summaryrefslogtreecommitdiff
path: root/apps/plugins/mpegplayer/idct_arm.S
diff options
context:
space:
mode:
authorJens Arnold <amiconn@rockbox.org>2007-10-17 00:29:44 +0000
committerJens Arnold <amiconn@rockbox.org>2007-10-17 00:29:44 +0000
commit45d4345405d06360370c90f2b3e4036e7df968d1 (patch)
tree981fe8735085d2af9aa4f6a0737b8fe893f7c5a2 /apps/plugins/mpegplayer/idct_arm.S
parent4829f7835a9f1965263438f2ee9587dc19408397 (diff)
downloadrockbox-45d4345405d06360370c90f2b3e4036e7df968d1.tar.gz
rockbox-45d4345405d06360370c90f2b3e4036e7df968d1.zip
Mpegplayer: Convert the assembler optimised ARM IDCT into a proper assembler file. Clean up the CPU selection in idct.c.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@15158 a1c6a512-1295-4272-9138-f99709370657
Diffstat (limited to 'apps/plugins/mpegplayer/idct_arm.S')
-rw-r--r--apps/plugins/mpegplayer/idct_arm.S440
1 files changed, 440 insertions, 0 deletions
diff --git a/apps/plugins/mpegplayer/idct_arm.S b/apps/plugins/mpegplayer/idct_arm.S
new file mode 100644
index 0000000000..a340f40a07
--- /dev/null
+++ b/apps/plugins/mpegplayer/idct_arm.S
@@ -0,0 +1,440 @@
1/***************************************************************************
2 * __________ __ ___.
3 * Open \______ \ ____ ____ | | _\_ |__ _______ ___
4 * Source | _// _ \_/ ___\| |/ /| __ \ / _ \ \/ /
5 * Jukebox | | ( <_> ) \___| < | \_\ ( <_> > < <
6 * Firmware |____|_ /\____/ \___ >__|_ \|___ /\____/__/\_ \
7 * \/ \/ \/ \/ \/
8 * $Id $
9 *
10 * Copyright (C) 2007 by Michael Sevakis
11 *
12 * All files in this archive are subject to the GNU General Public License.
13 * See the file COPYING in the source tree root for full license agreement.
14 *
15 * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
16 * KIND, either express or implied.
17 *
18 ****************************************************************************/
19
20 .global mpeg2_idct_copy_arm
21 .type mpeg2_idct_copy_arm, %function
22 .global mpeg2_idct_add_arm
23 .type mpeg2_idct_add_arm, %function
24
25
26/* Custom calling convention:
27 * r0 contains block pointer and is non-volatile
28 * all non-volatile c context saved and restored on its behalf
29 */
30.idct:
31 add r12, r0, #128
321:
33 ldrsh r1, [r0, #0] /* d0 */
34 ldrsh r2, [r0, #2] /* d1 */
35 ldrsh r3, [r0, #4] /* d2 */
36 ldrsh r4, [r0, #6] /* d3 */
37 ldrsh r5, [r0, #8] /* d0 */
38 ldrsh r6, [r0, #10] /* d1 */
39 ldrsh r7, [r0, #12] /* d2 */
40 ldrsh r8, [r0, #14] /* d3 */
41 orrs r9, r2, r3
42 orreqs r9, r4, r5
43 orreqs r9, r6, r7
44 cmpeq r8, #0
45 bne 2f
46 mov r1, r1, asl #15
47 bic r1, r1, #0x8000
48 orr r1, r1, r1, lsr #16
49 str r1, [r0], #4
50 str r1, [r0], #4
51 str r1, [r0], #4
52 str r1, [r0], #4
