 v0 y0 u0 \n\
230 movdqu %%xmm2, (%0) # Store low UYVY \n\
231 movdqu %%xmm1, %%xmm2 # u3 v3 u2 v2 u1 v1 u0 v0 \n\
232 punpckhbw %%xmm0, %%xmm2 # y3 v1 y2 u1 y1 v0 y0 u0 \n\
233 movdqu %%xmm2, 16(%0) # Store high UYVY \n\
234 movdqu %%xmm1, %%xmm4 # u3 v3 u2 v2 u1 v1 u0 v0 \n\
235 punpcklbw %%xmm3, %%xmm4 # Y3 v1 Y2 u1 Y1 v0 Y0 u0 \n\
236 movdqu %%xmm4, (%1) # Store low UYVY \n\
237 punpckhbw %%xmm3, %%xmm1 # Y7 v3 Y6 u3 Y5 v2 Y4 u2 \n\
238 movdqu %%xmm1, 16(%1) # Store high UYVY \n\
243 /* Used in both accelerated and C modules */
245 #define C_YUV420_YVYU( ) \
246 *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++; \
247 *(p_line1)++ = *(p_line2)++ = *(p_v)++; \
248 *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++; \
249 *(p_line1)++ = *(p_line2)++ = *(p_u)++; \
251 #define C_YUV420_Y211( ) \
252 *(p_line1)++ = *(p_y1); p_y1 += 2; \
253 *(p_line2)++ = *(p_y2); p_y2 += 2; \
254 *(p_line1)++ = *(p_line2)++ = *(p_u) - 0x80; p_u += 2; \
255 *(p_line1)++ = *(p_y1); p_y1 += 2; \
256 *(p_line2)++ = *(p_y2); p_y2 += 2; \
257 *(p_line1)++ = *(p_line2)++ = *(p_v) - 0x80; p_v += 2; \
260 #define C_YUV420_YUYV( ) \
261 *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++; \
262 *(p_line1)++ = *(p_line2)++ = *(p_u)++; \
263 *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++; \
264 *(p_line1)++ = *(p_line2)++ = *(p_v)++; \
266 #define C_YUV420_UYVY( ) \
267 *(p_line1)++ = *(p_line2)++ = *(p_u)++; \
268 *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++; \
269 *(p_line1)++ = *(p_line2)++ = *(p_v)++; \
270 *(p_line1)++ = *(p_y1)++; *(p_line2)++ = *(p_y2)++; \