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
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305

use core::arch::aarch64::*;

use crate::fft::{
    Complex32,
    butterflies::ops::{
        complex_mul, complex_mul_i, load_neg_imag_mask, load_scale_neon, v8x_neon, w8x_neon,
    },
};

/// Performs a single radix-8 Stockham butterfly stage for stride=1 (out-of-place, NEON).
///
/// This is a specialized version for the stride=1 case (first stage) that uses sequential
/// stores. For stride=1, output indices are sequential: j=8*i.
#[target_feature(enable = "neon")]
pub(super) unsafe fn butterfly_radix8_stride1_neon(
    src: &[Complex32],
    dst: &mut [Complex32],
    stage_twiddles: &[Complex32],
) {
    let samples = src.len();
    let eighth_samples = samples >> 3;
    let simd_iters = (eighth_samples >> 1) << 1;

    unsafe {
        let neg_imag_mask = load_neg_imag_mask();
        let scale = load_scale_neon();

        for i in (0..simd_iters).step_by(2) {
            // Load 2 complex numbers from each eighth.
            let z0_ptr = src.as_ptr().add(i) as *const f32;
            let z0 = vld1q_f32(z0_ptr);

            let z1_ptr = src.as_ptr().add(i + eighth_samples) as *const f32;
            let z1 = vld1q_f32(z1_ptr);

            let z2_ptr = src.as_ptr().add(i + eighth_samples * 2) as *const f32;
            let z2 = vld1q_f32(z2_ptr);

            let z3_ptr = src.as_ptr().add(i + eighth_samples * 3) as *const f32;
            let z3 = vld1q_f32(z3_ptr);

            let z4_ptr = src.as_ptr().add(i + eighth_samples * 4) as *const f32;
            let z4 = vld1q_f32(z4_ptr);

            let z5_ptr = src.as_ptr().add(i + eighth_samples * 5) as *const f32;
            let z5 = vld1q_f32(z5_ptr);

            let z6_ptr = src.as_ptr().add(i + eighth_samples * 6) as *const f32;
            let z6 = vld1q_f32(z6_ptr);

            let z7_ptr = src.as_ptr().add(i + eighth_samples * 7) as *const f32;
            let z7 = vld1q_f32(z7_ptr);

            // Stride=1 optimization: all twiddles are identity (1+0i), skip multiplication.
            let t1 = z1;
            let t2 = z2;
            let t3 = z3;
            let t4 = z4;
            let t5 = z5;
            let t6 = z6;
            let t7 = z7;

            // Compute radix-4 DFT on even indices (z0, t2, t4, t6).
            let even_a0 = vaddq_f32(z0, t4);
            let even_a1 = vsubq_f32(z0, t4);
            let even_a2 = vaddq_f32(t2, t6);
            let t2_sub_t6 = vsubq_f32(t2, t6);
            let even_a3 = complex_mul_i(t2_sub_t6, neg_imag_mask);

            let x_even_0 = vaddq_f32(even_a0, even_a2);
            let x_even_2 = vsubq_f32(even_a0, even_a2);
            let x_even_1 = vaddq_f32(even_a1, even_a3);
            let x_even_3 = vsubq_f32(even_a1, even_a3);

            // Compute radix-4 DFT on odd indices (t1, t3, t5, t7).
            let odd_a0 = vaddq_f32(t1, t5);
            let odd_a1 = vsubq_f32(t1, t5);
            let odd_a2 = vaddq_f32(t3, t7);
            let t3_sub_t7 = vsubq_f32(t3, t7);
            let odd_a3 = complex_mul_i(t3_sub_t7, neg_imag_mask);

            let x_odd_0 = vaddq_f32(odd_a0, odd_a2);
            let x_odd_2 = vsubq_f32(odd_a0, odd_a2);
            let x_odd_1 = vaddq_f32(odd_a1, odd_a3);
            let x_odd_3 = vsubq_f32(odd_a1, odd_a3);

            // Combine even and odd parts with W_8 twiddles.
            // out[0] = x_even[0] + x_odd[0]
            // out[4] = x_even[0] - x_odd[0]
            let out0 = vaddq_f32(x_even_0, x_odd_0);
            let out4 = vsubq_f32(x_even_0, x_odd_0);

            // out[1] = x_even[1] + W_8^1 * x_odd[1]
            // out[5] = x_even[1] - W_8^1 * x_odd[1]
            let w8_1_odd_1 = w8x_neon(x_odd_1, neg_imag_mask, scale);
            let out1 = vaddq_f32(x_even_1, w8_1_odd_1);
            let out5 = vsubq_f32(x_even_1, w8_1_odd_1);

            // out[2] = x_even[2] + W_8^2 * x_odd[2]
            // out[6] = x_even[2] - W_8^2 * x_odd[2]
            // W_8^2 = -i, so multiply by -i: (a+bi)*(-i) = (b, -a)
            let w8_2_odd_2 = complex_mul_i(x_odd_2, neg_imag_mask);
            let out2 = vaddq_f32(x_even_2, w8_2_odd_2);
            let out6 = vsubq_f32(x_even_2, w8_2_odd_2);

