1 @/***************************************************************************** 2 @* 3 @* Copyright (C) 2012 Ittiam Systems Pvt Ltd, Bangalore 4 @* 5 @* Licensed under the Apache License, Version 2.0 (the "License"); 6 @* you may not use this file except in compliance with the License. 7 @* You may obtain a copy of the License at: 8 @* 9 @* http://www.apache.org/licenses/LICENSE-2.0 10 @* 11 @* Unless required by applicable law or agreed to in writing, software 12 @* distributed under the License is distributed on an "AS IS" BASIS, 13 @* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 @* See the License for the specific language governing permissions and 15 @* limitations under the License. 16 @* 17 @*****************************************************************************/ 18 @/** 19 @******************************************************************************* 20 @* @file 21 @* ihevc_intra_pred_chroma_mode_19_to_25.s 22 @* 23 @* @brief 24 @* contains function definitions for intra prediction dc filtering. 25 @* functions are coded using neon intrinsics and can be compiled using 26 27 @* rvct 28 @* 29 @* @author 30 @* naveen sr 31 @* 32 @* @par list of functions: 33 @* 34 @* 35 @* @remarks 36 @* none 37 @* 38 @******************************************************************************* 39 @*/ 40 @/** 41 @******************************************************************************* 42 @* 43 @* @brief 44 @* chroma intraprediction filter for dc input 45 @* 46 @* @par description: 47 @* 48 @* @param[in] pu1_ref 49 @* uword8 pointer to the source 50 @* 51 @* @param[out] pu1_dst 52 @* uword8 pointer to the destination 53 @* 54 @* @param[in] src_strd 55 @* integer source stride 56 @* 57 @* @param[in] dst_strd 58 @* integer destination stride 59 @* 60 @* @param[in] nt 61 @* size of tranform block 62 @* 63 @* @param[in] mode 64 @* type of filtering 65 @* 66 @* @returns 67 @* 68 @* @remarks 69 @* none 70 @* 71 @******************************************************************************* 72 @*/ 73 74 @void ihevc_intra_pred_chroma_mode_19_to_25(uword8* pu1_ref, 75 @ word32 src_strd, 76 @ uword8* pu1_dst, 77 @ word32 dst_strd, 78 @ word32 nt, 79 @ word32 mode) 80 @ 81 @**************variables vs registers***************************************** 82 @r0 => *pu1_ref 83 @r1 => src_strd 84 @r2 => *pu1_dst 85 @r3 => dst_strd 86 87 @stack contents from #40 88 @ nt 89 @ mode 90 91 .text 92 .align 4 93 94 95 96 97 .globl ihevc_intra_pred_chroma_mode_19_to_25_a9q 98 .extern gai4_ihevc_ang_table 99 .extern gai4_ihevc_inv_ang_table 100 .extern gau1_ihevc_planar_factor 101 102 gai4_ihevc_inv_ang_table_addr: 103 .long gai4_ihevc_inv_ang_table - ulbl1 - 8 104 105 gau1_ihevc_planar_factor_addr: 106 .long gau1_ihevc_planar_factor - ulbl2 - 8 107 108 gai4_ihevc_ang_table_addr_1: 109 .long gai4_ihevc_ang_table - ulbl3 - 8 110 111 gai4_ihevc_ang_table_addr_2: 112 .long gai4_ihevc_ang_table - ulbl4 - 8 113 114 .type ihevc_intra_pred_chroma_mode_19_to_25_a9q, %function 115 116 ihevc_intra_pred_chroma_mode_19_to_25_a9q: 117 118 stmfd sp!, {r4-r12, r14} @stack stores the values of the arguments 119 120 ldr r4,[sp,#40] @loads nt 121 ldr r7, gai4_ihevc_ang_table_addr_1 122 ulbl3: 123 add r7,r7,pc 124 125 ldr r5,[sp,#44] @mode (19 to 25) 126 ldr r8, gai4_ihevc_inv_ang_table_addr 127 ulbl1: 128 add r8,r8,pc 129 130 add r7, r7, r5, lsl #2 @gai4_ihevc_ang_table[mode] 131 add r8, r8, r5, lsl #2 @gai4_ihevc_inv_ang_table 132 sub r8, r8, #48 @gai4_ihevc_inv_ang_table[mode - 12] 133 134 ldr r7, [r7] @intra_pred_ang 135 sub sp, sp, #132 @ref_temp[2 * max_cu_size + 2] 136 137 ldr r8, [r8] @inv_ang 138 add r6, sp, r4 , lsl #1 @ref_temp + 2 * nt 139 140 mul r9, r4, r7 @nt*intra_pred_ang 141 142 sub r6, r6, #2 @ref_temp + 2*nt - 2 143 144 add r1, r0, r4, lsl #2 @r1 = &src[4nt] 145 vdup.8 d30, r7 @intra_pred_ang 146 147 mov r7, r4 148 149 asr r9, r9, #5 150 151 vld1.32 d0,[r1]! @ pu1_ref[two_nt + k] 152 153 vst1.32 d0,[r6]! @ref_temp[k + nt - 1] = pu1_ref[two_nt + k]@ 154 155 subs r7, r7, #4 156 beq end_loop_copy 157 subs r7,r7,#4 158 beq loop_copy_8 159 subs r7,r7,#8 160 beq loop_copy_16 161 162 loop_copy_32: 163 vld1.8 {d0,d1,d2,d3},[r1]! 164 vld1.8 {d4,d5,d6},[r1]! 165 166 vst1.8 {d0,d1,d2,d3},[r6]! 167 168 169 vst1.8 {d4,d5,d6},[r6]! 170 b end_loop_copy 171 172 loop_copy_16: 173 vld1.8 {d0,d1,d2},[r1]! 174 vst1.8 {d0,d1,d2},[r6]! 175 176 b end_loop_copy 177 178 loop_copy_8: 179 vld1.8 d0,[r1]! 180 vst1.8 d0,[r6]! 181 182 end_loop_copy: 183 184 ldrh r11, [r1] 185 strh r11, [r6] 186 187 cmp r9, #-1 188 bge linear_filtering 189 190 add r6, sp, r4 ,lsl #1 @ref_temp + 2 * nt 191 sub r6, r6, #4 @ref_temp + 2 * nt - 2 - 2 192 193 mov r12, #0xffffffff 194 195 rsb r9, r9, r12 @count to take care off ref_idx 196 197 add r1, r0, r4, lsl #2 @r1 = &src[2nt] 198 199 mov r7, #128 @inv_ang_sum 200 201 loop_copy_ref_idx: 202 203 add r7, r7, r8 @inv_ang_sum += inv_ang 204 mov r0,r7, lsr #8 205 mov r0,r0, lsl #1 206 ldrh r11, [r1, -r0] 207 strh r11, [r6], #-2 208 209 subs r9, r9, #1 210 211 bne loop_copy_ref_idx 212 213 214 linear_filtering: 215 @ after copy 216 @ below code is taken from mode 27 to 33 and modified 217 218 ldr r6,gai4_ihevc_ang_table_addr_2 @loads word32 gai4_ihevc_ang_table[35] 219 ulbl4: 220 add r6,r6,pc 221 222 lsl r7,r4,#2 @four_nt 223 