Torque2D/engine/source/math/mMathSSE.cc

//-----------------------------------------------------------------------------
// Copyright (c) 2013 GarageGames, LLC
//
// Permission is hereby granted, free of charge, to any person obtaining a copy
// of this software and associated documentation files (the "Software"), to
// deal in the Software without restriction, including without limitation the
// rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
// sell copies of the Software, and to permit persons to whom the Software is
// furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included in
// all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
// FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
// IN THE SOFTWARE.
//-----------------------------------------------------------------------------

#include "math/mMathFn.h"
#include "math/mPlane.h"
#include "math/mMatrix.h"


#if defined(TORQUE_SUPPORTS_VC_INLINE_X86_ASM)
#define ADD_SSE_FN
// inlined version here.
void SSE_MatrixF_x_MatrixF(const F32 *matA, const F32 *matB, F32 *result)
{
   __asm
   {
      mov         edx, matA
      mov         ecx, matB
      mov         eax, result

      movss       xmm0, [edx]
      movups      xmm1, [ecx]
      shufps      xmm0, xmm0, 0
      movss       xmm2, [edx + 4]
      mulps       xmm0, xmm1
      shufps      xmm2, xmm2, 0
      movups      xmm3, [ecx + 10h]
      movss       xmm7, [edx + 8]
      mulps       xmm2, xmm3
      shufps      xmm7, xmm7, 0
      addps       xmm0, xmm2
      movups      xmm4, [ecx + 20h]
      movss       xmm2, [edx + 0Ch]
      mulps       xmm7, xmm4
      shufps      xmm2, xmm2, 0
      addps       xmm0, xmm7
      movups      xmm5, [ecx + 30h]
      movss       xmm6, [edx + 10h]
      mulps       xmm2, xmm5
      movss       xmm7, [edx + 14h]
      shufps      xmm6, xmm6, 0
      addps       xmm0, xmm2
      shufps      xmm7, xmm7, 0
      movlps[eax], xmm0
      movhps[eax + 8], xmm0
      mulps       xmm7, xmm3
      movss       xmm0, [edx + 18h]
      mulps       xmm6, xmm1
      shufps      xmm0, xmm0, 0
      addps       xmm6, xmm7
      mulps       xmm0, xmm4
      movss       xmm2, [edx + 24h]
      addps       xmm6, xmm0
      movss       xmm0, [edx + 1Ch]
      movss       xmm7, [edx + 20h]
      shufps      xmm0, xmm0, 0
      shufps      xmm7, xmm7, 0
      mulps       xmm0, xmm5
      mulps       xmm7, xmm1
      addps       xmm6, xmm0
      shufps      xmm2, xmm2, 0
      movlps[eax + 10h], xmm6
      movhps[eax + 18h], xmm6
      mulps       xmm2, xmm3
      movss       xmm6, [edx + 28h]
      addps       xmm7, xmm2
      shufps      xmm6, xmm6, 0
      movss       xmm2, [edx + 2Ch]
      mulps       xmm6, xmm4
      shufps      xmm2, xmm2, 0
      addps       xmm7, xmm6
      mulps       xmm2, xmm5
      movss       xmm0, [edx + 34h]
      addps       xmm7, xmm2
      shufps      xmm0, xmm0, 0
      movlps[eax + 20h], xmm7
      movss       xmm2, [edx + 30h]
      movhps[eax + 28h], xmm7
      mulps       xmm0, xmm3
      shufps      xmm2, xmm2, 0
      movss       xmm6, [edx + 38h]
      mulps       xmm2, xmm1
      shufps      xmm6, xmm6, 0
      addps       xmm2, xmm0
      mulps       xmm6, xmm4
      movss       xmm7, [edx + 3Ch]
      shufps      xmm7, xmm7, 0
      addps       xmm2, xmm6
      mulps       xmm7, xmm5
      addps       xmm2, xmm7
      movups[eax + 30h], xmm2
   }
}
void SSE_MatrixF_x_MatrixF_Aligned(const F32 *matA, const F32 *matB, F32 *result)
{
   __asm
   {
      mov         edx, matA
      mov         ecx, matB
      mov         eax, result

