mirror of
https://github.com/blawar/GLideN64.git
synced 2024-07-02 09:03:37 +00:00
arm neon: Cleanup InverseTransformVectorNormalize
Add a arm asm multivector function of InverseTransformVectorNormalize which replaces 2x and 4x functions.
This commit is contained in:
gizmo98
Sergey Lipskiy
parent
37772ca51e
commit
542d083840
|
|
@ -88,16 +88,10 @@ End:
|
|||
#endif // WIN32_ASM
|
||||
}
|
||||
|
||||
void InverseTransformVectorNormalize2x(float src[2][3], float dst[2][3], float mtx[4][4] )
|
||||
void InverseTransformVectorNormalizeN(float src[][3], float dst[][3], float mtx[4][4], s32 count)
|
||||
{
|
||||
InverseTransformVectorNormalize((float(*))src[0], (float(*))dst[0], mtx);
|
||||
InverseTransformVectorNormalize((float(*))src[1], (float(*))dst[1], mtx);
|
||||
}
|
||||
|
||||
void InverseTransformVectorNormalize4x(float src[4][3], float dst[4][3], float mtx[4][4] )
|
||||
{
|
||||
InverseTransformVectorNormalize((float(*))src[0], (float(*))dst[0], mtx);
|
||||
InverseTransformVectorNormalize((float(*))src[1], (float(*))dst[1], mtx);
|
||||
InverseTransformVectorNormalize((float(*))src[2], (float(*))dst[2], mtx);
|
||||
InverseTransformVectorNormalize((float(*))src[3], (float(*))dst[3], mtx);
|
||||
for (s32 i = 0; i < count; i++)
|
||||
{
|
||||
InverseTransformVectorNormalize((float(*))src[i], (float(*))dst[i], mtx);
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -2,13 +2,13 @@
|
|||
#define _3DMATH_H
|
||||
#include <memory.h>
|
||||
#include <string.h>
|
||||
#include <Types.h>
|
||||
|
||||
void MultMatrix(float m0[4][4], float m1[4][4], float dest[4][4]);
|
||||
void MultMatrix2(float m0[4][4], float m1[4][4]);
|
||||
void TransformVectorNormalize(float vec[3], float mtx[4][4]);
|
||||
void InverseTransformVectorNormalize(float src[3], float dst[3], float mtx[4][4]);
|
||||
void InverseTransformVectorNormalize2x(float src[2][3], float dst[2][3], float mtx[4][4] );
|
||||
void InverseTransformVectorNormalize4x(float src[4][3], float dst[4][3], float mtx[4][4] );
|
||||
void InverseTransformVectorNormalizeN(float src[][3], float dst[][3], float mtx[4][4], s32 count);
|
||||
void Normalize(float v[3]);
|
||||
float DotProduct(const float v0[3], const float v1[3]);
|
||||
|
||||
|
|
|
|||
|
|
@ -140,98 +140,206 @@ void Normalize(float v[3])
|
|||
);
|
||||
}
|
||||
|
||||
void InverseTransformVectorNormalize2x(float src[2][3], float dst[2][3], float mtx[4][4] )
|
||||