53 cmp r0, r12
54 blo 1b
55 b 3f
562:
57 mov r1, r1, asl #11 /* r1 = d0 = (block[0] << 11) + 2048 */
58 add r1, r1, #2048
59 add r1, r1, r3, asl #11 /* r1 = t0 = d0 + (block[2] << 11) */
60 sub r3, r1, r3, asl #12 /* r3 = t1 = d0 - (block[2] << 11) */
61
62 add r9, r2, r4 /* r9 = tmp = (d1+d3)*(1108/4) */
63 add r10, r9, r9, asl #2
64 add r10, r10, r9, asl #4
65 add r9, r10, r9, asl #8
66
67 add r10, r2, r2, asl #4 /* r2 = t2 = tmp + (d1*(1568/32)*8) */
68 add r2, r10, r2, asl #5
69 add r2, r9, r2, asl #3
70
71 add r10, r4, r4, asl #2 /* r4 = t3 = tmp - (d3*(3784/8)*2) */
72 rsb r10, r10, r4, asl #6
73 add r4, r4, r10, asl #3
74 sub r4, r9, r4, asl #1
75 /* t2 & t3 are 1/4 final value here */
76 add r1, r1, r2, asl #2 /* r1 = a0 = t0 + t2 */
77 sub r2, r1, r2, asl #3 /* r2 = a3 = t0 - t2 */
78 add r3, r3, r4, asl #2 /* r3 = a1 = t1 + t3 */
79 sub r4, r3, r4, asl #3 /* r4 = a2 = t1 - t3 */
80
81 add r9, r8, r5 /* r9 = tmp = 565*(d3 + d0) */
82 add r10, r9, r9, asl #4
83 add r10, r10, r10, asl #5
84 add r9, r10, r9, asl #2
85
86 add r10, r5, r5, asl #4 /* r5 = t0 = tmp + (((2276/4)*d0)*4) */
87 add r10, r10, r10, asl #5
88 add r5, r10, r5, asl #3
89 add r5, r9, r5, asl #2
90
91 add r10, r8, r8, asl #2 /* r8 = t1 = tmp - (((3406/2)*d3)*2) */
92 add r10, r10, r10, asl #4
93 add r10, r10, r8, asl #7
94 rsb r8, r8, r10, asl #3
95 sub r8, r9, r8, asl #1
96
97 add r9, r6, r7 /* r9 = tmp = (2408/8)*(d1 + d2) */
98 add r10, r9, r9, asl #3
99 add r10, r10, r10, asl #5
100 add r9, r10, r9, asl #2
101
102 add r10, r7, r7, asl #3 /* r7 = t2 = (tmp*8) - 799*d2 */
103 add r10, r10, r7, asl #4
104 rsb r7, r7, r10, asl #5
105 rsb r7, r7, r9, asl #3
106
107 sub r10, r6, r6, asl #4 /* r6 = t3 = (tmp*8) - 4017*d1 */
108 sub r10, r10, r6, asl #6
109 add r10, r10, r6, asl #12
110 add r6, r10, r6
111 rsb r6, r6, r9, asl #3
112 /* t0 = r5, t1 = r8, t2 = r7, t3 = r6*/
113 add r9, r5, r7 /* r9 = b0 = t0 + t2 */
114 add r10, r8, r6 /* r10 = b3 = t1 + t3 */
115 sub r5, r5, r7 /* t0 -= t2 */
116 sub r8, r8, r6 /* t1 -= t3 */
117 add r6, r5, r8 /* r6 = t0 + t1 */
118 sub r7, r5, r8 /* r7 = t0 - t1 */
119
120 add r11, r6, r6, asr #2 /* r6 = b1 = r6*(181/128) */
121 add r11, r11, r11, asr #5
122 add r6, r11, r6, asr #3
123 add r11, r7, r7, asr #2 /* r7 = b2 = r7*(181/128) */
124 add r11, r11, r11, asr #5
125 add r7, r11, r7, asr #3
126 /* r1 = a0, r3 = a1, r4 = a2, r2 = a3 */
127 /* r9 = b0, r6 = b1*2, r7 = b2*2, r10 = b3 */
128 add r5, r1, r9 /* block[0] = (a0 + b0) >> 12 */
129 mov r5, r5, asr #12
130 strh r5, [r0], #2
131 add r8, r3, r6, asr #1 /* block[1] = (a1 + b1) >> 12 */
132 mov r8, r8, asr #12
133 strh r8, [r0], #2
134 add r5, r4, r7, asr #1 /* block[2] = (a2 + b2) >> 12 */