            // out[3] = x_even[3] + W_8^3 * x_odd[3]
            // out[7] = x_even[3] - W_8^3 * x_odd[3]
            let w8_3_odd_3 = v8x_neon(x_odd_3, neg_imag_mask, scale);
            let out3 = vaddq_f32(x_even_3, w8_3_odd_3);
            let out7 = vsubq_f32(x_even_3, w8_3_odd_3);

            // Sequential stores for stride=1.
            // We have 8 outputs (out0-out7), each containing 2 complex numbers.
            // Need to store them sequentially: [out0[0], out1[0], out2[0], out3[0], out4[0], out5[0], out6[0], out7[0], out0[1], ...]

            let j = 8 * i;
            let dst_ptr = dst.as_mut_ptr().add(j) as *mut f32;

            // Extract low and high halves of each output.
            let out0_0 = vget_low_f32(out0);
            let out0_1 = vget_high_f32(out0);
            let out1_0 = vget_low_f32(out1);
            let out1_1 = vget_high_f32(out1);
            let out2_0 = vget_low_f32(out2);
            let out2_1 = vget_high_f32(out2);
            let out3_0 = vget_low_f32(out3);
            let out3_1 = vget_high_f32(out3);
            let out4_0 = vget_low_f32(out4);
            let out4_1 = vget_high_f32(out4);
            let out5_0 = vget_low_f32(out5);
            let out5_1 = vget_high_f32(out5);
            let out6_0 = vget_low_f32(out6);
            let out6_1 = vget_high_f32(out6);
            let out7_0 = vget_low_f32(out7);
            let out7_1 = vget_high_f32(out7);

            // Store first iteration outputs.
            vst1_f32(dst_ptr, out0_0);
            vst1_f32(dst_ptr.add(2), out1_0);
            vst1_f32(dst_ptr.add(4), out2_0);
            vst1_f32(dst_ptr.add(6), out3_0);
            vst1_f32(dst_ptr.add(8), out4_0);
            vst1_f32(dst_ptr.add(10), out5_0);
            vst1_f32(dst_ptr.add(12), out6_0);
            vst1_f32(dst_ptr.add(14), out7_0);

            // Store second iteration outputs.
            vst1_f32(dst_ptr.add(16), out0_1);
            vst1_f32(dst_ptr.add(18), out1_1);
            vst1_f32(dst_ptr.add(20), out2_1);
            vst1_f32(dst_ptr.add(22), out3_1);
            vst1_f32(dst_ptr.add(24), out4_1);
            vst1_f32(dst_ptr.add(26), out5_1);
            vst1_f32(dst_ptr.add(28), out6_1);
            vst1_f32(dst_ptr.add(30), out7_1);
        }
    }

    super::butterfly_radix8_scalar::<2>(src, dst, stage_twiddles, 1, simd_iters);
}

/// Performs a single radix-8 Stockham butterfly stage for p>1 (out-of-place, NEON).
///
/// Generic version for p>1 cases. Uses direct SIMD stores with scattered outputs.
#[target_feature(enable = "neon")]
pub(super) unsafe fn butterfly_radix8_generic_neon(
    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 eighth_samples = samples >> 3;
    let simd_iters = (eighth_samples >> 1) << 1;

    unsafe {
        let neg_imag_mask = load_neg_imag_mask();
        let scale = load_scale_neon();

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

            // Load 2 complex numbers from each eighth.
            let z0_ptr = src.as_ptr().add(i) as *const f32;
            let z0 = vld1q_f32(z0_ptr);

            let z1_ptr = src.as_ptr().add(i + eighth_samples) as *const f32;
            let z1 = vld1q_f32(z1_ptr);

            let z2_ptr = src.as_ptr().add(i + eighth_samples * 2) as *const f32;
            let z2 = vld1q_f32(z2_ptr);

            let z3_ptr = src.as_ptr().add(i + eighth_samples * 3) as *const f32;
            let z3 = vld1q_f32(z3_ptr);

            let z4_ptr = src.as_ptr().add(i + eighth_samples * 4) as *const f32;
            let z4 = vld1q_f32(z4_ptr);

            let z5_ptr = src.as_ptr().add(i + eighth_samples * 5) as *const f32;
            let z5 = vld1q_f32(z5_ptr);

            let z6_ptr = src.as_ptr().add(i + eighth_samples * 6) as *const f32;
            let z6 = vld1q_f32(z6_ptr);

            let z7_ptr = src.as_ptr().add(i + eighth_samples * 7) as *const f32;
            let z7 = vld1q_f32(z7_ptr);

