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
|
#[cfg(target_arch = "x86")]
use core::arch::x86::*;
#[cfg(target_arch = "x86_64")]
use core::arch::x86_64::*;
use super::float::*;
use crate::core::{
storage::XYZ,
traits::{quaternion::Quaternion, scalar::*, vector::*},
};
impl Quaternion<f32> for __m128 {
type SIMDVector3 = __m128;
#[inline(always)]
fn conjugate(self) -> Self {
const SIGN: __m128 = const_f32x4!([-0.0, -0.0, -0.0, 0.0]);
unsafe { _mm_xor_ps(self, SIGN) }
}
#[inline]
fn lerp(self, end: Self, s: f32) -> Self {
glam_assert!(FloatVector4::is_normalized(self));
glam_assert!(FloatVector4::is_normalized(end));
unsafe {
const NEG_ZERO: __m128 = const_f32x4!([-0.0; 4]);
let start = self;
let end = end;
let dot = Vector4::dot_into_vec(start, end);
// Calculate the bias, if the dot product is positive or zero, there is no bias
// but if it is negative, we want to flip the 'end' rotation XYZW components
let bias = _mm_and_ps(dot, NEG_ZERO);
let interpolated = _mm_add_ps(
_mm_mul_ps(_mm_sub_ps(_mm_xor_ps(end, bias), start), _mm_set_ps1(s)),
start,
);
FloatVector4::normalize(interpolated)
}
}
#[inline]
fn slerp(self, end: Self, s: f32) -> Self {
// http://number-none.com/product/Understanding%20Slerp,%20Then%20Not%20Using%20It/
glam_assert!(FloatVector4::is_normalized(self));
glam_assert!(FloatVector4::is_normalized(end));
const DOT_THRESHOLD: f32 = 0.9995;
let dot = Vector4::dot(self, end);
if dot > DOT_THRESHOLD {
// assumes lerp returns a normalized quaternion
self.lerp(end, s)
} else {
// assumes scalar_acos clamps the input to [-1.0, 1.0]
let theta = dot.acos_approx();
let x = 1.0 - s;
let y = s;
let z = 1.0;
unsafe {
let tmp = _mm_mul_ps(_mm_set_ps1(theta), _mm_set_ps(0.0, z, y, x));
let tmp = m128_sin(tmp);
let scale1 = _mm_shuffle_ps(tmp, tmp, 0b00_00_00_00);
let scale2 = _mm_shuffle_ps(tmp, tmp, 0b01_01_01_01);
let theta_sin = _mm_shuffle_ps(tmp, tmp, 0b10_10_10_10);
self.mul(scale1).add(end.mul(scale2)).div(theta_sin)
}
}
}
#[inline]
fn mul_quaternion(self, other: Self) -> Self {
glam_assert!(FloatVector4::is_normalized(self));
glam_assert!(FloatVector4::is_normalized(other));
unsafe {
// Based on https://github.com/nfrechette/rtm `rtm::quat_mul`
let lhs = self;
let rhs = other;
const CONTROL_WZYX: __m128 = const_f32x4!([1.0, -1.0, 1.0, -1.0]);
const CONTROL_ZWXY: __m128 = const_f32x4!([1.0, 1.0, -1.0, -1.0]);
const CONTROL_YXWZ: __m128 = const_f32x4!([-1.0, 1.0, 1.0, -1.0]);
let r_xxxx = _mm_shuffle_ps(lhs, lhs, 0b00_00_00_00);
let r_yyyy = _mm_shuffle_ps(lhs, lhs, 0b01_01_01_01);
let r_zzzz = _mm_shuffle_ps(lhs, lhs, 0b10_10_10_10);
let r_wwww = _mm_shuffle_ps(lhs, lhs, 0b11_11_11_11);
let lxrw_lyrw_lzrw_lwrw = _mm_mul_ps(r_wwww, rhs);
let l_wzyx = _mm_shuffle_ps(rhs, rhs, 0b00_01_10_11);
let lwrx_lzrx_lyrx_lxrx = _mm_mul_ps(r_xxxx, l_wzyx);
let l_zwxy = _mm_shuffle_ps(l_wzyx, l_wzyx, 0b10_11_00_01);
let lwrx_nlzrx_lyrx_nlxrx = _mm_mul_ps(lwrx_lzrx_lyrx_lxrx, CONTROL_WZYX);
let lzry_lwry_lxry_lyry = _mm_mul_ps(r_yyyy, l_zwxy);
let l_yxwz = _mm_shuffle_ps(l_zwxy, l_zwxy, 0b00_01_10_11);
let lzry_lwry_nlxry_nlyry = _mm_mul_ps(lzry_lwry_lxry_lyry, CONTROL_ZWXY);
let lyrz_lxrz_lwrz_lzrz = _mm_mul_ps(r_zzzz, l_yxwz);
let result0 = _mm_add_ps(lxrw_lyrw_lzrw_lwrw, lwrx_nlzrx_lyrx_nlxrx);
let nlyrz_lxrz_lwrz_wlzrz = _mm_mul_ps(lyrz_lxrz_lwrz_lzrz, CONTROL_YXWZ);
let result1 = _mm_add_ps(lzry_lwry_nlxry_nlyry, nlyrz_lxrz_lwrz_wlzrz);
_mm_add_ps(result0, result1)
}
}
#[inline]
fn mul_vector3(self, other: XYZ<f32>) -> XYZ<f32> {
self.mul_float4_as_vector3(other.into()).into()
}
#[inline]
fn mul_float4_as_vector3(self, other: __m128) -> __m128 {
glam_assert!(FloatVector4::is_normalized(self));
unsafe {
const TWO: __m128 = const_f32x4!([2.0; 4]);
let w = _mm_shuffle_ps(self, self, 0b11_11_11_11);
let b = self;
let b2 = Vector3::dot_into_vec(b, b);
other
.mul(w.mul(w).sub(b2))
.add(b.mul(Vector3::dot_into_vec(other, b).mul(TWO)))
.add(b.cross(other).mul(w.mul(TWO)))
}
}
}
|
