External/cglm
1
0
Fork 0
mirror of https://github.com/recp/cglm.git synced 2025-12-24 12:32:40 +00:00
Code Issues Packages Projects Releases Wiki Activity

new matrix multiplication helper for rotation matrices

Browse Source
This commit is contained in:
Recep Aslantas
2018-04-18 14:05:09 +03:00
parent d648f5772d
commit be0e3fc9f2
2 changed files with 86 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

54
include/cglm/affine-mat.h
View File

@@ -81,6 +81,60 @@ glm_mul(mat4 m1, mat4 m2, mat4 dest) {
#endif
}
/*!
* @brief this is similar to glm_mat4_mul but specialized to affine transform
*
* Right Matrix format should be:
* R R R 0
* R R R 0
* R R R 0
* 0 0 0 1
*
* this reduces some multiplications. It should be faster than mat4_mul.
* if you are not sure about matrix format then DON'T use this! use mat4_mul
*
* @param[in] m1 affine matrix 1
* @param[in] m2 affine matrix 2
* @param[out] dest result matrix
*/
CGLM_INLINE
void
glm_mul_rot(mat4 m1, mat4 m2, mat4 dest) {
#if defined( __SSE__ ) || defined( __SSE2__ )
glm_mul_rot_sse2(m1, m2, dest);
#else
float a00 = m1[0][0], a01 = m1[0][1], a02 = m1[0][2], a03 = m1[0][3],
a10 = m1[1][0], a11 = m1[1][1], a12 = m1[1][2], a13 = m1[1][3],
a20 = m1[2][0], a21 = m1[2][1], a22 = m1[2][2], a23 = m1[2][3],
a30 = m1[3][0], a31 = m1[3][1], a32 = m1[3][2], a33 = m1[3][3],
b00 = m2[0][0], b01 = m2[0][1], b02 = m2[0][2],
b10 = m2[1][0], b11 = m2[1][1], b12 = m2[1][2],
b20 = m2[2][0], b21 = m2[2][1], b22 = m2[2][2],
b33 = m2[3][3];
dest[0][0] = a00 * b00 + a10 * b01 + a20 * b02;
dest[0][1] = a01 * b00 + a11 * b01 + a21 * b02;
dest[0][2] = a02 * b00 + a12 * b01 + a22 * b02;
dest[0][3] = a03 * b00 + a13 * b01 + a23 * b02;
dest[1][0] = a00 * b10 + a10 * b11 + a20 * b12;
dest[1][1] = a01 * b10 + a11 * b11 + a21 * b12;
dest[1][2] = a02 * b10 + a12 * b11 + a22 * b12;
dest[1][3] = a03 * b10 + a13 * b11 + a23 * b12;
dest[2][0] = a00 * b20 + a10 * b21 + a20 * b22;
dest[2][1] = a01 * b20 + a11 * b21 + a21 * b22;
dest[2][2] = a02 * b20 + a12 * b21 + a22 * b22;
dest[2][3] = a03 * b20 + a13 * b21 + a23 * b22;
dest[3][0] = a30;
dest[3][1] = a31;
dest[3][2] = a32;
dest[3][3] = a33;
#endif
}
/*!
* @brief inverse orthonormal rotation + translation matrix (ridig-body)
*

32
include/cglm/simd/sse2/affine.h
View File

@@ -49,6 +49,38 @@ glm_mul_sse2(mat4 m1, mat4 m2, mat4 dest) {
_mm_mul_ps(_mm_shuffle1_ps1(r, 3), l3))));
}
CGLM_INLINE
void
glm_mul_rot_sse2(mat4 m1, mat4 m2, mat4 dest) {
/* D = R * L (Column-Major) */
__m128 l0, l1, l2, l3, r;
l0 = _mm_load_ps(m1[0]);
l1 = _mm_load_ps(m1[1]);
l2 = _mm_load_ps(m1[2]);
l3 = _mm_load_ps(m1[3]);
r = _mm_load_ps(m2[0]);
_mm_store_ps(dest[0],
_mm_add_ps(_mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps1(r, 0), l0),
_mm_mul_ps(_mm_shuffle1_ps1(r, 1), l1)),
_mm_mul_ps(_mm_shuffle1_ps1(r, 2), l2)));
r = _mm_load_ps(m2[1]);
_mm_store_ps(dest[1],
_mm_add_ps(_mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps1(r, 0), l0),
_mm_mul_ps(_mm_shuffle1_ps1(r, 1), l1)),
_mm_mul_ps(_mm_shuffle1_ps1(r, 2), l2)));
r = _mm_load_ps(m2[2]);
_mm_store_ps(dest[2],
_mm_add_ps(_mm_add_ps(_mm_mul_ps(_mm_shuffle1_ps1(r, 0), l0),
_mm_mul_ps(_mm_shuffle1_ps1(r, 1), l1)),
_mm_mul_ps(_mm_shuffle1_ps1(r, 2), l2)));
_mm_store_ps(dest[3], l3);
}
CGLM_INLINE
void
glm_inv_tr_sse2(mat4 mat) {