File size: 13,553 Bytes
8ead80b |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 |
/*
C-like prototype :
void j_rev_dct_arm(DCTBLOCK data)
With DCTBLOCK being a pointer to an array of 64 'signed shorts'
Copyright (c) 2001 Lionel Ulmer ([email protected] / [email protected])
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
*/
#include "libavutil/arm/asm.S"
#define FIX_0_298631336 2446
#define FIX_0_541196100 4433
#define FIX_0_765366865 6270
#define FIX_1_175875602 9633
#define FIX_1_501321110 12299
#define FIX_2_053119869 16819
#define FIX_3_072711026 25172
#define FIX_M_0_390180644 -3196
#define FIX_M_0_899976223 -7373
#define FIX_M_1_847759065 -15137
#define FIX_M_1_961570560 -16069
#define FIX_M_2_562915447 -20995
#define FIX_0xFFFF 0xFFFF
#define FIX_0_298631336_ID 0
#define FIX_0_541196100_ID 4
#define FIX_0_765366865_ID 8
#define FIX_1_175875602_ID 12
#define FIX_1_501321110_ID 16
#define FIX_2_053119869_ID 20
#define FIX_3_072711026_ID 24
#define FIX_M_0_390180644_ID 28
#define FIX_M_0_899976223_ID 32
#define FIX_M_1_847759065_ID 36
#define FIX_M_1_961570560_ID 40
#define FIX_M_2_562915447_ID 44
#define FIX_0xFFFF_ID 48
function ff_j_rev_dct_arm, export=1
push {r0, r4 - r11, lr}
mov lr, r0 @ lr = pointer to the current row
mov r12, #8 @ r12 = row-counter
movrel r11, const_array @ r11 = base pointer to the constants array
row_loop:
ldrsh r0, [lr, # 0] @ r0 = 'd0'
ldrsh r2, [lr, # 2] @ r2 = 'd2'
@ Optimization for row that have all items except the first set to 0
@ (this works as the int16_t are always 4-byte aligned)
ldr r5, [lr, # 0]
ldr r6, [lr, # 4]
ldr r3, [lr, # 8]
ldr r4, [lr, #12]
orr r3, r3, r4
orr r3, r3, r6
orrs r5, r3, r5
beq end_of_row_loop @ nothing to be done as ALL of them are '0'
orrs r3, r3, r2
beq empty_row
ldrsh r1, [lr, # 8] @ r1 = 'd1'
ldrsh r4, [lr, # 4] @ r4 = 'd4'
ldrsh r6, [lr, # 6] @ r6 = 'd6'
ldr r3, [r11, #FIX_0_541196100_ID]
add r7, r2, r6
ldr r5, [r11, #FIX_M_1_847759065_ID]
mul r7, r3, r7 @ r7 = z1
ldr r3, [r11, #FIX_0_765366865_ID]
mla r6, r5, r6, r7 @ r6 = tmp2
add r5, r0, r4 @ r5 = tmp0
mla r2, r3, r2, r7 @ r2 = tmp3
sub r3, r0, r4 @ r3 = tmp1
add r0, r2, r5, lsl #13 @ r0 = tmp10
rsb r2, r2, r5, lsl #13 @ r2 = tmp13
add r4, r6, r3, lsl #13 @ r4 = tmp11
rsb r3, r6, r3, lsl #13 @ r3 = tmp12
