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
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
|
#if GLM_ARCH & GLM_ARCH_SSE2_BIT
#include "type_mat4x4.hpp"
#include "../geometric.hpp"
#include "../simd/matrix.h"
#include <cstring>
namespace glm{
namespace detail
{
# if GLM_CONFIG_ALIGNED_GENTYPES == GLM_ENABLE
template<qualifier Q>
struct compute_matrixCompMult<4, 4, float, Q, true>
{
GLM_STATIC_ASSERT(detail::is_aligned<Q>::value, "Specialization requires aligned");
GLM_FUNC_QUALIFIER static mat<4, 4, float, Q> call(mat<4, 4, float, Q> const& x, mat<4, 4, float, Q> const& y)
{
mat<4, 4, float, Q> Result;
glm_mat4_matrixCompMult(
*static_cast<glm_vec4 const (*)[4]>(&x[0].data),
*static_cast<glm_vec4 const (*)[4]>(&y[0].data),
*static_cast<glm_vec4(*)[4]>(&Result[0].data));
return Result;
}
};
# endif
template<qualifier Q>
struct compute_transpose<4, 4, float, Q, true>
{
GLM_FUNC_QUALIFIER static mat<4, 4, float, Q> call(mat<4, 4, float, Q> const& m)
{
mat<4, 4, float, Q> Result;
glm_mat4_transpose(&m[0].data, &Result[0].data);
return Result;
}
};
template<qualifier Q>
struct compute_determinant<4, 4, float, Q, true>
{
GLM_FUNC_QUALIFIER static float call(mat<4, 4, float, Q> const& m)
{
return _mm_cvtss_f32(glm_mat4_determinant(&m[0].data));
}
};
template<qualifier Q>
struct compute_inverse<4, 4, float, Q, true>
{
GLM_FUNC_QUALIFIER static mat<4, 4, float, Q> call(mat<4, 4, float, Q> const& m)
{
mat<4, 4, float, Q> Result;
glm_mat4_inverse(&m[0].data, &Result[0].data);
return Result;
}
};
}//namespace detail
# if GLM_CONFIG_ALIGNED_GENTYPES == GLM_ENABLE
template<>
GLM_FUNC_QUALIFIER mat<4, 4, float, aligned_lowp> outerProduct<4, 4, float, aligned_lowp>(vec<4, float, aligned_lowp> const& c, vec<4, float, aligned_lowp> const& r)
{
__m128 NativeResult[4];
glm_mat4_outerProduct(c.data, r.data, NativeResult);
mat<4, 4, float, aligned_lowp> Result;
std::memcpy(&Result[0], &NativeResult[0], sizeof(Result));
return Result;
}
template<>
GLM_FUNC_QUALIFIER mat<4, 4, float, aligned_mediump> outerProduct<4, 4, float, aligned_mediump>(vec<4, float, aligned_mediump> const& c, vec<4, float, aligned_mediump> const& r)
{
__m128 NativeResult[4];
glm_mat4_outerProduct(c.data, r.data, NativeResult);
mat<4, 4, float, aligned_mediump> Result;
std::memcpy(&Result[0], &NativeResult[0], sizeof(Result));
return Result;
}
template<>
GLM_FUNC_QUALIFIER mat<4, 4, float, aligned_highp> outerProduct<4, 4, float, aligned_highp>(vec<4, float, aligned_highp> const& c, vec<4, float, aligned_highp> const& r)
{
__m128 NativeResult[4];
glm_mat4_outerProduct(c.data, r.data, NativeResult);
mat<4, 4, float, aligned_highp> Result;
std::memcpy(&Result[0], &NativeResult[0], sizeof(Result));