void InverseTransformVectorNormalizeN(float src[][3], float dst[][3], float mtx[4][4], int count)
|
||||
{
|
||||
void *ptr = (void*)&count;
|
||||
asm volatile (
|
||||
// Load src
|
||||
"vld3.32 {d0,d1,d2}, [%1] \n\t" // load vector 0-1
|
||||
"ldr r4, [%3] \n\t"
|
||||
"b 4f \n\t"
|
||||
"7: \n\t"
|
||||
// Load four src vectors
|
||||
" vld3.32 {d0,d2,d4}, [%1]! \n\t" // load vector 0-1
|
||||
" vld3.32 {d1,d3,d5}, [%1]! \n\t" // load vector 2-3
|
||||
" vld3.32 {d24,d26,d28}, [%1]! \n\t" // load vector 4-5
|
||||
" vld3.32 {d25[0],d27[0],d29[0]}, [%1]!\n\t" // load vector 6
|
||||
|
||||
// Load mtx
|
||||
" vld1.32 {d6-d9}, [%0]! \n\t" // load row 0-1
|
||||
" vld1.32 {d10-d11}, [%0] \n\t" // load row 2
|
||||
" sub %0, %0, #32 \n\t"
|
||||
|
||||
// Multiply and add
|
||||
" vmul.f32 q6, q0, d6[0] \n\t" //dst[0] = v[0]*mtx[0][0]
|
||||
" vmul.f32 q7, q0, d8[0] \n\t" //dst[1] = v[0]*mtx[1][0]
|
||||
" vmul.f32 q8, q0, d10[0] \n\t" //dst[2] = v[0]*mtx[2][0]
|
||||
" vmul.f32 q9, q12, d6[0] \n\t" //dst[0] = v[0]*mtx[0][0]
|
||||
" vmul.f32 q10, q12, d8[0] \n\t" //dst[1] = v[0]*mtx[1][0]
|
||||
" vmul.f32 q11, q12, d10[0] \n\t" //dst[2] = v[0]*mtx[2][0]
|
||||
" vmla.f32 q6, q1, d6[1] \n\t" //dst[0] += v[1]*mtx[0][1]
|
||||
" vmla.f32 q7, q1, d8[1] \n\t" //dst[1] += v[1]*mtx[1][1]
|
||||
" vmla.f32 q8, q1, d10[1] \n\t" //dst[2] += v[1]*mtx[2][1]
|
||||
" vmla.f32 q9, q13, d6[1] \n\t" //dst[0] += v[1]*mtx[0][1]
|
||||
" vmla.f32 q10, q13, d8[1] \n\t" //dst[1] += v[1]*mtx[1][1]
|
||||
" vmla.f32 q11, q13, d10[1] \n\t" //dst[2] += v[1]*mtx[2][1]
|
||||
" vmla.f32 q6, q2, d7[0] \n\t" //dst[0] += v[0]*mtx[2][1]
|
||||
" vmla.f32 q7, q2, d9[0] \n\t" //dst[1] += v[2]*mtx[2][1]
|
||||
" vmla.f32 q8, q2, d11[0] \n\t" //dst[2] += v[2]*mtx[2][2]
|
||||
" vmla.f32 q9, q14, d7[0] \n\t" //dst[0] += v[0]*mtx[2][1]
|
||||
" vmla.f32 q10, q14, d9[0] \n\t" //dst[1] += v[2]*mtx[2][1]
|
||||
" vmla.f32 q11, q14, d11[0] \n\t" //dst[2] += v[2]*mtx[2][2]
|
||||
|
||||
// Normalize 4x
|
||||
" vmul.f32 q0, q6, q6 \n\t" //len = dst[0]*dst[0]
|
||||
" vmul.f32 q4, q9, q9 \n\t" //len = dst[0]*dst[0]
|
||||
" vmla.f32 q0, q7, q7 \n\t" //len += dst[1]*dst[1]
|
||||
" vmla.f32 q4, q10, q10 \n\t" //len += dst[1]*dst[1]
|
||||
" vmla.f32 q0, q8, q8 \n\t" //len += dst[2]*dst[2]
|
||||
" vmla.f32 q4, q11, q11 \n\t" //len += dst[2]*dst[2]
|
||||
|
||||
" vmov.f32 q1, q0 \n\t" //q1 = len
|
||||
" vrsqrte.f32 q0, q0 \n\t" //q0 = ~ 1.0 / sqrt(q0)
|
||||
