[glide64] remove 3dmathSIMD.asm.cpp (from Merge Glide64 changes from the previous attempt)
e110f50489This commit is contained in:
zilmar
parent
1203155d37
commit
d0f45f17ab
|
|
@ -204,71 +204,229 @@ void MulMatricesC(float m1[4][4],float m2[4][4],float r[4][4])
|
|||
|
||||
// 2008.03.29 H.Morii - added SSE 3DNOW! 3x3 1x3 matrix multiplication
|
||||
// and 3DNOW! 4x4 4x4 matrix multiplication
|
||||
// 2011-01-03 Balrog - removed because is in NASM format and not 64-bit compatible
|
||||
// This will need fixing.
|
||||
MULMATRIX MulMatrices = MulMatricesC;
|
||||
TRANSFORMVECTOR TransformVector = TransformVectorC;
|
||||
TRANSFORMVECTOR InverseTransformVector = InverseTransformVectorC;
|
||||
DOTPRODUCT DotProduct = DotProductC;
|
||||
NORMALIZEVECTOR NormalizeVector = NormalizeVectorC;
|
||||
|
||||
extern "C" void TransformVector3DNOW(float *src, float *dst, float mat[4][4]);
|
||||
extern "C" void InverseTransformVector3DNOW(float *src, float *dst, float mat[4][4]);
|
||||
extern "C" void MulMatricesSSE(float m1[4][4],float m2[4][4],float r[4][4]);
|
||||
extern "C" void MulMatrices3DNOW(float m1[4][4],float m2[4][4],float r[4][4]);
|
||||
extern "C" float DotProductSSE3(register float *v1, register float *v2);
|
||||
extern "C" float DotProduct3DNOW(register float *v1, register float *v2);
|
||||
extern "C" void NormalizeVectorSSE(float *v);
|
||||
extern "C" void NormalizeVector3DNOW(float *v);
|
||||
|
||||
extern "C" void DetectSIMD(int function, int * iedx, int * iecx);
|
||||
|
||||
void math_init()
|
||||
void MulMatricesSSE(float m1[4][4],float m2[4][4],float r[4][4])
|
||||
{
|
||||
#ifndef _DEBUG
|
||||
int iecx = 0, iedx = 0;
|
||||
#if defined(__GNUC__) && !defined(NO_ASM) && !defined(NOSSE)
|
||||
/* [row][col]*/
|
||||
typedef float v4sf __attribute__ ((vector_size (16)));
|
||||
v4sf row0 = __builtin_ia32_loadups(m2[0]);
|
||||
v4sf row1 = __builtin_ia32_loadups(m2[1]);
|
||||
v4sf row2 = __builtin_ia32_loadups(m2[2]);
|
||||
v4sf row3 = __builtin_ia32_loadups(m2[3]);
|
||||
|
||||
GLIDE64_TRY
|
||||
for (int i = 0; i < 4; ++i)
|
||||
{
|
||||
DetectSIMD(0x0000001, &iedx, &iecx);
|
||||
v4sf leftrow = __builtin_ia32_loadups(m1[i]);
|
||||
|
||||
// Fill tmp with four copies of leftrow[0]
|
||||
v4sf tmp = leftrow;
|
||||
tmp = _mm_shuffle_ps (tmp, tmp, 0);
|
||||
// Calculate the four first summands
|
||||
v4sf destrow = tmp * row0;
|
||||
|
||||
// Fill tmp with four copies of leftrow[1]
|
||||
tmp = leftrow;
|
||||
tmp = _mm_shuffle_ps (tmp, tmp, 1 + (1 << 2) + (1 << 4) + (1 << 6));
|
||||
destrow += tmp * row1;
|
||||
|
||||
// Fill tmp with four copies of leftrow[2]
|
||||
tmp = leftrow;
|
||||
tmp = _mm_shuffle_ps (tmp, tmp, 2 + (2 << 2) + (2 << 4) + (2 << 6));
|
||||
destrow += tmp * row2;
|
||||
|
||||
// Fill tmp with four copies of leftrow[3]
|
||||
tmp = leftrow;
|
||||
tmp = _mm_shuffle_ps (tmp, tmp, 3 + (3 << 2) + (3 << 4) + (3 << 6));
|
||||
destrow += tmp * row3;
|
||||
|
||||
__builtin_ia32_storeups(r[i], destrow);
|
||||
}
|
||||
GLIDE64_CATCH
|
||||
#elif !defined(NO_ASM) && !defined(NOSSE)
|
||||
__asm
|
||||
{
|
||||
return;
|
||||
