Files
another-boids-in-rust/vendor/moxcms/src/conversions/neon/cube.rs

443 lines
15 KiB
Rust

/*
* // Copyright (c) Radzivon Bartoshyk 3/2025. All rights reserved.
* //
* // Redistribution and use in source and binary forms, with or without modification,
* // are permitted provided that the following conditions are met:
* //
* // 1. Redistributions of source code must retain the above copyright notice, this
* // list of conditions and the following disclaimer.
* //
* // 2. Redistributions in binary form must reproduce the above copyright notice,
* // this list of conditions and the following disclaimer in the documentation
* // and/or other materials provided with the distribution.
* //
* // 3. Neither the name of the copyright holder nor the names of its
* // contributors may be used to endorse or promote products derived from
* // this software without specific prior written permission.
* //
* // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
* // AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* // IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
* // DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
* // FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* // DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
* // SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
* // CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
* // OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
use crate::conversions::neon::interpolator::NeonVector;
use crate::math::{FusedMultiplyAdd, FusedMultiplyNegAdd};
use std::arch::aarch64::*;
use std::ops::{Add, Mul, Sub};
/// 3D CLUT NEON helper
///
/// Represents hexahedron.
pub(crate) struct CubeNeon<'a> {
array: &'a [f32],
x_stride: u32,
y_stride: u32,
grid_size: [u8; 3],
}
struct HexahedronFetch3<'a> {
array: &'a [f32],
x_stride: u32,
y_stride: u32,
}
trait CubeFetch<T> {
fn fetch(&self, x: i32, y: i32, z: i32) -> T;
}
impl CubeFetch<NeonVector> for HexahedronFetch3<'_> {
#[inline(always)]
fn fetch(&self, x: i32, y: i32, z: i32) -> NeonVector {
let start = (x as u32 * self.x_stride + y as u32 * self.y_stride + z as u32) as usize * 3;
unsafe {
let k = self.array.get_unchecked(start..);
let lo = vld1_f32(k.as_ptr());
let hi = vld1_lane_f32::<0>(k.get_unchecked(2..).as_ptr(), vdup_n_f32(0.));
NeonVector {
v: vcombine_f32(lo, hi),
}
}
}
}
impl<'a> CubeNeon<'a> {
pub(crate) fn new(arr: &'a [f32], grid: [u8; 3], components: usize) -> Self {
// This is safety precondition, array size must be not less than full grid size * components.
// Needs to ensure that it is not missed somewhere else
assert_eq!(
grid[0] as usize * grid[1] as usize * grid[2] as usize * components,
arr.len()
);
let y_stride = grid[1] as u32;
let x_stride = y_stride * grid[0] as u32;
CubeNeon {
array: arr,
x_stride,
y_stride,
grid_size: grid,
}
}
#[inline(always)]