" vmov.f32 q5, q4 \n\t" //q5 = len
|
||||
" vrsqrte.f32 q4, q4 \n\t" //q4 = ~ 1.0 / sqrt(q4)
|
||||
" vmul.f32 q2, q0, q1 \n\t" //q2 = q0 * q1
|
||||
" vmul.f32 q12, q4, q5 \n\t" //q12 = q4 * q5
|
||||
" vrsqrts.f32 q3, q2, q0 \n\t" //q3 = (3 - q0 * q2) / 2
|
||||
" vrsqrts.f32 q13, q12, q4 \n\t" //q13 = (3 - q4 * q12) / 2
|
||||
" vmul.f32 q0, q0, q3 \n\t" //q0 = q0 * q3
|
||||
" vmul.f32 q4, q4, q13 \n\t" //q4 = q4 * q13
|
||||
" vmul.f32 q2, q0, q1 \n\t" //q2 = q0 * q1
|
||||
" vmul.f32 q12, q4, q5 \n\t" //q12 = q4 * q5
|
||||
" vrsqrts.f32 q3, q2, q0 \n\t" //q3 = (3 - q0 * q2) / 2
|
||||
" vrsqrts.f32 q13, q12, q4 \n\t" //q13 = (3 - q4 * q12) / 2
|
||||
" vmul.f32 q0, q0, q3 \n\t" //q0 = q0 * q3
|
||||
" vmul.f32 q4, q4, q13 \n\t" //q4 = q4 * q13
|
||||
|
||||
" vmul.f32 q6, q6, q0 \n\t" //dst[0] = dst[0]*(1/len)
|
||||
" vmul.f32 q7, q7, q0 \n\t" //dst[1] = dst[1]*(1/len)
|
||||
" vmul.f32 q8, q8, q0 \n\t" //dst[2] = dst[2]*(1/len)
|
||||
|
||||
" vmul.f32 q9, q9, q4 \n\t" //dst[0] = dst[0]*(1/len)
|
||||
" vmul.f32 q10, q10, q4 \n\t" //dst[1] = dst[1]*(1/len)
|
||||
" vmul.f32 q11, q11, q4 \n\t" //dst[2] = dst[2]*(1/len)
|
||||
|
||||
// Store four dst vectors
|
||||
" vst3.32 {d12, d14, d16}, [%2]! \n\t" // store vector 0-1
|
||||
" vst3.32 {d13, d15, d17}, [%2]! \n\t" // store vector 2-3
|
||||
" vst3.32 {d18, d20, d22}, [%2]! \n\t" // store vector 4-5
|
||||
" vst3.32 {d19[0],d21[0],d23[0]}, [%2]!\n\t" // store vector 6
|
||||
" sub r4, r4, #7 \n\t"
|
||||
"4: \n\t"
|
||||
" cmp r4, #6 \n\t"
|
||||
" bgt 7b \n\t"
|
||||
" cmp r4, #3 \n\t"
|
||||
" ble 2f \n\t"
|
||||
// Load four src vectors
|
||||
" vld3.32 {d0,d2,d4}, [%1]! \n\t" // load vector 0-1
|
||||
" vld3.32 {d1,d3,d5}, [%1]! \n\t" // load vector 2-3
|
||||
|
||||
// Load mtx
|
||||
"vld1.32 {d6-d9}, [%0]! \n\t" // load row 0-1
|
||||
"vld1.32 {d10-d11}, [%0] \n\t" // load row 2
|
||||
// Load mtx
|
||||
" vld1.32 {d6-d9}, [%0]! \n\t" // load row 0-1
|
||||
" vld1.32 {d10-d11}, [%0] \n\t" // load row 2
|
||||
" sub %0, %0, #32 \n\t"
|
||||
|
||||
// Multiply and add
|
||||
" vmul.f32 q8, q0, d6[0] \n\t" //dst[0] = v[0]*mtx[0][0]
|
||||
" vmul.f32 q9, q0, d8[0] \n\t" //dst[1] = v[0]*mtx[1][0]
|
||||
" vmul.f32 q10, q0, d10[0] \n\t" //dst[2] = v[0]*mtx[2][0]
|
||||
" vmla.f32 q8, q1, d6[1] \n\t" //dst[0] += v[1]*mtx[0][1]
|
||||
" vmla.f32 q9, q1, d8[1] \n\t" //dst[1] += v[1]*mtx[1][1]
|
||||
" vmla.f32 q10, q1, d10[1] \n\t" //dst[2] += v[1]*mtx[2][1]
|
||||
