| @@ -54,10 +54,10 @@ function ff_hscale8to15_X8_neon, export=1 · 53 unmodified lines |
| 54 | movi v1.2d, #0 // val sum part 2 (for dst[1]) | 54 | movi v1.2d, #0 // val sum part 2 (for dst[1]) |
| 55 | movi v2.2d, #0 // val sum part 3 (for dst[2]) | 55 | movi v2.2d, #0 // val sum part 3 (for dst[2]) |
| 56 | movi v3.2d, #0 // val sum part 4 (for dst[3]) | 56 | movi v3.2d, #0 // val sum part 4 (for dst[3]) |
| 57 | add x17, x3, w8, UXTW // srcp + filterPos[0] | 57 | add x17, x3, w8, uxtw // srcp + filterPos[0] |
| 58 | add x8, x3, w0, UXTW // srcp + filterPos[1] | 58 | add x8, x3, w0, uxtw // srcp + filterPos[1] |
| 59 | add x0, x3, w11, UXTW // srcp + filterPos[2] | 59 | add x0, x3, w11, uxtw // srcp + filterPos[2] |
| 60 | add x11, x3, w9, UXTW // srcp + filterPos[3] | 60 | add x11, x3, w9, uxtw // srcp + filterPos[3] |
| 61 | mov w15, w6 // filterSize counter | 61 | mov w15, w6 // filterSize counter |
| 62 | 2: ld1 {v4.8b}, [x17], #8 // srcp[filterPos[0] + {0..7}] | 62 | 2: ld1 {v4.8b}, [x17], #8 // srcp[filterPos[0] + {0..7}] |
| 63 | ld1 {v5.8h}, [x16], #16 // load 8x16-bit filter values, part 1 | 63 | ld1 {v5.8h}, [x16], #16 // load 8x16-bit filter values, part 1 |
| @@ -231,14 +231,14 @@ function ff_hscale8to15_4_neon, export=1 · 167 unmodified lines |
| 231 | add x5, x5, #32 // advance filterPos | 231 | add x5, x5, #32 // advance filterPos |
| 232 | | 232 | |
| 233 | // gather random access data from src into contiguous memory | 233 | // gather random access data from src into contiguous memory |
| 234 | ldr w8, [x3, w8, UXTW] // src[filterPos[idx + 0]][0..3] | 234 | ldr w8, [x3, w8, uxtw] // src[filterPos[idx + 0]][0..3] |
| 235 | ldr w9, [x3, w9, UXTW] // src[filterPos[idx + 1]][0..3] | 235 | ldr w9, [x3, w9, uxtw] // src[filterPos[idx + 1]][0..3] |
| 236 | ldr w10, [x3, w10, UXTW] // src[filterPos[idx + 2]][0..3] | 236 | ldr w10, [x3, w10, uxtw] // src[filterPos[idx + 2]][0..3] |
| 237 | ldr w11, [x3, w11, UXTW] // src[filterPos[idx + 3]][0..3] | 237 | ldr w11, [x3, w11, uxtw] // src[filterPos[idx + 3]][0..3] |
| 238 | ldr w12, [x3, w12, UXTW] // src[filterPos[idx + 4]][0..3] | 238 | ldr w12, [x3, w12, uxtw] // src[filterPos[idx + 4]][0..3] |
| 239 | ldr w13, [x3, w13, UXTW] // src[filterPos[idx + 5]][0..3] | 239 | ldr w13, [x3, w13, uxtw] // src[filterPos[idx + 5]][0..3] |
| 240 | ldr w14, [x3, w14, UXTW] // src[filterPos[idx + 6]][0..3] | 240 | ldr w14, [x3, w14, uxtw] // src[filterPos[idx + 6]][0..3] |
| 241 | ldr w15, [x3, w15, UXTW] // src[filterPos[idx + 7]][0..3] | 241 | ldr w15, [x3, w15, uxtw] // src[filterPos[idx + 7]][0..3] |
