zenpng 0.1.4

PNG encoding and decoding with zencodec trait integration
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

//! Up filter: `row[i] = row[i].wrapping_add(prev[i])`
//!
//! Trivially parallel — no inter-pixel dependency. Process 32 bytes (AVX2)
//! or 16 bytes (SSE2) at a time.

use archmage::prelude::*;
#[cfg(target_arch = "aarch64")]
use safe_unaligned_simd::aarch64::{vld1q_u8, vst1q_u8};
#[cfg(target_arch = "wasm32")]
use safe_unaligned_simd::wasm32::{v128_load, v128_store};
#[cfg(target_arch = "x86_64")]
use safe_unaligned_simd::x86_64::{
    _mm_loadu_si128, _mm_storeu_si128, _mm256_loadu_si256, _mm256_storeu_si256,
};

pub(crate) fn unfilter_up(row: &mut [u8], prev: &[u8]) {
    incant!(unfilter_up_impl(row, prev), [v3, v1, neon, wasm128, scalar])
}

#[cfg(target_arch = "x86_64")]
#[arcane]
fn unfilter_up_impl_v3(_token: Desktop64, row: &mut [u8], prev: &[u8]) {
    let len = row.len().min(prev.len());
    let mut i = 0;

    while i + 32 <= len {
        let vr = _mm256_loadu_si256(<&[u8; 32]>::try_from(&row[i..i + 32]).unwrap());
        let vp = _mm256_loadu_si256(<&[u8; 32]>::try_from(&prev[i..i + 32]).unwrap());
        let sum = _mm256_add_epi8(vr, vp);
        _mm256_storeu_si256(<&mut [u8; 32]>::try_from(&mut row[i..i + 32]).unwrap(), sum);
        i += 32;
    }

    // SSE2 pass for 16–31 remaining bytes (AVX2 implies SSE2).
    while i + 16 <= len {
        let vr = _mm_loadu_si128(<&[u8; 16]>::try_from(&row[i..i + 16]).unwrap());
        let vp = _mm_loadu_si128(<&[u8; 16]>::try_from(&prev[i..i + 16]).unwrap());
        let sum = _mm_add_epi8(vr, vp);
        _mm_storeu_si128(<&mut [u8; 16]>::try_from(&mut row[i..i + 16]).unwrap(), sum);
        i += 16;
    }

    while i < len {
        row[i] = row[i].wrapping_add(prev[i]);
        i += 1;
    }
}

#[cfg(target_arch = "x86_64")]
#[arcane]
fn unfilter_up_impl_v1(_token: X64V1Token, row: &mut [u8], prev: &[u8]) {
    let len = row.len().min(prev.len());
    let mut i = 0;

    while i + 16 <= len {
        let vr = _mm_loadu_si128(<&[u8; 16]>::try_from(&row[i..i + 16]).unwrap());
        let vp = _mm_loadu_si128(<&[u8; 16]>::try_from(&prev[i..i + 16]).unwrap());
        let sum = _mm_add_epi8(vr, vp);
        _mm_storeu_si128(<&mut [u8; 16]>::try_from(&mut row[i..i + 16]).unwrap(), sum);
        i += 16;
    }

    while i < len {
        row[i] = row[i].wrapping_add(prev[i]);
        i += 1;
    }
}

// ── NEON (aarch64) ──────────────────────────────────────────────────

#[cfg(target_arch = "aarch64")]
#[arcane]
fn unfilter_up_impl_neon(_token: NeonToken, row: &mut [u8], prev: &[u8]) {
    let len = row.len().min(prev.len());
    let mut i = 0;

    while i + 16 <= len {
        let vr = vld1q_u8(<&[u8; 16]>::try_from(&row[i..i + 16]).unwrap());
        let vp = vld1q_u8(<&[u8; 16]>::try_from(&prev[i..i + 16]).unwrap());
        let sum = vaddq_u8(vr, vp);
        vst1q_u8(<&mut [u8; 16]>::try_from(&mut row[i..i + 16]).unwrap(), sum);
        i += 16;
    }

    while i < len {
        row[i] = row[i].wrapping_add(prev[i]);
        i += 1;
    }
}

// ── WASM SIMD128 ───────────────────────────────────────────────────

#[cfg(target_arch = "wasm32")]
#[arcane]
fn unfilter_up_impl_wasm128(_token: Wasm128Token, row: &mut [u8], prev: &[u8]) {
    let len = row.len().min(prev.len());
    let mut i = 0;

    while i + 16 <= len {
        let vr = v128_load(<&[u8; 16]>::try_from(&row[i..i + 16]).unwrap());
        let vp = v128_load(<&[u8; 16]>::try_from(&prev[i..i + 16]).unwrap());
        let sum = i8x16_add(vr, vp);
        v128_store(<&mut [u8; 16]>::try_from(&mut row[i..i + 16]).unwrap(), sum);
        i += 16;
    }

    while i < len {
        row[i] = row[i].wrapping_add(prev[i]);
        i += 1;
    }
}

fn unfilter_up_impl_scalar(_token: ScalarToken, row: &mut [u8], prev: &[u8]) {
    let len = row.len().min(prev.len());
    for i in 0..len {
        row[i] = row[i].wrapping_add(prev[i]);
    }
}

#[cfg(test)]
mod tests {
    use archmage::testing::{CompileTimePolicy, for_each_token_permutation};

    fn scalar_up(row: &mut [u8], prev: &[u8]) {
        for i in 0..row.len().min(prev.len()) {
            row[i] = row[i].wrapping_add(prev[i]);
        }
    }

    #[test]
    fn up_all_tiers() {
        let report = for_each_token_permutation(CompileTimePolicy::Warn, |_perm| {
            for &len in &[0, 1, 3, 4, 15, 16, 17, 31, 32, 33, 100, 4096, 65536] {
                let prev: Vec<u8> = (0..len).map(|i| (i * 7 + 13) as u8).collect();
                let filtered: Vec<u8> = (0..len).map(|i| (i * 3 + 5) as u8).collect();

                let mut expected = filtered.clone();
                scalar_up(&mut expected, &prev);

                let mut actual = filtered.clone();
                super::unfilter_up(&mut actual, &prev);

                assert_eq!(actual, expected, "up mismatch at len={len}");
            }
        });
        eprintln!("up filter: {report}");
    }
}