224 add r8,r6,r5,lsl #2 @*gai4_ihevc_ang_table[mode] 225 ldr r9,[r8] @intra_pred_ang = gai4_ihevc_ang_table[mode] 226 ldr r1,gau1_ihevc_planar_factor_addr @used for ((row + 1) * intra_pred_ang) row values 227 ulbl2: 228 add r1,r1,pc 229 add r6,r1,#1 230 231 add r8, sp, r4, lsl #1 @ref_temp + 2 * nt 232 sub r8,#2 @ref_temp + 2*nt -2 233 234 mov lr,#0 @row 235 mov r12,r4 236 lsl r4,r4,#1 237 238 core_loop_8: 239 add r8,r8,#2 @pu1_ref_main_idx += (four_nt + 1) 240 vdup.8 d0,r9 @intra_pred_ang 241 mov r12,r4,lsr #4 @divide by 8 242 243 vmov.i8 d1,#32 244 mul r7,r4,r12 245 246 vmov.i16 q3,#31 247 248 249 mov r1,r8 250 251 mov r5,r4 252 mov r11,#2 253 254 prologue: 255 vld1.8 {d3},[r6] @loads the row value 256 vmull.s8 q1,d3,d0 @pos = ((row + 1) * intra_pred_ang) 257 vand q2,q1,q3 @dup_const_fract(fract = pos & (31)) 258 vmovn.i16 d4,q2 259 vshrn.s16 d5,q1,#5 @idx = pos >> 5 260 vshl.s8 d5,d5,#1 261 262 vdup.8 d31,d4[0] 263 add r0,r2,r3 264 265 vmov.u32 lr,d5[0] @(i row)extract idx to the r register 266 @ lsl lr,lr,#1 267 268 vdup.8 d29,d4[1] @(ii) 269 sbfx r9,lr,#0,#8 270 271 add r10,r8,r9 @(i row)*pu1_ref[ref_main_idx] 272 273 vld1.8 {d8},[r10],r11 @(i row)ref_main_idx 274 sbfx r9,lr,#8,#8 275 276 vld1.8 {d9},[r10] @(i row)ref_main_idx_1 277 add r12,r8,r9 @(ii)*pu1_ref[ref_main_idx] 278 279 sbfx r9,lr,#16,#8 280 vsub.u8 d30,d1,d31 @32-fract(dup_const_32_fract) 281 add r10,r8,r9 @(iii)*pu1_ref[ref_main_idx] 282 283 vld1.8 {d12},[r12],r11 @(ii)ref_main_idx 284 vmull.u8 q5,d8,d30 @(i row)vmull_u8(ref_main_idx, dup_const_32_fract) 285 286 vld1.8 {d13},[r12] @(ii)ref_main_idx_1 287 vmlal.u8 q5,d9,d31 @(i row)vmull_u8(ref_main_idx_1, dup_const_fract) 288 289 vdup.8 d27,d4[2] @(iii) 290 vsub.u8 d28,d1,d29 @(ii)32-fract(dup_const_32_fract) 291 sbfx r9,lr,#24,#8 292 293 vdup.8 d25,d4[3] @(iv) 294 vmull.u8 q7,d12,d28 @(ii)vmull_u8(ref_main_idx, dup_const_32_fract) 295 add r12,r8,r9 @(iv)*pu1_ref[ref_main_idx] 296 297 vld1.8 {d16},[r10],r11 @(iii)ref_main_idx 298 vmlal.u8 q7,d13,d29 @(ii)vmull_u8(ref_main_idx_1, dup_const_fract) 299 300 vld1.8 {d17},[r10] @(iii)ref_main_idx_1 301 vrshrn.i16 d10,q5,#5 @(i row)shift_res = vrshrn_n_u16(add_res, 5) 302 303 vld1.8 {d20},[r12],r11 @(iv)ref_main_idx 304 vsub.u8 d26,d1,d27 @(iii)32-fract(dup_const_32_fract) 305 306 vld1.8 {d21},[r12] @(iv)ref_main_idx_1 307 308 vdup.8 d31,d4[4] @(v) 309 vmull.u8 q9,d16,d26 @(iii)vmull_u8(ref_main_idx, dup_const_32_fract) 310 311 vmov.u32 lr,d5[1] @extract idx to the r register 312 vmlal.u8 q9,d17,d27 @(iii)vmull_u8(ref_main_idx_1, dup_const_fract) 313 @ lsl lr,lr,#1 314 315 vst1.8 {d10},[r2]! @(i row) 316 vrshrn.i16 d14,q7,#5 @(ii)shift_res = vrshrn_n_u16(add_res, 5) 317 318 sbfx r9,lr,#0,#8 319 vdup.8 d29,d4[5] @(vi) 320 add r10,r8,r9 @(v)*pu1_ref[ref_main_idx] 321 322 vld1.8 {d8},[r10],r11 @(v)ref_main_idx 323 vsub.u8 d24,d1,d25 @(iv)32-fract(dup_const_32_fract) 324 325 vmull.u8 q11,d20,d24 @(iv)vmull_u8(ref_main_idx, dup_const_32_fract) 326 sbfx r9,lr,#8,#8 327 328 vld1.8 {d9},[r10] @(v)ref_main_idx_1 329 vmlal.u8 q11,d21,d25 @(iv)vmull_u8(ref_main_idx_1, dup_const_fract) 330 331 vst1.8 {d14},[r0],r3 @(ii) 332 vrshrn.i16 d18,q9,#5 @(iii)shift_res = vrshrn_n_u16(add_res, 5) 333 334 add r12,r8,r9 @(vi)*pu1_ref[ref_main_idx] 335 vdup.8 d27,d4[6] @(vii) 336 337 sbfx r9,lr,#16,#8 338 vsub.u8 d30,d1,d31 @(v)32-fract(dup_const_32_fract) 339 add r10,r8,r9 @(vii)*pu1_ref[ref_main_idx] 340 341 vld1.8 {d12},[r12],r11 @(vi)ref_main_idx 342 vmull.u8 q5,d8,d30 @(v)vmull_u8(ref_main_idx, dup_const_32_fract) 343 344 vld1.8 {d13},[r12] @(vi)ref_main_idx_1 345 vmlal.u8 q5,d9,d31 @(v)vmull_u8(ref_main_idx_1, dup_const_fract) 346 347 vst1.8 {d18},[r0],r3 @(iii) 348 vrshrn.i16 d22,q11,#5 @(iv)shift_res = vrshrn_n_u16(add_res, 5) 349 350 vdup.8 d25,d4[7] @(viii) 351 sbfx r9,lr,#24,#8 352 353 vld1.8 {d16},[r10],r11 @(vii)ref_main_idx 354 vsub.u8 d28,d1,d29 @(vi)32-fract(dup_const_32_fract) 355 356 vld1.8 {d17},[r10] @(vii)ref_main_idx_1 357 vmull.u8 q7,d12,d28 @(vi)vmull_u8(ref_main_idx, dup_const_32_fract) 358 359 add r12,r8,r9 @(viii)*pu1_ref[ref_main_idx] 360 vmlal.u8 q7,d13,d29 @(vi)vmull_u8(ref_main_idx_1, dup_const_fract) 361 subs r7,r7,#8 362 363 vst1.8 {d22},[r0],r3 @(iv) 364 cmp r4,#8 @ go to end if 4x4 365 beq end_loops 366 367 vrshrn.i16 d10,q5,#5 @(v)shift_res = vrshrn_n_u16(add_res, 5) 368 369 vld1.8 {d20},[r12],r11 @(viii)ref_main_idx 370 vsub.u8 d26,d1,d27 @(vii)32-fract(dup_const_32_fract) 371 372 vld1.8 {d21},[r12] @(viii)ref_main_idx_1 373 vmull.u8 q9,d16,d26 @(vii)vmull_u8(ref_main_idx, dup_const_32_fract) 374 375 addgt r8,r8,#8 376 vmlal.u8 q9,d17,d27 @(vii)vmull_u8(ref_main_idx_1, dup_const_fract) 377 subgt r4,r4,#8 378 379 vst1.8 {d10},[r0],r3 @(v) 380 vrshrn.i16 d14,q7,#5 @(vi)shift_res = vrshrn_n_u16(add_res, 5) 381 382 beq epilogue 383 384 vld1.8 {d5},[r6] @loads the row value 385 vmull.s8 q1,d5,d0 @pos = ((row + 1) * intra_pred_ang) 386 vand q2,q1,q3 @dup_const_fract(fract = pos & (31)) 387 vmovn.i16 d4,q2 388 vshrn.s16 d3,q1,#5 @idx = pos >> 5 389 vshl.s8 d3,d3,#1 390 vmov.u32 lr,d3[0] @(i)extract idx to the r register 391 @ lsl lr,lr,#1 392 sbfx r9,lr,#0,#8 393 add r10,r8,r9 @(i)*pu1_ref[ref_main_idx] 394 395 kernel_8_rows: 396 vdup.8 d31,d4[0] 397 subs r4,r4,#8 398 sbfx r9,lr,#8,#8 399 400 vld1.8 {d8},[r10],r11 @(i)ref_main_idx 401 vsub.u8 