| 242 | stp w8, w9, [sp] // *scratch_mem = { src[filterPos[idx + 0]][0..3], src[filterPos[idx + 1]][0..3] } | 242 | stp w8, w9, [sp] // *scratch_mem = { src[filterPos[idx + 0]][0..3], src[filterPos[idx + 1]][0..3] } |
| 243 | stp w10, w11, [sp, #8] // *scratch_mem = { src[filterPos[idx + 2]][0..3], src[filterPos[idx + 3]][0..3] } | 243 | stp w10, w11, [sp, #8] // *scratch_mem = { src[filterPos[idx + 2]][0..3], src[filterPos[idx + 3]][0..3] } |
| 244 | stp w12, w13, [sp, #16] // *scratch_mem = { src[filterPos[idx + 4]][0..3], src[filterPos[idx + 5]][0..3] } | 244 | stp w12, w13, [sp, #16] // *scratch_mem = { src[filterPos[idx + 4]][0..3], src[filterPos[idx + 5]][0..3] } |
| @@ -263,21 +263,21 @@ function ff_hscale8to15_4_neon, export=1 · 18 unmodified lines |
| 263 | // interleaved SIMD and prefetching intended to keep ld/st and vector pipelines busy | 263 | // interleaved SIMD and prefetching intended to keep ld/st and vector pipelines busy |
| 264 | uxtl v16.8h, v16.8b // unsigned extend long, covert src data to 16-bit | 264 | uxtl v16.8h, v16.8b // unsigned extend long, covert src data to 16-bit |
| 265 | uxtl v17.8h, v17.8b // unsigned extend long, covert src data to 16-bit | 265 | uxtl v17.8h, v17.8b // unsigned extend long, covert src data to 16-bit |
| 266 | ldr w8, [x3, w8, UXTW] // src[filterPos[idx + 0]], next iteration | 266 | ldr w8, [x3, w8, uxtw] // src[filterPos[idx + 0]], next iteration |
| 267 | ldr w9, [x3, w9, UXTW] // src[filterPos[idx + 1]], next iteration | 267 | ldr w9, [x3, w9, uxtw] // src[filterPos[idx + 1]], next iteration |
| 268 | uxtl v18.8h, v18.8b // unsigned extend long, covert src data to 16-bit | 268 | uxtl v18.8h, v18.8b // unsigned extend long, covert src data to 16-bit |
| 269 | uxtl v19.8h, v19.8b // unsigned extend long, covert src data to 16-bit | 269 | uxtl v19.8h, v19.8b // unsigned extend long, covert src data to 16-bit |
| 270 | ldr w10, [x3, w10, UXTW] // src[filterPos[idx + 2]], next iteration | 270 | ldr w10, [x3, w10, uxtw] // src[filterPos[idx + 2]], next iteration |
| 271 | ldr w11, [x3, w11, UXTW] // src[filterPos[idx + 3]], next iteration | 271 | ldr w11, [x3, w11, uxtw] // src[filterPos[idx + 3]], next iteration |
| 272 | | 272 | |
| 273 | smlal v0.4s, v1.4h, v16.4h // multiply accumulate inner loop j = 0, idx = 0..3 | 273 | smlal v0.4s, v1.4h, v16.4h // multiply accumulate inner loop j = 0, idx = 0..3 |
| 274 | smlal v0.4s, v2.4h, v17.4h // multiply accumulate inner loop j = 1, idx = 0..3 | 274 | smlal v0.4s, v2.4h, v17.4h // multiply accumulate inner loop j = 1, idx = 0..3 |
