History

aarch64: Lowercase UXTW/SXTW and similar flags

Martin Storsjö committed 3 years ago

Commit 93cda5a1 parent 184103b6 files · +105 −105Browse files

Unified · Split

6 files
  1. libavcodec/aarch64/h264cmc_neon.S +2 −2
  2. libavcodec/aarch64/h264idct_neon.S +1 −1
  3. libavfilter/aarch64/vf_bwdif_neon.S +3 −3
  4. libavfilter/aarch64/vf_nlmeans_neon.S +4 −4
  5. libswscale/aarch64/hscale.S +88 −88
  6. libswscale/aarch64/yuv2rgb_neon.S +7 −7
libavcodec/aarch64/h264cmc_neon.S+2−2
@@ -38,7 +38,7 @@ function ff_\type\()_\codec\()_chroma_mc8_neon, export=1 · 37 unmodified lines
38 lsl w9, w9, #338 lsl w9, w9, #3
39 lsl w10, w10, #139 lsl w10, w10, #1
40 add w9, w9, w1040 add w9, w9, w10
41 add x6, x6, w9, UXTW41 add x6, x6, w9, uxtw
42 ld1r {v22.8h}, [x6]42 ld1r {v22.8h}, [x6]
43 .endif43 .endif
44 .ifc \codec,vc144 .ifc \codec,vc1
@@ -208,7 +208,7 @@ function ff_\type\()_\codec\()_chroma_mc4_neon, export=1 · 163 unmodified lines
208 lsl w9, w9, #3208 lsl w9, w9, #3
209 lsl w10, w10, #1209 lsl w10, w10, #1
210 add w9, w9, w10210 add w9, w9, w10
211 add x6, x6, w9, UXTW211 add x6, x6, w9, uxtw
212 ld1r {v22.8h}, [x6]212 ld1r {v22.8h}, [x6]
213 .endif213 .endif
214 .ifc \codec,vc1214 .ifc \codec,vc1
libavcodec/aarch64/h264idct_neon.S+1−1
@@ -385,7 +385,7 @@ function ff_h264_idct8_add4_neon, export=1 · 384 unmodified lines
385 movrel x14, .L_ff_h264_idct8_add_neon385 movrel x14, .L_ff_h264_idct8_add_neon
3861: ldrb w9, [x7], #43861: ldrb w9, [x7], #4
387 ldrsw x0, [x5], #16387 ldrsw x0, [x5], #16
388 ldrb w9, [x4, w9, UXTW]388 ldrb w9, [x4, w9, uxtw]
389 subs w9, w9, #1389 subs w9, w9, #1
390 b.lt 2f390 b.lt 2f
391 ldrsh w11, [x1]391 ldrsh w11, [x1]
libavfilter/aarch64/vf_bwdif_neon.S+3−3
@@ -186,13 +186,13 @@ function ff_bwdif_filter_line3_neon, export=1 · 185 unmodified lines
186 mov w10, w6 // w10 = loop count186 mov w10, w6 // w10 = loop count
187 neg w9, w5 // w9 = mref187 neg w9, w5 // w9 = mref
188 lsl w8, w9, #1 // w8 = mref2188 lsl w8, w9, #1 // w8 = mref2
189 add w7, w9, w9, LSL #1 // w7 = mref3189 add w7, w9, w9, lsl #1 // w7 = mref3
190 lsl w6, w9, #2 // w6 = mref4190 lsl w6, w9, #2 // w6 = mref4
191 mov w11, w5 // w11 = pref191 mov w11, w5 // w11 = pref
192 lsl w12, w5, #1 // w12 = pref2192 lsl w12, w5, #1 // w12 = pref2
193 add w13, w5, w5, LSL #1 // w13 = pref3193 add w13, w5, w5, lsl #1 // w13 = pref3
194 lsl w14, w5, #2 // w14 = pref4194 lsl w14, w5, #2 // w14 = pref4
195 add w15, w5, w5, LSL #2 // w15 = pref5195 add w15, w5, w5, lsl #2 // w15 = pref5
196 add w16, w14, w12 // w16 = pref6196 add w16, w14, w12 // w16 = pref6
197 197
198 lsl w5, w1, #1 // w5 = d_stride * 2198 lsl w5, w1, #1 // w5 = d_stride * 2
libavfilter/aarch64/vf_nlmeans_neon.S+4−4
@@ -35,10 +35,10 @@ · 34 unmodified lines
35 35
36function ff_compute_safe_ssd_integral_image_neon, export=136function ff_compute_safe_ssd_integral_image_neon, export=1
37 movi v26.4s, #0 // used as zero for the "rotations" in acc_sum_store37 movi v26.4s, #0 // used as zero for the "rotations" in acc_sum_store
38 sub x3, x3, w6, UXTW // s1 padding (s1_linesize - w)38 sub x3, x3, w6, uxtw // s1 padding (s1_linesize - w)
39 sub x5, x5, w6, UXTW // s2 padding (s2_linesize - w)39 sub x5, x5, w6, uxtw // s2 padding (s2_linesize - w)