return Result;
}
# endif
}//namespace glm
#elif GLM_ARCH & GLM_ARCH_NEON_BIT
namespace glm {
#if GLM_LANG & GLM_LANG_CXX11_FLAG
template <qualifier Q>
GLM_FUNC_QUALIFIER
typename std::enable_if<detail::is_aligned<Q>::value, mat<4, 4, float, Q>>::type
operator*(mat<4, 4, float, Q> const & m1, mat<4, 4, float, Q> const & m2)
{
auto MulRow = [&](int l) {
float32x4_t const SrcA = m2[l].data;
float32x4_t r = neon::mul_lane(m1[0].data, SrcA, 0);
r = neon::madd_lane(r, m1[1].data, SrcA, 1);
r = neon::madd_lane(r, m1[2].data, SrcA, 2);
r = neon::madd_lane(r, m1[3].data, SrcA, 3);
return r;
};
mat<4, 4, float, aligned_highp> Result;
Result[0].data = MulRow(0);
Result[1].data = MulRow(1);
Result[2].data = MulRow(2);
Result[3].data = MulRow(3);
return Result;
}
#endif // CXX11
template<qualifier Q>
struct detail::compute_inverse<4, 4, float, Q, true>
{
GLM_FUNC_QUALIFIER static mat<4, 4, float, Q> call(mat<4, 4, float, Q> const& m)
{
float32x4_t const& m0 = m[0].data;
float32x4_t const& m1 = m[1].data;
float32x4_t const& m2 = m[2].data;
float32x4_t const& m3 = m[3].data;
// m[2][2] * m[3][3] - m[3][2] * m[2][3];
// m[2][2] * m[3][3] - m[3][2] * m[2][3];
// m[1][2] * m[3][3] - m[3][2] * m[1][3];
// m[1][2] * m[2][3] - m[2][2] * m[1][3];
float32x4_t Fac0;
{
float32x4_t w0 = vcombine_f32(neon::dup_lane(m2, 2), neon::dup_lane(m1, 2));
float32x4_t w1 = neon::copy_lane(neon::dupq_lane(m3, 3), 3, m2, 3);
float32x4_t w2 = neon::copy_lane(neon::dupq_lane(m3, 2), 3, m2, 2);
float32x4_t w3 = vcombine_f32(neon::dup_lane(m2, 3), neon::dup_lane(m1, 3));
Fac0 = w0 * w1 - w2 * w3;
}
// m[2][1] * m[3][3] - m[3][1] * m[2][3];
// m[2][1] * m[3][3] - m[3][1] * m[2][3];
// m[1][1] * m[3][3] - m[3][1] * m[1][3];
// m[1][1] * m[2][3] - m[2][1] * m[1][3];
float32x4_t Fac1;
{
float32x4_t w0 = vcombine_f32(neon::dup_lane(m2, 1), neon::dup_lane(m1, 1));
float32x4_t w1 = neon::copy_lane(neon::dupq_lane(m3, 3), 3, m2, 3);
float32x4_t w2 = neon::copy_lane(neon::dupq_lane(m3, 1), 3, m2, 1);
float32x4_t w3 = vcombine_f32(neon::dup_lane(m2, 3), neon::dup_lane(m1, 3));
Fac1 = w0 * w1 - w2 * w3;
}
// m[2][1] * m[3][2] - m[3][1] * m[2][2];
// m[2][1] * m[3][2] - m[3][1] * m[2][2];
// m[1][1] * m[3][2] - m[3][1] * m[1][2];
// m[1][1] * m[2][2] - m[2][1] * m[1][2];
float32x4_t Fac2;
{
float32x4_t w0 = vcombine_f32(neon::dup_lane(m2, 1), neon::dup_lane(m1, 1));
float32x4_t w1 = neon::copy_lane(neon::dupq_lane(m3, 2), 3, m2, 2);
float32x4_t w2 = neon::copy_lane(neon::dupq_lane(m3, 1), 3, m2, 1);
float32x4_t w3 = vcombine_f32(neon::dup_lane(m2, 2), neon::dup_lane(m1, 2));
Fac2 = w0 * w1 - w2 * w3;
}
// m[2][0] * m[3][3] - m[3][0] * m[2][3];
// m[2][0] * m[3][3] - m[3][0] * m[2][3];
// m[1][0] * m[3][3] - m[3][0] * m[1][3];