mov eax, dword ptr [r]
|
||||
mov ecx, dword ptr [m1]
|
||||
mov edx, dword ptr [m2]
|
||||
|
||||
movaps xmm0,[edx]
|
||||
movaps xmm1,[edx+16]
|
||||
movaps xmm2,[edx+32]
|
||||
movaps xmm3,[edx+48]
|
||||
|
||||
// r[0][0],r[0][1],r[0][2],r[0][3]
|
||||
|
||||
movaps xmm4,xmmword ptr[ecx]
|
||||
movaps xmm5,xmm4
|
||||
movaps xmm6,xmm4
|
||||
movaps xmm7,xmm4
|
||||
|
||||
shufps xmm4,xmm4,00000000b
|
||||
shufps xmm5,xmm5,01010101b
|
||||
shufps xmm6,xmm6,10101010b
|
||||
shufps xmm7,xmm7,11111111b
|
||||
|
||||
mulps xmm4,xmm0
|
||||
mulps xmm5,xmm1
|
||||
mulps xmm6,xmm2
|
||||
mulps xmm7,xmm3
|
||||
|
||||
addps xmm4,xmm5
|
||||
addps xmm4,xmm6
|
||||
addps xmm4,xmm7
|
||||
|
||||
movaps xmmword ptr[eax],xmm4
|
||||
|
||||
// r[1][0],r[1][1],r[1][2],r[1][3]
|
||||
|
||||
movaps xmm4,xmmword ptr[ecx+16]
|
||||
movaps xmm5,xmm4
|
||||
movaps xmm6,xmm4
|
||||
movaps xmm7,xmm4
|
||||
|
||||
shufps xmm4,xmm4,00000000b
|
||||
shufps xmm5,xmm5,01010101b
|
||||
shufps xmm6,xmm6,10101010b
|
||||
shufps xmm7,xmm7,11111111b
|
||||
|
||||
mulps xmm4,xmm0
|
||||
mulps xmm5,xmm1
|
||||
mulps xmm6,xmm2
|
||||
mulps xmm7,xmm3
|
||||
|
||||
addps xmm4,xmm5
|
||||
addps xmm4,xmm6
|
||||
addps xmm4,xmm7
|
||||
|
||||
movaps xmmword ptr[eax+16],xmm4
|
||||
|
||||
|
||||
// r[2][0],r[2][1],r[2][2],r[2][3]
|
||||
|
||||
movaps xmm4,xmmword ptr[ecx+32]
|
||||
movaps xmm5,xmm4
|
||||
movaps xmm6,xmm4
|
||||
movaps xmm7,xmm4
|
||||
|
||||
shufps xmm4,xmm4,00000000b
|
||||
shufps xmm5,xmm5,01010101b
|
||||
shufps xmm6,xmm6,10101010b
|
||||
shufps xmm7,xmm7,11111111b
|
||||
|
||||
mulps xmm4,xmm0
|
||||
mulps xmm5,xmm1
|
||||
mulps xmm6,xmm2
|
||||
mulps xmm7,xmm3
|
||||
|
||||
addps xmm4,xmm5
|
||||
addps xmm4,xmm6
|
||||
addps xmm4,xmm7
|
||||
|
||||
movaps xmmword ptr[eax+32],xmm4
|
||||
|
||||
// r[3][0],r[3][1],r[3][2],r[3][3]
|
||||
|
||||
movaps xmm4,xmmword ptr[ecx+48]
|
||||
movaps xmm5,xmm4
|
||||
movaps xmm6,xmm4
|
||||
movaps xmm7,xmm4
|
||||
|
||||
shufps xmm4,xmm4,00000000b
|
||||
shufps xmm5,xmm5,01010101b
|
||||
shufps xmm6,xmm6,10101010b
|
||||
shufps xmm7,xmm7,11111111b
|
||||
|
||||
mulps xmm4,xmm0
|
||||
mulps xmm5,xmm1
|
||||
mulps xmm6,xmm2
|
||||
mulps xmm7,xmm3
|
||||
|
||||
addps xmm4,xmm5
|
||||
addps xmm4,xmm6
|
||||
addps xmm4,xmm7
|
||||
|
||||
movaps xmmword ptr[eax+48],xmm4
|
||||
}
|
||||
#endif // _WIN32
|
||||
}
|
||||
if (iedx & 0x2000000) //SSE
|
||||
|
||||
|
||||
|
||||
void math_init()
|
||||
{
|
||||
MulMatrices = MulMatricesSSE;
|
||||
//InverseTransformVector = InverseTransformVectorSSE;
|
||||
//NormalizeVector = NormalizeVectorSSE; /* not ready yet */
|
||||
LOG("SSE detected.\n");
|
||||
}
|
||||
if (iedx & 0x4000000) // SSE2
|
||||
{
|
||||
LOG("SSE2 detected.\n");
|
||||
}
|
||||
if (iecx & 0x1) // SSE3
|
||||
{
|
||||
//DotProduct = DotProductSSE3; /* not ready yet */
|
||||
LOG("SSE3 detected.\n");
|
||||
}
|
||||
// the 3dnow version is faster than sse
|
||||
iecx = 0;
|
||||
iedx = 0;
|
||||
GLIDE64_TRY
|
||||
{
|
||||
DetectSIMD(0x80000001, &iedx, &iecx);
|
||||
}
|
||||
GLIDE64_CATCH
|
||||
{
|
||||
return;
|
||||
}
|
||||
if (iedx & 0x80000000) //3DNow!
