xbmc/rendering/MatrixGL.neon.cpp

   1 /*
   2  *  Copyright (C) 2005-2018 Team Kodi
   3  *  This file is part of Kodi - https://kodi.tv
   4  *
   5  *  SPDX-License-Identifier: GPL-2.0-or-later
   6  *  See LICENSES/README.md for more information.
   7  */
   8
   9
  10 void Matrix4Mul(float* src_mat_1, const float* src_mat_2)
  11 {
  12   asm volatile (
  13     // Store A & B leaving room at top of registers for result (q0-q3)
  14     "vldmia %0, { q4-q7 }  \n\t"
  15     "vldmia %1, { q8-q11 } \n\t"
  16
  17     // result = first column of B x first row of A
  18     "vmul.f32 q0, q8, d8[0]\n\t"
  19     "vmul.f32 q1, q8, d10[0]\n\t"
  20     "vmul.f32 q2, q8, d12[0]\n\t"
  21     "vmul.f32 q3, q8, d14[0]\n\t"
  22
  23     // result += second column of B x second row of A
  24     "vmla.f32 q0, q9, d8[1]\n\t"
  25     "vmla.f32 q1, q9, d10[1]\n\t"
  26     "vmla.f32 q2, q9, d12[1]\n\t"
  27     "vmla.f32 q3, q9, d14[1]\n\t"
  28
  29     // result += third column of B x third row of A
  30     "vmla.f32 q0, q10, d9[0]\n\t"
  31     "vmla.f32 q1, q10, d11[0]\n\t"
  32     "vmla.f32 q2, q10, d13[0]\n\t"
  33     "vmla.f32 q3, q10, d15[0]\n\t"
  34
  35     // result += last column of B x last row of A
  36     "vmla.f32 q0, q11, d9[1]\n\t"
  37     "vmla.f32 q1, q11, d11[1]\n\t"
  38     "vmla.f32 q2, q11, d13[1]\n\t"
  39     "vmla.f32 q3, q11, d15[1]\n\t"
  40
  41     // output = result registers
  42     "vstmia %1, { q0-q3 }"
  43     : //no output
  44     : "r" (src_mat_2), "r" (src_mat_1)       // input - note *value* of pointer doesn't change
  45     : "memory", "q0", "q1", "q2", "q3", "q4", "q5", "q6", "q7", "q8", "q9", "q10", "q11" //clobber
  46     );
  47 }