push {r0, r2, r3, r4} @ save on the stack tmp10, tmp13, tmp12, tmp11
ldrsh r3, [lr, #10] @ r3 = 'd3'
ldrsh r5, [lr, #12] @ r5 = 'd5'
ldrsh r7, [lr, #14] @ r7 = 'd7'
add r0, r3, r5 @ r0 = 'z2'
add r2, r1, r7 @ r2 = 'z1'
add r4, r3, r7 @ r4 = 'z3'
add r6, r1, r5 @ r6 = 'z4'
ldr r9, [r11, #FIX_1_175875602_ID]
add r8, r4, r6 @ r8 = z3 + z4
ldr r10, [r11, #FIX_M_0_899976223_ID]
mul r8, r9, r8 @ r8 = 'z5'
ldr r9, [r11, #FIX_M_2_562915447_ID]
mul r2, r10, r2 @ r2 = 'z1'
ldr r10, [r11, #FIX_M_1_961570560_ID]
mul r0, r9, r0 @ r0 = 'z2'
ldr r9, [r11, #FIX_M_0_390180644_ID]
mla r4, r10, r4, r8 @ r4 = 'z3'
ldr r10, [r11, #FIX_0_298631336_ID]
mla r6, r9, r6, r8 @ r6 = 'z4'
ldr r9, [r11, #FIX_2_053119869_ID]
mla r7, r10, r7, r2 @ r7 = tmp0 + z1
ldr r10, [r11, #FIX_3_072711026_ID]
mla r5, r9, r5, r0 @ r5 = tmp1 + z2
ldr r9, [r11, #FIX_1_501321110_ID]
mla r3, r10, r3, r0 @ r3 = tmp2 + z2
add r7, r7, r4 @ r7 = tmp0
mla r1, r9, r1, r2 @ r1 = tmp3 + z1
add r5, r5, r6 @ r5 = tmp1
add r3, r3, r4 @ r3 = tmp2
add r1, r1, r6 @ r1 = tmp3
pop {r0, r2, r4, r6} @ r0 = tmp10 / r2 = tmp13 / r4 = tmp12 / r6 = tmp11
@ r1 = tmp3 / r3 = tmp2 / r5 = tmp1 / r7 = tmp0
@ Compute DESCALE(tmp10 + tmp3, CONST_BITS-PASS1_BITS)
add r8, r0, r1
add r8, r8, #(1<<10)
mov r8, r8, asr #11
strh r8, [lr, # 0]
@ Compute DESCALE(tmp10 - tmp3, CONST_BITS-PASS1_BITS)
sub r8, r0, r1
add r8, r8, #(1<<10)
mov r8, r8, asr #11
strh r8, [lr, #14]
@ Compute DESCALE(tmp11 + tmp2, CONST_BITS-PASS1_BITS)
add r8, r6, r3
add r8, r8, #(1<<10)
mov r8, r8, asr #11
strh r8, [lr, # 2]
@ Compute DESCALE(tmp11 - tmp2, CONST_BITS-PASS1_BITS)
sub r8, r6, r3
add r8, r8, #(1<<10)
mov r8, r8, asr #11
strh r8, [lr, #12]
@ Compute DESCALE(tmp12 + tmp1, CONST_BITS-PASS1_BITS)
add r8, r4, r5
add r8, r8, #(1<<10)
mov r8, r8, asr #11
strh r8, [lr, # 4]
@ Compute DESCALE(tmp12 - tmp1, CONST_BITS-PASS1_BITS)
sub r8, r4, r5
add r8, r8, #(1<<10)
mov r8, r8, asr #11
strh r8, [lr, #10]
@ Compute DESCALE(tmp13 + tmp0, CONST_BITS-PASS1_BITS)
add r8, r2, r7
add r8, r8, #(1<<10)
mov r8, r8, asr #11
strh r8, [lr, # 6]
@ Compute DESCALE(tmp13 - tmp0, CONST_BITS-PASS1_BITS)
sub r8, r2, r7
add r8, r8, #(1<<10)
mov r8, r8, asr #11
strh r8, [lr, # 8]
@ End of row loop
add lr, lr, #16
subs r12, r12, #1
bne row_loop
beq start_column_loop
empty_row:
ldr r1, [r11, #FIX_0xFFFF_ID]
