Mercurial > libavcodec.hg
annotate i386/snowdsp_mmx.c @ 3207:33110c1008a4 libavcodec
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
Patch by Robert Edele < yartrebo AH earthlink POIS net >
Original thread:
Date: Mar 20, 2006 5:54 PM
Subject: [Ffmpeg-devel] [PATCH] snow mmx + sse2 part 3
author | gpoirier |
---|---|
date | Mon, 20 Mar 2006 22:27:59 +0000 |
parents | |
children | 81cafbc23b8d |
rev | line source |
---|---|
3207
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
1 /* |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
2 * MMX and SSE2 optimized snow DSP utils |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
3 * Copyright (c) 2005-2006 Robert Edele <yartrebo@earthlink.net> |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
4 * |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
5 * This library is free software; you can redistribute it and/or |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
6 * modify it under the terms of the GNU Lesser General Public |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
7 * License as published by the Free Software Foundation; either |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
8 * version 2 of the License, or (at your option) any later version. |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
9 * |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
10 * This library is distributed in the hope that it will be useful, |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
13 * Lesser General Public License for more details. |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
14 * |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
15 * You should have received a copy of the GNU Lesser General Public |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
16 * License along with this library; if not, write to the Free Software |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
18 */ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
19 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
20 #include "../avcodec.h" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
21 #include "../snow.h" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
22 #include "mmx.h" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
23 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
24 #define snow_vertical_compose_sse2_load_add(op,r,t0,t1,t2,t3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
25 ""op" (%%"r",%%"REG_d",4), %%"t0" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
26 ""op" 16(%%"r",%%"REG_d",4), %%"t1" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
27 ""op" 32(%%"r",%%"REG_d",4), %%"t2" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
28 ""op" 48(%%"r",%%"REG_d",4), %%"t3" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
29 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
30 #define snow_vertical_compose_sse2_load(r,t0,t1,t2,t3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
31 snow_vertical_compose_sse2_load_add("movdqa",r,t0,t1,t2,t3) |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
32 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
33 #define snow_vertical_compose_sse2_add(r,t0,t1,t2,t3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
34 snow_vertical_compose_sse2_load_add("paddd",r,t0,t1,t2,t3) |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
35 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
36 #define snow_vertical_compose_sse2_sub(s0,s1,s2,s3,t0,t1,t2,t3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
37 "psubd %%"s0", %%"t0" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
38 "psubd %%"s1", %%"t1" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
39 "psubd %%"s2", %%"t2" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
40 "psubd %%"s3", %%"t3" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
41 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
42 #define snow_vertical_compose_sse2_store(w,s0,s1,s2,s3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
43 "movdqa %%"s0", (%%"w",%%"REG_d",4) \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
44 "movdqa %%"s1", 16(%%"w",%%"REG_d",4) \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
45 "movdqa %%"s2", 32(%%"w",%%"REG_d",4) \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
46 "movdqa %%"s3", 48(%%"w",%%"REG_d",4) \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
47 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
48 #define snow_vertical_compose_sse2_sra(n,t0,t1,t2,t3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
49 "psrad $"n", %%"t0" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
50 "psrad $"n", %%"t1" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
51 "psrad $"n", %%"t2" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
52 "psrad $"n", %%"t3" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
53 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
54 #define snow_vertical_compose_sse2_r2r_add(s0,s1,s2,s3,t0,t1,t2,t3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
55 "paddd %%"s0", %%"t0" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
56 "paddd %%"s1", %%"t1" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
57 "paddd %%"s2", %%"t2" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
58 "paddd %%"s3", %%"t3" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
59 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
60 #define snow_vertical_compose_sse2_sll(n,t0,t1,t2,t3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
61 "pslld $"n", %%"t0" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