|
||||
{
|
||||
MulMatrices = MulMatrices3DNOW;
|
||||
TransformVector = TransformVector3DNOW;
|
||||
InverseTransformVector = InverseTransformVector3DNOW;
|
||||
//DotProduct = DotProduct3DNOW; //not ready yet
|
||||
NormalizeVector = NormalizeVector3DNOW; // not ready yet
|
||||
LOG("3DNOW! detected.\n");
|
||||
}
|
||||
#ifndef _DEBUG
|
||||
int IsSSE = FALSE;
|
||||
int edx, eax;
|
||||
#if defined(__GNUC__) && !defined(NO_ASM) && !defined(NOSSE)
|
||||
GLIDE64_TRY
|
||||
{
|
||||
#if defined(__x86_64__)
|
||||
asm volatile(" cpuid; "
|
||||
: "=a"(eax), "=d"(edx)
|
||||
: "0"(1)
|
||||
: "rbx", "rcx"
|
||||
);
|
||||
#else
|
||||
asm volatile(" push %%ebx; "
|
||||
" push %%ecx; "
|
||||
" cpuid; "
|
||||
" pop %%ecx; "
|
||||
" pop %%ebx; "
|
||||
: "=a"(eax), "=d"(edx)
|
||||
: "0"(1)
|
||||
:
|
||||
);
|
||||
#endif
|
||||
}
|
||||
GLIDE64_CATCH
|
||||
{ return; }
|
||||
// Check for SSE
|
||||
if (edx & (1 << 25))
|
||||
IsSSE = TRUE;
|
||||
#elif !defined(NO_ASM) && !defined(NOSSE)
|
||||
DWORD dwEdx;
|
||||
__try
|
||||
{
|
||||
__asm
|
||||
{
|
||||
mov eax,1
|
||||
cpuid
|
||||
mov dwEdx,edx
|
||||
}
|
||||
}
|
||||
__except(EXCEPTION_EXECUTE_HANDLER)
|
||||
{
|
||||
return;
|
||||
}
|
||||
|
||||
if (dwEdx & (1<<25))
|
||||
{
|
||||
if (dwEdx & (1<<24))
|
||||
{
|
||||
__try
|
||||
{
|
||||
__asm xorps xmm0, xmm0
|
||||
IsSSE = TRUE;
|
||||
}
|
||||
__except(EXCEPTION_EXECUTE_HANDLER)
|
||||
{
|
||||
return;
|
||||
}
|
||||
}
|
||||
}
|
||||
#endif // _WIN32
|
||||
if (IsSSE)
|
||||
{
|
||||
MulMatrices = MulMatricesSSE;
|
||||
LOG("3DNOW! detected.\n");
|
||||
}
|
||||
|
||||
#endif //_DEBUG
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -1,492 +0,0 @@
|
|||
;/*
|
||||
;* Glide64 - Glide video plugin for Nintendo 64 emulators.
|
||||
;*
|
||||
;* This program is free software; you can redistribute it and/or modify
|
||||
;* it under the terms of the GNU General Public License as published by
|
||||
;* the Free Software Foundation; either version 2 of the License, or
|
||||
;* any later version.
|
||||
;*
|
||||
;* This program is distributed in the hope that it will be useful,
|
||||
;* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
||||
;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
||||
;* GNU General Public License for more details.