| 275 | ldr w12, [x3, w12, UXTW] // src[filterPos[idx + 4]], next iteration | 275 | ldr w12, [x3, w12, uxtw] // src[filterPos[idx + 4]], next iteration |
| 276 | ldr w13, [x3, w13, UXTW] // src[filterPos[idx + 5]], next iteration | 276 | ldr w13, [x3, w13, uxtw] // src[filterPos[idx + 5]], next iteration |
| 277 | smlal v0.4s, v3.4h, v18.4h // multiply accumulate inner loop j = 2, idx = 0..3 | 277 | smlal v0.4s, v3.4h, v18.4h // multiply accumulate inner loop j = 2, idx = 0..3 |
| 278 | smlal v0.4s, v4.4h, v19.4h // multiply accumulate inner loop j = 3, idx = 0..3 | 278 | smlal v0.4s, v4.4h, v19.4h // multiply accumulate inner loop j = 3, idx = 0..3 |
| 279 | ldr w14, [x3, w14, UXTW] // src[filterPos[idx + 6]], next iteration | 279 | ldr w14, [x3, w14, uxtw] // src[filterPos[idx + 6]], next iteration |
| 280 | ldr w15, [x3, w15, UXTW] // src[filterPos[idx + 7]], next iteration | 280 | ldr w15, [x3, w15, uxtw] // src[filterPos[idx + 7]], next iteration |
| 281 | | 281 | |
| 282 | smlal2 v5.4s, v1.8h, v16.8h // multiply accumulate inner loop j = 0, idx = 4..7 | 282 | smlal2 v5.4s, v1.8h, v16.8h // multiply accumulate inner loop j = 0, idx = 4..7 |
| 283 | smlal2 v5.4s, v2.8h, v17.8h // multiply accumulate inner loop j = 1, idx = 4..7 | 283 | smlal2 v5.4s, v2.8h, v17.8h // multiply accumulate inner loop j = 1, idx = 4..7 |
| @@ -331,7 +331,7 @@ function ff_hscale8to15_4_neon, export=1 · 47 unmodified lines |
| 331 | 2: | 331 | 2: |
| 332 | // load src | 332 | // load src |
| 333 | ldr w8, [x5], #4 // filterPos[i] | 333 | ldr w8, [x5], #4 // filterPos[i] |
| 334 | add x9, x3, w8, UXTW // calculate the address for src load | 334 | add x9, x3, w8, uxtw // calculate the address for src load |
| 335 | ld1 {v5.s}[0], [x9] // src[filterPos[i] + 0..3] | 335 | ld1 {v5.s}[0], [x9] // src[filterPos[i] + 0..3] |
| 336 | // load filter | 336 | // load filter |
| 337 | ld1 {v6.4h}, [x4], #8 // filter[filterSize * i + 0..3] | 337 | ld1 {v6.4h}, [x4], #8 // filter[filterSize * i + 0..3] |
| @@ -372,14 +372,14 @@ function ff_hscale8to19_4_neon, export=1 · 34 unmodified lines |
| 372 | add x5, x5, #32 | 372 | add x5, x5, #32 |
| 373 | | 373 | |
| 374 | // load data from | 374 | // load data from |
| 375 | ldr w8, [x3, w8, UXTW] | 375 | ldr w8, [x3, w8, uxtw] |
| 376 | ldr w9, [x3, w9, UXTW] | 376 | ldr w9, [x3, w9, uxtw] |
| 377 | ldr w10, [x3, w10, UXTW] | 377 | ldr w10, [x3, w10, uxtw] |
| 378 | ldr w11, [x3, w11, UXTW] | 378 | ldr w11, [x3, w11, uxtw] |
| 379 | ldr w12, [x3, w12, UXTW] | 379 | ldr w12, [x3, w12, uxtw] |
