flaw 0.6.1

Embedded signal filtering, no-std and no-alloc compatible.
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

#![doc = include_str!("../README.md")]
#![cfg_attr(not(feature = "std"), no_std)]
pub mod fir;
pub mod fractional_delay;
pub mod iir;
pub mod median;
pub mod sos;

use core::ops::{Index, IndexMut};
use crunchy::unroll;
pub use fir::SisoFirFilter;
pub use fractional_delay::polynomial_fractional_delay;
pub use iir::SisoIirFilter;
pub use median::MedianFilter;
pub use sos::SisoSosFilter;

pub mod generated;
pub use generated::butter::butter1::butter1;
pub use generated::butter::butter2::butter2;
pub use generated::butter::butter3::butter3;
pub use generated::butter::butter4::butter4;
pub use generated::butter::butter5::butter5;
pub use generated::butter::butter6::butter6;

use num_traits::{MulAdd, Num};

/// A simple array with large memory alignment because it will be accessed
/// often in a loop, with methods specialized for filter evaluation.
#[derive(Clone, Copy, Debug)]
#[repr(align(8))]
pub struct AlignedArray<T, const N: usize>([T; N]);

impl<T: Copy + Num, const N: usize> Default for AlignedArray<T, N> {
    fn default() -> Self {
        Self([T::zero(); N])
    }
}

impl<T: Copy + Num + MulAdd<Output = T>, const N: usize> AlignedArray<T, N> {
    /// Multiply-and-sum between this array and a target ring buffer.
    ///
    /// A starting value can be provided for the summation,
    /// which can be helpful for fine-tuning floating-point error.
    #[inline]
    pub fn dot(&self, buf: &Ring<T, N>, start: T) -> T {
        const { assert!(N < 129, "N > 128 not supported") }

        // Multiply-and-sum loops could be turned into chained mul-add,
        // but microcontrollers mostly don't have mul-add instructions
        // as of the year 2025, so using mul_add here would cause severe
        // performance regression.
        let other = buf.buf();
        let mut acc = start;

        #[cfg(not(feature = "fma"))]
        unroll! {
            for i < 128 in 0..N {
                acc = acc + self.0[i] * other[i];
            }
        }

        #[cfg(feature = "fma")]
        unroll! {
            for i < 128 in 0..N {
                acc = self.0[i].mul_add(other[i], acc);
            }
        }

        acc
    }
}

impl<T, const N: usize> Index<usize> for AlignedArray<T, N> {
    type Output = T;
    #[inline]
    fn index(&self, idx: usize) -> &Self::Output {
        &self.0[idx]
    }
}

impl<T, const N: usize> IndexMut<usize> for AlignedArray<T, N> {
    #[inline]
    fn index_mut(&mut self, idx: usize) -> &mut Self::Output {
        &mut self.0[idx]
    }
}

/// Ring buffer.
/// Most recent sample is stored last.
#[derive(Clone, Copy, Default, Debug)]
#[repr(transparent)]
pub struct Ring<T: Num + Copy, const N: usize> {
    buf: AlignedArray<T, N>,
}

impl<T: Num + Copy, const N: usize> Ring<T, N> {
    /// Initialize with buffer populated with constant value
    pub fn new(value: T) -> Self {
        Self {
            buf: AlignedArray([value; N]),
        }
    }

    /// Replace the oldest value in the buffer with a new value.
    /// Despite the unnecessary copies compared to a true ring buffer, this is
    /// faster overall in filtering applications where one dot product is performed
    /// per push, and the cost of splitting the dot product into two segments is large
    /// compared to the savings from avoiding copies.
    pub fn push(&mut self, value: T) {
        unroll! {
            for i < 128 in 0..(N-1) {
                self.buf.0[i] = self.buf.0[i + 1];
            }
        }
        self.buf.0[N - 1] = value;
    }

    /// The whole internal buffer, with no indication of the current index
    fn buf(&self) -> &[T; N] {
        &self.buf.0
    }
}

/// `std` is required for tests, but is not a default feature.
/// To allow the library to compile with default features,
/// tests that require `std` are feature-gated.
/// This test makes sure we do not skip the real tests.
#[cfg(test)]
#[cfg(not(feature = "std"))]
mod test {
    #[test]
    fn require_std_for_tests() {
        panic!("`std` feature is required for tests")
    }
}

#[cfg(feature = "std")]
#[cfg(test)]
mod test {
    /// Test initialization to a given input value
    #[test]
    fn test_initialize() {
        let e = core::f64::consts::E as f32;
        let mut f = super::butter2(0.2).unwrap();
        f.set_steady_state(e);
        assert!((e - f.update(e)).abs() / e < 1e-6);
    }
}