fn trilinear<
T: Copy
+ From<f32>
+ Sub<T, Output = T>
+ Mul<T, Output = T>
+ Add<T, Output = T>
+ FusedMultiplyNegAdd<T>
+ FusedMultiplyAdd<T>,
>(
&self,
lin_x: f32,
lin_y: f32,
lin_z: f32,
fetch: impl CubeFetch<T>,
) -> T {
let lin_x = lin_x.max(0.0).min(1.0);
let lin_y = lin_y.max(0.0).min(1.0);
let lin_z = lin_z.max(0.0).min(1.0);
let scale_x = (self.grid_size[0] as i32 - 1) as f32;
let scale_y = (self.grid_size[1] as i32 - 1) as f32;
let scale_z = (self.grid_size[2] as i32 - 1) as f32;
let x = (lin_x * scale_x).floor() as i32;
let y = (lin_y * scale_y).floor() as i32;
let z = (lin_z * scale_z).floor() as i32;
let x_n = (lin_x * scale_x).ceil() as i32;
let y_n = (lin_y * scale_y).ceil() as i32;
let z_n = (lin_z * scale_z).ceil() as i32;
let x_d = T::from(lin_x * scale_x - x as f32);
let y_d = T::from(lin_y * scale_y - y as f32);
let z_d = T::from(lin_z * scale_z - z as f32);
let c000 = fetch.fetch(x, y, z);
let c100 = fetch.fetch(x_n, y, z);
let c010 = fetch.fetch(x, y_n, z);
let c110 = fetch.fetch(x_n, y_n, z);
let c001 = fetch.fetch(x, y, z_n);
let c101 = fetch.fetch(x_n, y, z_n);
let c011 = fetch.fetch(x, y_n, z_n);
let c111 = fetch.fetch(x_n, y_n, z_n);
let c00 = c000.neg_mla(c000, x_d).mla(c100, x_d);
let c10 = c010.neg_mla(c010, x_d).mla(c110, x_d);
let c01 = c001.neg_mla(c001, x_d).mla(c101, x_d);
let c11 = c011.neg_mla(c011, x_d).mla(c111, x_d);
let c0 = c00.neg_mla(c00, y_d).mla(c10, y_d);
let c1 = c01.neg_mla(c01, y_d).mla(c11, y_d);
c0.neg_mla(c0, z_d).mla(c1, z_d)
}
#[cfg(feature = "options")]
#[inline]
fn pyramid<
T: Copy
+ From<f32>
+ Sub<T, Output = T>
+ Mul<T, Output = T>
+ Add<T, Output = T>
+ FusedMultiplyAdd<T>,
>(
&self,
lin_x: f32,
lin_y: f32,
lin_z: f32,
fetch: impl CubeFetch<T>,
) -> T {
let lin_x = lin_x.max(0.0).min(1.0);
let lin_y = lin_y.max(0.0).min(1.0);
let lin_z = lin_z.max(0.0).min(1.0);
let scale_x = (self.grid_size[0] as i32 - 1) as f32;
let scale_y = (self.grid_size[1] as i32 - 1) as f32;
let scale_z = (self.grid_size[2] as i32 - 1) as f32;
let x = (lin_x * scale_x).floor() as i32;
let y = (lin_y * scale_y).floor() as i32;
let z = (lin_z * scale_z).floor() as i32;
let x_n = (lin_x * scale_x).ceil() as i32;
let y_n = (lin_y * scale_y).ceil() as i32;
let z_n = (lin_z * scale_z).ceil() as i32;
let dr = lin_x * scale_x - x as f32;
let dg = lin_y * scale_y - y as f32;
let db = lin_z * scale_z - z as f32;
let c0 = fetch.fetch(x, y, z);
if dr > db && dg > db {
let x0 = fetch.fetch(x_n, y_n, z_n);
let x1 = fetch.fetch(x_n, y_n, z);
let x2 = fetch.fetch(x_n, y, z);
let x3 = fetch.fetch(x, y_n, z);
let c1 = x0 - x1;
let c2 = x2 - c0;
let c3 = x3 - c0;
let c4 = c0 - x3 - x2 + x1;
let s0 = c0.mla(c1, T::from(db));
let s1 = s0.mla(c2, T::from(dr));
let s2 = s1.mla(c3, T::from(dg));
s2.mla(c4, T::from(dr * dg))
} else if db > dr && dg > dr {
let x0 = fetch.fetch(x, y, z_n);
