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
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
|
From: Mans Rullgard <mans@mansr.com>
Date: Fri, 13 Jun 2008 01:21:58 +0000 (+0100)
Subject: ARM: NEON optimised put_pixels functions
X-Git-Url: http://git.mansr.com/?p=ffmpeg.mru;a=commitdiff_plain;h=86410ed1948118a29c70946d5294df9feb04dfef
ARM: NEON optimised put_pixels functions
---
diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index d91185e..27746df 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -433,6 +433,10 @@ ASM_OBJS-$(HAVE_ARMV5TE) += armv4l/simple_idct_armv5te.o \
ASM_OBJS-$(HAVE_ARMV6) += armv4l/simple_idct_armv6.o \
+OBJS-$(HAVE_NEON) += armv4l/dsputil_neon.o \
+
+ASM_OBJS-$(HAVE_NEON) += armv4l/dsputil_neon_s.o \
+
OBJS-$(HAVE_VIS) += sparc/dsputil_vis.o \
sparc/simple_idct_vis.o \
diff --git a/libavcodec/armv4l/dsputil_arm.c b/libavcodec/armv4l/dsputil_arm.c
index 100b89e..89b51e7 100644
--- a/libavcodec/armv4l/dsputil_arm.c
+++ b/libavcodec/armv4l/dsputil_arm.c
@@ -26,6 +26,7 @@
extern void dsputil_init_iwmmxt(DSPContext* c, AVCodecContext *avctx);
extern void ff_float_init_arm_vfp(DSPContext* c, AVCodecContext *avctx);
+extern void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx);
extern void j_rev_dct_ARM(DCTELEM *data);
extern void simple_idct_ARM(DCTELEM *data);
@@ -302,4 +303,7 @@ void dsputil_init_armv4l(DSPContext* c, AVCodecContext *avctx)
#ifdef HAVE_ARMVFP
ff_float_init_arm_vfp(c, avctx);
#endif
+#ifdef HAVE_NEON
+ ff_dsputil_init_neon(c, avctx);
+#endif
}
diff --git a/libavcodec/armv4l/dsputil_neon.c b/libavcodec/armv4l/dsputil_neon.c
new file mode 100644
index 0000000..8a10dde
--- /dev/null
+++ b/libavcodec/armv4l/dsputil_neon.c
@@ -0,0 +1,67 @@
+/*
+ * ARM NEON optimised DSP functions
+ * Copyright (c) 2008 Mans Rullgard <mans@mansr.com>
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include <stdint.h>
+
+#include "libavcodec/avcodec.h"
+#include "libavcodec/dsputil.h"
+
+void ff_put_pixels16_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels16_x2_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels16_y2_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels16_xy2_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels8_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels8_x2_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels8_y2_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels8_xy2_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels16_x2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels16_y2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels16_xy2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels8_x2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels8_y2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
+void ff_put_pixels8_xy2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
+
+void ff_put_h264_qpel16_mc00_neon(uint8_t *, uint8_t *, int);
+void ff_put_h264_qpel8_mc00_neon(uint8_t *, uint8_t *, int);
+