" vmla.f32 q8, q2, d7[0] \n\t" //dst[0] += v[0]*mtx[2][1]
|
||||
" vmla.f32 q9, q2, d9[0] \n\t" //dst[1] += v[2]*mtx[2][1]
|
||||
" vmla.f32 q10, q2, d11[0] \n\t" //dst[2] += v[2]*mtx[2][2]
|
||||
|
||||
// Normalize 4x
|
||||
" vmul.f32 q0, q8, q8 \n\t" //len = dst[0]*dst[0]
|
||||
" vmla.f32 q0, q9, q9 \n\t" //len += dst[1]*dst[1]
|
||||
" vmla.f32 q0, q10, q10 \n\t" //len += dst[2]*dst[2]
|
||||
|
||||
" vmov.f32 q1, q0 \n\t" //d1 = len
|
||||
" vrsqrte.f32 q0, q0 \n\t" //q0 = ~ 1.0 / sqrt(q0)
|
||||
" vmul.f32 q7, q0, q1 \n\t" //q7 = q0 * q1
|
||||
" vrsqrts.f32 q6, q7, q0 \n\t" //q6 = (3 - q0 * q7) / 2
|
||||
" vmul.f32 q0, q0, q6 \n\t" //d0 = d0 * d15
|
||||
" vmul.f32 q7, q0, q1 \n\t" //q7 = q0 * q1
|
||||
" vrsqrts.f32 q6, q7, q0 \n\t" //q6 = (3 - q0 * q7) / 2
|
||||
" vmul.f32 q0, q0, q6 \n\t" //q0 = q0 * q6
|
||||
|
||||
" vmul.f32 q8, q8, q0 \n\t" //dst[0] = dst[0]*(1/len)
|
||||
" vmul.f32 q9, q9, q0 \n\t" //dst[1] = dst[1]*(1/len)
|
||||
" vmul.f32 q10, q10, q0 \n\t" //dst[2] = dst[2]*(1/len)
|
||||
|
||||
// Store four dst vectors
|
||||
" vst3.32 {d16, d18, d20}, [%2]! \n\t" // store vector 0-1
|
||||
" vst3.32 {d17, d19, d21}, [%2]! \n\t" // store vector 2-3
|
||||
" sub r4, r4, #4 \n\t"
|
||||
"2: \n\t"
|
||||
" cmp r4, #1 \n\t"
|
||||
" ble 1f \n\t"
|
||||
// Load src
|
||||
" vld3.32 {d0,d1,d2}, [%1]! \n\t" // load vector 0-1
|
||||
|
||||
// Load mtx
|
||||
" vld1.32 {d6-d9}, [%0]! \n\t" // load row 0-1
|
||||
" vld1.32 {d10-d11}, [%0] \n\t" // load row 2
|
||||
" sub %0, %0, #32 \n\t"
|
||||
|
||||
// Multiply and add
|
||||
"vmul.f32 d3, d0, d6[0] \n\t" //dst[0] = v[0]*mtx[0][0]
|
||||
"vmul.f32 d4, d0, d8[0] \n\t" //dst[1] = v[0]*mtx[1][0]
|
||||
"vmul.f32 d5, d0, d10[0] \n\t" //dst[2] = v[0]*mtx[2][0]
|
||||
"vmla.f32 d3, d1, d6[1] \n\t" //dst[0] = v[1]*mtx[0][1]
|
||||
"vmla.f32 d4, d1, d8[1] \n\t" //dst[1] += v[1]*mtx[1][1]
|
||||
"vmla.f32 d5, d1, d10[1] \n\t" //dst[2] += v[1]*mtx[2][1]
|
||||
"vmla.f32 d3, d2, d7[0] \n\t" //dst[0] += v[2]*mtx[0][2]
|
||||
"vmla.f32 d4, d2, d9[0] \n\t" //dst[1] += v[2]*mtx[1][2]
|
||||
"vmla.f32 d5, d2, d11[0] \n\t" //dst[2] += v[2]*mtx[2][2]
|
||||
// Multiply and add
|
||||
" vmul.f32 d3, d0, d6[0] \n\t" //dst[0] = v[0]*mtx[0][0]
|
||||
" vmul.f32 d4, d0, d8[0] \n\t" //dst[1] = v[0]*mtx[1][0]
|
||||
" vmul.f32 d5, d0, d10[0] \n\t" //dst[2] = v[0]*mtx[2][0]
|
||||
" vmla.f32 d3, d1, d6[1] \n\t" //dst[0] = v[1]*mtx[0][1]
|
||||
