1 /*
2 * Copyright (c), Recep Aslantas.
3 *
4 * MIT License (MIT), http://opensource.org/licenses/MIT
5 * Full license can be found in the LICENSE file
6 */
7
8 #ifndef cglm_mat2_neon_h
9 #define cglm_mat2_neon_h
10 #if defined(__ARM_NEON_FP)
11
12 #include "../../common.h"
13 #include "../intrin.h"
14
15 CGLM_INLINE
16 void
glm_mat2_mul_neon(mat2 m1,mat2 m2,mat2 dest)17 glm_mat2_mul_neon(mat2 m1, mat2 m2, mat2 dest) {
18 float32x4x2_t a1;
19 glmm_128 x0, x1, x2;
20 float32x2_t dc, ba;
21
22 x1 = glmm_load(m1[0]); /* d c b a */
23 x2 = glmm_load(m2[0]); /* h g f e */
24
25 dc = vget_high_f32(x1);
26 ba = vget_low_f32(x1);
27
28 /* g g e e, h h f f */
29 a1 = vtrnq_f32(x2, x2);
30
31 /*
32 dest[0][0] = a * e + c * f;
33 dest[0][1] = b * e + d * f;
34 dest[1][0] = a * g + c * h;
35 dest[1][1] = b * g + d * h;
36 */
37 x0 = glmm_fmadd(vcombine_f32(ba, ba), a1.val[0],
38 vmulq_f32(vcombine_f32(dc, dc), a1.val[1]));
39
40 glmm_store(dest[0], x0);
41 }
42
43 #endif
44 #endif /* cglm_mat2_neon_h */
45