|
||||
;*
|
||||
;* You should have received a copy of the GNU General Public License
|
||||
;* along with this program; if not, write to the Free Software
|
||||
;* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
||||
;
|
||||
;****************************************************************
|
||||
;
|
||||
; Glide64 - Glide Plugin for Nintendo 64 emulators
|
||||
; Project started on December 29th, 2001
|
||||
;
|
||||
; Authors:
|
||||
; Dave2001, original author, founded the project in 2001, left it in 2002
|
||||
; Gugaman, joined the project in 2002, left it in 2002
|
||||
; Sergey 'Gonetz' Lipski, joined the project in 2002, main author since fall of 2002
|
||||
; Hiroshi 'KoolSmoky' Morii, joined the project in 2007
|
||||
;
|
||||
;****************************************************************
|
||||
;
|
||||
; To modify Glide64:
|
||||
; * Write your name and (optional)email, commented by your work, so I know who did it, and so that you can find which parts you modified when it comes time to send it to me.
|
||||
; * Do NOT send me the whole project or file that you modified. Take out your modified code sections, and tell me where to put them. If people sent the whole thing, I would have many different versions, but no idea how to combine them all.
|
||||
;
|
||||
;****************************************************************
|
||||
|
||||
%include "inc/c32.mac"
|
||||
|
||||
segment .text
|
||||
*/
|
||||
|
||||
extern "C" void __declspec(naked) DetectSIMD(int func, int * iedx, int * iecx)
|
||||
{
|
||||
_asm {
|
||||
push ebp
|
||||
mov ebp,esp
|
||||
mov eax,[func]
|
||||
cpuid
|
||||
mov eax,[iedx]
|
||||
mov [eax],edx
|
||||
mov eax,[iecx]
|
||||
mov [eax],ecx
|
||||
leave
|
||||
ret
|
||||
}
|
||||
}
|
||||
|
||||
/****************************************************************
|
||||
;
|
||||
; ******** SSE ********
|
||||
;
|
||||
;****************************************************************/
|
||||
|
||||
extern "C" void __declspec(naked) MulMatricesSSE(float m1[4][4],float m2[4][4],float r[4][4])
|
||||
{
|
||||
__asm
|
||||
{
|
||||
push ebp
|
||||
mov ebp,esp
|
||||
|
||||
mov eax,[r]
|
||||
mov ecx,[m1]
|
||||
mov edx,[m2]
|
||||
|
||||
movaps xmm0,[edx]
|
||||
movaps xmm1,[edx+16]
|
||||
movaps xmm2,[edx+32]
|
||||
movaps xmm3,[edx+48]
|
||||
|
||||
; r[0][0],r[0][1],r[0][2],r[0][3]
|
||||
|
||||
movaps xmm4,[ecx]
|
||||
movaps xmm5,xmm4
|
||||
movaps xmm6,xmm4
|
||||
movaps xmm7,xmm4
|
||||
|
||||
shufps xmm4,xmm4,00000000b
|
||||
shufps xmm5,xmm5,01010101b
|
||||
shufps xmm6,xmm6,10101010b
|
||||
shufps xmm7,xmm7,11111111b
|
||||
|
||||
mulps xmm4,xmm0
|
||||
mulps xmm5,xmm1
|
||||
mulps xmm6,xmm2
|
||||
