Showing the changes to libavcodec (2 of 3 files). Show all
Unified · Split
| @@ -28,6 +28,10 @@ · 27 unmodified lines | ||
| 28 | 28 | void ff_ac3_exponent_min_neon(uint8_t *exp, int num_reuse_blocks, int nb_coefs); |
| 29 | 29 | void ff_ac3_extract_exponents_neon(uint8_t *exp, int32_t *coef, int nb_coefs); |
| 30 | 30 | void ff_float_to_fixed24_neon(int32_t *dst, const float *src, size_t len); |
| 31 | +void ff_ac3_sum_square_butterfly_int32_neon(int64_t sum[4], | |
| 32 | + const int32_t *coef0, | |
| 33 | + const int32_t *coef1, | |
| 34 | + int len); | |
| 31 | 35 | |
| 32 | 36 | av_cold void ff_ac3dsp_init_aarch64(AC3DSPContext *c) |
| 33 | 37 | { |
| @@ -37,4 +41,5 @@ av_cold void ff_ac3dsp_init_aarch64(AC3DSPContext *c) · 3 unmodified lines | ||
| 37 | 41 | c->ac3_exponent_min = ff_ac3_exponent_min_neon; |
| 38 | 42 | c->extract_exponents = ff_ac3_extract_exponents_neon; |
| 39 | 43 | c->float_to_fixed24 = ff_float_to_fixed24_neon; |
| 44 | + c->sum_square_butterfly_int32 = ff_ac3_sum_square_butterfly_int32_neon; | |
| 40 | 45 | } |
| @@ -64,3 +64,26 @@ function ff_float_to_fixed24_neon, export=1 · 63 unmodified lines | ||
| 64 | 64 | b.ne 1b |
| 65 | 65 | ret |
| 66 | 66 | endfunc |
| 67 | + | |
| 68 | +function ff_ac3_sum_square_butterfly_int32_neon, export=1 | |
| 69 | + movi v0.2d, #0 | |
| 70 | + movi v1.2d, #0 | |
| 71 | + movi v2.2d, #0 | |
| 72 | + movi v3.2d, #0 | |
| 73 | +1: ld1 {v4.2s}, [x1], #8 | |
| 74 | + ld1 {v5.2s}, [x2], #8 | |
| 75 | + add v6.2s, v4.2s, v5.2s | |
| 76 | + sub v7.2s, v4.2s, v5.2s | |
| 77 | + smlal v0.2d, v4.2s, v4.2s | |
| 78 | + smlal v1.2d, v5.2s, v5.2s | |
| 79 | + smlal v2.2d, v6.2s, v6.2s | |
| 80 | + smlal v3.2d, v7.2s, v7.2s | |
| 81 | + subs w3, w3, #2 | |
| 82 | + b.gt 1b | |
| 83 | + addp d0, v0.2d | |
| 84 | + addp d1, v1.2d | |
| 85 | + addp d2, v2.2d | |
| 86 | + addp d3, v3.2d | |
| 87 | + st1 {v0.1d-v3.1d}, [x0] | |
| 88 | + ret | |
| 89 | +endfunc | |