Created
September 14, 2018 02:49
-
-
Save djg/9ca44f6e9872dd4a457a24a61b8b63ca to your computer and use it in GitHub Desktop.
SIMD in Rust is kinda gross.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
// Type your code here, or load an example. | |
#[cfg(target_arch = "x86")] | |
use std::arch::x86::__m128; | |
#[cfg(target_arch = "x86_64")] | |
use std::arch::x86_64::__m128; | |
#[cfg(all(any(target_arch = "x86", target_arch = "x86_64"), target_feature = "sse2"))] | |
pub struct Ray { | |
pub pt: __m128, | |
pub dir: __m128, | |
} | |
#[cfg(all(any(target_arch = "x86", target_arch = "x86_64"), target_feature = "sse2"))] | |
pub struct Aabb { | |
pub min: __m128, | |
pub max: __m128, | |
} | |
macro_rules! _mm_shuffle { | |
($w:expr, $z:expr, $y:expr, $x:expr) => ( | |
($w << 6) | ($z << 4) | ($y << 2) | $x | |
) | |
} | |
#[cfg(all(any(target_arch = "x86", target_arch = "x86_64"), target_feature = "sse2"))] | |
pub fn hit(aabb: &Aabb, r: &Ray, t_min: f32, t_max: f32) -> bool { | |
#[cfg(target_arch = "x86")] | |
use std::arch::x86::*; | |
#[cfg(target_arch = "x86_64")] | |
use std::arch::x86_64::*; | |
unsafe { | |
let inv_d = _mm_div_ps(_mm_set1_ps(1.), r.dir); | |
let t0 = _mm_mul_ps(_mm_sub_ps(aabb.min, r.pt), inv_d); | |
let t1 = _mm_mul_ps(_mm_sub_ps(aabb.max, r.pt), inv_d); | |
// swap | |
let nm = _mm_cmplt_ps(inv_d, _mm_setzero_ps()); | |
let (t0, t1) = (_mm_or_ps(_mm_and_ps(t1, nm), _mm_andnot_ps(nm, t0)), | |
_mm_or_ps(_mm_and_ps(t0, nm), _mm_andnot_ps(nm, t1))); | |
let t_min = _mm_move_ss(_mm_shuffle_ps(t0, t0, _mm_shuffle!(2, 1, 0, 0)), _mm_set_ss(t_min)); | |
let t_max = _mm_move_ss(_mm_shuffle_ps(t1, t1, _mm_shuffle!(2, 1, 0, 0)), _mm_set_ss(t_max)); | |
let t_min = { | |
let v = _mm_max_ps(t_min, _mm_shuffle_ps(t_min, t_min, _mm_shuffle!(0, 0, 3, 2))); | |
let v = _mm_max_ps(v, _mm_shuffle_ps(v, v, _mm_shuffle!(0, 0, 0, 1))); | |
_mm_cvtss_f32(v) | |
}; | |
let t_max = { | |
let v = _mm_min_ps(t_max, _mm_shuffle_ps(t_max, t_max, _mm_shuffle!(0, 0, 3, 2))); | |
let v = _mm_min_ps(v, _mm_shuffle_ps(v, v, _mm_shuffle!(0, 0, 0, 1))); | |
_mm_cvtss_f32(v) | |
}; | |
t_min < t_max | |
} | |
} |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment