1 ;******************************************************************************
2 ;* VC1 deblocking optimizations
3 ;* Copyright (c) 2009 David Conrad
5 ;* This file is part of FFmpeg.
7 ;* FFmpeg is free software; you can redistribute it and/or
8 ;* modify it under the terms of the GNU Lesser General Public
9 ;* License as published by the Free Software Foundation; either
10 ;* version 2.1 of the License, or (at your option) any later version.
12 ;* FFmpeg is distributed in the hope that it will be useful,
13 ;* but WITHOUT ANY WARRANTY; without even the implied warranty of
14 ;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 ;* Lesser General Public License for more details.
17 ;* You should have received a copy of the GNU Lesser General Public
18 ;* License along with FFmpeg; if not, write to the Free Software
19 ;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 ;******************************************************************************
22 %include "libavutil/x86/x86util.asm"
29 ; dst_low, dst_high (src), zero
30 ; zero-extends one vector from 8 to 16 bits
37 %macro STORE_4_WORDS 6
60 ; in: p1 p0 q0 q1, clobbers p0
61 ; out: p1 = (2*(p1 - q1) - 5*(p0 - q0) + 4) >> 3
62 %macro VC1_LOOP_FILTER_A0 4
82 pcmpgtw m6, m3 ; if (a2 < a0 || a1 < a0)
84 pmullw m3, [pw_5] ; 5*(a3 - a0)
86 psraw m2, 3 ; abs(d/8)
87 pxor m7, m3 ; d_sign ^= a0_sign
95 pcmpgtw m3, m4 ; if (a0 < pq)
102 pxor m3, m7 ; d_sign ^ clip_sign
104 pminsw m2, m4 ; min(d, clip)
106 pand m6, m4 ; filt3 (C return value)
108 ; each set of 4 pixels is not filtered if the 3rd is not
119 pand m3, m2 ; d final
130 ; 1st param: size of filter
131 ; 2nd param: mov suffix equivalent to the filter size
132 %macro VC1_V_LOOP_FILTER 2
143 VC1_LOOP_FILTER_A0 m6, m4, m7, m0
149 VC1_LOOP_FILTER_A0 m7, m4, m1, m2
155 VC1_LOOP_FILTER_A0 m5, m2, m3, m4
162 ; 1st param: size of filter
163 ; NOTE: UNPACK_8TO16 this number of 8 bit numbers are in half a register
164 ; 2nd (optional) param: temp register to use for storing words
165 %macro VC1_H_LOOP_FILTER 1-2
171 TRANSPOSE4x4B 0, 1, 2, 3, 4
185 TRANSPOSE4x4W 0, 1, 2, 3, 4
189 UNPACK_8TO16 bw, 6, 0, 5
190 UNPACK_8TO16 bw, 7, 1, 5
191 VC1_LOOP_FILTER_A0 m6, m0, m7, m1
192 UNPACK_8TO16 bw, 4, 2, 5
193 mova m0, m1 ; m0 = p0
194 VC1_LOOP_FILTER_A0 m7, m1, m4, m2
195 UNPACK_8TO16 bw, 1, 3, 5
197 VC1_LOOP_FILTER_A0 m5, m2, m1, m3
203 STORE_4_WORDS [r0-1], [r0+r1-1], [r0+2*r1-1], [r0+r3-1], m0, %2
206 STORE_4_WORDS [r4-1], [r4+r1-1], [r4+2*r1-1], [r4+r3-1], m0, %2
209 STORE_4_WORDS [r0-1], [r0+r1-1], [r0+2*r1-1], [r0+r3-1], m0, 0
210 STORE_4_WORDS [r4-1], [r4+r1-1], [r4+2*r1-1], [r4+r3-1], m0, 4
215 %macro START_V_FILTER 0
223 %macro START_H_FILTER 1
232 cglobal vc1_v_loop_filter_internal
233 VC1_V_LOOP_FILTER 4, d
236 cglobal vc1_h_loop_filter_internal
237 VC1_H_LOOP_FILTER 4, r4
240 ; void ff_vc1_v_loop_filter4_mmxext(uint8_t *src, int stride, int pq)
241 cglobal vc1_v_loop_filter4, 3,5,0
243 call vc1_v_loop_filter_internal
246 ; void ff_vc1_h_loop_filter4_mmxext(uint8_t *src, int stride, int pq)
247 cglobal vc1_h_loop_filter4, 3,5,0
249 call vc1_h_loop_filter_internal
252 ; void ff_vc1_v_loop_filter8_mmxext(uint8_t *src, int stride, int pq)
253 cglobal vc1_v_loop_filter8, 3,5,0
255 call vc1_v_loop_filter_internal
258 call vc1_v_loop_filter_internal
261 ; void ff_vc1_h_loop_filter8_mmxext(uint8_t *src, int stride, int pq)
262 cglobal vc1_h_loop_filter8, 3,5,0
264 call vc1_h_loop_filter_internal
266 call vc1_h_loop_filter_internal
274 ; void ff_vc1_v_loop_filter8_sse2(uint8_t *src, int stride, int pq)
275 cglobal vc1_v_loop_filter8, 3,5,8
277 VC1_V_LOOP_FILTER 8, q
280 ; void ff_vc1_h_loop_filter8_sse2(uint8_t *src, int stride, int pq)
281 cglobal vc1_h_loop_filter8, 3,6,8
283 VC1_H_LOOP_FILTER 8, r5
287 ; void ff_vc1_v_loop_filter4_ssse3(uint8_t *src, int stride, int pq)
288 cglobal vc1_v_loop_filter4, 3,5,0
290 VC1_V_LOOP_FILTER 4, d
293 ; void ff_vc1_h_loop_filter4_ssse3(uint8_t *src, int stride, int pq)
294 cglobal vc1_h_loop_filter4, 3,5,0
296 VC1_H_LOOP_FILTER 4, r4
300 ; void ff_vc1_v_loop_filter8_ssse3(uint8_t *src, int stride, int pq)
301 cglobal vc1_v_loop_filter8, 3,5,8
303 VC1_V_LOOP_FILTER 8, q
306 ; void ff_vc1_h_loop_filter8_ssse3(uint8_t *src, int stride, int pq)
307 cglobal vc1_h_loop_filter8, 3,6,8
309 VC1_H_LOOP_FILTER 8, r5
313 ; void ff_vc1_h_loop_filter8_sse4(uint8_t *src, int stride, int pq)
314 cglobal vc1_h_loop_filter8, 3,5,8