d24,d1,d25 @(viii)32-fract(dup_const_32_fract) 402 403 addle r6,r6,#8 @increment the row value 404 add r12,r8,r9 @(ii)*pu1_ref[ref_main_idx] 405 406 vld1.8 {d9},[r10] @(i)ref_main_idx_1 407 vmull.u8 q11,d20,d24 @(viii)vmull_u8(ref_main_idx, dup_const_32_fract) 408 409 vld1.8 {d5},[r6] @loads the row value 410 vmlal.u8 q11,d21,d25 @(viii)vmull_u8(ref_main_idx_1, dup_const_fract) 411 412 vdup.8 d29,d4[1] @(ii) 413 vrshrn.i16 d18,q9,#5 @(vii)shift_res = vrshrn_n_u16(add_res, 5) 414 415 sbfx r9,lr,#16,#8 416 417 vst1.8 {d14},[r0],r3 @(vi) 418 vsub.u8 d30,d1,d31 @(i)32-fract(dup_const_32_fract) 419 420 add r10,r8,r9 @(iii)*pu1_ref[ref_main_idx] 421 422 vld1.8 {d12},[r12],r11 @(ii)ref_main_idx 423 vmull.u8 q5,d8,d30 @(i)vmull_u8(ref_main_idx, dup_const_32_fract) 424 425 vld1.8 {d13},[r12] @(ii)ref_main_idx_1 426 vmlal.u8 q5,d9,d31 @(i)vmull_u8(ref_main_idx_1, dup_const_fract) 427 428 sbfx r9,lr,#24,#8 429 movle r4,r5 @reload nt 430 431 vmov.u32 lr,d3[1] @extract idx to the r register 432 vrshrn.i16 d22,q11,#5 @(viii)shift_res = vrshrn_n_u16(add_res, 5) 433 434 vdup.8 d27,d4[2] @(iii) 435 vsub.u8 d28,d1,d29 @(ii)32-fract(dup_const_32_fract) 436 add r12,r8,r9 @(iv)*pu1_ref[ref_main_idx] 437 438 vld1.8 {d16},[r10],r11 @(iii)ref_main_idx 439 vmull.u8 q7,d12,d28 @(ii)vmull_u8(ref_main_idx, dup_const_32_fract) 440 441 vst1.8 {d18},[r0],r3 @(vii) 442 vmlal.u8 q7,d13,d29 @(ii)vmull_u8(ref_main_idx_1, dup_const_fract) 443 444 vld1.8 {d17},[r10] @(iii)ref_main_idx_1 445 vrshrn.i16 d10,q5,#5 @(i)shift_res = vrshrn_n_u16(add_res, 5) 446 447 vdup.8 d25,d4[3] @(iv) 448 vmull.s8 q1,d5,d0 @pos = ((row + 1) * intra_pred_ang) 449 450 vst1.8 {d22},[r0] @(viii) 451 vsub.u8 d26,d1,d27 @(iii)32-fract(dup_const_32_fract) 452 453 vld1.8 {d20},[r12],r11 @(iv)ref_main_idx 454 vmull.u8 q9,d16,d26 @(iii)vmull_u8(ref_main_idx, dup_const_32_fract) 455 @ lsl lr,lr,#1 456 457 vld1.8 {d21},[r12] @(iv)ref_main_idx_1 458 vmlal.u8 q9,d17,d27 @(iii)vmull_u8(ref_main_idx_1, dup_const_fract) 459 460 sbfx r9,lr,#0,#8 461 add r0,r2,r3 462 463 vdup.8 d31,d4[4] @(v) 464 vrshrn.i16 d14,q7,#5 @(ii)shift_res = vrshrn_n_u16(add_res, 5) 465 466 add r10,r8,r9 @(v)*pu1_ref[ref_main_idx] 467 sbfx r9,lr,#8,#8 468 469 vst1.8 {d10},[r2]! @(i) 470 vsub.u8 d24,d1,d25 @(iv)32-fract(dup_const_32_fract) 471 472 vdup.8 d29,d4[5] @(vi) 473 vmull.u8 q11,d20,d24 @(iv)vmull_u8(ref_main_idx, dup_const_32_fract) 474 475 vdup.8 d27,d4[6] @(vii) 476 vmlal.u8 q11,d21,d25 @(iv)vmull_u8(ref_main_idx_1, dup_const_fract) 477 478 add r12,r8,r9 @(vi)*pu1_ref[ref_main_idx] 479 sbfx r9,lr,#16,#8 480 481 vdup.8 d25,d4[7] @(viii) 482 vrshrn.i16 d18,q9,#5 @(iii)shift_res = vrshrn_n_u16(add_res, 5) 