" vmla.f32 d4, d1, d8[1] \n\t" //dst[1] += v[1]*mtx[1][1]
|
||||
" vmla.f32 d5, d1, d10[1] \n\t" //dst[2] += v[1]*mtx[2][1]
|
||||
" vmla.f32 d3, d2, d7[0] \n\t" //dst[0] += v[2]*mtx[0][2]
|
||||
" vmla.f32 d4, d2, d9[0] \n\t" //dst[1] += v[2]*mtx[1][2]
|
||||
" vmla.f32 d5, d2, d11[0] \n\t" //dst[2] += v[2]*mtx[2][2]
|
||||
|
||||
// Normalize 2x
|
||||
"vmul.f32 d0, d3, d3 \n\t" //len = dst[0]*dst[0]
|
||||
"vmla.f32 d0, d4, d4 \n\t" //len += dst[1]*dst[1]
|
||||
"vmla.f32 d0, d5, d5 \n\t" //len += dst[2]*dst[2]
|
||||
// Normalize 2x
|
||||
" vmul.f32 d0, d3, d3 \n\t" //len = dst[0]*dst[0]
|
||||
" vmla.f32 d0, d4, d4 \n\t" //len += dst[1]*dst[1]
|
||||
" vmla.f32 d0, d5, d5 \n\t" //len += dst[2]*dst[2]
|
||||
|
||||
"vmov.f32 d1, d0 \n\t" //d1 = d0
|
||||
"vrsqrte.f32 d0, d0 \n\t" //d0 = ~ 1.0 / sqrt(d0)
|
||||
"vmul.f32 d2, d0, d1 \n\t" //d2 = d0 * d1
|
||||
"vrsqrts.f32 d6, d2, d0 \n\t" //d3 = (3 - d0 * d2) / 2
|
||||
"vmul.f32 d0, d0, d6 \n\t" //d0 = d0 * d3
|
||||
"vmul.f32 d2, d0, d1 \n\t" //d2 = d0 * d1
|
||||
"vrsqrts.f32 d6, d2, d0 \n\t" //d3 = (3 - d0 * d3) / 2
|
||||
"vmul.f32 d0, d0, d6 \n\t" //d0 = d0 * d3
|
||||
" vmov.f32 d1, d0 \n\t" //d1 = d0
|
||||
" vrsqrte.f32 d0, d0 \n\t" //d0 = ~ 1.0 / sqrt(d0)
|
||||
" vmul.f32 d2, d0, d1 \n\t" //d2 = d0 * d1
|
||||
" vrsqrts.f32 d6, d2, d0 \n\t" //d3 = (3 - d0 * d2) / 2
|
||||
" vmul.f32 d0, d0, d6 \n\t" //d0 = d0 * d3
|
||||
" vmul.f32 d2, d0, d1 \n\t" //d2 = d0 * d1
|
||||
" vrsqrts.f32 d6, d2, d0 \n\t" //d3 = (3 - d0 * d3) / 2
|
||||
" vmul.f32 d0, d0, d6 \n\t" //d0 = d0 * d3
|
||||
|
||||
"vmul.f32 d3, d3, d0 \n\t" //dst[0] = dst[0]*(1/len)
|
||||
"vmul.f32 d4, d4, d0 \n\t" //dst[1] = dst[1]*(1/len)
|
||||
"vmul.f32 d5, d5, d0 \n\t" //dst[2] = dst[2]*(1/len)
|
||||
" vmul.f32 d3, d3, d0 \n\t" //dst[0] = dst[0]*(1/len)
|
||||
" vmul.f32 d4, d4, d0 \n\t" //dst[1] = dst[1]*(1/len)
|
||||
" vmul.f32 d5, d5, d0 \n\t" //dst[2] = dst[2]*(1/len)
|
||||
|
||||
// Store two dst vectors
|
||||
"vst3.32 {d3, d4, d5}, [%2] \n\t" // store vector 0-1
|
||||
: "+r"(mtx), "+r"(src), "+r"(dst) :
|
||||
: "d0","d1","d2","d3","d4","d5","d6","d7","d8","d9","d10","d11", "memory"
|
||||
);
|
||||
}
|
||||
// Store two dst vectors
|
||||
" vst3.32 {d3, d4, d5}, [%2]! \n\t" // store vector 0-1
|
||||
" sub r4, r4, #2 \n\t"
|
||||
"1: \n\t"
|
||||
" cmp r4, #1 \n\t"
|
||||
" blt 0f \n\t"
|
||||
" vld1.32 {d0}, [%1]! \n\t" //d0 = src[0] & src[1]
|
||||
" vld1.32 {d1[0]}, [%1]! \n\t" //d1[0] = src[2]