            // Load prepackaged twiddles in packed format.
            let tw_ptr = stage_twiddles.as_ptr().add(i * 7) as *const f32;
            let w1 = vld1q_f32(tw_ptr);
            let w2 = vld1q_f32(tw_ptr.add(4));
            let w3 = vld1q_f32(tw_ptr.add(8));
            let w4 = vld1q_f32(tw_ptr.add(12));
            let w5 = vld1q_f32(tw_ptr.add(16));
            let w6 = vld1q_f32(tw_ptr.add(20));
            let w7 = vld1q_f32(tw_ptr.add(24));

            // Apply twiddle factors.
            let t1 = complex_mul(w1, z1);
            let t2 = complex_mul(w2, z2);
            let t3 = complex_mul(w3, z3);
            let t4 = complex_mul(w4, z4);
            let t5 = complex_mul(w5, z5);
            let t6 = complex_mul(w6, z6);
            let t7 = complex_mul(w7, z7);

            // Compute radix-4 DFT on even indices.
            let even_a0 = vaddq_f32(z0, t4);
            let even_a1 = vsubq_f32(z0, t4);
            let even_a2 = vaddq_f32(t2, t6);
            let t2_sub_t6 = vsubq_f32(t2, t6);
            let even_a3 = complex_mul_i(t2_sub_t6, neg_imag_mask);

            let x_even_0 = vaddq_f32(even_a0, even_a2);
            let x_even_2 = vsubq_f32(even_a0, even_a2);
            let x_even_1 = vaddq_f32(even_a1, even_a3);
            let x_even_3 = vsubq_f32(even_a1, even_a3);

            // Compute radix-4 DFT on odd indices.
            let odd_a0 = vaddq_f32(t1, t5);
            let odd_a1 = vsubq_f32(t1, t5);
            let odd_a2 = vaddq_f32(t3, t7);
            let t3_sub_t7 = vsubq_f32(t3, t7);
            let odd_a3 = complex_mul_i(t3_sub_t7, neg_imag_mask);

            let x_odd_0 = vaddq_f32(odd_a0, odd_a2);
            let x_odd_2 = vsubq_f32(odd_a0, odd_a2);
            let x_odd_1 = vaddq_f32(odd_a1, odd_a3);
            let x_odd_3 = vsubq_f32(odd_a1, odd_a3);

            // Combine even and odd parts.
            let out0 = vaddq_f32(x_even_0, x_odd_0);
            let out4 = vsubq_f32(x_even_0, x_odd_0);

            let w8_1_odd_1 = w8x_neon(x_odd_1, neg_imag_mask, scale);
            let out1 = vaddq_f32(x_even_1, w8_1_odd_1);
            let out5 = vsubq_f32(x_even_1, w8_1_odd_1);

            let w8_2_odd_2 = complex_mul_i(x_odd_2, neg_imag_mask);
            let out2 = vaddq_f32(x_even_2, w8_2_odd_2);
            let out6 = vsubq_f32(x_even_2, w8_2_odd_2);

            let w8_3_odd_3 = v8x_neon(x_odd_3, neg_imag_mask, scale);
            let out3 = vaddq_f32(x_even_3, w8_3_odd_3);
            let out7 = vsubq_f32(x_even_3, w8_3_odd_3);

            // Calculate output indices with wraparound.
            let j0 = 8 * i - 7 * k0;
            let j1 = 8 * (i + 1) - 7 * k1;

            // Direct scattered stores using 64-bit operations.
            let dst_ptr = dst.as_mut_ptr().add(j0) as *mut f32;

            // Store first iteration (k0) outputs.
            vst1_f32(dst_ptr, vget_low_f32(out0));
            vst1_f32(dst_ptr.add(stride * 2), vget_low_f32(out1));
            vst1_f32(dst_ptr.add(stride * 4), vget_low_f32(out2));
            vst1_f32(dst_ptr.add(stride * 6), vget_low_f32(out3));
            vst1_f32(dst_ptr.add(stride * 8), vget_low_f32(out4));
            vst1_f32(dst_ptr.add(stride * 10), vget_low_f32(out5));
            vst1_f32(dst_ptr.add(stride * 12), vget_low_f32(out6));
            vst1_f32(dst_ptr.add(stride * 14), vget_low_f32(out7));

            // Store second iteration (k1) outputs.
            let dst_ptr1 = dst.as_mut_ptr().add(j1) as *mut f32;
            vst1_f32(dst_ptr1, vget_high_f32(out0));
            vst1_f32(dst_ptr1.add(stride * 2), vget_high_f32(out1));
            vst1_f32(dst_ptr1.add(stride * 4), vget_high_f32(out2));
            vst1_f32(dst_ptr1.add(stride * 6), vget_high_f32(out3));
            vst1_f32(dst_ptr1.add(stride * 8), vget_high_f32(out4));
            vst1_f32(dst_ptr1.add(stride * 10), vget_high_f32(out5));
            vst1_f32(dst_ptr1.add(stride * 12), vget_high_f32(out6));
            vst1_f32(dst_ptr1.add(stride * 14), vget_high_f32(out7));
        }
    }

    super::butterfly_radix8_scalar::<2>(src, dst, stage_twiddles, stride, simd_iters);
}