62 "pslld $"n", %%"t1" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
63 "pslld $"n", %%"t2" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
64 "pslld $"n", %%"t3" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
65 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
66 #define snow_vertical_compose_sse2_move(s0,s1,s2,s3,t0,t1,t2,t3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
67 "movdqa %%"s0", %%"t0" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
68 "movdqa %%"s1", %%"t1" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
69 "movdqa %%"s2", %%"t2" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
70 "movdqa %%"s3", %%"t3" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
71 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
72 void ff_snow_vertical_compose97i_sse2(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, DWTELEM *b3, DWTELEM *b4, DWTELEM *b5, int width){ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
73 long i = width; |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
74 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
75 while(i & 0xF) |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
76 { |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
77 i--; |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
78 b4[i] -= (W_DM*(b3[i] + b5[i])+W_DO)>>W_DS; |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
79 b3[i] -= (W_CM*(b2[i] + b4[i])+W_CO)>>W_CS; |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
80 b2[i] += (W_BM*(b1[i] + b3[i])+4*b2[i]+W_BO)>>W_BS; |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
81 b1[i] += (W_AM*(b0[i] + b2[i])+W_AO)>>W_AS; |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
82 } |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
83 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
84 asm volatile ( |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
85 "jmp 2f \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
86 "1: \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
87 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
88 "mov %6, %%"REG_a" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
89 "mov %4, %%"REG_b" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
90 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
91 snow_vertical_compose_sse2_load(REG_b,"xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
92 snow_vertical_compose_sse2_add(REG_a,"xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
93 snow_vertical_compose_sse2_move("xmm0","xmm2","xmm4","xmm6","xmm1","xmm3","xmm5","xmm7") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
94 snow_vertical_compose_sse2_sll("1","xmm0","xmm2","xmm4","xmm6")\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
95 snow_vertical_compose_sse2_r2r_add("xmm1","xmm3","xmm5","xmm7","xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
96 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
97 "pcmpeqd %%xmm1, %%xmm1 \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
98 "pslld $31, %%xmm1 \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
99 "psrld $29, %%xmm1 \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
100 "mov %5, %%"REG_a" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
101 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
102 snow_vertical_compose_sse2_r2r_add("xmm1","xmm1","xmm1","xmm1","xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
103 snow_vertical_compose_sse2_sra("3","xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
104 snow_vertical_compose_sse2_load(REG_a,"xmm1","xmm3","xmm5","xmm7") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
105 snow_vertical_compose_sse2_sub("xmm0","xmm2","xmm4","xmm6","xmm1","xmm3","xmm5","xmm7") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
106 snow_vertical_compose_sse2_store(REG_a,"xmm1","xmm3","xmm5","xmm7") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
107 "mov %3, %%"REG_c" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
108 snow_vertical_compose_sse2_load(REG_b,"xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
109 snow_vertical_compose_sse2_add(REG_c,"xmm1","xmm3","xmm5","xmm7") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
110 snow_vertical_compose_sse2_sub("xmm1","xmm3","xmm5","xmm7","xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
111 snow_vertical_compose_sse2_store(REG_b,"xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
112 "mov %2, %%"REG_a" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
113 snow_vertical_compose_sse2_load(REG_c,"xmm1","xmm3","xmm5","xmm7") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
114 snow_vertical_compose_sse2_add(REG_a,"xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
115 snow_vertical_compose_sse2_sll("2","xmm1","xmm3","xmm5","xmm7")\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
116 snow_vertical_compose_sse2_r2r_add("xmm1","xmm3","xmm5","xmm7","xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
117 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
118 "pcmpeqd %%xmm1, %%xmm1 \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
119 "pslld $31, %%xmm1 \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
120 "psrld $28, %%xmm1 \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
121 "mov %1, %%"REG_b" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
122 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
123 snow_vertical_compose_sse2_r2r_add("xmm1","xmm1","xmm1","xmm1","xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
124 snow_vertical_compose_sse2_sra("4","xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