40 sub x9, x0, w1, UXTW #2 // dst_top40 sub x9, x0, w1, uxtw #2 // dst_top
41 sub x1, x1, w6, UXTW // dst padding (dst_linesize_32 - w)41 sub x1, x1, w6, uxtw // dst padding (dst_linesize_32 - w)
42 lsl x1, x1, #2 // dst padding expressed in bytes42 lsl x1, x1, #2 // dst padding expressed in bytes
431: mov w10, w6 // width copy for each line431: mov w10, w6 // width copy for each line
44 sub x0, x0, #16 // beginning of the dst line minus 4 sums44 sub x0, x0, #16 // beginning of the dst line minus 4 sums
libswscale/aarch64/hscale.S+88−88
@@ -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 counter61 mov w15, w6 // filterSize counter
622: ld1 {v4.8b}, [x17], #8 // srcp[filterPos[0] + {0..7}]622: ld1 {v4.8b}, [x17], #8 // srcp[filterPos[0] + {0..7}]
63 ld1 {v5.8h}, [x16], #16 // load 8x16-bit filter values, part 163 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 filterPos231 add x5, x5, #32 // advance filterPos
232 232
233 // gather random access data from src into contiguous memory233 // 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 busy263 // 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-bit264 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-bit265 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 iteration266 ldr w8, [x3, w8, uxtw] // src[filterPos[idx + 0]], next iteration
267 ldr w9, [x3, w9, UXTW] // src[filterPos[idx + 1]], next iteration267 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-bit268 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-bit269 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 iteration270 ldr w10, [x3, w10, uxtw] // src[filterPos[idx + 2]], next iteration
271 ldr w11, [x3, w11, UXTW] // src[filterPos[idx + 3]], next iteration271 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..3273 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..3274 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 iteration275 ldr w12, [x3, w12, uxtw] // src[filterPos[idx + 4]], next iteration
276 ldr w13, [x3, w13, UXTW] // src[filterPos[idx + 5]], next iteration276 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..3277 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..3278 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 iteration279 ldr w14, [x3, w14, uxtw] // src[filterPos[idx + 6]], next iteration
280 ldr w15, [x3, w15, UXTW] // src[filterPos[idx + 7]], next iteration280 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..7282 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..7283 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
3312:3312:
332 // load src332 // 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 load334 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 filter336 // 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, #32372 add x5, x5, #32
373 373
374 // load data from374 // 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, #32384 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, #32400 add x5, x5, #32
401 uxtl v0.8h, v0.8b401 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 src403 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, v28.8h405 smull2 v6.4s, v0.8h, v28.8h
406 stp w8, w9, [sp]406 stp w8, w9, [sp]
407 407