| 380 | ldr w13, [x3, w13, UXTW] | 380 | ldr w13, [x3, w13, uxtw] |
| 381 | ldr w14, [x3, w14, UXTW] | 381 | ldr w14, [x3, w14, uxtw] |
| 382 | ldr w15, [x3, w15, UXTW] | 382 | ldr w15, [x3, w15, uxtw] |
| 383 | | 383 | |
| 384 | sub sp, sp, #32 | 384 | sub sp, sp, #32 |
| 385 | | 385 | |
| @@ -399,30 +399,30 @@ function ff_hscale8to19_4_neon, export=1 · 13 unmodified lines |
| 399 | ldp w14, w15, [x5, #24] // filterPos[6], filterPos[7] | 399 | ldp w14, w15, [x5, #24] // filterPos[6], filterPos[7] |
| 400 | add x5, x5, #32 | 400 | add x5, x5, #32 |
| 401 | uxtl v0.8h, v0.8b | 401 | uxtl v0.8h, v0.8b |
| 402 | ldr w8, [x3, w8, UXTW] | 402 | ldr w8, [x3, w8, uxtw] |
| 403 | smull v5.4s, v0.4h, v28.4h // multiply first column of src | 403 | smull v5.4s, v0.4h, v28.4h // multiply first column of src |
| 404 | ldr w9, [x3, w9, UXTW] | 404 | ldr w9, [x3, w9, uxtw] |
| 405 | smull2 v6.4s, v0.8h, v28.8h | 405 | smull2 v6.4s, v0.8h, v28.8h |
| 406 | stp w8, w9, [sp] | 406 | stp w8, w9, [sp] |
| 407 | | 407 | |
| 408 | uxtl v1.8h, v1.8b | 408 | uxtl v1.8h, v1.8b |
| 409 | ldr w10, [x3, w10, UXTW] | 409 | ldr w10, [x3, w10, uxtw] |
| 410 | smlal v5.4s, v1.4h, v29.4h // multiply second column of src | 410 | smlal v5.4s, v1.4h, v29.4h // multiply second column of src |
| 411 | ldr w11, [x3, w11, UXTW] | 411 | ldr w11, [x3, w11, uxtw] |
| 412 | smlal2 v6.4s, v1.8h, v29.8h | 412 | smlal2 v6.4s, v1.8h, v29.8h |
| 413 | stp w10, w11, [sp, #8] | 413 | stp w10, w11, [sp, #8] |
| 414 | | 414 | |
| 415 | uxtl v2.8h, v2.8b | 415 | uxtl v2.8h, v2.8b |
| 416 | ldr w12, [x3, w12, UXTW] | 416 | ldr w12, [x3, w12, uxtw] |
| 417 | smlal v5.4s, v2.4h, v30.4h // multiply third column of src | 417 | smlal v5.4s, v2.4h, v30.4h // multiply third column of src |
| 418 | ldr w13, [x3, w13, UXTW] | 418 | ldr w13, [x3, w13, uxtw] |
| 419 | smlal2 v6.4s, v2.8h, v30.8h | 419 | smlal2 v6.4s, v2.8h, v30.8h |
| 420 | stp w12, w13, [sp, #16] | 420 | stp w12, w13, [sp, #16] |
| 421 | | 421 | |
| 422 | uxtl v3.8h, v3.8b | 422 | uxtl v3.8h, v3.8b |
| 423 | ldr w14, [x3, w14, UXTW] | 423 | ldr w14, [x3, w14, uxtw] |
| 424 | smlal v5.4s, v3.4h, v31.4h // multiply fourth column of src | 424 | smlal v5.4s, v3.4h, v31.4h // multiply fourth column of src |
| 425 | ldr w15, [x3, w15, UXTW] | 425 | ldr w15, [x3, w15, uxtw] |
| 426 | smlal2 v6.4s, v3.8h, v31.8h | 426 | smlal2 v6.4s, v3.8h, v31.8h |
| 427 | stp w14, w15, [sp, #24] | 427 | stp w14, w15, [sp, #24] |
| 428 | | 428 | |
| @@ -468,7 +468,7 @@ function ff_hscale8to19_4_neon, export=1 · 39 unmodified lines |
| 468 | | 468 | |
| 469 | 2: | 469 | 2: |
| 470 | ldr w8, [x5], #4 // load filterPos | 470 | ldr w8, [x5], #4 // load filterPos |
| 471 | add x9, x3, w8, UXTW // src + filterPos | 471 | add x9, x3, w8, uxtw // src + filterPos |
| 472 | ld1 {v0.s}[0], [x9] // load 4 * uint8_t* into one single | 472 | ld1 {v0.s}[0], [x9] // load 4 * uint8_t* into one single |
| 473 | ld1 {v31.4h}, [x4], #8 | 473 | ld1 {v31.4h}, [x4], #8 |
| 474 | uxtl v0.8h, v0.8b | 474 | uxtl v0.8h, v0.8b |
| @@ -503,10 +503,10 @@ function ff_hscale8to19_X8_neon, export=1 · 28 unmodified lines |
| 503 | movi v1.2d, #0 // val sum part 2 (for dst[1]) | 503 | movi v1.2d, #0 // val sum part 2 (for dst[1]) |
| 504 | movi v2.2d, #0 // val sum part 3 (for dst[2]) | 504 | movi v2.2d, #0 // val sum part 3 (for dst[2]) |
| 505 | movi v3.2d, #0 // val sum part 4 (for dst[3]) | 505 | movi v3.2d, #0 // val sum part 4 (for dst[3]) |
| 506 | add x17, x3, w8, UXTW // srcp + filterPos[0] | 506 | add x17, x3, w8, uxtw // srcp + filterPos[0] |
| 507 | add x8, x3, w0, UXTW // srcp + filterPos[1] | 507 | add x8, x3, w0, uxtw // srcp + filterPos[1] |
| 508 | add x0, x3, w11, UXTW // srcp + filterPos[2] | 508 | add x0, x3, w11, uxtw // srcp + filterPos[2] |
| 509 | add x11, x3, w9, UXTW // srcp + filterPos[3] | 509 | add x11, x3, w9, uxtw // srcp + filterPos[3] |
| 510 | mov w15, w6 // filterSize counter | 510 | mov w15, w6 // filterSize counter |
| 511 | 2: ld1 {v4.8b}, [x17], #8 // srcp[filterPos[0] + {0..7}] | 511 | 2: ld1 {v4.8b}, [x17], #8 // srcp[filterPos[0] + {0..7}] |
| 512 | ld1 {v5.8h}, [x16], #16 // load 8x16-bit filter values, part 1 | 512 | ld1 {v5.8h}, [x16], #16 // load 8x16-bit filter values, part 1 |
| @@ -567,13 +567,13 @@ function ff_hscale8to19_X4_neon, export=1 · 54 unmodified lines |
| 567 | | 567 | |
| 568 | mov x12, x4 // filter + 0 | 568 | mov x12, x4 // filter + 0 |
| 569 | add x13, x4, x7 // filter + 1 | 569 | add x13, x4, x7 // filter + 1 |
| 570 | add x8, x3, w8, UXTW // srcp + filterPos 0 | 570 | add x8, x3, w8, uxtw // srcp + filterPos 0 |
| 571 | add x14, x13, x7 // filter + 2 | 571 | add x14, x13, x7 // filter + 2 |
| 572 | add x9, x3, w9, UXTW // srcp + filterPos 1 | 572 | add x9, x3, w9, uxtw // srcp + filterPos 1 |
| 573 | add x15, x14, x7 // filter + 3 | 573 | add x15, x14, x7 // filter + 3 |
| 574 | add x10, x3, w10, UXTW // srcp + filterPos 2 | 574 | add x10, x3, w10, uxtw // srcp + filterPos 2 |
| 575 | mov w0, w6 // save the filterSize to temporary variable | 575 | mov w0, w6 // save the filterSize to temporary variable |