let x1 = fetch.fetch(x_n, y_n, z_n);
let x2 = fetch.fetch(x, y_n, z_n);
let x3 = fetch.fetch(x, y_n, z);
let c1 = x0 - c0;
let c2 = x1 - x2;
let c3 = x3 - c0;
let c4 = c0 - x3 - x0 + x2;
let s0 = c0.mla(c1, T::from(db));
let s1 = s0.mla(c2, T::from(dr));
let s2 = s1.mla(c3, T::from(dg));
s2.mla(c4, T::from(dg * db))
} else {
let x0 = fetch.fetch(x, y, z_n);
let x1 = fetch.fetch(x_n, y, z);
let x2 = fetch.fetch(x_n, y, z_n);
let x3 = fetch.fetch(x_n, y_n, z_n);
let c1 = x0 - c0;
let c2 = x1 - c0;
let c3 = x3 - x2;
let c4 = c0 - x1 - x0 + x2;
let s0 = c0.mla(c1, T::from(db));
let s1 = s0.mla(c2, T::from(dr));
let s2 = s1.mla(c3, T::from(dg));
s2.mla(c4, T::from(db * dr))
}
}
#[cfg(feature = "options")]
#[inline]
fn tetra<
T: Copy
+ From<f32>
+ Sub<T, Output = T>
+ Mul<T, Output = T>
+ Add<T, Output = T>
+ FusedMultiplyAdd<T>,
>(
&self,
lin_x: f32,
lin_y: f32,
lin_z: f32,
fetch: impl CubeFetch<T>,
) -> T {
let lin_x = lin_x.max(0.0).min(1.0);
let lin_y = lin_y.max(0.0).min(1.0);
let lin_z = lin_z.max(0.0).min(1.0);
let scale_x = (self.grid_size[0] as i32 - 1) as f32;
let scale_y = (self.grid_size[1] as i32 - 1) as f32;
let scale_z = (self.grid_size[2] as i32 - 1) as f32;
let x = (lin_x * scale_x).floor() as i32;
let y = (lin_y * scale_y).floor() as i32;
let z = (lin_z * scale_z).floor() as i32;
let x_n = (lin_x * scale_x).ceil() as i32;
let y_n = (lin_y * scale_y).ceil() as i32;
let z_n = (lin_z * scale_z).ceil() as i32;
let rx = lin_x * scale_x - x as f32;
let ry = lin_y * scale_y - y as f32;
let rz = lin_z * scale_z - z as f32;
let c0 = fetch.fetch(x, y, z);
let c2;
let c1;
let c3;
if rx >= ry {
if ry >= rz {
//rx >= ry && ry >= rz
c1 = fetch.fetch(x_n, y, z) - c0;
c2 = fetch.fetch(x_n, y_n, z) - fetch.fetch(x_n, y, z);
c3 = fetch.fetch(x_n, y_n, z_n) - fetch.fetch(x_n, y_n, z);
} else if rx >= rz {
//rx >= rz && rz >= ry
c1 = fetch.fetch(x_n, y, z) - c0;
c2 = fetch.fetch(x_n, y_n, z_n) - fetch.fetch(x_n, y, z_n);
c3 = fetch.fetch(x_n, y, z_n) - fetch.fetch(x_n, y, z);
} else {
//rz > rx && rx >= ry
c1 = fetch.fetch(x_n, y, z_n) - fetch.fetch(x, y, z_n);
c2 = fetch.fetch(x_n, y_n, z_n) - fetch.fetch(x_n, y, z_n);
c3 = fetch.fetch(x, y, z_n) - c0;
}
} else if rx >= rz {
//ry > rx && rx >= rz
c1 = fetch.fetch(x_n, y_n, z) - fetch.fetch(x, y_n, z);
c2 = fetch.fetch(x, y_n, z) - c0;
c3 = fetch.fetch(x_n, y_n, z_n) - fetch.fetch(x_n, y_n, z);
} else if ry >= rz {
//ry >= rz && rz > rx
c1 = fetch.fetch(x_n, y_n, z_n) - fetch.fetch(x, y_n, z_n);
c2 = fetch.fetch(x, y_n, z) - c0;
c3 = fetch.fetch(x, y_n, z_n) - fetch.fetch(x, y_n, z);
} else {
//rz > ry && ry > rx
c1 = fetch.fetch(x_n, y_n, z_n) - fetch.fetch(x, y_n, z_n);
c2 = fetch.fetch(x, y_n, z_n) - fetch.fetch(x, y, z_n);
c3 = fetch.fetch(x, y, z_n) - c0;
}
let s0 = c0.mla(c1, T::from(rx));
let s1 = s0.mla(c2, T::from(ry));
