2
0
mirror of https://github.com/rehlds/rehlds.git synced 2025-03-25 11:50:31 +03:00
rehlds/rehlds/engine/mathlib_sse.cpp
2017-12-09 17:07:39 +07:00

376 lines
10 KiB
C++

/*
*
* 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 (at
* your option) 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
*
* In addition, as a special exception, the author gives permission to
* link the code of this program with the Half-Life Game Engine ("HL
* Engine") and Modified Game Libraries ("MODs") developed by Valve,
* L.L.C ("Valve"). You must obey the GNU General Public License in all
* respects for all of the code used other than the HL Engine and MODs
* from Valve. If you modify this file, you may extend this exception
* to your version of the file, but you are not obligated to do so. If
* you do not wish to do so, delete this exception statement from your
* version.
*
*/
#include "precompiled.h"
#if defined(REHLDS_SSE)
// Intrisics guide: https://software.intel.com/sites/landingpage/IntrinsicsGuide/
// Shufps calculator: http://wurstcaptures.untergrund.net/assembler_tricks.html
// aligned vec4_t
typedef ALIGN16 vec4_t avec4_t;
typedef ALIGN16 unsigned int aivec4_t[4];
// conversion multiplier
const avec4_t deg2rad =
{
(float)M_PI / 180.f,
(float)M_PI / 180.f,
(float)M_PI / 180.f,
(float)M_PI / 180.f
};
const aivec4_t negmask[4] =
{
0x80000000,
0x80000000,
0x80000000,
0x80000000
};
const aivec4_t negmask_1001 =
{
0x80000000,
0,
0,
0x80000000
};
const aivec4_t negmask_0010 =
{
0,
0,
0x80000000,
0
};
// save 4d xmm to 3d vector. we can't optimize many simple vector3 functions because saving back to 3d is slow.
inline void xmm2vec(vec_t *v, const __m128 m)
{
_mm_storel_pi((__m64*)v, m);
_mm_store_ss(v + 2, _mm_shuffle_ps(m, m, 0x02));
}
FUNC_TARGET("sse4.1")
inline __m128 dotProduct3D(__m128 v1, __m128 v2)
{
if (cpuinfo.sse4_1)
return _mm_dp_ps(v1, v2, 0x71);
__m128 v = _mm_mul_ps(v1, v2);
return _mm_add_ps(_mm_movehl_ps(v, v), _mm_hadd_ps(v, v)); // SSE3
}
inline __m128 crossProduct3D(__m128 a, __m128 b)
{
__m128 tmp1 = _mm_mul_ps(a, _mm_shuffle_ps(b, b, _MM_SHUFFLE(3, 0, 2, 1)));
__m128 tmp2 = _mm_mul_ps(b, _mm_shuffle_ps(a, a, _MM_SHUFFLE(3, 0, 2, 1)));
__m128 m = _mm_sub_ps(tmp1, tmp2);
return _mm_shuffle_ps(m, m, _MM_SHUFFLE(3, 0, 2, 1));
}
inline __m128 length3D(__m128 v)
{
return _mm_sqrt_ps(dotProduct3D(v, v));
}
inline __m128 length2D(__m128 v)
{
v = _mm_mul_ps(v, v);
return _mm_sqrt_ps(_mm_hadd_ps(v, v)); // hadd = SSE3
}
int BoxOnPlaneSide(vec_t *emins, vec_t *emaxs, mplane_t *p)
{
double dist1, dist2;
int sides = 0;
__m128 emin = _mm_loadu_ps(emins);
__m128 emax = _mm_loadu_ps(emaxs);
avec4_t d1, d2;
// general case
switch (p->signbits)
{
case 0:
_mm_store_ps(d1, emax);
_mm_store_ps(d2, emin);