| 576 | add x11, x3, w11, UXTW // srcp + filterPos 3 | 576 | add x11, x3, w11, uxtw // srcp + filterPos 3 |
| 577 | add x5, x5, #16 // advance filter position | 577 | add x5, x5, #16 // advance filter position |
| 578 | mov x16, xzr // clear the register x16 used for offsetting the filter values | 578 | mov x16, xzr // clear the register x16 used for offsetting the filter values |
| 579 | | 579 | |
| @@ -674,14 +674,14 @@ function ff_hscale16to15_4_neon_asm, export=1 · 94 unmodified lines |
| 674 | lsl x15, x15, #1 | 674 | lsl x15, x15, #1 |
| 675 | | 675 | |
| 676 | // load src with given offset | 676 | // load src with given offset |
| 677 | ldr x8, [x3, w8, UXTW] | 677 | ldr x8, [x3, w8, uxtw] |
| 678 | ldr x9, [x3, w9, UXTW] | 678 | ldr x9, [x3, w9, uxtw] |
| 679 | ldr x10, [x3, w10, UXTW] | 679 | ldr x10, [x3, w10, uxtw] |
| 680 | ldr x11, [x3, w11, UXTW] | 680 | ldr x11, [x3, w11, uxtw] |
| 681 | ldr x12, [x3, w12, UXTW] | 681 | ldr x12, [x3, w12, uxtw] |
| 682 | ldr x13, [x3, w13, UXTW] | 682 | ldr x13, [x3, w13, uxtw] |
| 683 | ldr x14, [x3, w14, UXTW] | 683 | ldr x14, [x3, w14, uxtw] |
| 684 | ldr x15, [x3, w15, UXTW] | 684 | ldr x15, [x3, w15, uxtw] |
| 685 | | 685 | |
| 686 | sub sp, sp, #64 | 686 | sub sp, sp, #64 |
| 687 | // push src on stack so it can be loaded into vectors later | 687 | // push src on stack so it can be loaded into vectors later |
| @@ -754,14 +754,14 @@ function ff_hscale16to15_4_neon_asm, export=1 · 66 unmodified lines |
| 754 | lsl x14, x14, #1 | 754 | lsl x14, x14, #1 |
| 755 | lsl x15, x15, #1 | 755 | lsl x15, x15, #1 |
| 756 | | 756 | |
| 757 | ldr x8, [x3, w8, UXTW] | 757 | ldr x8, [x3, w8, uxtw] |
| 758 | ldr x9, [x3, w9, UXTW] | 758 | ldr x9, [x3, w9, uxtw] |
| 759 | ldr x10, [x3, w10, UXTW] | 759 | ldr x10, [x3, w10, uxtw] |
| 760 | ldr x11, [x3, w11, UXTW] | 760 | ldr x11, [x3, w11, uxtw] |
| 761 | ldr x12, [x3, w12, UXTW] | 761 | ldr x12, [x3, w12, uxtw] |
| 762 | ldr x13, [x3, w13, UXTW] | 762 | ldr x13, [x3, w13, uxtw] |
| 763 | ldr x14, [x3, w14, UXTW] | 763 | ldr x14, [x3, w14, uxtw] |
| 764 | ldr x15, [x3, w15, UXTW] | 764 | ldr x15, [x3, w15, uxtw] |
| 765 | | 765 | |
| 766 | stp x8, x9, [sp] | 766 | stp x8, x9, [sp] |
| 767 | stp x10, x11, [sp, #16] | 767 | stp x10, x11, [sp, #16] |
| @@ -819,7 +819,7 @@ function ff_hscale16to15_4_neon_asm, export=1 · 51 unmodified lines |
| 819 | 2: | 819 | 2: |
| 820 | ldr w8, [x5], #4 // load filterPos | 820 | ldr w8, [x5], #4 // load filterPos |
| 821 | lsl w8, w8, #1 | 821 | lsl w8, w8, #1 |
| 822 | add x9, x3, w8, UXTW // src + filterPos | 822 | add x9, x3, w8, uxtw // src + filterPos |
