resampler 0.5.1

A small audio resampling library
Documentation 
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

use core::arch::x86_64::*;

use super::SQRT3_2;
use crate::fft::{
    Complex32,
    butterflies::ops::{complex_mul_avx, complex_mul_sqrt3_i_avx},
};

/// Performs a single radix-3 Stockham butterfly stage (out-of-place, AVX+FMA) for stride=1 (first stage).
#[target_feature(enable = "avx,fma")]
pub(super) unsafe fn butterfly_radix3_stride1_avx_fma(
    src: &[Complex32],
    dst: &mut [Complex32],
    stage_twiddles: &[Complex32],
) {
    let samples = src.len();
    let third_samples = samples / 3;
    let simd_iters = (third_samples >> 2) << 2;

    unsafe {
        let sqrt3_2_vec = _mm256_set_ps(
            -SQRT3_2, SQRT3_2, -SQRT3_2, SQRT3_2, -SQRT3_2, SQRT3_2, -SQRT3_2, SQRT3_2,
        );

        let half_vec = _mm256_set1_ps(0.5);

        for i in (0..simd_iters).step_by(4) {
            let z0_ptr = src.as_ptr().add(i) as *const f32;
            let z0 = _mm256_loadu_ps(z0_ptr);

            let z1_ptr = src.as_ptr().add(i + third_samples) as *const f32;
            let z1 = _mm256_loadu_ps(z1_ptr);

            let z2_ptr = src.as_ptr().add(i + third_samples * 2) as *const f32;
            let z2 = _mm256_loadu_ps(z2_ptr);

            // Identity twiddles: t1 = (1+0i) * z1 = z1, t2 = (1+0i) * z2 = z2 (skip twiddle load and multiply)
            let t1 = z1;
            let t2 = z2;

            // sum_t = t1 + t2, diff_t = t1 - t2
            let sum_t = _mm256_add_ps(t1, t2);
            let diff_t = _mm256_sub_ps(t1, t2);

            // out0 = z0 + sum_t
            let out0 = _mm256_add_ps(z0, sum_t);

            // re_part = z0 - 0.5 * sum_t
            let re_im_part = _mm256_fnmadd_ps(sum_t, half_vec, z0);

            // sqrt3 multiplication
            let sqrt3_diff = complex_mul_sqrt3_i_avx(diff_t, sqrt3_2_vec);

            let out1 = _mm256_add_ps(re_im_part, sqrt3_diff);
            let out2 = _mm256_sub_ps(re_im_part, sqrt3_diff);

            // Interleave for sequential stores: [out0[0], out1[0], out2[0], out0[1], ...]
            let out0_pd = _mm256_castps_pd(out0);
            let out1_pd = _mm256_castps_pd(out1);
            let out2_pd = _mm256_castps_pd(out2);

            // Extract 128-bit lanes (each containing 2 complex numbers).
            let out0_lo = _mm256_castpd256_pd128(out0_pd);
            let out0_hi = _mm256_extractf128_pd(out0_pd, 1);
            let out1_lo = _mm256_castpd256_pd128(out1_pd);
            let out1_hi = _mm256_extractf128_pd(out1_pd, 1);
            let out2_lo = _mm256_castpd256_pd128(out2_pd);
            let out2_hi = _mm256_extractf128_pd(out2_pd, 1);

            // Build result0: [out0[0], out1[0], out2[0], out0[1]]
            let temp0_lo = _mm_unpacklo_pd(out0_lo, out1_lo); // [out0[0], out1[0]]
            let temp0_hi = _mm_shuffle_pd::<2>(out2_lo, out0_lo); // [out2[0], out0[1]]
            let result0 = _mm256_castpd_ps(_mm256_set_m128d(temp0_hi, temp0_lo));

            // Build result1: [out1[1], out2[1], out0[2], out1[2]]
            let temp1_lo = _mm_unpackhi_pd(out1_lo, out2_lo); // [out1[1], out2[1]]
            let temp1_hi = _mm_unpacklo_pd(out0_hi, out1_hi); // [out0[2], out1[2]]
            let result1 = _mm256_castpd_ps(_mm256_set_m128d(temp1_hi, temp1_lo));

            // Build result2: [out2[2], out0[3], out1[3], out2[3]]
            let temp2_lo = _mm_shuffle_pd::<2>(out2_hi, out0_hi); // [out2[2], out0[3]]
            let temp2_hi = _mm_unpackhi_pd(out1_hi, out2_hi); // [out1[3], out2[3]]
            let result2 = _mm256_castpd_ps(_mm256_set_m128d(temp2_hi, temp2_lo));