mulps xmm7,xmm3
|
||||
|
||||
addps xmm4,xmm5
|
||||
addps xmm4,xmm6
|
||||
addps xmm4,xmm7
|
||||
|
||||
movaps [eax],xmm4
|
||||
|
||||
; r[1][0],r[1][1],r[1][2],r[1][3]
|
||||
|
||||
movaps xmm4,[ecx+16]
|
||||
movaps xmm5,xmm4
|
||||
movaps xmm6,xmm4
|
||||
movaps xmm7,xmm4
|
||||
|
||||
shufps xmm4,xmm4,00000000b
|
||||
shufps xmm5,xmm5,01010101b
|
||||
shufps xmm6,xmm6,10101010b
|
||||
shufps xmm7,xmm7,11111111b
|
||||
|
||||
mulps xmm4,xmm0
|
||||
mulps xmm5,xmm1
|
||||
mulps xmm6,xmm2
|
||||
mulps xmm7,xmm3
|
||||
|
||||
addps xmm4,xmm5
|
||||
addps xmm4,xmm6
|
||||
addps xmm4,xmm7
|
||||
|
||||
movaps [eax+16],xmm4
|
||||
|
||||
|
||||
; r[2][0],r[2][1],r[2][2],r[2][3]
|
||||
|
||||
movaps xmm4,[ecx+32]
|
||||
movaps xmm5,xmm4
|
||||
movaps xmm6,xmm4
|
||||
movaps xmm7,xmm4
|
||||
|
||||
shufps xmm4,xmm4,00000000b
|
||||
shufps xmm5,xmm5,01010101b
|
||||
shufps xmm6,xmm6,10101010b
|
||||
shufps xmm7,xmm7,11111111b
|
||||
|
||||
mulps xmm4,xmm0
|
||||
mulps xmm5,xmm1
|
||||
mulps xmm6,xmm2
|
||||
mulps xmm7,xmm3
|
||||
|
||||
addps xmm4,xmm5
|
||||
addps xmm4,xmm6
|
||||
addps xmm4,xmm7
|
||||
|
||||
movaps [eax+32],xmm4
|
||||
|
||||
; r[3][0],r[3][1],r[3][2],r[3][3]
|
||||
|
||||
movaps xmm4,[ecx+48]
|
||||
movaps xmm5,xmm4
|
||||
movaps xmm6,xmm4
|
||||
movaps xmm7,xmm4
|
||||
|
||||
shufps xmm4,xmm4,00000000b
|
||||
shufps xmm5,xmm5,01010101b
|
||||
shufps xmm6,xmm6,10101010b
|
||||
shufps xmm7,xmm7,11111111b
|
||||
|
||||
mulps xmm4,xmm0
|
||||
mulps xmm5,xmm1
|
||||
mulps xmm6,xmm2
|
||||
mulps xmm7,xmm3
|
||||
|
||||
addps xmm4,xmm5
|
||||
addps xmm4,xmm6
|
||||
addps xmm4,xmm7
|
||||
|
||||
movaps [eax+48],xmm4
|
||||
leave
|
||||
ret
|
||||
}
|
||||
}
|
||||
|
||||
extern "C" void __declspec(naked) NormalizeVectorSSE (float *v)
|
||||
{
|
||||
_asm
|
||||
{
|
||||
push ebp
|
||||
mov ebp,esp
|
||||
|
||||
mov edx, [v]
|
||||
movaps xmm0, [edx] ; x y z 0
|
||||
movaps xmm2, xmm0 ; x y z 0
|
||||
mulps xmm0, xmm0 ; x*x y*y z*z 0
|
||||
movaps xmm1, xmm0 ; x*x y*y z*z 0
|
||||
shufps xmm0, xmm1, 0x4e ; z*z 0 x*x y*y
|
||||
addps xmm0, xmm1 ; x*x+z*z y*y z*z+x*x y*y
|
||||
movaps xmm1, xmm0 ; x*x+z*z y*y z*z+x*x y*y
|
||||
shufps xmm1, xmm1, 0x11 ; y*y z*z+x*x y*y z*z+x*x
|
||||
addps xmm0, xmm1 ; x*x+z*z+y*y
|
||||
rsqrtps xmm0, xmm0 ; 1.0/sqrt(x*x+z*z+y*y)
|
||||
mulps xmm2, xmm0 ; x/sqrt(x*x+z*z+y*y) y/sqrt(x*x+z*z+y*y) z/sqrt(x*x+z*z+y*y) 0
|
||||
movaps [edx], xmm2
|
||||
leave
|
||||
ret
|
||||
}
|
||||
}
|
||||
|
||||
/*****************************************************************
|
||||
;
|
||||
; ******** SSE3 ********
|
||||
;
|
||||
;****************************************************************/
|
||||
|
||||