| 823 | ld1 {v0.4h}, [x9] // load 4 * uint16_t | 823 | ld1 {v0.4h}, [x9] // load 4 * uint16_t |
| 824 | ld1 {v31.4h}, [x4], #8 | 824 | ld1 {v31.4h}, [x4], #8 |
| 825 | | 825 | |
| @@ -869,10 +869,10 @@ function ff_hscale16to15_X8_neon_asm, export=1 · 43 unmodified lines |
| 869 | movi v1.2d, #0 // val sum part 2 (for dst[1]) | 869 | movi v1.2d, #0 // val sum part 2 (for dst[1]) |
| 870 | movi v2.2d, #0 // val sum part 3 (for dst[2]) | 870 | movi v2.2d, #0 // val sum part 3 (for dst[2]) |
| 871 | movi v3.2d, #0 // val sum part 4 (for dst[3]) | 871 | movi v3.2d, #0 // val sum part 4 (for dst[3]) |
| 872 | add x17, x3, w8, UXTW // srcp + filterPos[0] | 872 | add x17, x3, w8, uxtw // srcp + filterPos[0] |
| 873 | add x8, x3, w10, UXTW // srcp + filterPos[1] | 873 | add x8, x3, w10, uxtw // srcp + filterPos[1] |
| 874 | add x10, x3, w11, UXTW // srcp + filterPos[2] | 874 | add x10, x3, w11, uxtw // srcp + filterPos[2] |
| 875 | add x11, x3, w9, UXTW // srcp + filterPos[3] | 875 | add x11, x3, w9, uxtw // srcp + filterPos[3] |
| 876 | mov w15, w6 // filterSize counter | 876 | mov w15, w6 // filterSize counter |
| 877 | 2: ld1 {v4.8h}, [x17], #16 // srcp[filterPos[0] + {0..7}] | 877 | 2: ld1 {v4.8h}, [x17], #16 // srcp[filterPos[0] + {0..7}] |
| 878 | ld1 {v5.8h}, [x16], #16 // load 8x16-bit filter values, part 1 | 878 | ld1 {v5.8h}, [x16], #16 // load 8x16-bit filter values, part 1 |
| @@ -1082,14 +1082,14 @@ function ff_hscale16to19_4_neon_asm, export=1 · 203 unmodified lines |
| 1082 | lsl x15, x15, #1 | 1082 | lsl x15, x15, #1 |
| 1083 | | 1083 | |
| 1084 | // load src with given offset | 1084 | // load src with given offset |
| 1085 | ldr x8, [x3, w8, UXTW] | 1085 | ldr x8, [x3, w8, uxtw] |
| 1086 | ldr x9, [x3, w9, UXTW] | 1086 | ldr x9, [x3, w9, uxtw] |
| 1087 | ldr x10, [x3, w10, UXTW] | 1087 | ldr x10, [x3, w10, uxtw] |
| 1088 | ldr x11, [x3, w11, UXTW] | 1088 | ldr x11, [x3, w11, uxtw] |
| 1089 | ldr x12, [x3, w12, UXTW] | 1089 | ldr x12, [x3, w12, uxtw] |
| 1090 | ldr x13, [x3, w13, UXTW] | 1090 | ldr x13, [x3, w13, uxtw] |
| 1091 | ldr x14, [x3, w14, UXTW] | 1091 | ldr x14, [x3, w14, uxtw] |
| 1092 | ldr x15, [x3, w15, UXTW] | 1092 | ldr x15, [x3, w15, uxtw] |
| 1093 | | 1093 | |
| 1094 | sub sp, sp, #64 | 1094 | sub sp, sp, #64 |
| 1095 | // push src on stack so it can be loaded into vectors later | 1095 | // push src on stack so it can be loaded into vectors later |
| @@ -1160,14 +1160,14 @@ function ff_hscale16to19_4_neon_asm, export=1 · 64 unmodified lines |
| 1160 | lsl x14, x14, #1 | 1160 | lsl x14, x14, #1 |
