TorqueGameEngines/Torque3D
1
0
Fork 0
mirror of https://github.com/TorqueGameEngines/Torque3D.git synced 2026-03-19 12:20:57 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix batch on neon

Browse source
This commit is contained in:
marauder2k7 2026-03-05 20:04:33 +00:00
parent add7f2a5d7
commit c09d5a4579
4 changed files with 21 additions and 21 deletions
Download patch file Download diff file Expand all files Collapse all files

4
Engine/source/math/impl/mat44_impl.inl
View file

@ -392,9 +392,9 @@ namespace math_backend::mat44
// MATRIX BATCH FUNCTIONS
//--------------------------------------------------
inline void mat44_batch_mul_pos3(const float* m, const float* points, size_t count, float* result)
inline void mat44_batch_mul_pos3(const float* m, const float* points, int count, float* result)
{
size_t i = 0;
int i = 0;
f32x4x4 ma = m_load(m);
// AVX has 8 lanes to play with

6
Engine/source/math/impl/math_c.cpp
View file

@ -488,11 +488,11 @@ namespace math_backend::mat44::dispatch
};
gMat44.batch_mul_pos3 = [](const float* m, const float* pts, size_t count, float* result_ptrs) {
size_t i = 0;
gMat44.batch_mul_pos3 = [](const float* m, const float* pts, int count, float* result_ptrs) {
int i = 0;
for (; i < count; i++)
{
size_t idx = i * 3;
int idx = i * 3;
gMat44.mul_pos3(m, &pts[idx], &result_ptrs[idx]);
}
};

30
Engine/source/math/isa/neon/neon_intrinsics.h
View file

@ -422,9 +422,9 @@ namespace
{
vec4_batch4 r;
r.x = (f32x4){ ptr[9], ptr[6], ptr[3], ptr[0] };
r.y = (f32x4){ ptr[10], ptr[7], ptr[4], ptr[1] };
r.z = (f32x4){ ptr[11], ptr[8], ptr[5], ptr[2] };
r.x = (f32x4){ ptr[0], ptr[3], ptr[6], ptr[9] };
r.y = (f32x4){ ptr[1], ptr[4], ptr[7], ptr[10] };
r.z = (f32x4){ ptr[2], ptr[5], ptr[8], ptr[11] };
if (fillW)
{
@ -456,20 +456,20 @@ namespace
{
vec4_batch4 r;
float32x4_t m00 = vdupq_n_f32(m.r0.m128_f32[0]);
float32x4_t m01 = vdupq_n_f32(m.r0.m128_f32[1]);
float32x4_t m02 = vdupq_n_f32(m.r0.m128_f32[2]);
float32x4_t m03 = vdupq_n_f32(m.r0.m128_f32[3]);
float32x4_t m00 = vdupq_n_f32(m.r0[0]);
float32x4_t m01 = vdupq_n_f32(m.r0[1]);
float32x4_t m02 = vdupq_n_f32(m.r0[2]);
float32x4_t m03 = vdupq_n_f32(m.r0[3]);
float32x4_t m10 = vdupq_n_f32(m.r1.m128_f32[0]);
float32x4_t m11 = vdupq_n_f32(m.r1.m128_f32[1]);
float32x4_t m12 = vdupq_n_f32(m.r1.m128_f32[2]);
float32x4_t m13 = vdupq_n_f32(m.r1.m128_f32[3]);
float32x4_t m10 = vdupq_n_f32(m.r1[0]);
float32x4_t m11 = vdupq_n_f32(m.r1[1]);
float32x4_t m12 = vdupq_n_f32(m.r1[2]);
float32x4_t m13 = vdupq_n_f32(m.r1[3]);
float32x4_t m20 = vdupq_n_f32(m.r2.m128_f32[0]);
float32x4_t m21 = vdupq_n_f32(m.r2.m128_f32[1]);
float32x4_t m22 = vdupq_n_f32(m.r2.m128_f32[2]);
float32x4_t m23 = vdupq_n_f32(m.r2.m128_f32[3]);
float32x4_t m20 = vdupq_n_f32(m.r2[0]);
float32x4_t m21 = vdupq_n_f32(m.r2[1]);
float32x4_t m22 = vdupq_n_f32(m.r2[2]);
float32x4_t m23 = vdupq_n_f32(m.r2[3]);
// row0 dot
r.x = vaddq_f32(

2
Engine/source/math/public/mat44_dispatch.h
View file

@ -21,7 +21,7 @@ namespace math_backend::mat44::dispatch
void (*scale)(float*, const float*) = nullptr;
void (*get_scale)(const float*, float*) = nullptr;
void (*batch_mul_pos3)(const float* m, const float* pts, size_t count, float* result_ptrs) = nullptr;
void (*batch_mul_pos3)(const float* m, const float* pts, int count, float* result_ptrs) = nullptr;
};
// Global dispatch table