|
||||
|
||||
void InverseTransformVectorNormalize4x(float src[4][3], float dst[4][3], float mtx[4][4] )
|
||||
{
|
||||
asm volatile (
|
||||
// Load four src vectors
|
||||
"vld3.32 {d0,d2,d4}, [%1]! \n\t" // load vector 0-1
|
||||
"vld3.32 {d1,d3,d5}, [%1] \n\t" // load vector 2-3
|
||||
|
||||
// Load mtx
|
||||
"vld1.32 {d6-d9}, [%0]! \n\t" // load row 0-1
|
||||
"vld1.32 {d10-d11}, [%0] \n\t" // load row 2
|
||||
|
||||
// Multiply and add
|
||||
"vmul.f32 q8, q0, d6[0] \n\t" //dst[0] = v[0]*mtx[0][0]
|
||||
"vmul.f32 q9, q0, d8[0] \n\t" //dst[1] = v[0]*mtx[1][0]
|
||||
"vmul.f32 q10, q0, d10[0] \n\t" //dst[2] = v[0]*mtx[2][0]
|
||||
"vmla.f32 q8, q1, d6[1] \n\t" //dst[0] += v[1]*mtx[0][1]
|
||||
"vmla.f32 q9, q1, d8[1] \n\t" //dst[1] += v[1]*mtx[1][1]
|
||||
"vmla.f32 q10, q1, d10[1] \n\t" //dst[2] += v[1]*mtx[2][1]
|
||||
"vmla.f32 q8, q2, d7[0] \n\t" //dst[0] += v[0]*mtx[2][1]
|
||||
"vmla.f32 q9, q2, d9[0] \n\t" //dst[1] += v[2]*mtx[2][1]
|
||||
"vmla.f32 q10, q2, d11[0] \n\t" //dst[2] += v[2]*mtx[2][2]
|
||||
// Load mtx
|
||||
" vld4.32 {d18, d20, d22, d24}, [%0]! \n\t"
|
||||
" vld4.32 {d19, d21, d23, d25}, [%0]! \n\t"
|
||||
" vmul.f32 q2, q9, d0[0] \n\t" //q2 = q9*d0[0]
|
||||
" vmla.f32 q2, q10, d0[1] \n\t" //q2 += q1*d0[1]
|
||||
" vmla.f32 q2, q11, d1[0] \n\t" //q2 += q2*d1[0]*/
|
||||
|
||||
// Normalize 4x
|
||||
"vmul.f32 q0, q8, q8 \n\t" //len = dst[0]*dst[0]
|
||||
"vmla.f32 q0, q9, q9 \n\t" //len += dst[1]*dst[1]
|
||||
"vmla.f32 q0, q10, q10 \n\t" //len += dst[2]*dst[2]
|
||||
" vmul.f32 d0, d4, d4 \n\t" //d0 = d4*d4
|
||||
" vpadd.f32 d0, d0, d0 \n\t" //d0 = d[0] + d[1]
|
||||
" vmla.f32 d0, d5, d5 \n\t" //d0 += d5*d5
|
||||
|
||||
"vmov.f32 q1, q0 \n\t" //d1 = len
|
||||
"vrsqrte.f32 q0, q0 \n\t" //q0 = ~ 1.0 / sqrt(q0)
|
||||
"vmul.f32 q7, q0, q1 \n\t" //q7 = q0 * q1
|
||||
"vrsqrts.f32 q6, q7, q0 \n\t" //q6 = (3 - q0 * q7) / 2
|
||||
"vmul.f32 q0, q0, q6 \n\t" //d0 = d0 * d15
|
||||
"vmul.f32 q7, q0, q1 \n\t" //q7 = q0 * q1
|
||||
"vrsqrts.f32 q6, q7, q0 \n\t" //q6 = (3 - q0 * q7) / 2
|
||||
"vmul.f32 q0, q0, q6 \n\t" //q0 = q0 * q6
|
||||
" vmov.f32 d1, d0 \n\t" //d1 = d0
|
||||
" vrsqrte.f32 d0, d0 \n\t" //d0 = ~ 1.0 / sqrt(d0)
|
||||
" vmul.f32 d2, d0, d1 \n\t" //d2 = d0 * d1
|
||||
" vrsqrts.f32 d3, d2, d0 \n\t" //d3 = (3 - d0 * d2) / 2
|
||||
" vmul.f32 d0, d0, d3 \n\t" //d0 = d0 * d3
|
||||
" vmul.f32 d2, d0, d1 \n\t" //d2 = d0 * d1
|
||||