125 snow_vertical_compose_sse2_add(REG_c,"xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
126 snow_vertical_compose_sse2_store(REG_c,"xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
127 snow_vertical_compose_sse2_add(REG_b,"xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
128 snow_vertical_compose_sse2_move("xmm0","xmm2","xmm4","xmm6","xmm1","xmm3","xmm5","xmm7") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
129 snow_vertical_compose_sse2_sll("1","xmm0","xmm2","xmm4","xmm6")\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
130 snow_vertical_compose_sse2_r2r_add("xmm1","xmm3","xmm5","xmm7","xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
131 snow_vertical_compose_sse2_sra("1","xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
132 snow_vertical_compose_sse2_add(REG_a,"xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
133 snow_vertical_compose_sse2_store(REG_a,"xmm0","xmm2","xmm4","xmm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
134 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
135 "2: \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
136 "sub $16, %%"REG_d" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
137 "jge 1b \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
138 :"+d"(i) |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
139 : |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
140 "m"(b0),"m"(b1),"m"(b2),"m"(b3),"m"(b4),"m"(b5): |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
141 "%"REG_a"","%"REG_b"","%"REG_c""); |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
142 } |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
143 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
144 #define snow_vertical_compose_mmx_load_add(op,r,t0,t1,t2,t3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
145 ""op" (%%"r",%%"REG_d",4), %%"t0" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
146 ""op" 8(%%"r",%%"REG_d",4), %%"t1" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
147 ""op" 16(%%"r",%%"REG_d",4), %%"t2" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
148 ""op" 24(%%"r",%%"REG_d",4), %%"t3" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
149 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
150 #define snow_vertical_compose_mmx_load(r,t0,t1,t2,t3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
151 snow_vertical_compose_mmx_load_add("movq",r,t0,t1,t2,t3) |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
152 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
153 #define snow_vertical_compose_mmx_add(r,t0,t1,t2,t3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
154 snow_vertical_compose_mmx_load_add("paddd",r,t0,t1,t2,t3) |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
155 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
156 #define snow_vertical_compose_mmx_sub(s0,s1,s2,s3,t0,t1,t2,t3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
157 snow_vertical_compose_sse2_sub(s0,s1,s2,s3,t0,t1,t2,t3) |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
158 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
159 #define snow_vertical_compose_mmx_store(w,s0,s1,s2,s3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
160 "movq %%"s0", (%%"w",%%"REG_d",4) \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
161 "movq %%"s1", 8(%%"w",%%"REG_d",4) \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
162 "movq %%"s2", 16(%%"w",%%"REG_d",4) \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
163 "movq %%"s3", 24(%%"w",%%"REG_d",4) \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
164 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
165 #define snow_vertical_compose_mmx_sra(n,t0,t1,t2,t3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
166 snow_vertical_compose_sse2_sra(n,t0,t1,t2,t3) |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
167 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
168 #define snow_vertical_compose_mmx_r2r_add(s0,s1,s2,s3,t0,t1,t2,t3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
169 snow_vertical_compose_sse2_r2r_add(s0,s1,s2,s3,t0,t1,t2,t3) |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
170 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
171 #define snow_vertical_compose_mmx_sll(n,t0,t1,t2,t3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
172 snow_vertical_compose_sse2_sll(n,t0,t1,t2,t3) |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
173 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
174 #define snow_vertical_compose_mmx_move(s0,s1,s2,s3,t0,t1,t2,t3)\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
175 "movq %%"s0", %%"t0" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
176 "movq %%"s1", %%"t1" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
177 "movq %%"s2", %%"t2" \n\t"\ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
178 "movq %%"s3", %%"t3" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
179 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
180 void ff_snow_vertical_compose97i_mmx(DWTELEM *b0, DWTELEM *b1, DWTELEM *b2, DWTELEM *b3, DWTELEM *b4, DWTELEM *b5, int width){ |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
181 long i = width; |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
182 while(i & 0x7) |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
183 { |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
184 i--; |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
185 b4[i] -= (W_DM*(b3[i] + b5[i])+W_DO)>>W_DS; |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
186 b3[i] -= (W_CM*(b2[i] + b4[i])+W_CO)>>W_CS; |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