// m[1][0] * m[2][3] - m[2][0] * m[1][3];
float32x4_t Fac3;
{
float32x4_t w0 = vcombine_f32(neon::dup_lane(m2, 0), neon::dup_lane(m1, 0));
float32x4_t w1 = neon::copy_lane(neon::dupq_lane(m3, 3), 3, m2, 3);
float32x4_t w2 = neon::copy_lane(neon::dupq_lane(m3, 0), 3, m2, 0);
float32x4_t w3 = vcombine_f32(neon::dup_lane(m2, 3), neon::dup_lane(m1, 3));
Fac3 = w0 * w1 - w2 * w3;
}
// m[2][0] * m[3][2] - m[3][0] * m[2][2];
// m[2][0] * m[3][2] - m[3][0] * m[2][2];
// m[1][0] * m[3][2] - m[3][0] * m[1][2];
// m[1][0] * m[2][2] - m[2][0] * m[1][2];
float32x4_t Fac4;
{
float32x4_t w0 = vcombine_f32(neon::dup_lane(m2, 0), neon::dup_lane(m1, 0));
float32x4_t w1 = neon::copy_lane(neon::dupq_lane(m3, 2), 3, m2, 2);
float32x4_t w2 = neon::copy_lane(neon::dupq_lane(m3, 0), 3, m2, 0);
float32x4_t w3 = vcombine_f32(neon::dup_lane(m2, 2), neon::dup_lane(m1, 2));
Fac4 = w0 * w1 - w2 * w3;
}
// m[2][0] * m[3][1] - m[3][0] * m[2][1];
// m[2][0] * m[3][1] - m[3][0] * m[2][1];
// m[1][0] * m[3][1] - m[3][0] * m[1][1];
// m[1][0] * m[2][1] - m[2][0] * m[1][1];
float32x4_t Fac5;
{
float32x4_t w0 = vcombine_f32(neon::dup_lane(m2, 0), neon::dup_lane(m1, 0));
float32x4_t w1 = neon::copy_lane(neon::dupq_lane(m3, 1), 3, m2, 1);
float32x4_t w2 = neon::copy_lane(neon::dupq_lane(m3, 0), 3, m2, 0);
float32x4_t w3 = vcombine_f32(neon::dup_lane(m2, 1), neon::dup_lane(m1, 1));
Fac5 = w0 * w1 - w2 * w3;
}
float32x4_t Vec0 = neon::copy_lane(neon::dupq_lane(m0, 0), 0, m1, 0); // (m[1][0], m[0][0], m[0][0], m[0][0]);
float32x4_t Vec1 = neon::copy_lane(neon::dupq_lane(m0, 1), 0, m1, 1); // (m[1][1], m[0][1], m[0][1], m[0][1]);
float32x4_t Vec2 = neon::copy_lane(neon::dupq_lane(m0, 2), 0, m1, 2); // (m[1][2], m[0][2], m[0][2], m[0][2]);
float32x4_t Vec3 = neon::copy_lane(neon::dupq_lane(m0, 3), 0, m1, 3); // (m[1][3], m[0][3], m[0][3], m[0][3]);
float32x4_t Inv0 = Vec1 * Fac0 - Vec2 * Fac1 + Vec3 * Fac2;
float32x4_t Inv1 = Vec0 * Fac0 - Vec2 * Fac3 + Vec3 * Fac4;
float32x4_t Inv2 = Vec0 * Fac1 - Vec1 * Fac3 + Vec3 * Fac5;
float32x4_t Inv3 = Vec0 * Fac2 - Vec1 * Fac4 + Vec2 * Fac5;
float32x4_t r0 = float32x4_t{-1, +1, -1, +1} * Inv0;
float32x4_t r1 = float32x4_t{+1, -1, +1, -1} * Inv1;
float32x4_t r2 = float32x4_t{-1, +1, -1, +1} * Inv2;
float32x4_t r3 = float32x4_t{+1, -1, +1, -1} * Inv3;
float32x4_t det = neon::mul_lane(r0, m0, 0);
det = neon::madd_lane(det, r1, m0, 1);
det = neon::madd_lane(det, r2, m0, 2);
det = neon::madd_lane(det, r3, m0, 3);
float32x4_t rdet = vdupq_n_f32(1 / vgetq_lane_f32(det, 0));
mat<4, 4, float, Q> r;
r[0].data = vmulq_f32(r0, rdet);
r[1].data = vmulq_f32(r1, rdet);
r[2].data = vmulq_f32(r2, rdet);
r[3].data = vmulq_f32(r3, rdet);
return r;
}
};
}//namespace glm
#endif
|