408 uxtl v1.8h, v1.8b408 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 src410 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, v29.8h412 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, v2.8b415 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 src417 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, v30.8h419 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, v3.8b422 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 src424 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, v31.8h426 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
4692:4692:
470 ldr w8, [x5], #4 // load filterPos470 ldr w8, [x5], #4 // load filterPos
471 add x9, x3, w8, UXTW // src + filterPos471 add x9, x3, w8, uxtw // src + filterPos
472 ld1 {v0.s}[0], [x9] // load 4 * uint8_t* into one single472 ld1 {v0.s}[0], [x9] // load 4 * uint8_t* into one single
473 ld1 {v31.4h}, [x4], #8473 ld1 {v31.4h}, [x4], #8
474 uxtl v0.8h, v0.8b474 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 counter510 mov w15, w6 // filterSize counter
5112: ld1 {v4.8b}, [x17], #8 // srcp[filterPos[0] + {0..7}]5112: ld1 {v4.8b}, [x17], #8 // srcp[filterPos[0] + {0..7}]
512 ld1 {v5.8h}, [x16], #16 // load 8x16-bit filter values, part 1512 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 + 0568 mov x12, x4 // filter + 0
569 add x13, x4, x7 // filter + 1569 add x13, x4, x7 // filter + 1
570 add x8, x3, w8, UXTW // srcp + filterPos 0570 add x8, x3, w8, uxtw // srcp + filterPos 0
571 add x14, x13, x7 // filter + 2571 add x14, x13, x7 // filter + 2
572 add x9, x3, w9, UXTW // srcp + filterPos 1572 add x9, x3, w9, uxtw // srcp + filterPos 1
573 add x15, x14, x7 // filter + 3573 add x15, x14, x7 // filter + 3
574 add x10, x3, w10, UXTW // srcp + filterPos 2574 add x10, x3, w10, uxtw // srcp + filterPos 2
575 mov w0, w6 // save the filterSize to temporary variable575 mov w0, w6 // save the filterSize to temporary variable
576 add x11, x3, w11, UXTW // srcp + filterPos 3576 add x11, x3, w11, uxtw // srcp + filterPos 3
577 add x5, x5, #16 // advance filter position577 add x5, x5, #16 // advance filter position
578 mov x16, xzr // clear the register x16 used for offsetting the filter values578 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, #1674 lsl x15, x15, #1
675 675
676 // load src with given offset676 // 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, #64686 sub sp, sp, #64
687 // push src on stack so it can be loaded into vectors later687 // 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, #1754 lsl x14, x14, #1
755 lsl x15, x15, #1755 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
8192:8192:
820 ldr w8, [x5], #4 // load filterPos820 ldr w8, [x5], #4 // load filterPos
821 lsl w8, w8, #1821 lsl w8, w8, #1
822 add x9, x3, w8, UXTW // src + filterPos822 add x9, x3, w8, uxtw // src + filterPos
823 ld1 {v0.4h}, [x9] // load 4 * uint16_t823 ld1 {v0.4h}, [x9] // load 4 * uint16_t
824 ld1 {v31.4h}, [x4], #8824 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 counter876 mov w15, w6 // filterSize counter
8772: ld1 {v4.8h}, [x17], #16 // srcp[filterPos[0] + {0..7}]8772: ld1 {v4.8h}, [x17], #16 // srcp[filterPos[0] + {0..7}]
878 ld1 {v5.8h}, [x16], #16 // load 8x16-bit filter values, part 1878 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, #11082 lsl x15, x15, #1
1083 1083
