youmuliu/stgsharp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

daily update on 20230829 by nitload youmu

Browse Source
This commit is contained in:
Nitload
2023-08-29 22:29:47 +08:00
parent 8aac9fee87
commit 1fcbfb972b
125 changed files with 6505 additions and 2077 deletions
Download Patch File Download Diff File Expand all files Collapse all files
源代码/StgShqrpGraphic/mat4.c
+26 -75
View File
@@ -3,73 +3,7 @@
#include <pmmintrin.h>
SSGC_API matrix4map* add_mat4(matrix4map* mat1, matrix4map* mat2)
{
matrix4map* ret = _aligned_malloc(sizeof(matrix4), 16);
if (ret != 0)
{
ret->colum[0] = _mm_add_ps(mat1->colum[0], mat2->colum[0]);
ret->colum[1] = _mm_add_ps(mat1->colum[1], mat2->colum[1]);
ret->colum[2] = _mm_add_ps(mat1->colum[2], mat2->colum[2]);
ret->colum[3] = _mm_add_ps(mat1->colum[3], mat2->colum[3]);
}
return ret;
}
SSGC_API matrix4map* sub_mat4(matrix4map* mat1, matrix4map* mat2)
{
matrix4map* ret = _aligned_malloc(sizeof(matrix4), 16);
if (ret != 0)
{
ret->colum[0] = _mm_sub_ps(mat1->colum[0], mat2->colum[0]);
ret->colum[1] = _mm_sub_ps(mat1->colum[1], mat2->colum[1]);
ret->colum[2] = _mm_sub_ps(mat1->colum[2], mat2->colum[2]);
ret->colum[3] = _mm_sub_ps(mat1->colum[3], mat2->colum[3]);
}
return ret;
}
SSGC_API __inline matrix4map* __cdecl mul_const(matrix4map* mat, float value)
{
matrix4map* ret = _aligned_malloc(sizeof(matrix4),16);
if (NULL != ret)
{
float cstarray[4] = { value, value, value, value };
__m128 cst128 = _mm_loadu_ps(&cstarray);
ret->colum[0] = _mm_mul_ps(mat->colum[0], cst128);
ret->colum[1] = _mm_mul_ps(mat->colum[1], cst128);
ret->colum[2] = _mm_mul_ps(mat->colum[2], cst128);
ret->colum[3] = _mm_mul_ps(mat->colum[3], cst128);
}
return ret;
}
//left is the transposed matrix of real left
SSGC_API __inline matrix4* __cdecl mul_4_4(matrix4map* left, matrix4map* right)
{
matrix4map* retPtr = _aligned_malloc(sizeof(matrix4map), 16);
if (NULL != retPtr)
{
for (size_t i = 0; i < 4; i++)
{
for (size_t t = 0; t < 4; t++)
{
retPtr->m[i][t] = mulps_sum_internal(
ALIGN( left->colum[i]),ALIGN(right->colum[t]));
}
}
}
return retPtr;
}
SSGC_API __inline void __cdecl transpose4to2(matrix4map* source, matrix4map* target)
{
@@ -94,18 +28,35 @@ SSGC_API __inline void __cdecl transpose4to4(matrix4map* source, matrix4map* tar
}
SSGC_API __inline float __cdecl det_mat4(matrix4* matPtr)
SSGC_API __inline void __cdecl transpose4to3(matrix4map* source, matrix3map* target)
{
__m128 t0 = _mm_shuffle_ps(ALIGN(source->colum[0]), ALIGN(source->colum[1]), _MM_SHUFFLE(1, 0, 1, 0));
__m128 t1 = _mm_shuffle_ps(ALIGN(source->colum[0]), ALIGN(source->colum[1]), _MM_SHUFFLE(3, 2, 3, 2));
__m128 t2 = _mm_shuffle_ps(ALIGN(source->colum[2]), ALIGN(source->colum[3]), _MM_SHUFFLE(1, 0, 1, 0));
__m128 t3 = _mm_shuffle_ps(ALIGN(source->colum[2]), ALIGN(source->colum[3]), _MM_SHUFFLE(3, 2, 3, 2));
target->colum[0] = _mm_shuffle_ps(t0, t2, _MM_SHUFFLE(2, 0, 2, 0));
target->colum[1] = _mm_shuffle_ps(t0, t2, _MM_SHUFFLE(3, 1, 3, 1));
target->colum[2] = _mm_shuffle_ps(t1, t3, _MM_SHUFFLE(2, 0, 2, 0));
}
SSGC_API __inline float __cdecl det_mat4(matrix4map* matPtr, matrix4map* transpose)
{
__m128 mc0 = ALIGN(matPtr->colum[0]);
__m128 mc1 = ALIGN(matPtr->colum[1]);
__m128 mc2 = ALIGN(matPtr->colum[2]);
__m128 mc3 = ALIGN(matPtr->colum[3]);
__m128 cogeneration = _mm_set_ps(
det_mat3_internal(matPtr->matrix.colum[1], matPtr->matrix.colum[2], matPtr->matrix.colum[3]),
det_mat3_internal(matPtr->matrix.colum[0], matPtr->matrix.colum[2], matPtr->matrix.colum[3]),
det_mat3_internal(matPtr->matrix.colum[0], matPtr->matrix.colum[1], matPtr->matrix.colum[3]),
det_mat3_internal(matPtr->matrix.colum[0], matPtr->matrix.colum[1], matPtr->matrix.colum[2])
det_mat3_internal(mc0, mc1, mc2),
det_mat3_internal(mc0, mc1, mc3),
det_mat3_internal(mc0, mc2, mc3),
det_mat3_internal(mc1, mc2, mc3)
);
printf("%f\n\n", det_mat3_internal(matPtr->matrix.colum[1], matPtr->matrix.colum[2], matPtr->matrix.colum[3]));
__m128 retvec = _mm_mul_ps((matPtr->transpose).colum[3],cogeneration);
__m128 retvec = _mm_mul_ps(ALIGN(transpose->colum[3]),cogeneration);
retvec = _mm_hsub_ps(retvec,retvec);
retvec = _mm_hadd_ps(retvec,retvec);
return -retvec.m128_f32[0];