float __declspec(naked) DotProductSSE3(register float *v1, register float *v2)
|
||||
{
|
||||
_asm {
|
||||
push ebp
|
||||
mov ebp,esp
|
||||
|
||||
mov eax,[v1]
|
||||
mov edx,[v2]
|
||||
movaps xmm0, [eax]
|
||||
mulps xmm0, [edx]
|
||||
haddps xmm0, xmm0
|
||||
haddps xmm0, xmm0
|
||||
; movss eax, xmm0
|
||||
leave
|
||||
ret
|
||||
}
|
||||
}
|
||||
|
||||
/****************************************************************
|
||||
;
|
||||
; ******** 3DNOW ********
|
||||
;
|
||||
;****************************************************************/
|
||||
|
||||
extern "C" void __declspec(naked) TransformVector3DNOW(float *src, float *dst, float mat[4][4])
|
||||
{
|
||||
_asm {
|
||||
push ebp
|
||||
mov ebp,esp
|
||||
|
||||
femms
|
||||
mov ecx,[src]
|
||||
mov eax,[dst]
|
||||
mov edx,[mat]
|
||||
movq mm0,[ecx] ; src[1] src[0]
|
||||
movd mm2,[ecx+8] ; 0 src[2]
|
||||
movq mm1,mm0 ; src[1] src[0]
|
||||
punpckldq mm0,mm0 ; src[0] src[0]
|
||||
punpckhdq mm1,mm1 ; src[1] src[1]
|
||||
punpckldq mm2,mm2 ; src[2] src[2]
|
||||
movq mm3,mm0 ; src[0] src[0]
|
||||
movq mm4,mm1 ; src[1] src[1]
|
||||
movq mm5,mm2 ; src[2] src[2]
|
||||
pfmul mm0,[edx] ; src[0]*mat[0][1] src[0]*mat[0][0]
|
||||
pfmul mm3,[edx+8] ; 0 src[0]*mat[0][2]
|
||||
pfmul mm1,[edx+16] ; src[1]*mat[1][1] src[1]*mat[1][0]
|
||||
pfmul mm4,[edx+24] ; 0 src[1]*mat[1][2]
|
||||
pfmul mm2,[edx+32] ; src[2]*mat[2][1] src[2]*mat[2][0]
|
||||
pfmul mm5,[edx+40] ; 0 src[2]*mat[2][2]
|
||||
pfadd mm0,mm1 ; src[0]*mat[0][1]+src[1]*mat[1][1] src[0]*mat[0][0]+src[1]*mat[1][0]
|
||||
pfadd mm3,mm4 ; 0 src[0]*mat[0][2]+src[1]*mat[1][2]
|
||||
pfadd mm0,mm2 ; src[0]*mat[0][1]+src[1]*mat[1][1]+src[2]*mat[2][1] src[0]*mat[0][0]+src[1]*mat[1][0]+src[2]*mat[2][0]
|
||||
pfadd mm3,mm5 ; 0 src[0]*mat[0][2]+src[1]*mat[1][2]+src[2]*mat[2][2]
|
||||
movq [eax],mm0 ; mat[0][1]*src[0]+mat[1][1]*src[1]+mat[2][1]*src[2] mat[0][0]*src[0]+mat[1][0]*src[1]+mat[2][0]*src[2]
|
||||
movd [eax+8],mm3 ; mat[0][2]*src[0]+mat[1][2]*src[1]+mat[2][2]*src[2]
|
||||
femms
|
||||
leave
|
||||
ret
|
||||
}
|
||||
}
|
||||
|
||||
extern "C" void __declspec(naked) InverseTransformVector3DNOW(float *src, float *dst, float mat[4][4])
|
||||
{
|
||||
_asm {
|
||||
push ebp
|
||||
mov ebp,esp
|
||||
|
||||
femms
|
||||
mov ecx,[src]
|
||||
mov eax,[dst]
|
||||
mov edx,[mat]
|
||||
movq mm0,[ecx] ; src[1] src[0]
|
||||
movd mm4,[ecx+8] ; 0 src[2]
|
||||
movq mm1,mm0 ; src[1] src[0]
|
||||
pfmul mm0,[edx] ; src[1]*mat[0][1] src[0]*mat[0][0]
|
||||
movq mm5,mm4 ; 0 src[2]
|
||||
pfmul mm4,[edx+8] ; 0 src[2]*mat[0][2]
|
||||
movq mm2,mm1 ; src[1] src[0]
|
||||
pfmul mm1,[edx+16] ; src[1]*mat[1][1] src[0]*mat[1][0]
|
||||