" vrsqrts.f32 d3, d2, d0 \n\t" //d3 = (3 - d0 * d3) / 2
|
||||
" vmul.f32 d0, d0, d3 \n\t" //d0 = d0 * d3
|
||||
|
||||
"vmul.f32 q8, q8, q0 \n\t" //dst[0] = dst[0]*(1/len)
|
||||
"vmul.f32 q9, q9, q0 \n\t" //dst[1] = dst[1]*(1/len)
|
||||
"vmul.f32 q10, q10, q0 \n\t" //dst[2] = dst[2]*(1/len)
|
||||
" vmul.f32 q2, q2, d0[0] \n\t" //q2 = q2*d0[0]
|
||||
|
||||
// Store four dst vectors
|
||||
"vst3.32 {d16, d18, d20}, [%2]! \n\t" // store vector 0-1
|
||||
"vst3.32 {d17, d19, d21}, [%2] \n\t" // store vector 2-3
|
||||
|
||||
: "+r"(mtx), "+r"(src), "+r"(dst) :
|
||||
" vst1.32 {d4}, [%2]! \n\t" //dst = d4
|
||||
" vst1.32 {d5[0]}, [%2]! \n\t" //dst += d5[0]
|
||||
"0: \n\t"
|
||||
: "+r"(mtx), "+r"(src), "+r"(dst), "+r"(ptr) :
|
||||
: "d0","d1","d2","d3","d4","d5","d6","d7","d8","d9","d10","d11","d12","d13","d14","d15","d16","d17","d18","d19",
|
||||
"d20","d21", "memory"
|
||||
"d20","d21", "d22","d23","d24","d25","d26","d27","d28","d29", "r4", "memory"
|
||||
);
|
||||
}
|
||||
|
|
|
|||
19
src/gSP.cpp
19
src/gSP.cpp
|
|
@ -926,19 +926,8 @@ void gSPLookAt( u32 _l, u32 _n )
|
|||
static
|
||||
void gSPUpdateLightVectors()
|
||||
{
|
||||
s32 count = gSP.numLights-1;
|
||||
while (count >= 3) {
|
||||
InverseTransformVectorNormalize4x(&gSP.lights.xyz[count-3],&gSP.lights.i_xyz[count-3],
|
||||
gSP.matrix.modelView[gSP.matrix.modelViewi]);
|
||||
count -= 4;
|
||||
}
|
||||
if (count >= 1){
|
||||
InverseTransformVectorNormalize2x(&gSP.lights.xyz[count-1], &gSP.lights.i_xyz[count-1],
|
||||
gSP.matrix.modelView[gSP.matrix.modelViewi]);
|
||||
count -= 2;
|
||||
}
|
||||
if (count == 0)
|
||||
InverseTransformVectorNormalize(gSP.lights.xyz[0], gSP.lights.i_xyz[0], gSP.matrix.modelView[gSP.matrix.modelViewi]);
|
||||
InverseTransformVectorNormalizeN(&gSP.lights.xyz[0], &gSP.lights.i_xyz[0],
|
||||
gSP.matrix.modelView[gSP.matrix.modelViewi], gSP.numLights);
|
||||
gSP.changed ^= CHANGED_LIGHT;
|
||||
gSP.changed |= CHANGED_HW_LIGHT;
|
||||
}
|
||||
|
|
@ -947,8 +936,8 @@ static
|
|||
void gSPUpdateLookatVectors()
|
||||
{
|
||||
if (gSP.lookatEnable) {
|
||||
InverseTransformVectorNormalize2x(&gSP.lookat.xyz[0], &gSP.lookat.i_xyz[0],
|
||||
gSP.matrix.modelView[gSP.matrix.modelViewi]);
|
||||
InverseTransformVectorNormalizeN(&gSP.lookat.xyz[0], &gSP.lookat.i_xyz[0],
|
||||
gSP.matrix.modelView[gSP.matrix.modelViewi], 2);
|
||||
}
|
||||
gSP.changed ^= CHANGED_LOOKAT;
|
||||
}
|
||||
|
|
|
|||
Loading…
Reference in New Issue
Block a user