      movss       xmm0, [edx]
      movaps      xmm1, [ecx]
      shufps      xmm0, xmm0, 0
      movss       xmm2, [edx + 4]
      mulps       xmm0, xmm1
      shufps      xmm2, xmm2, 0
      movaps      xmm3, [ecx + 10h]
      movss       xmm7, [edx + 8]
      mulps       xmm2, xmm3
      shufps      xmm7, xmm7, 0
      addps       xmm0, xmm2
      movaps      xmm4, [ecx + 20h]
      movss       xmm2, [edx + 0Ch]
      mulps       xmm7, xmm4
      shufps      xmm2, xmm2, 0
      addps       xmm0, xmm7
      movaps      xmm5, [ecx + 30h]
      movss       xmm6, [edx + 10h]
      mulps       xmm2, xmm5
      movss       xmm7, [edx + 14h]
      shufps      xmm6, xmm6, 0
      addps       xmm0, xmm2
      shufps      xmm7, xmm7, 0
      movlps[eax], xmm0
      movhps[eax + 8], xmm0
      mulps       xmm7, xmm3
      movss       xmm0, [edx + 18h]
      mulps       xmm6, xmm1
      shufps      xmm0, xmm0, 0
      addps       xmm6, xmm7
      mulps       xmm0, xmm4
      movss       xmm2, [edx + 24h]
      addps       xmm6, xmm0
      movss       xmm0, [edx + 1Ch]
      movss       xmm7, [edx + 20h]
      shufps      xmm0, xmm0, 0
      shufps      xmm7, xmm7, 0
      mulps       xmm0, xmm5
      mulps       xmm7, xmm1
      addps       xmm6, xmm0
      shufps      xmm2, xmm2, 0
      movlps[eax + 10h], xmm6
      movhps[eax + 18h], xmm6
      mulps       xmm2, xmm3
      movss       xmm6, [edx + 28h]
      addps       xmm7, xmm2
      shufps      xmm6, xmm6, 0
      movss       xmm2, [edx + 2Ch]
      mulps       xmm6, xmm4
      shufps      xmm2, xmm2, 0
      addps       xmm7, xmm6
      mulps       xmm2, xmm5
      movss       xmm0, [edx + 34h]
      addps       xmm7, xmm2
      shufps      xmm0, xmm0, 0
      movlps[eax + 20h], xmm7
      movss       xmm2, [edx + 30h]
      movhps[eax + 28h], xmm7
      mulps       xmm0, xmm3
      shufps      xmm2, xmm2, 0
      movss       xmm6, [edx + 38h]
      mulps       xmm2, xmm1
      shufps      xmm6, xmm6, 0
      addps       xmm2, xmm0
      mulps       xmm6, xmm4
      movss       xmm7, [edx + 3Ch]
      shufps      xmm7, xmm7, 0
      addps       xmm2, xmm6
      mulps       xmm7, xmm5
      addps       xmm2, xmm7
      movaps[eax + 30h], xmm2
   }
}
// if we set our flag, we always try to build the inlined asm.
// EXCEPT if we're in an old version of Codewarrior that can't handle SSE code.
// TODO: the NASM implementation of SSE_MatrixF_x_MatrixF_Aligned is missing,
// so we temporary disable this until fixed (needed for linux dedicated build)
#elif defined(TORQUE_SUPPORTS_NASM)
#define ADD_SSE_FN
extern "C"
{
   void SSE_MatrixF_x_MatrixF(const F32 *matA, const F32 *matB, F32 *result);
   void SSE_MatrixF_x_MatrixF_Aligned(const F32 *matA, const F32 *matB, F32 *result);
}

#elif defined( TORQUE_COMPILER_GCC ) && (defined( TORQUE_CPU_X86 ) || defined( TORQUE_CPU_X64 ) && !defined(TORQUE_OS_IOS))
#define ADD_SSE_FN