1084 // load src with given offset1084 // 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, #641094 sub sp, sp, #64
1095 // push src on stack so it can be loaded into vectors later1095 // 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, #11160 lsl x14, x14, #1
1161 lsl x15, x15, #11161 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
12242:12242:
1225 ldr w8, [x5], #4 // load filterPos1225 ldr w8, [x5], #4 // load filterPos
1226 lsl w8, w8, #11226 lsl w8, w8, #1
1227 add x9, x3, w8, UXTW // src + filterPos1227 add x9, x3, w8, uxtw // src + filterPos
1228 ld1 {v0.4h}, [x9] // load 4 * uint16_t1228 ld1 {v0.4h}, [x9] // load 4 * uint16_t
1229 ld1 {v31.4h}, [x4], #81229 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 counter1281 mov w15, w6 // filterSize counter
12822: ld1 {v4.8h}, [x17], #16 // srcp[filterPos[0] + {0..7}]12822: ld1 {v4.8h}, [x17], #16 // srcp[filterPos[0] + {0..7}]
1283 ld1 {v5.8h}, [x16], #16 // load 8x16-bit filter values, part 11283 ld1 {v5.8h}, [x16], #16 // load 8x16-bit filter values, part 1
libswscale/aarch64/yuv2rgb_neon.S+7−7
@@ -102,7 +102,7 @@ · 101 unmodified lines
102.macro increment_nv12102.macro increment_nv12
103 ands w15, w1, #1103 ands w15, w1, #1
104 csel w16, w7, w11, ne // incC = (h & 1) ? paddincC : -width104 csel w16, w7, w11, ne // incC = (h & 1) ? paddincC : -width
105 add x6, x6, w16, SXTW // srcC += incC105 add x6, x6, w16, sxtw // srcC += incC
106.endm106.endm
107 107
108.macro increment_nv21108.macro increment_nv21
@@ -113,13 +113,13 @@ · 4 unmodified lines
113 ands w15, w1, #1113 ands w15, w1, #1
114 csel w16, w7, w11, ne // incU = (h & 1) ? paddincU : -width/2114 csel w16, w7, w11, ne // incU = (h & 1) ? paddincU : -width/2
115 csel w17, w14, w11, ne // incV = (h & 1) ? paddincV : -width/2115 csel w17, w14, w11, ne // incV = (h & 1) ? paddincV : -width/2
116 add x6, x6, w16, SXTW // srcU += incU116 add x6, x6, w16, sxtw // srcU += incU
117 add x13, x13, w17, SXTW // srcV += incV117 add x13, x13, w17, sxtw // srcV += incV
118.endm118.endm
119 119
120.macro increment_yuv422p120.macro increment_yuv422p
121 add x6, x6, w7, SXTW // srcU += incU121 add x6, x6, w7, sxtw // srcU += incU
122 add x13, x13, w14, SXTW // srcV += incV122 add x13, x13, w14, sxtw // srcV += incV
123.endm123.endm
124 124
125.macro compute_rgba r1 g1 b1 a1 r2 g2 b2 a2125.macro compute_rgba r1 g1 b1 a1 r2 g2 b2 a2
@@ -189,8 +189,8 @@ function ff_\ifmt\()_to_\ofmt\()_neon, export=1 · 63 unmodified lines
189 st4 {v16.8b,v17.8b,v18.8b,v19.8b}, [x2], #32189 st4 {v16.8b,v17.8b,v18.8b,v19.8b}, [x2], #32
190 subs w8, w8, #16 // width -= 16190 subs w8, w8, #16 // width -= 16
191 b.gt 2b191 b.gt 2b
192 add x2, x2, w3, SXTW // dst += padding192 add x2, x2, w3, sxtw // dst += padding
193 add x4, x4, w5, SXTW // srcY += paddingY193 add x4, x4, w5, sxtw // srcY += paddingY
194 increment_\ifmt194 increment_\ifmt
195 subs w1, w1, #1 // height -= 1195 subs w1, w1, #1 // height -= 1
196 b.gt 1b196 b.gt 1b

Keyboard shortcuts

General

Show keyboard shortcuts
?
Focus the current search
/

Issues and pull requests

Open filters
f

Issues, pull requests and Inbox

Select next item
j↓
Select previous item
k↑
Open selected item
Enter
Dismiss filters or clear selection
Esc

Repository code

Go to file
t