| 1161 | lsl x15, x15, #1 | 1161 | lsl x15, x15, #1 |
| 1162 | | 1162 | |
| 1163 | ldr x8, [x3, w8, UXTW] | 1163 | ldr x8, [x3, w8, uxtw] |
| 1164 | ldr x9, [x3, w9, UXTW] | 1164 | ldr x9, [x3, w9, uxtw] |
| 1165 | ldr x10, [x3, w10, UXTW] | 1165 | ldr x10, [x3, w10, uxtw] |
| 1166 | ldr x11, [x3, w11, UXTW] | 1166 | ldr x11, [x3, w11, uxtw] |
| 1167 | ldr x12, [x3, w12, UXTW] | 1167 | ldr x12, [x3, w12, uxtw] |
| 1168 | ldr x13, [x3, w13, UXTW] | 1168 | ldr x13, [x3, w13, uxtw] |
| 1169 | ldr x14, [x3, w14, UXTW] | 1169 | ldr x14, [x3, w14, uxtw] |
| 1170 | ldr x15, [x3, w15, UXTW] | 1170 | ldr x15, [x3, w15, uxtw] |
| 1171 | | 1171 | |
| 1172 | stp x8, x9, [sp] | 1172 | stp x8, x9, [sp] |
| 1173 | stp x10, x11, [sp, #16] | 1173 | stp x10, x11, [sp, #16] |
| @@ -1224,7 +1224,7 @@ function ff_hscale16to19_4_neon_asm, export=1 · 50 unmodified lines |
| 1224 | 2: | 1224 | 2: |
| 1225 | ldr w8, [x5], #4 // load filterPos | 1225 | ldr w8, [x5], #4 // load filterPos |
| 1226 | lsl w8, w8, #1 | 1226 | lsl w8, w8, #1 |
| 1227 | add x9, x3, w8, UXTW // src + filterPos | 1227 | add x9, x3, w8, uxtw // src + filterPos |
| 1228 | ld1 {v0.4h}, [x9] // load 4 * uint16_t | 1228 | ld1 {v0.4h}, [x9] // load 4 * uint16_t |
| 1229 | ld1 {v31.4h}, [x4], #8 | 1229 | ld1 {v31.4h}, [x4], #8 |
| 1230 | | 1230 | |
| @@ -1274,10 +1274,10 @@ function ff_hscale16to19_X8_neon_asm, export=1 · 43 unmodified lines |
| 1274 | movi v1.2d, #0 // val sum part 2 (for dst[1]) | 1274 | movi v1.2d, #0 // val sum part 2 (for dst[1]) |
| 1275 | movi v2.2d, #0 // val sum part 3 (for dst[2]) | 1275 | movi v2.2d, #0 // val sum part 3 (for dst[2]) |
| 1276 | movi v3.2d, #0 // val sum part 4 (for dst[3]) | 1276 | movi v3.2d, #0 // val sum part 4 (for dst[3]) |
| 1277 | add x17, x3, w8, UXTW // srcp + filterPos[0] | 1277 | add x17, x3, w8, uxtw // srcp + filterPos[0] |
| 1278 | add x8, x3, w10, UXTW // srcp + filterPos[1] | 1278 | add x8, x3, w10, uxtw // srcp + filterPos[1] |
| 1279 | add x10, x3, w11, UXTW // srcp + filterPos[2] | 1279 | add x10, x3, w11, uxtw // srcp + filterPos[2] |
| 1280 | add x11, x3, w9, UXTW // srcp + filterPos[3] | 1280 | add x11, x3, w9, uxtw // srcp + filterPos[3] |
| 1281 | mov w15, w6 // filterSize counter | 1281 | mov w15, w6 // filterSize counter |
| 1282 | 2: ld1 {v4.8h}, [x17], #16 // srcp[filterPos[0] + {0..7}] | 1282 | 2: ld1 {v4.8h}, [x17], #16 // srcp[filterPos[0] + {0..7}] |
| 1283 | ld1 {v5.8h}, [x16], #16 // load 8x16-bit filter values, part 1 | 1283 | ld1 {v5.8h}, [x16], #16 // load 8x16-bit filter values, part 1 |