mov r0, r0, lsl #2
and r0, r0, r1
add r0, r0, r0, lsl #16
str r0, [lr, # 0]
str r0, [lr, # 4]
str r0, [lr, # 8]
str r0, [lr, #12]
end_of_row_loop:
@ End of loop
add lr, lr, #16
subs r12, r12, #1
bne row_loop
start_column_loop:
@ Start of column loop
pop {lr}
mov r12, #8
column_loop:
ldrsh r0, [lr, #( 0*8)] @ r0 = 'd0'
ldrsh r2, [lr, #( 4*8)] @ r2 = 'd2'
ldrsh r4, [lr, #( 8*8)] @ r4 = 'd4'
ldrsh r6, [lr, #(12*8)] @ r6 = 'd6'
ldr r3, [r11, #FIX_0_541196100_ID]
add r1, r2, r6
ldr r5, [r11, #FIX_M_1_847759065_ID]
mul r1, r3, r1 @ r1 = z1
ldr r3, [r11, #FIX_0_765366865_ID]
mla r6, r5, r6, r1 @ r6 = tmp2
add r5, r0, r4 @ r5 = tmp0
mla r2, r3, r2, r1 @ r2 = tmp3
sub r3, r0, r4 @ r3 = tmp1
add r0, r2, r5, lsl #13 @ r0 = tmp10
rsb r2, r2, r5, lsl #13 @ r2 = tmp13
add r4, r6, r3, lsl #13 @ r4 = tmp11
rsb r6, r6, r3, lsl #13 @ r6 = tmp12
ldrsh r1, [lr, #( 2*8)] @ r1 = 'd1'
ldrsh r3, [lr, #( 6*8)] @ r3 = 'd3'
ldrsh r5, [lr, #(10*8)] @ r5 = 'd5'
ldrsh r7, [lr, #(14*8)] @ r7 = 'd7'
@ Check for empty odd column (happens about 20 to 25 % of the time according to my stats)
orr r9, r1, r3
orr r10, r5, r7
orrs r10, r9, r10
beq empty_odd_column
push {r0, r2, r4, r6} @ save on the stack tmp10, tmp13, tmp12, tmp11
add r0, r3, r5 @ r0 = 'z2'
add r2, r1, r7 @ r2 = 'z1'
add r4, r3, r7 @ r4 = 'z3'
add r6, r1, r5 @ r6 = 'z4'
ldr r9, [r11, #FIX_1_175875602_ID]
add r8, r4, r6
ldr r10, [r11, #FIX_M_0_899976223_ID]
mul r8, r9, r8 @ r8 = 'z5'
ldr r9, [r11, #FIX_M_2_562915447_ID]
mul r2, r10, r2 @ r2 = 'z1'
ldr r10, [r11, #FIX_M_1_961570560_ID]
mul r0, r9, r0 @ r0 = 'z2'
ldr r9, [r11, #FIX_M_0_390180644_ID]
mla r4, r10, r4, r8 @ r4 = 'z3'
ldr r10, [r11, #FIX_0_298631336_ID]
mla r6, r9, r6, r8 @ r6 = 'z4'
ldr r9, [r11, #FIX_2_053119869_ID]
mla r7, r10, r7, r2 @ r7 = tmp0 + z1
ldr r10, [r11, #FIX_3_072711026_ID]
mla r5, r9, r5, r0 @ r5 = tmp1 + z2
ldr r9, [r11, #FIX_1_501321110_ID]
mla r3, r10, r3, r0 @ r3 = tmp2 + z2
add r7, r7, r4 @ r7 = tmp0
mla r1, r9, r1, r2 @ r1 = tmp3 + z1
add r5, r5, r6 @ r5 = tmp1
add r3, r3, r4 @ r3 = tmp2
add r1, r1, r6 @ r1 = tmp3
pop {r0, r2, r4, r6} @ r0 = tmp10 / r2 = tmp13 / r4 = tmp11 / r6 = tmp12
@ r1 = tmp3 / r3 = tmp2 / r5 = tmp1 / r7 = tmp0
@ Compute DESCALE(tmp10 + tmp3, CONST_BITS+PASS1_BITS+3)
add r8, r0, r1
add r8, r8, #(1<<17)
mov r8, r8, asr #18