135 mov r5, r5, asr #12
136 strh r5, [r0], #2
137 add r8, r2, r10 /* block[3] = (a3 + b3) >> 12 */
138 mov r8, r8, asr #12
139 strh r8, [r0], #2
140 sub r5, r2, r10 /* block[4] = (a3 - b3) >> 12 */
141 mov r5, r5, asr #12
142 strh r5, [r0], #2
143 sub r8, r4, r7, asr #1 /* block[5] = (a2 - b2) >> 12 */
144 mov r8, r8, asr #12
145 strh r8, [r0], #2
146 sub r5, r3, r6, asr #1 /* block[6] = (a1 - b1) >> 12 */
147 mov r5, r5, asr #12
148 strh r5, [r0], #2
149 sub r8, r1, r9 /* block[7] = (a0 - b0) >> 12 */
150 mov r8, r8, asr #12
151 strh r8, [r0], #2
152 cmp r0, r12
153 blo 1b
1543:
155 sub r0, r0, #128
156 add r12, r0, #16
1574:
158 ldrsh r1, [r0, #0*8] /* d0 */
159 ldrsh r2, [r0, #2*8] /* d1 */
160 ldrsh r3, [r0, #4*8] /* d2 */
161 ldrsh r4, [r0, #6*8] /* d3 */
162 ldrsh r5, [r0, #8*8] /* d0 */
163 ldrsh r6, [r0, #10*8] /* d1 */
164 ldrsh r7, [r0, #12*8] /* d2 */
165 ldrsh r8, [r0, #14*8] /* d3 */
166
167 mov r1, r1, asl #11 /* r1 = d0 = (block[0] << 11) + 2048 */
168 add r1, r1, #65536
169 add r1, r1, r3, asl #11 /* r1 = t0 = d0 + d2:(block[2] << 11) */
170 sub r3, r1, r3, asl #12 /* r3 = t1 = d0 - d2:(block[2] << 11) */
171
172 add r9, r2, r4 /* r9 = tmp = (d1+d3)*(1108/4) */
173 add r10, r9, r9, asl #2
174 add r10, r10, r9, asl #4
175 add r9, r10, r9, asl #8
176
177 add r11, r2, r2, asl #4 /* r2 = t2 = tmp + (d1*(1568/32)*8) */
178 add r2, r11, r2, asl #5
179 add r2, r9, r2, asl #3
180
181 add r10, r4, r4, asl #2 /* r4 = t3 = tmp - (d3*(3784/8)*2) */
182 rsb r10, r10, r4, asl #6
183 add r4, r4, r10, asl #3
184 sub r4, r9, r4, asl #1
185 /* t2 & t3 are 1/4 final value here */
186 add r1, r1, r2, asl #2 /* r1 = a0 = t0 + t2 */
187 sub r2, r1, r2, asl #3 /* r2 = a3 = t0 - t2 */
188 add r3, r3, r4, asl #2 /* r3 = a1 = t1 + t3 */
189 sub r4, r3, r4, asl #3 /* r4 = a2 = t1 - t3 */
190
191 add r9, r8, r5 /* r9 = tmp = 565*(d3 + d0) */
192 add r10, r9, r9, asl #4
193 add r10, r10, r10, asl #5
194 add r9, r10, r9, asl #2
195
196 add r10, r5, r5, asl #4 /* r5 = t0 = tmp + (((2276/4)*d0)*4) */
197 add r10, r10, r10, asl #5
198 add r5, r10, r5, asl #3
199 add r5, r9, r5, asl #2
200
201 add r10, r8, r8, asl #2 /* r8 = t1 = tmp - (((3406/2)*d3)*2) */
202 add r10, r10, r10, asl #4
203 add r10, r10, r8, asl #7
204 rsb r8, r8, r10, asl #3
205 sub r8, r9, r8, asl #1
206
207 add r9, r6, r7 /* r9 = tmp = (2408/8)*(d1 + d2) */
208 add r10, r9, r9, asl #3
209 add r10, r10, r10, asl #5
210 add r9, r10, r9, asl #2
211
212 add r10, r7, r7, asl #3 /* r7 = t2 = (tmp*8) - 799*d2 */
213 add r10, r10, r7, asl #4
214 rsb r7, r7, r10, asl #5
215 rsb r7, r7, r9, asl #3
216
217 sub r10, r6, r6, asl #4 /* r6 = t3 = (tmp*8) - 4017*d1 */
218 sub r10, r10, r6, asl #6
219 add r10, r10, r6, asl #12