movq mm6,mm5 ; 0 src[2]
|
||||
pfmul mm5,[edx+24] ; 0 src[2]*mat[1][2]
|
||||
movq mm3,mm2 ; src[1] src[0]
|
||||
pfmul mm2,[edx+32] ; src[1]*mat[2][1] src[0]*mat[2][0]
|
||||
movq mm7,mm6 ; 0 src[2]
|
||||
pfmul mm6,[edx+40] ; 0 src[2]*mat[2][2]
|
||||
pfacc mm0,mm4 ; src[2]*mat[0][2] src[1]*mat[0][1]+src[0]*mat[0][0]
|
||||
pfacc mm1,mm5 ; src[2]*mat[1][2] src[1]*mat[1][1]+src[0]*mat[1][0]
|
||||
pfacc mm2,mm6 ; src[2]*mat[2][2] src[1]*mat[2][1]+src[0]*mat[2][0]
|
||||
pfacc mm0,mm1 ; src[2]*mat[1][2]+src[1]*mat[1][1]+src[0]*mat[1][0] src[2]*mat[0][2]+src[1]*mat[0][1]+src[0]*mat[0][0]
|
||||
pfacc mm2,mm3 ; 0 src[2]*mat[2][2]+src[1]*mat[2][1]+src[0]*mat[2][0]
|
||||
movq [eax],mm0 ; mat[1][0]*src[0]+mat[1][1]*src[1]+mat[1][2]*src[2] mat[0][0]*src[0]+mat[0][1]*src[1]+mat[0][2]*src[2]
|
||||
movd [eax+8],mm2 ; mat[2][0]*src[0]+mat[2][1]*src[1]+mat[2][2]*src[2]
|
||||
femms
|
||||
leave
|
||||
ret
|
||||
}
|
||||
}
|
||||
|
||||
extern "C" void __declspec(naked) MulMatrices3DNOW(float m1[4][4],float m2[4][4],float r[4][4])
|
||||
{
|
||||
_asm {
|
||||
push ebp
|
||||
mov ebp,esp
|
||||
|
||||
femms
|
||||
mov ecx,[m1]
|
||||
mov eax,[r]
|
||||
mov edx,[m2]
|
||||
|
||||
movq mm0,[ecx]
|
||||
movq mm1,[ecx+8]
|
||||
movq mm4,[edx]
|
||||
punpckhdq mm2,mm0
|
||||
movq mm5,[edx+16]
|
||||
punpckhdq mm3,mm1
|
||||
movq mm6,[edx+32]
|
||||
punpckldq mm0,mm0
|
||||
punpckldq mm1,mm1
|
||||
pfmul mm4,mm0
|
||||
punpckhdq mm2,mm2
|
||||
pfmul mm0,[edx+8]
|
||||
movq mm7,[edx+48]
|
||||
pfmul mm5,mm2
|
||||
punpckhdq mm3,mm3
|
||||
pfmul mm2,[edx+24]
|
||||
pfmul mm6,mm1
|
||||
pfadd mm5,mm4
|
||||
pfmul mm1,[edx+40]
|
||||
pfadd mm2,mm0
|
||||
pfmul mm7,mm3
|
||||
pfadd mm6,mm5
|
||||
pfmul mm3,[edx+56]
|
||||
pfadd mm2,mm1
|
||||
pfadd mm7,mm6
|
||||
movq mm0,[ecx+16]
|
||||
pfadd mm3,mm2
|
||||
movq mm1,[ecx+24]
|
||||
movq [eax],mm7
|
||||
movq mm4,[edx]
|
||||
movq [eax+8],mm3
|
||||
|
||||
punpckhdq mm2,mm0
|
||||
movq mm5,[edx+16]
|
||||
punpckhdq mm3,mm1
|
||||
movq mm6,[edx+32]
|
||||
punpckldq mm0,mm0
|
||||
punpckldq mm1,mm1
|
||||
pfmul mm4,mm0
|
||||
punpckhdq mm2,mm2
|
||||
pfmul mm0,[edx+8]
|
||||
movq mm7,[edx+48]
|
||||
pfmul mm5,mm2
|
||||
punpckhdq mm3,mm3
|
||||
pfmul mm2,[edx+24]
|
||||
pfmul mm6,mm1
|
||||
pfadd mm5,mm4
|
||||
pfmul mm1,[edx+40]
|
||||
pfadd mm2,mm0
|
||||
pfmul mm7,mm3
|
||||
pfadd mm6,mm5
|
||||
pfmul mm3,[edx+56]
|
||||
pfadd mm2,mm1
|
||||
pfadd mm7,mm6
|
||||
movq mm0,[ecx+32]
|
||||
pfadd mm3,mm2
|
||||
movq mm1,[ecx+40]
|
||||
movq [eax+16],mm7
|
||||
movq mm4,[edx]
|
||||
movq [eax+24],mm3
|
||||
|
||||
punpckhdq mm2,mm0
|
||||
movq mm5,[edx+16]
|
||||
punpckhdq mm3,mm1
|
||||
movq mm6,[edx+32]
|
||||
punpckldq mm0,mm0