+void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx)
+{
+ c->put_pixels_tab[0][0] = ff_put_pixels16_neon;
+ c->put_pixels_tab[0][1] = ff_put_pixels16_x2_neon;
+ c->put_pixels_tab[0][2] = ff_put_pixels16_y2_neon;
+ c->put_pixels_tab[0][3] = ff_put_pixels16_xy2_neon;
+ c->put_pixels_tab[1][0] = ff_put_pixels8_neon;
+ c->put_pixels_tab[1][1] = ff_put_pixels8_x2_neon;
+ c->put_pixels_tab[1][2] = ff_put_pixels8_y2_neon;
+ c->put_pixels_tab[1][3] = ff_put_pixels8_xy2_neon;
+
+ c->put_no_rnd_pixels_tab[0][0] = ff_put_pixels16_neon;
+ c->put_no_rnd_pixels_tab[0][1] = ff_put_pixels16_x2_no_rnd_neon;
+ c->put_no_rnd_pixels_tab[0][2] = ff_put_pixels16_y2_no_rnd_neon;
+ c->put_no_rnd_pixels_tab[0][3] = ff_put_pixels16_xy2_no_rnd_neon;
+ c->put_no_rnd_pixels_tab[1][0] = ff_put_pixels8_neon;
+ c->put_no_rnd_pixels_tab[1][1] = ff_put_pixels8_x2_no_rnd_neon;
+ c->put_no_rnd_pixels_tab[1][2] = ff_put_pixels8_y2_no_rnd_neon;
+ c->put_no_rnd_pixels_tab[1][3] = ff_put_pixels8_xy2_no_rnd_neon;
+
+ c->put_h264_qpel_pixels_tab[0][0] = ff_put_h264_qpel16_mc00_neon;
+ c->put_h264_qpel_pixels_tab[1][0] = ff_put_h264_qpel8_mc00_neon;
+}
diff --git a/libavcodec/armv4l/dsputil_neon_s.S b/libavcodec/armv4l/dsputil_neon_s.S
new file mode 100644
index 0000000..fc5e401
--- /dev/null
+++ b/libavcodec/armv4l/dsputil_neon_s.S
@@ -0,0 +1,254 @@
+/*
+ * ARM NEON optimised DSP functions
+ * Copyright (c) 2008 Mans Rullgard <mans@mansr.com>
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+ .fpu neon
+ .text
+
+ .macro put_pixels16
+ dmb
+1: vld1.64 {d0, d1}, [r1], r2
+ vld1.64 {d2, d3}, [r1], r2
+ vld1.64 {d4, d5}, [r1], r2
+ vld1.64 {d6, d7}, [r1], r2
+ pld [r1]
+ subs r3, r3, #4
+ vst1.64 {d0, d1}, [r0,:128], r2
+ vst1.64 {d2, d3}, [r0,:128], r2
+ vst1.64 {d4, d5}, [r0,:128], r2
+ vst1.64 {d6, d7}, [r0,:128], r2
+ bne 1b
+ bx lr
+ .endm
+
+ .macro put_pixels16_x2 vhadd=vrhadd.u8
+ dmb
+1: vld1.64 {d0-d2}, [r1], r2
+ vld1.64 {d4-d6}, [r1], r2
+ pld [r1]
+ subs r3, r3, #2
+ vext.8 q1, q0, q1, #1
+ vext.8 q3, q2, q3, #1
+ \vhadd q0, q0, q1
+ \vhadd q2, q2, q3
+ vst1.64 {d0, d1}, [r0,:128], r2
+ vst1.64 {d4, d5}, [r0,:128], r2
+ bne 1b
+ bx lr
+ .endm
+
+ .macro put_pixels16_y2 vhadd=vrhadd.u8
+ push {lr}
+ add ip, r1, r2
+ lsl lr, r2, #1
+ vld1.64 {d0, d1}, [r1], lr
+ vld1.64 {d2, d3}, [ip], lr
+ dmb
+1: subs r3, r3, #2
+ \vhadd q2, q0, q1
+ vld1.64 {d0, d1}, [r1], lr
+ vst1.64 {d4, d5}, [r0,:128], r2
+ \vhadd q2, q0, q1
+ vld1.64 {d2, d3}, [ip], lr
+ vst1.64 {d4, d5}, [r0,:128], r2
+ bne 1b
+ pop {pc}
+ .endm
+
+ .macro put_pixels16_xy2 vshrn=vrshrn.u16 no_rnd=0
+ push {lr}
+ lsl lr, r2, #1
+ add ip, r1, r2
+ vld1.64 {d0-d2}, [r1], lr
+ vld1.64 {d4-d6}, [ip], lr
+ .if \no_rnd
+ vmov.i16 q13, #1
+ .endif
+ pld [r1]
+ pld [ip]
+ vext.8 q1, q0, q1, #1
+ vext.8 q3, q2, q3, #1
+ vaddl.u8 q8, d0, d2
+ vaddl.u8 q10, d1, d3
+ vaddl.u8 q9, d4, d6
+ vaddl.u8 q11, d5, d7
+ dmb
+1: subs r3, r3, #2
+ vld1.64 {d0-d2}, [r1], lr
+ vadd.u16 q12, q8, q9
+ pld [r1]
+ .if \no_rnd
+ vadd.u16 q12, q12, q13
+ .endif
+ vext.8 q15, q0, q1, #1
