actions/itgmania212121
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fast matrix and vector multiply on Mac OS X.

Browse Source
This commit is contained in:
Steve Checkoway
2005-06-21 08:04:42 +00:00
parent 518da6b626
commit 0bb1e19481
1 changed files with 25 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
stepmania/src/RageMath.cpp
+25 -3
View File
@@ -13,6 +13,10 @@
#include "arch/Dialog/Dialog.h"
#include <float.h>
#if defined(DARWIN)
# include <vecLib/vBLAS.h>
#endif
void RageVec3ClearBounds( RageVector3 &mins, RageVector3 &maxs )
{
mins = RageVector3( FLT_MAX, FLT_MAX, FLT_MAX );
@@ -77,6 +81,11 @@ void RageVec3TransformNormal( RageVector3* pOut, const RageVector3* pV, const Ra
void RageVec4TransformCoord( RageVector4* pOut, const RageVector4* pV, const RageMatrix* pM )
{
#if defined(DARWIN)
// (M^t * v)^t = v^t * M
cblas_sgemv(CblasRowMajor, CblasTrans, 4, 4, 1, &pM->m00, 4, &pV->x, 1,
0, &pOut->x, 1);
#else
const RageMatrix &a = *pM;
const RageVector4 &v = *pV;
*pOut = RageVector4(
@@ -84,6 +93,7 @@ void RageVec4TransformCoord( RageVector4* pOut, const RageVector4* pV, const Rag
a.m01*v.x+a.m11*v.y+a.m21*v.z+a.m31*v.w,
a.m02*v.x+a.m12*v.y+a.m22*v.z+a.m32*v.w,
a.m03*v.x+a.m13*v.y+a.m23*v.z+a.m33*v.w );
#endif
}
RageMatrix::RageMatrix( float v00, float v01, float v02, float v03,
@@ -99,11 +109,14 @@ RageMatrix::RageMatrix( float v00, float v01, float v02, float v03,
void RageMatrixIdentity( RageMatrix* pOut )
{
*pOut = RageMatrix(
static float identity[16] = {1,0,0,0,0,1,0,0,0,0,1,0,0,0,0,1};
memcpy(&pOut->m00, identity, sizeof(identity));
/* *pOut = RageMatrix(
1,0,0,0,
0,1,0,0,
0,0,1,0,
0,0,0,1 );
*/
}
RageMatrix RageMatrix::GetTranspose() const
@@ -113,6 +126,10 @@ RageMatrix RageMatrix::GetTranspose() const
void RageMatrixMultiply( RageMatrix* pOut, const RageMatrix* pA, const RageMatrix* pB )
{
#if defined(DARWIN)
cblas_sgemm(CblasRowMajor, CblasNoTrans, CblasNoTrans, 4, 4, 4, 1,
&pB->m00, 4, &pA->m00, 4, 0, &pOut->m00, 4);
#else
//#if defined(_WINDOWS) || defined(_XBOX)
// // <30 cycles for theirs versus >100 for ours.
// D3DXMatrixMultiply( (D3DMATRIX*)pOut, (D3DMATRIX*)pA, (D3DMATRIX*)pB );
@@ -140,6 +157,7 @@ void RageMatrixMultiply( RageMatrix* pOut, const RageMatrix* pA, const RageMatri
);
// phew!
//#endif
#endif
}
void RageMatrixTranslation( RageMatrix* pOut, float x, float y, float z )
@@ -433,7 +451,9 @@ void RageQuatSlerp(RageVector4 *pOut, const RageVector4 &from, const RageVector4
to1[1] = - to.y;
to1[2] = - to.z;
to1[3] = - to.w;
} else {
}
else
{
to1[0] = to.x;
to1[1] = to.y;
to1[2] = to.z;
@@ -449,7 +469,9 @@ void RageQuatSlerp(RageVector4 *pOut, const RageVector4 &from, const RageVector4
float sinom = RageFastSin(omega);
scale0 = RageFastSin((1.0f - t) * omega) / sinom;
scale1 = RageFastSin(t * omega) / sinom;
} else {
}
else
{
// "from" and "to" quaternions are very close
// ... so we can do a linear interpolation
scale0 = 1.0f - t;