strh r8, [lr, #( 0*8)]
@ Compute DESCALE(tmp10 - tmp3, CONST_BITS+PASS1_BITS+3)
sub r8, r0, r1
add r8, r8, #(1<<17)
mov r8, r8, asr #18
strh r8, [lr, #(14*8)]
@ Compute DESCALE(tmp11 + tmp2, CONST_BITS+PASS1_BITS+3)
add r8, r4, r3
add r8, r8, #(1<<17)
mov r8, r8, asr #18
strh r8, [lr, #( 2*8)]
@ Compute DESCALE(tmp11 - tmp2, CONST_BITS+PASS1_BITS+3)
sub r8, r4, r3
add r8, r8, #(1<<17)
mov r8, r8, asr #18
strh r8, [lr, #(12*8)]
@ Compute DESCALE(tmp12 + tmp1, CONST_BITS+PASS1_BITS+3)
add r8, r6, r5
add r8, r8, #(1<<17)
mov r8, r8, asr #18
strh r8, [lr, #( 4*8)]
@ Compute DESCALE(tmp12 - tmp1, CONST_BITS+PASS1_BITS+3)
sub r8, r6, r5
add r8, r8, #(1<<17)
mov r8, r8, asr #18
strh r8, [lr, #(10*8)]
@ Compute DESCALE(tmp13 + tmp0, CONST_BITS+PASS1_BITS+3)
add r8, r2, r7
add r8, r8, #(1<<17)
mov r8, r8, asr #18
strh r8, [lr, #( 6*8)]
@ Compute DESCALE(tmp13 - tmp0, CONST_BITS+PASS1_BITS+3)
sub r8, r2, r7
add r8, r8, #(1<<17)
mov r8, r8, asr #18
strh r8, [lr, #( 8*8)]
@ End of row loop
add lr, lr, #2
subs r12, r12, #1
bne column_loop
beq the_end
empty_odd_column:
@ Compute DESCALE(tmp10 + tmp3, CONST_BITS+PASS1_BITS+3)
@ Compute DESCALE(tmp10 - tmp3, CONST_BITS+PASS1_BITS+3)
add r0, r0, #(1<<17)
mov r0, r0, asr #18
strh r0, [lr, #( 0*8)]
strh r0, [lr, #(14*8)]
@ Compute DESCALE(tmp11 + tmp2, CONST_BITS+PASS1_BITS+3)
@ Compute DESCALE(tmp11 - tmp2, CONST_BITS+PASS1_BITS+3)
add r4, r4, #(1<<17)
mov r4, r4, asr #18
strh r4, [lr, #( 2*8)]
strh r4, [lr, #(12*8)]
@ Compute DESCALE(tmp12 + tmp1, CONST_BITS+PASS1_BITS+3)
@ Compute DESCALE(tmp12 - tmp1, CONST_BITS+PASS1_BITS+3)
add r6, r6, #(1<<17)
mov r6, r6, asr #18
strh r6, [lr, #( 4*8)]
strh r6, [lr, #(10*8)]
@ Compute DESCALE(tmp13 + tmp0, CONST_BITS+PASS1_BITS+3)
@ Compute DESCALE(tmp13 - tmp0, CONST_BITS+PASS1_BITS+3)
add r2, r2, #(1<<17)
mov r2, r2, asr #18
strh r2, [lr, #( 6*8)]
strh r2, [lr, #( 8*8)]
@ End of row loop
add lr, lr, #2
subs r12, r12, #1
bne column_loop
the_end:
@ The end....
pop {r4 - r11, pc}
endfunc
const const_array
.word FIX_0_298631336
.word FIX_0_541196100
.word FIX_0_765366865
.word FIX_1_175875602
.word FIX_1_501321110
.word FIX_2_053119869
.word FIX_3_072711026
.word FIX_M_0_390180644
.word FIX_M_0_899976223
.word FIX_M_1_847759065
.word FIX_M_1_961570560
.word FIX_M_2_562915447
.word FIX_0xFFFF
endconst
|