void SSE_MatrixF_x_MatrixF(const F32 *matA, const F32 *matB, F32 *result)
{
   asm
   (
      "movss      (%%edx),%%xmm0\n"
      "movups     (%%ecx),%%xmm1\n"
      "shufps     $0,%%xmm0,%%xmm0\n"
      "movss      4(%%edx),%%xmm2\n"
      "mulps      %%xmm1,%%xmm0\n"
      "shufps     $0,%%xmm2,%%xmm2\n"
      "movups     0x10(%%ecx),%%xmm3\n"
      "movss      8(%%edx),%%xmm7\n"
      "mulps      %%xmm3,%%xmm2\n"
      "shufps     $0,%%xmm7,%%xmm7\n"
      "addps      %%xmm2,%%xmm0\n"
      "movups     0x20(%%ecx),%%xmm4\n"
      "movss      0x0c(%%edx),%%xmm2\n"
      "mulps      %%xmm4,%%xmm7\n"
      "shufps     $0,%%xmm2,%%xmm2\n"
      "addps      %%xmm7,%%xmm0\n"
      "movups     0x30(%%ecx),%%xmm5\n"
      "movss      0x10(%%edx),%%xmm6\n"
      "mulps      %%xmm5,%%xmm2\n"
      "movss      0x14(%%edx),%%xmm7\n"
      "shufps     $0,%%xmm6,%%xmm6\n"
      "addps      %%xmm2,%%xmm0\n"
      "shufps     $0,%%xmm7,%%xmm7\n"
      "movlps     %%xmm0,(%%eax)\n"
      "movhps     %%xmm0,8(%%eax)\n"
      "mulps      %%xmm3,%%xmm7\n"
      "movss      0x18(%%edx),%%xmm0\n"
      "mulps      %%xmm1,%%xmm6\n"
      "shufps     $0,%%xmm0,%%xmm0\n"
      "addps      %%xmm7,%%xmm6\n"
      "mulps      %%xmm4,%%xmm0\n"
      "movss      0x24(%%edx),%%xmm2\n"
      "addps      %%xmm0,%%xmm6\n"
      "movss      0x1c(%%edx),%%xmm0\n"
      "movss      0x20(%%edx),%%xmm7\n"
      "shufps     $0,%%xmm0,%%xmm0\n"
      "shufps     $0,%%xmm7,%%xmm7\n"
      "mulps      %%xmm5,%%xmm0\n"
      "mulps      %%xmm1,%%xmm7\n"
      "addps      %%xmm0,%%xmm6\n"
      "shufps     $0,%%xmm2,%%xmm2\n"
      "movlps     %%xmm6,0x10(%%eax)\n"
      "movhps     %%xmm6,0x18(%%eax)\n"
      "mulps      %%xmm3,%%xmm2\n"
      "movss      0x28(%%edx),%%xmm6\n"
      "addps      %%xmm2,%%xmm7\n"
      "shufps     $0,%%xmm6,%%xmm6\n"
      "movss      0x2c(%%edx),%%xmm2\n"
      "mulps      %%xmm4,%%xmm6\n"
      "shufps     $0,%%xmm2,%%xmm2\n"
      "addps      %%xmm6,%%xmm7\n"
      "mulps      %%xmm5,%%xmm2\n"
      "movss      0x34(%%edx),%%xmm0\n"
      "addps      %%xmm2,%%xmm7\n"
      "shufps     $0,%%xmm0,%%xmm0\n"
      "movlps     %%xmm7,0x20(%%eax)\n"
      "movss      0x30(%%edx),%%xmm2\n"
      "movhps     %%xmm7,0x28(%%eax)\n"
      "mulps      %%xmm3,%%xmm0\n"
      "shufps     $0,%%xmm2,%%xmm2\n"
      "movss      0x38(%%edx),%%xmm6\n"
      "mulps      %%xmm1,%%xmm2\n"
      "shufps     $0,%%xmm6,%%xmm6\n"
      "addps      %%xmm0,%%xmm2\n"
      "mulps      %%xmm4,%%xmm6\n"
      "movss      0x3c(%%edx),%%xmm7\n"
      "shufps     $0,%%xmm7,%%xmm7\n"
      "addps      %%xmm6,%%xmm2\n"
      "mulps      %%xmm5,%%xmm7\n"
      "addps      %%xmm7,%%xmm2\n"
      "movups     %%xmm2,0x30(%%eax)\n"

