colconv 0.1.0

SIMD-dispatched color-conversion kernels covering the FFmpeg AVPixelFormat space, with a Sink-based API so consumers pick which derived outputs (RGB / Luma / HSV / custom) they want without paying for the ones they don't.
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

//! Runtime SIMD dispatchers for planar GBR sources (Tier 10).
//!
//! Three kernels:
//! - [`gbr_to_rgb_row`] — interleave G/B/R → packed `R, G, B`.
//! - [`gbra_to_rgba_row`] — interleave G/B/R/A → packed `R, G, B, A`
//!   (real source α).
//! - [`gbr_to_rgba_opaque_row`] — interleave G/B/R → packed
//!   `R, G, B, 0xFF` (`Gbrp` standalone with_rgba path).
//!
//! Each function follows the same `cfg_select!` pattern used across
//! `dispatch::*`: platform arm at compile time, best available backend
//! at runtime via the `*_available()` helpers.

#[cfg(any(
  target_arch = "aarch64",
  target_arch = "x86_64",
  target_arch = "wasm32"
))]
use crate::row::arch;
#[cfg(target_arch = "aarch64")]
use crate::row::neon_available;
#[cfg(target_arch = "wasm32")]
use crate::row::simd128_available;
#[cfg(target_arch = "x86_64")]
use crate::row::{avx2_available, avx512_available, sse41_available};
use crate::row::{rgb_row_bytes, rgba_row_bytes, scalar};

/// Interleaves three planar G/B/R rows into packed `R, G, B` bytes.
///
/// `use_simd = false` forces the scalar reference path.
#[cfg_attr(not(tarpaulin), inline(always))]
pub fn gbr_to_rgb_row(
  g: &[u8],
  b: &[u8],
  r: &[u8],
  rgb_out: &mut [u8],
  width: usize,
  use_simd: bool,
) {
  let rgb_min = rgb_row_bytes(width);
  assert!(g.len() >= width, "g row too short");
  assert!(b.len() >= width, "b row too short");
  assert!(r.len() >= width, "r row too short");
  assert!(rgb_out.len() >= rgb_min, "rgb_out row too short");

  if use_simd {
    cfg_select! {
      target_arch = "aarch64" => {
        if neon_available() {
          // SAFETY: NEON verified available.
          unsafe { arch::neon::gbr_to_rgb_row(g, b, r, rgb_out, width); }
          return;
        }
      },
      target_arch = "x86_64" => {
        if avx512_available() {
          // SAFETY: AVX-512BW verified available.
          unsafe { arch::x86_avx512::gbr_to_rgb_row(g, b, r, rgb_out, width); }
          return;
        }
        if avx2_available() {
          // SAFETY: AVX2 verified available.
          unsafe { arch::x86_avx2::gbr_to_rgb_row(g, b, r, rgb_out, width); }
          return;
        }
        if sse41_available() {
          // SAFETY: SSE4.1 verified available.
          unsafe { arch::x86_sse41::gbr_to_rgb_row(g, b, r, rgb_out, width); }
          return;
        }
      },
      target_arch = "wasm32" => {
        if simd128_available() {
          // SAFETY: simd128 compile-time enabled.
          unsafe { arch::wasm_simd128::gbr_to_rgb_row(g, b, r, rgb_out, width); }
          return;
        }
      },
      _ => {}
    }
  }

  scalar::gbr_to_rgb_row(g, b, r, rgb_out, width);
}

/// Interleaves four planar G/B/R/A rows into packed `R, G, B, A` bytes
/// (real source α).
///
/// `use_simd = false` forces the scalar reference path.
#[cfg_attr(not(tarpaulin), inline(always))]
#[allow(clippy::too_many_arguments)]
pub fn gbra_to_rgba_row(
  g: &[u8],
  b: &[u8],
  r: &[u8],
  a: &[u8],
  rgba_out: &mut [u8],
  width: usize,
  use_simd: bool,
) {
  let rgba_min = rgba_row_bytes(width);
  assert!(g.len() >= width, "g row too short");
  assert!(b.len() >= width, "b row too short");
  assert!(r.len() >= width, "r row too short");
  assert!(a.len() >= width, "a row too short");
  assert!(rgba_out.len() >= rgba_min, "rgba_out row too short");

  if use_simd {
    cfg_select! {
      target_arch = "aarch64" => {
        if neon_available() {
          // SAFETY: NEON verified available.
          unsafe { arch::neon::gbra_to_rgba_row(g, b, r, a, rgba_out, width); }
          return;
        }
      },
      target_arch = "x86_64" => {
        if avx512_available() {
          // SAFETY: AVX-512BW verified available.
          unsafe { arch::x86_avx512::gbra_to_rgba_row(g, b, r, a, rgba_out, width); }
          return;
        }
        if avx2_available() {
          // SAFETY: AVX2 verified available.
          unsafe { arch::x86_avx2::gbra_to_rgba_row(g, b, r, a, rgba_out, width); }
          return;
        }
        if sse41_available() {
          // SAFETY: SSE4.1 verified available.
          unsafe { arch::x86_sse41::gbra_to_rgba_row(g, b, r, a, rgba_out, width); }
          return;
        }
      },
      target_arch = "wasm32" => {
        if simd128_available() {
          // SAFETY: simd128 compile-time enabled.
          unsafe { arch::wasm_simd128::gbra_to_rgba_row(g, b, r, a, rgba_out, width); }
          return;
        }
      },
      _ => {}
    }
  }

  scalar::gbra_to_rgba_row(g, b, r, a, rgba_out, width);
}

/// Interleaves three planar G/B/R rows into packed `R, G, B, A` bytes
/// with constant α = `0xFF` (used by `Gbrp` for the standalone
/// `with_rgba` path).
///
/// `use_simd = false` forces the scalar reference path.
#[cfg_attr(not(tarpaulin), inline(always))]
pub fn gbr_to_rgba_opaque_row(
  g: &[u8],
  b: &[u8],
  r: &[u8],
  rgba_out: &mut [u8],
  width: usize,
  use_simd: bool,
) {
  let rgba_min = rgba_row_bytes(width);
  assert!(g.len() >= width, "g row too short");
  assert!(b.len() >= width, "b row too short");
  assert!(r.len() >= width, "r row too short");
  assert!(rgba_out.len() >= rgba_min, "rgba_out row too short");

  if use_simd {
    cfg_select! {
      target_arch = "aarch64" => {
        if neon_available() {
          // SAFETY: NEON verified available.
          unsafe { arch::neon::gbr_to_rgba_opaque_row(g, b, r, rgba_out, width); }
          return;
        }
      },
      target_arch = "x86_64" => {
        if avx512_available() {
          // SAFETY: AVX-512BW verified available.
          unsafe { arch::x86_avx512::gbr_to_rgba_opaque_row(g, b, r, rgba_out, width); }
          return;
        }
        if avx2_available() {
          // SAFETY: AVX2 verified available.
          unsafe { arch::x86_avx2::gbr_to_rgba_opaque_row(g, b, r, rgba_out, width); }
          return;
        }
        if sse41_available() {
          // SAFETY: SSE4.1 verified available.
          unsafe { arch::x86_sse41::gbr_to_rgba_opaque_row(g, b, r, rgba_out, width); }
          return;
        }
      },
      target_arch = "wasm32" => {
        if simd128_available() {
          // SAFETY: simd128 compile-time enabled.
          unsafe { arch::wasm_simd128::gbr_to_rgba_opaque_row(g, b, r, rgba_out, width); }
          return;
        }
      },
      _ => {}
    }
  }

  scalar::gbr_to_rgba_opaque_row(g, b, r, rgba_out, width);
}