|
||||
punpckldq mm1,mm1
|
||||
pfmul mm4,mm0
|
||||
punpckhdq mm2,mm2
|
||||
pfmul mm0,[edx+8]
|
||||
movq mm7,[edx+48]
|
||||
pfmul mm5,mm2
|
||||
punpckhdq mm3,mm3
|
||||
pfmul mm2,[edx+24]
|
||||
pfmul mm6,mm1
|
||||
pfadd mm5,mm4
|
||||
pfmul mm1,[edx+40]
|
||||
pfadd mm2,mm0
|
||||
pfmul mm7,mm3
|
||||
pfadd mm6,mm5
|
||||
pfmul mm3,[edx+56]
|
||||
pfadd mm2,mm1
|
||||
pfadd mm7,mm6
|
||||
movq mm0,[ecx+48]
|
||||
pfadd mm3,mm2
|
||||
movq mm1,[ecx+56]
|
||||
movq [eax+32],mm7
|
||||
movq mm4,[edx]
|
||||
movq [eax+40],mm3
|
||||
|
||||
punpckhdq mm2,mm0
|
||||
movq mm5,[edx+16]
|
||||
punpckhdq mm3,mm1
|
||||
movq mm6,[edx+32]
|
||||
punpckldq mm0,mm0
|
||||
punpckldq mm1,mm1
|
||||
pfmul mm4,mm0
|
||||
punpckhdq mm2,mm2
|
||||
pfmul mm0,[edx+8]
|
||||
movq mm7,[edx+48]
|
||||
pfmul mm5,mm2
|
||||
punpckhdq mm3,mm3
|
||||
pfmul mm2,[edx+24]
|
||||
pfmul mm6,mm1
|
||||
pfadd mm5,mm4
|
||||
pfmul mm1,[edx+40]
|
||||
pfadd mm2,mm0
|
||||
pfmul mm7,mm3
|
||||
pfadd mm6,mm5
|
||||
pfmul mm3,[edx+56]
|
||||
pfadd mm2,mm1
|
||||
pfadd mm7,mm6
|
||||
pfadd mm3,mm2
|
||||
movq [eax+48],mm7
|
||||
movq [eax+56],mm3
|
||||
femms
|
||||
leave
|
||||
ret
|
||||
}
|
||||
}
|
||||
|
||||
extern "C" float __declspec(naked) DotProduct3DNOW(register float *v1, register float *v2)
|
||||
{
|
||||
_asm {
|
||||
push ebp
|
||||
mov ebp,esp
|
||||
|
||||
femms
|
||||
mov edx,[v1]
|
||||
mov eax,[v2]
|
||||
movq mm0,[edx]
|
||||
movq mm3,[eax]
|
||||
pfmul mm0,mm3
|
||||
movq mm2,[edx+8]
|
||||
movq mm1,[eax+8]
|
||||
pfacc mm0,mm0
|
||||
pfmul mm1,mm2
|
||||
pfadd mm0,mm1
|
||||
movd eax,mm0
|
||||
femms
|
||||
leave
|
||||
ret
|
||||
}
|
||||
}
|
||||
|
||||
extern "C" void __declspec(naked) NormalizeVector3DNOW(float *v)
|
||||
{
|
||||
_asm {
|
||||
push ebp
|
||||
mov ebp,esp
|
||||
femms
|
||||
mov edx,[v]
|
||||
movq mm0,[edx]
|
||||
movq mm3,[edx+8]
|
||||
movq mm1,mm0
|
||||
movq mm2,mm3
|
||||
pfmul mm0,mm0
|
||||
pfmul mm3,mm3
|
||||
pfacc mm0,mm0
|
||||
pfadd mm0,mm3
|
||||
;movq mm4,mm0 ; prepare for 24bit precision
|
||||
;punpckldq mm4,mm4 ; prepare for 24bit precision
|
||||
pfrsqrt mm0,mm0 ; 15bit precision 1/sqrtf(v)
|
||||
;movq mm3,mm0
|
||||
;pfmul mm0,mm0
|
||||
;pfrsqit1 mm0,mm4
|
||||
;pfrcpit2 mm0,mm3 ; 24bit precision 1/sqrtf(v)
|
||||
pfmul mm1,mm0
|
||||
pfmul mm2,mm0
|
||||
movq [edx],mm1
|
||||
movq [edx+8],mm2
|
||||
femms
|
||||
leave
|
||||
ret
|
||||
}
|
||||
}
|
||||
|
|
@ -356,10 +356,6 @@
|
|||
RelativePath="3dmath.h"
|
||||
>
|
||||
</File>
|
||||
<File
|
||||
RelativePath=".\3dmathSIMD.asm.cpp"
|
||||
>
|
||||
</File>
|
||||
<File
|
||||
RelativePath="Combine.cpp"
|
||||
>
|
||||
|
|
|
|||
Loading…
Reference in New Issue