            // Sequential stores.
            let j = 3 * i;
            let dst_ptr = dst.as_mut_ptr().add(j) as *mut f32;
            _mm256_storeu_ps(dst_ptr, result0);
            _mm256_storeu_ps(dst_ptr.add(8), result1);
            _mm256_storeu_ps(dst_ptr.add(16), result2);
        }
    }

    super::butterfly_radix3_scalar::<4>(src, dst, stage_twiddles, 1, simd_iters);
}

/// Performs a single radix-3 Stockham butterfly stage (out-of-place, AVX+FMA) for generic p.
#[target_feature(enable = "avx,fma")]
pub(super) unsafe fn butterfly_radix3_generic_avx_fma(
    src: &[Complex32],
    dst: &mut [Complex32],
    stage_twiddles: &[Complex32],
    stride: usize,
) {
    // We convince the compiler here that stride can't be 0 to optimize better.
    if stride == 0 {
        return;
    }

    let samples = src.len();
    let third_samples = samples / 3;
    let simd_iters = (third_samples >> 2) << 2;

    unsafe {
        let sqrt3_2_vec = _mm256_set_ps(
            -SQRT3_2, SQRT3_2, -SQRT3_2, SQRT3_2, -SQRT3_2, SQRT3_2, -SQRT3_2, SQRT3_2,
        );

        let half_vec = _mm256_set1_ps(0.5);

        for i in (0..simd_iters).step_by(4) {
            let k = i % stride;
            let k0 = k;
            let k1 = k + 1 - ((k + 1 >= stride) as usize) * stride;
            let k2 = k + 2 - ((k + 2 >= stride) as usize) * stride;
            let k3 = k + 3 - ((k + 3 >= stride) as usize) * stride;

            // Load z0
            let z0_ptr = src.as_ptr().add(i) as *const f32;
            let z0 = _mm256_loadu_ps(z0_ptr);

            // Load z1, z2
            let z1_ptr = src.as_ptr().add(i + third_samples) as *const f32;
            let z1 = _mm256_loadu_ps(z1_ptr);

            let z2_ptr = src.as_ptr().add(i + third_samples * 2) as *const f32;
            let z2 = _mm256_loadu_ps(z2_ptr);

            // Load prepackaged twiddles directly (no shuffle needed).
            let tw_ptr = stage_twiddles.as_ptr().add(i * 2) as *const f32;
            let w1 = _mm256_loadu_ps(tw_ptr); // w1[i..i+4]
            let w2 = _mm256_loadu_ps(tw_ptr.add(8)); // w2[i..i+4]

            // Complex multiply: t1 = w1 * z1, t2 = w2 * z2
            let t1 = complex_mul_avx(w1, z1);
            let t2 = complex_mul_avx(w2, z2);

            // sum_t = t1 + t2, diff_t = t1 - t2
            let sum_t = _mm256_add_ps(t1, t2);
            let diff_t = _mm256_sub_ps(t1, t2);

            // out0 = z0 + sum_t
            let out0 = _mm256_add_ps(z0, sum_t);

            // re_part = z0 - 0.5 * sum_t
            let re_im_part = _mm256_fnmadd_ps(sum_t, half_vec, z0);

            // sqrt3 multiplication
            let sqrt3_diff = complex_mul_sqrt3_i_avx(diff_t, sqrt3_2_vec);

            let out1 = _mm256_add_ps(re_im_part, sqrt3_diff);
            let out2 = _mm256_sub_ps(re_im_part, sqrt3_diff);

            // Calculate output indices.
            let j0 = 3 * i - 2 * k0;
            let j1 = 3 * (i + 1) - 2 * k1;
            let j2 = 3 * (i + 2) - 2 * k2;
            let j3 = 3 * (i + 3) - 2 * k3;

            // Direct SIMD stores (no stack spills)
            let out0_pd = _mm256_castps_pd(out0);
            let out1_pd = _mm256_castps_pd(out1);
            let out2_pd = _mm256_castps_pd(out2);

            let out0_lo = _mm256_castpd256_pd128(out0_pd);
            let out0_hi = _mm256_extractf128_pd(out0_pd, 1);
            let out1_lo = _mm256_castpd256_pd128(out1_pd);
            let out1_hi = _mm256_extractf128_pd(out1_pd, 1);
            let out2_lo = _mm256_castpd256_pd128(out2_pd);
            let out2_hi = _mm256_extractf128_pd(out2_pd, 1);

            let dst_ptr = dst.as_mut_ptr() as *mut f64;

            // Store iteration 0
            _mm_storel_pd(dst_ptr.add(j0), out0_lo);
            _mm_storel_pd(dst_ptr.add(j0 + stride), out1_lo);
            _mm_storel_pd(dst_ptr.add(j0 + stride * 2), out2_lo);

            // Store iteration 1
            _mm_storeh_pd(dst_ptr.add(j1), out0_lo);
            _mm_storeh_pd(dst_ptr.add(j1 + stride), out1_lo);
            _mm_storeh_pd(dst_ptr.add(j1 + stride * 2), out2_lo);

            // Store iteration 2
            _mm_storel_pd(dst_ptr.add(j2), out0_hi);
            _mm_storel_pd(dst_ptr.add(j2 + stride), out1_hi);
            _mm_storel_pd(dst_ptr.add(j2 + stride * 2), out2_hi);

            // Store iteration 3
            _mm_storeh_pd(dst_ptr.add(j3), out0_hi);
            _mm_storeh_pd(dst_ptr.add(j3 + stride), out1_hi);
            _mm_storeh_pd(dst_ptr.add(j3 + stride * 2), out2_hi);
        }
    }

    super::butterfly_radix3_scalar::<4>(src, dst, stage_twiddles, stride, simd_iters);
}