220 add r6, r10, r6
221 rsb r6, r6, r9, asl #3
222 /* t0=r5, t1=r8, t2=r7, t3=r6*/
223 add r9, r5, r7 /* r9 = b0 = t0 + t2 */
224 add r10, r8, r6 /* r10 = b3 = t1 + t3 */
225 sub r5, r5, r7 /* t0 -= t2 */
226 sub r8, r8, r6 /* t1 -= t3 */
227 add r6, r5, r8 /* r6 = t0 + t1 */
228 sub r7, r5, r8 /* r7 = t0 - t1 */
229
230 add r11, r6, r6, asr #2 /* r6 = b1 = r5*(181/128) */
231 add r11, r11, r11, asr #5
232 add r6, r11, r6, asr #3
233 add r11, r7, r7, asr #2 /* r7 = b2 = r6*(181/128) */
234 add r11, r11, r11, asr #5
235 add r7, r11, r7, asr #3
236 /* r1 = a0, r3 = a1, r4 = a2, r2 = a3 */
237 /* r9 = b0, r6 = b1*2, r7 = b2*2, r10 = b3 */
238 add r5, r1, r9 /* block[0] = (a0 + b0) >> 17 */
239 mov r5, r5, asr #17
240 strh r5, [r0, #0*8]
241 add r8, r3, r6, asr #1 /* block[1] = (a1 + b1) >> 17 */
242 mov r8, r8, asr #17
243 strh r8, [r0, #2*8]
244 add r5, r4, r7, asr #1 /* block[2] = (a2 + b2) >> 17 */
245 mov r5, r5, asr #17
246 strh r5, [r0, #4*8]
247 add r8, r2, r10 /* block[3] = (a3 + b3) >> 17 */
248 mov r8, r8, asr #17
249 strh r8, [r0, #6*8]
250 sub r5, r2, r10 /* block[4] = (a3 - b3) >> 17 */
251 mov r5, r5, asr #17
252 strh r5, [r0, #8*8]
253 sub r8, r4, r7, asr #1 /* block[5] = (a2 - b2) >> 17 */
254 mov r8, r8, asr #17
255 strh r8, [r0, #10*8]
256 sub r5, r3, r6, asr #1 /* block[6] = (a1 - b1) >> 17 */
257 mov r5, r5, asr #17
258 strh r5, [r0, #12*8]
259 sub r8, r1, r9 /* block[7] = (a0 - b0) >> 17 */
260 mov r8, r8, asr #17
261 strh r8, [r0, #14*8]
262 add r0, r0, #2
263 cmp r0, r12
264 blo 4b
265 sub r0, r0, #16
266 bx lr
267
268mpeg2_idct_copy_arm:
269 stmfd sp!, { r1-r2, r4-r12, lr }
270 bl .idct
271 ldmfd sp!, { r1-r2 }
272 mov r11, #0
273 add r12, r0, #128
2741:
275 ldrsh r3, [r0, #0]
276 ldrsh r4, [r0, #2]
277 ldrsh r5, [r0, #4]
278 ldrsh r6, [r0, #6]
279 ldrsh r7, [r0, #8]
280 ldrsh r8, [r0, #10]
281 ldrsh r9, [r0, #12]
282 ldrsh r10, [r0, #14]
283 cmp r3, #255
284 mvnhi r3, r3, asr #31
285 strb r3, [r1, #0]
286 str r11, [r0], #4
287 cmp r4, #255
288 mvnhi r4, r4, asr #31
289 strb r4, [r1, #1]
290 cmp r5, #255
291 mvnhi r5, r5, asr #31
292 strb r5, [r1, #2]
293 str r11, [r0], #4
294 cmp r6, #255
295 mvnhi r6, r6, asr #31
296 strb r6, [r1, #3]
297 cmp r7, #255
298 mvnhi r7, r7, asr #31
299 strb r7, [r1, #4]
300 str r11, [r0], #4
301 cmp r8, #255
302 mvnhi r8, r8, asr #31
303 strb r8, [r1, #5]
304 cmp r9, #255
305 mvnhi r9, r9, asr #31
306 strb r9, [r1, #6]
307 str r11, [r0], #4
308 cmp r10, #255
309 mvnhi r10, r10, asr #31
310 strb r10, [r1, #7]
311 add r1, r1, r2
312 cmp r0, r12
313 blo 1b
314 ldmfd sp!, { r4-r12, pc }
315
316mpeg2_idct_add_arm:
317 cmp r0, #129
318 mov r0, r1
319 ldreqsh r1, [r0, #0]
320 bne 1f