s1.mla(c3, T::from(rz))
}
#[cfg(feature = "options")]
#[inline]
fn prism<
T: Copy
+ From<f32>
+ Sub<T, Output = T>
+ Mul<T, Output = T>
+ Add<T, Output = T>
+ FusedMultiplyAdd<T>,
>(
&self,
lin_x: f32,
lin_y: f32,
lin_z: f32,
fetch: impl CubeFetch<T>,
) -> T {
let lin_x = lin_x.max(0.0).min(1.0);
let lin_y = lin_y.max(0.0).min(1.0);
let lin_z = lin_z.max(0.0).min(1.0);
let scale_x = (self.grid_size[0] as i32 - 1) as f32;
let scale_y = (self.grid_size[1] as i32 - 1) as f32;
let scale_z = (self.grid_size[2] as i32 - 1) as f32;
let x = (lin_x * scale_x).floor() as i32;
let y = (lin_y * scale_y).floor() as i32;
let z = (lin_z * scale_z).floor() as i32;
let x_n = (lin_x * scale_x).ceil() as i32;
let y_n = (lin_y * scale_y).ceil() as i32;
let z_n = (lin_z * scale_z).ceil() as i32;
let dr = lin_x * scale_x - x as f32;
let dg = lin_y * scale_y - y as f32;
let db = lin_z * scale_z - z as f32;
let c0 = fetch.fetch(x, y, z);
if db >= dr {
let x0 = fetch.fetch(x, y, z_n);
let x1 = fetch.fetch(x_n, y, z_n);
let x2 = fetch.fetch(x, y_n, z);
let x3 = fetch.fetch(x, y_n, z_n);
let x4 = fetch.fetch(x_n, y_n, z_n);
let c1 = x0 - c0;
let c2 = x1 - x0;
let c3 = x2 - c0;
let c4 = c0 - x2 - x0 + x3;
let c5 = x0 - x3 - x1 + x4;
let s0 = c0.mla(c1, T::from(db));
let s1 = s0.mla(c2, T::from(dr));
let s2 = s1.mla(c3, T::from(dg));
let s3 = s2.mla(c4, T::from(dg * db));
s3.mla(c5, T::from(dr * dg))
} else {
let x0 = fetch.fetch(x_n, y, z);
let x1 = fetch.fetch(x_n, y, z_n);
let x2 = fetch.fetch(x, y_n, z);
let x3 = fetch.fetch(x_n, y_n, z);
let x4 = fetch.fetch(x_n, y_n, z_n);
let c1 = x1 - x0;
let c2 = x0 - c0;
let c3 = x2 - c0;
let c4 = x0 - x3 - x1 + x4;
let c5 = c0 - x2 - x0 + x3;
let s0 = c0.mla(c1, T::from(db));
let s1 = s0.mla(c2, T::from(dr));
let s2 = s1.mla(c3, T::from(dg));
let s3 = s2.mla(c4, T::from(dg * db));
s3.mla(c5, T::from(dr * dg))
}
}
#[inline]
pub(crate) fn trilinear_vec3(&self, lin_x: f32, lin_y: f32, lin_z: f32) -> NeonVector {
self.trilinear(
lin_x,
lin_y,
lin_z,
HexahedronFetch3 {
array: self.array,
x_stride: self.x_stride,
y_stride: self.y_stride,
},
)
}
#[cfg(feature = "options")]
#[inline]
pub(crate) fn prism_vec3(&self, lin_x: f32, lin_y: f32, lin_z: f32) -> NeonVector {
self.prism(
lin_x,
lin_y,
lin_z,
HexahedronFetch3 {
array: self.array,
x_stride: self.x_stride,
y_stride: self.y_stride,
},
)
}
#[cfg(feature = "options")]
#[inline]
pub(crate) fn pyramid_vec3(&self, lin_x: f32, lin_y: f32, lin_z: f32) -> NeonVector {
self.pyramid(
lin_x,
lin_y,
lin_z,
HexahedronFetch3 {
array: self.array,
x_stride: self.x_stride,
y_stride: self.y_stride,
},
)
}
#[cfg(feature = "options")]
#[inline]
pub(crate) fn tetra_vec3(&self, lin_x: f32, lin_y: f32, lin_z: f32) -> NeonVector {
self.tetra(
lin_x,
lin_y,
lin_z,
HexahedronFetch3 {
array: self.array,
x_stride: self.x_stride,
y_stride: self.y_stride,
},
)
}
}