+ vadd.u16 q1 , q10, q11
+ \vshrn d28, q12, #2
+ .if \no_rnd
+ vadd.u16 q1, q1, q13
+ .endif
+ \vshrn d29, q1, #2
+ vaddl.u8 q8, d0, d30
+ vld1.64 {d2-d4}, [ip], lr
+ vaddl.u8 q10, d1, d31
+ vst1.64 {d28,d29}, [r0,:128], r2
+ vadd.u16 q12, q8, q9
+ pld [ip]
+ .if \no_rnd
+ vadd.u16 q12, q12, q13
+ .endif
+ vext.8 q2, q1, q2, #1
+ vadd.u16 q0, q10, q11
+ \vshrn d30, q12, #2
+ .if \no_rnd
+ vadd.u16 q0, q0, q13
+ .endif
+ \vshrn d31, q0, #2
+ vaddl.u8 q9, d2, d4
+ vaddl.u8 q11, d3, d5
+ vst1.64 {d30,d31}, [r0,:128], r2
+ bgt 1b
+ pop {pc}
+ .endm
+
+ .macro put_pixels8
+ dmb
+1: vld1.64 {d0}, [r1], r2
+ vld1.64 {d1}, [r1], r2
+ vld1.64 {d2}, [r1], r2
+ vld1.64 {d3}, [r1], r2
+ subs r3, r3, #4
+ vst1.64 {d0}, [r0,:64], r2
+ vst1.64 {d1}, [r0,:64], r2
+ vst1.64 {d2}, [r0,:64], r2
+ vst1.64 {d3}, [r0,:64], r2
+ bne 1b
+ bx lr
+ .endm
+
+ .macro put_pixels8_x2 vhadd=vrhadd.u8
+ dmb
+1: vld1.64 {d0, d1}, [r1], r2
+ vld1.64 {d2, d3}, [r1], r2
+ pld [r1]
+ subs r3, r3, #2
+ vext.8 d1, d0, d1, #1
+ vext.8 d3, d2, d3, #1
+ vswp d1, d2
+ \vhadd q0, q0, q1
+ vst1.64 {d0}, [r0,:64], r2
+ vst1.64 {d1}, [r0,:64], r2
+ bne 1b
+ bx lr
+ .endm
+
+ .macro put_pixels8_y2 vhadd=vrhadd.u8
+ push {lr}
+ add ip, r1, r2
+ lsl lr, r2, #1
+ vld1.64 {d0}, [r1], lr
+ vld1.64 {d1}, [ip], lr
+ dmb
+1: subs r3, r3, #2
+ \vhadd d4, d0, d1
+ vld1.64 {d0}, [r1], lr
+ vst1.64 {d4}, [r0,:64], r2
+ \vhadd d4, d0, d1
+ vld1.64 {d1}, [ip], lr
+ vst1.64 {d4}, [r0,:64], r2
+ bne 1b
+ pop {pc}
+ .endm
+
+ .macro put_pixels8_xy2 vshrn=vrshrn.u16 no_rnd=0
+ push {lr}
+ lsl lr, r2, #1
+ add ip, r1, r2
+ vld1.64 {d0, d1}, [r1], lr
+ vld1.64 {d2, d3}, [ip], lr
+ .if \no_rnd
+ vmov.i16 q11, #1
+ .endif
+ pld [r1]
+ pld [ip]
+ vext.8 d4, d0, d1, #1
+ vext.8 d6, d2, d3, #1
+ vaddl.u8 q8, d0, d4
+ vaddl.u8 q9, d2, d6
+ dmb
+1: subs r3, r3, #2
+ vld1.64 {d0, d1}, [r1], lr
+ pld [r1]
+ vadd.u16 q10, q8, q9
+ vext.8 d4, d0, d1, #1
+ .if \no_rnd
+ vadd.u16 q10, q10, q11
+ .endif
+ vaddl.u8 q8, d0, d4
+ \vshrn d5, q10, #2
+ vld1.64 {d2, d3}, [ip], lr
+ vadd.u16 q10, q8, q9
+ pld [ip]
+ .if \no_rnd
+ vadd.u16 q10, q10, q11
+ .endif
+ vst1.64 {d5}, [r0,:64], r2
+ \vshrn d7, q10, #2
+ vext.8 d6, d2, d3, #1
+ vaddl.u8 q9, d2, d6
+ vst1.64 {d7}, [r0,:64], r2
+ bgt 1b
+ pop {pc}
+ .endm
+
+ .macro extern name
+ .global \name
+ .type \name, %function
+ .func \name
+\name:
+ .endm
+
+ .macro defun name suf rnd_op args:vararg
+ extern ff_\name\suf\()_neon
+ \name \rnd_op \args
+ .endfunc
+ .endm
+
+ .macro defun2 name args:vararg
+ defun \name
+ defun \name \args
+ .endm
+
+ extern ff_put_h264_qpel16_mc00_neon
+ mov r3, #16
+ .endfunc
+
+ defun put_pixels16
+ defun2 put_pixels16_x2, _no_rnd, vhadd.u8
+ defun2 put_pixels16_y2, _no_rnd, vhadd.u8
+ defun2 put_pixels16_xy2, _no_rnd, vshrn.u16, 1
+
+ extern ff_put_h264_qpel8_mc00_neon
+ mov r3, #8
+ .endfunc
+
+ defun put_pixels8
+ defun2 put_pixels8_x2, _no_rnd, vhadd.u8
+ defun2 put_pixels8_y2, _no_rnd, vhadd.u8
+ defun2 put_pixels8_xy2, _no_rnd, vshrn.u16, 1
|