321 and r1, r1, #0x70
322 cmp r1, #0x40
323 bne 3f
3241:
325 stmfd sp!, { r2-r12, lr }
326 bl .idct
327 ldmfd sp!, { r1-r2 }
328 mov r11, #0
329 add r12, r0, #128
3302:
331 ldrb r3, [r1, #0]
332 ldrb r4, [r1, #1]
333 ldrb r5, [r1, #2]
334 ldrb r6, [r1, #3]
335 ldrsh r7, [r0, #0]
336 ldrsh r8, [r0, #2]
337 ldrsh r9, [r0, #4]
338 ldrsh r10, [r0, #6]
339 add r7, r7, r3
340 ldrb r3, [r1, #4]
341 cmp r7, #255
342 mvnhi r7, r7, asr #31
343 strb r7, [r1, #0]
344 ldrsh r7, [r0, #8]
345 add r8, r8, r4
346 ldrb r4, [r1, #5]
347 cmp r8, #255
348 mvnhi r8, r8, asr #31
349 strb r8, [r1, #1]
350 ldrsh r8, [r0, #10]
351 add r9, r9, r5
352 ldrb r5, [r1, #6]
353 cmp r9, #255
354 mvnhi r9, r9, asr #31
355 strb r9, [r1, #2]
356 ldrsh r9, [r0, #12]
357 add r10, r10, r6
358 ldrb r6, [r1, #7]
359 cmp r10, #255
360 mvnhi r10, r10, asr #31
361 strb r10, [r1, #3]
362 ldrsh r10, [r0, #14]
363 str r11, [r0], #4
364 add r7, r7, r3
365 cmp r7, #255
366 mvnhi r7, r7, asr #31
367 strb r7, [r1, #4]
368 str r11, [r0], #4
369 add r8, r8, r4
370 cmp r8, #255
371 mvnhi r8, r8, asr #31
372 strb r8, [r1, #5]
373 str r11, [r0], #4
374 add r9, r9, r5
375 cmp r9, #255
376 mvnhi r9, r9, asr #31
377 strb r9, [r1, #6]
378 add r10, r10, r6
379 cmp r10, #255
380 mvnhi r10, r10, asr #31
381 strb r10, [r1, #7]
382 str r11, [r0], #4
383 add r1, r1, r2
384 cmp r0, r12
385 blo 2b
386 ldmfd sp!, { r4-r12, pc }
3873:
388 stmfd sp!, { r4-r11 }
389 ldrsh r1, [r0, #0] /* r1 = block[0] */
390 mov r11, #0
391 strh r11, [r0, #0] /* block[0] = 0 */
392 strh r11, [r0, #126] /* block[63] = 0 */
393 add r1, r1, #64 /* r1 = DC << 7 */
394 add r0, r2, r3, asl #3
3954:
396 ldrb r4, [r2, #0]
397 ldrb r5, [r2, #1]
398 ldrb r6, [r2, #2]
399 ldrb r7, [r2, #3]
400 ldrb r8, [r2, #4]
401 ldrb r9, [r2, #5]
402 ldrb r10, [r2, #6]
403 ldrb r11, [r2, #7]
404 add r4, r4, r1, asr #7
405 cmp r4, #255
406 mvnhi r4, r4, asr #31
407 strb r4, [r2, #0]
408 add r5, r5, r1, asr #7
409 cmp r5, #255
410 mvnhi r5, r5, asr #31
411 strb r5, [r2, #1]
412 add r6, r6, r1, asr #7
413 cmp r6, #255
414 mvnhi r6, r6, asr #31
415 strb r6, [r2, #2]
416 add r7, r7, r1, asr #7
417 cmp r7, #255
418 mvnhi r7, r7, asr #31
419 strb r7, [r2, #3]
420 add r8, r8, r1, asr #7
421 cmp r8, #255
422 mvnhi r8, r8, asr #31
423 strb r8, [r2, #4]
424 add r9, r9, r1, asr #7
425 cmp r9, #255
426 mvnhi r9, r9, asr #31
427 strb r9, [r2, #5]
428 add r10, r10, r1, asr #7
429 cmp r10, #255
430 mvnhi r10, r10, asr #31
431 strb r10, [r2, #6]
432 add r11, r11, r1, asr #7
433 cmp r11, #255
434 mvnhi r11, r11, asr #31
435 strb r11, [r2, #7]
436 add r2, r2, r3
437 cmp r2, r0
438 blo 4b
439 ldmfd sp!, { r4-r11 }
440 bx lr
generated by cgit v1.2.3 (git 2.39.1) at 2024-09-30 19:23:32 +0000