483 484 vld1.8 {d8},[r10],r11 @(v)ref_main_idx 485 vand q2,q1,q3 @dup_const_fract(fract = pos & (31)) 486 487 vld1.8 {d9},[r10] @(v)ref_main_idx_1 488 vshrn.s16 d3,q1,#5 @idx = pos >> 5 489 490 vst1.8 {d14},[r0],r3 @(ii) 491 vrshrn.i16 d22,q11,#5 @(iv)shift_res = vrshrn_n_u16(add_res, 5) 492 493 add r10,r8,r9 @(vii)*pu1_ref[ref_main_idx] 494 sbfx r9,lr,#24,#8 495 496 vld1.8 {d12},[r12],r11 @(vi)ref_main_idx 497 vsub.u8 d30,d1,d31 @(v)32-fract(dup_const_32_fract) 498 499 vshl.s8 d3,d3,#1 500 501 vld1.8 {d13},[r12] @(vi)ref_main_idx_1 502 vmull.u8 q5,d8,d30 @(v)vmull_u8(ref_main_idx, dup_const_32_fract) 503 504 vmov.u32 lr,d3[0] @(i)extract idx to the r register 505 vmlal.u8 q5,d9,d31 @(v)vmull_u8(ref_main_idx_1, dup_const_fract) 506 507 add r12,r8,r9 @(viii)*pu1_ref[ref_main_idx] 508 movle r8,r1 @reload the source to pu1_src+2nt 509 510 vld1.8 {d16},[r10],r11 @(vii)ref_main_idx 511 vsub.u8 d28,d1,d29 @(vi)32-fract(dup_const_32_fract) 512 513 vst1.8 {d18},[r0],r3 @(iii) 514 vmull.u8 q7,d12,d28 @(vi)vmull_u8(ref_main_idx, dup_const_32_fract) 515 516 vld1.8 {d17},[r10] @(vii)ref_main_idx_1 517 vmlal.u8 q7,d13,d29 @(vi)vmull_u8(ref_main_idx_1, dup_const_fract) 518 519 vld1.8 {d20},[r12],r11 @(viii)ref_main_idx 520 vrshrn.i16 d10,q5,#5 @(v)shift_res = vrshrn_n_u16(add_res, 5) 521 522 vld1.8 {d21},[r12] @(viii)ref_main_idx_1 523 vsub.u8 d26,d1,d27 @(vii)32-fract(dup_const_32_fract) 524 525 addgt r8,r8,#8 @increment the source next set 8 columns in same row 526 lslle r12,r3,#3 527 suble r12,r12,r5 528 529 vst1.8 {d22},[r0],r3 @(iv) 530 vmull.u8 q9,d16,d26 @(vii)vmull_u8(ref_main_idx, dup_const_32_fract) 531 532 vst1.8 {d10},[r0],r3 @(v) 533 vmlal.u8 q9,d17,d27 @(vii)vmull_u8(ref_main_idx_1, dup_const_fract) 534 535 addle r2,r2,r12 @increment the dst pointer to 8*dst_strd - nt 536 sbfx r9,lr,#0,#8 537 538 vmovn.i16 d4,q2 539 vrshrn.i16 d14,q7,#5 @(vi)shift_res = vrshrn_n_u16(add_res, 5) 540 @ lsl lr,lr,#1 541 542 subs r7,r7,#8 543 add r10,r8,r9 @(i)*pu1_ref[ref_main_idx] 544 545 bne kernel_8_rows 546 547 epilogue: 548 vst1.8 {d14},[r0],r3 @(vi) 549 vrshrn.i16 d18,q9,#5 @(vii)shift_res = vrshrn_n_u16(add_res, 5) 550 551 vsub.u8 d24,d1,d25 @(viii)32-fract(dup_const_32_fract) 552 vmull.u8 q11,d20,d24 @(viii)vmull_u8(ref_main_idx, dup_const_32_fract) 553 vmlal.u8 q11,d21,d25 @(viii)vmull_u8(ref_main_idx_1, dup_const_fract) 554 555 vst1.8 {d18},[r0],r3 @(vii) 556 vrshrn.i16 d22,q11,#5 @(viii)shift_res = vrshrn_n_u16(add_res, 5) 557 558 vst1.8 {d22},[r0],r3 @(viii) 559 b end_loops 560 561 core_loop_4: 562 563 end_loops: 564 add sp, sp, #132 565 ldmfd sp!,{r4-r12,r15} @reload the registers from sp 566 567 568 569 570 571 572