187 b2[i] += (W_BM*(b1[i] + b3[i])+4*b2[i]+W_BO)>>W_BS; |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
188 b1[i] += (W_AM*(b0[i] + b2[i])+W_AO)>>W_AS; |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
189 } |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
190 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
191 asm volatile( |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
192 "jmp 2f \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
193 "1: \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
194 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
195 "mov %6, %%"REG_a" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
196 "mov %4, %%"REG_b" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
197 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
198 snow_vertical_compose_mmx_load(REG_b,"mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
199 snow_vertical_compose_mmx_add(REG_a,"mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
200 snow_vertical_compose_mmx_move("mm0","mm2","mm4","mm6","mm1","mm3","mm5","mm7") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
201 snow_vertical_compose_mmx_sll("1","mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
202 snow_vertical_compose_mmx_r2r_add("mm1","mm3","mm5","mm7","mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
203 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
204 "pcmpeqd %%mm1, %%mm1 \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
205 "pslld $31, %%mm1 \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
206 "psrld $29, %%mm1 \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
207 "mov %5, %%"REG_a" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
208 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
209 snow_vertical_compose_mmx_r2r_add("mm1","mm1","mm1","mm1","mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
210 snow_vertical_compose_mmx_sra("3","mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
211 snow_vertical_compose_mmx_load(REG_a,"mm1","mm3","mm5","mm7") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
212 snow_vertical_compose_mmx_sub("mm0","mm2","mm4","mm6","mm1","mm3","mm5","mm7") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
213 snow_vertical_compose_mmx_store(REG_a,"mm1","mm3","mm5","mm7") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
214 "mov %3, %%"REG_c" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
215 snow_vertical_compose_mmx_load(REG_b,"mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
216 snow_vertical_compose_mmx_add(REG_c,"mm1","mm3","mm5","mm7") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
217 snow_vertical_compose_mmx_sub("mm1","mm3","mm5","mm7","mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
218 snow_vertical_compose_mmx_store(REG_b,"mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
219 "mov %2, %%"REG_a" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
220 snow_vertical_compose_mmx_load(REG_c,"mm1","mm3","mm5","mm7") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
221 snow_vertical_compose_mmx_add(REG_a,"mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
222 snow_vertical_compose_mmx_sll("2","mm1","mm3","mm5","mm7") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
223 snow_vertical_compose_mmx_r2r_add("mm1","mm3","mm5","mm7","mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
224 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
225 "pcmpeqd %%mm1, %%mm1 \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
226 "pslld $31, %%mm1 \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
227 "psrld $28, %%mm1 \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
228 "mov %1, %%"REG_b" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
229 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
230 snow_vertical_compose_mmx_r2r_add("mm1","mm1","mm1","mm1","mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
231 snow_vertical_compose_mmx_sra("4","mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
232 snow_vertical_compose_mmx_add(REG_c,"mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
233 snow_vertical_compose_mmx_store(REG_c,"mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
234 snow_vertical_compose_mmx_add(REG_b,"mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
235 snow_vertical_compose_mmx_move("mm0","mm2","mm4","mm6","mm1","mm3","mm5","mm7") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
236 snow_vertical_compose_mmx_sll("1","mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
237 snow_vertical_compose_mmx_r2r_add("mm1","mm3","mm5","mm7","mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
238 snow_vertical_compose_mmx_sra("1","mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
239 snow_vertical_compose_mmx_add(REG_a,"mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
240 snow_vertical_compose_mmx_store(REG_a,"mm0","mm2","mm4","mm6") |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
241 |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
242 "2: \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
243 "sub $8, %%"REG_d" \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
244 "jge 1b \n\t" |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
245 :"+d"(i) |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
246 : |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
247 "m"(b0),"m"(b1),"m"(b2),"m"(b3),"m"(b4),"m"(b5): |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
248 "%"REG_a"","%"REG_b"","%"REG_c""); |
33110c1008a4
Add the mmx and sse2 implementations of ff_snow_vertical_compose().
gpoirier
parents:
diff
changeset
|
249 } |