      :
   : "d" (matA),
      "c" (matB),
      "a" (result)
      );
}

void SSE_MatrixF_x_MatrixF_Aligned(const F32 *matA, const F32 *matB, F32 *result)
{
   asm
   (
      "movss      (%%edx),%%xmm0\n"
      "movaps     (%%ecx),%%xmm1\n"
      "shufps     $0,%%xmm0,%%xmm0\n"
      "movss      4(%%edx),%%xmm2\n"
      "mulps      %%xmm1,%%xmm0\n"
      "shufps     $0,%%xmm2,%%xmm2\n"
      "movaps     0x10(%%ecx),%%xmm3\n"
      "movss      8(%%edx),%%xmm7\n"
      "mulps      %%xmm3,%%xmm2\n"
      "shufps     $0,%%xmm7,%%xmm7\n"
      "addps      %%xmm2,%%xmm0\n"
      "movaps     0x20(%%ecx),%%xmm4\n"
      "movss      0x0c(%%edx),%%xmm2\n"
      "mulps      %%xmm4,%%xmm7\n"
      "shufps     $0,%%xmm2,%%xmm2\n"
      "addps      %%xmm7,%%xmm0\n"
      "movaps     0x30(%%ecx),%%xmm5\n"
      "movss      0x10(%%edx),%%xmm6\n"
      "mulps      %%xmm5,%%xmm2\n"
      "movss      0x14(%%edx),%%xmm7\n"
      "shufps     $0,%%xmm6,%%xmm6\n"
      "addps      %%xmm2,%%xmm0\n"
      "shufps     $0,%%xmm7,%%xmm7\n"
      "movlps     %%xmm0,(%%eax)\n"
      "movhps     %%xmm0,8(%%eax)\n"
      "mulps      %%xmm3,%%xmm7\n"
      "movss      0x18(%%edx),%%xmm0\n"
      "mulps      %%xmm1,%%xmm6\n"
      "shufps     $0,%%xmm0,%%xmm0\n"
      "addps      %%xmm7,%%xmm6\n"
      "mulps      %%xmm4,%%xmm0\n"
      "movss      0x24(%%edx),%%xmm2\n"
      "addps      %%xmm0,%%xmm6\n"
      "movss      0x1c(%%edx),%%xmm0\n"
      "movss      0x20(%%edx),%%xmm7\n"
      "shufps     $0,%%xmm0,%%xmm0\n"
      "shufps     $0,%%xmm7,%%xmm7\n"
      "mulps      %%xmm5,%%xmm0\n"
      "mulps      %%xmm1,%%xmm7\n"
      "addps      %%xmm0,%%xmm6\n"
      "shufps     $0,%%xmm2,%%xmm2\n"
      "movlps     %%xmm6,0x10(%%eax)\n"
      "movhps     %%xmm6,0x18(%%eax)\n"
      "mulps      %%xmm3,%%xmm2\n"
      "movss      0x28(%%edx),%%xmm6\n"
      "addps      %%xmm2,%%xmm7\n"
      "shufps     $0,%%xmm6,%%xmm6\n"
      "movss      0x2c(%%edx),%%xmm2\n"
      "mulps      %%xmm4,%%xmm6\n"
      "shufps     $0,%%xmm2,%%xmm2\n"
      "addps      %%xmm6,%%xmm7\n"
      "mulps      %%xmm5,%%xmm2\n"
      "movss      0x34(%%edx),%%xmm0\n"
      "addps      %%xmm2,%%xmm7\n"
      "shufps     $0,%%xmm0,%%xmm0\n"
      "movlps     %%xmm7,0x20(%%eax)\n"
      "movss      0x30(%%edx),%%xmm2\n"
      "movhps     %%xmm7,0x28(%%eax)\n"
      "mulps      %%xmm3,%%xmm0\n"
      "shufps     $0,%%xmm2,%%xmm2\n"
      "movss      0x38(%%edx),%%xmm6\n"
      "mulps      %%xmm1,%%xmm2\n"
      "shufps     $0,%%xmm6,%%xmm6\n"
      "addps      %%xmm0,%%xmm2\n"
      "mulps      %%xmm4,%%xmm6\n"
      "movss      0x3c(%%edx),%%xmm7\n"
      "shufps     $0,%%xmm7,%%xmm7\n"
      "addps      %%xmm6,%%xmm2\n"
      "mulps      %%xmm5,%%xmm7\n"
      "addps      %%xmm7,%%xmm2\n"
      "movaps     %%xmm2,0x30(%%eax)\n"

      :
   : "d" (matA),
      "c" (matB),
      "a" (result)
      );
}

#endif

void mInstall_Library_SSE()
{
#if defined(ADD_SSE_FN)
   m_matF_x_matF = SSE_MatrixF_x_MatrixF;
   m_matF_x_matF_aligned = SSE_MatrixF_x_MatrixF_Aligned;
   // m_matF_x_point3F = Athlon_MatrixF_x_Point3F;
   // m_matF_x_vectorF = Athlon_MatrixF_x_VectorF;
#endif
}