break;
case 1:
_mm_store_ps(d1, emax);
_mm_store_ps(d2, emin);
d1[0] = emins[0];
d2[0] = emaxs[0];
break;
case 2:
_mm_store_ps(d1, emax);
_mm_store_ps(d2, emin);
d1[1] = emins[1];
d2[1] = emaxs[1];
break;
case 3:
_mm_store_ps(d1, emin);
_mm_store_ps(d2, emax);
d1[2] = emaxs[2];
d2[2] = emins[2];
break;
case 4:
_mm_store_ps(d1, emax);
_mm_store_ps(d2, emin);
d1[2] = emins[2];
d2[2] = emaxs[2];
break;
case 5:
_mm_store_ps(d1, emin);
_mm_store_ps(d2, emax);
d1[1] = emaxs[1];
d2[1] = emins[1];
break;
case 6:
_mm_store_ps(d1, emin);
_mm_store_ps(d2, emax);
d1[0] = emaxs[0];
d2[0] = emins[0];
break;
case 7:
_mm_store_ps(d1, emin);
_mm_store_ps(d2, emax);
break;
default:
BOPS_Error();
dist1 = dist2 = 0.0;
break;
}
dist1 = _DotProduct(p->normal, d1);
dist2 = _DotProduct(p->normal, d2);
if (dist1 >= p->dist)
sides = 1;
if (dist2 < p->dist)
sides |= 2;
return sides;
}
qboolean VectorCompare(const vec_t *v1, const vec_t *v2)
{
__m128 cmp = _mm_cmpneq_ps(_mm_loadu_ps(v1), _mm_loadu_ps(v2));
return !(_mm_movemask_ps(cmp) & (1|2|4));
}
void AngleVectors(const vec_t *angles, vec_t *forward, vec_t *right, vec_t *up)
{
#ifndef SWDS
g_engdstAddrs.pfnAngleVectors(&angles, &forward, &right, &up);
#endif // SWDS
__m128 s, c;
sincos_ps(_mm_mul_ps(_mm_loadu_ps(angles), _mm_load_ps(deg2rad)), &s, &c);
__m128 m1 = _mm_shuffle_ps(c, s, 0x90); // [cp][cp][sy][sr]
__m128 m2 = _mm_shuffle_ps(c, c, 0x09); // [cy][cr][cp][cp]
__m128 cp_mults = _mm_mul_ps(m1, m2); // [cp * cy][cp * cr][cp * sy][cp * sr];
m1 = _mm_shuffle_ps(c, s, 0x15); // [cy][cy][sy][sp]
m2 = _mm_shuffle_ps(s, c, 0xA0); // [sp][sp][cr][cr]
m1 = _mm_shuffle_ps(m1, m1, 0xC8); // [cy][sy][cy][sp]
__m128 m3 = _mm_shuffle_ps(s, s, 0x4A); // [sr][sr][sp][sy];
m3 = _mm_mul_ps(m3, _mm_mul_ps(m1, m2)); // [sp*cy*sr][sp*sy*sr][cr*cy*sp][cr*sp*sy]
m2 = _mm_shuffle_ps(s, c, 0x65); // [sy][sy][cr][cy]
m1 = _mm_shuffle_ps(c, s, 0xA6); // [cr][cy][sr][sr]
m2 = _mm_shuffle_ps(m2, m2, 0xD8); // [sy][cr][sy][cy]
m1 = _mm_xor_ps(m1, _mm_load_ps((float *)&negmask_1001)); // [-cr][cy][sr][-sr]
m1 = _mm_mul_ps(m1, m2); // [-cr*sy][cy*cr][sr*sy][-sr*cy]
m3 = _mm_add_ps(m3, m1);
if (forward)
{
_mm_storel_pi((__m64 *)forward, _mm_shuffle_ps(cp_mults, cp_mults, 0x08));
forward[2] = -_mm_cvtss_f32(s);
}
if (right)
{
__m128 r = _mm_shuffle_ps(m3, cp_mults, 0xF4); // [m3(0)][m3(1)][cp(3)][cp(3)]
xmm2vec(right, _mm_xor_ps(r, _mm_load_ps((float *)&negmask)));
}
if (up)
{
_mm_storel_pi((__m64 *)up, _mm_shuffle_ps(m3, m3, 0x0E));
up[2] = _mm_cvtss_f32(_mm_shuffle_ps(cp_mults, cp_mults, 0x01));
}
}
void AngleVectorsTranspose(const vec_t *angles, vec_t *forward, vec_t *right, vec_t *up)
{
__m128 s, c;
sincos_ps(_mm_mul_ps(_mm_loadu_ps(angles), _mm_load_ps(deg2rad)), &s, &c);
__m128 m1 = _mm_shuffle_ps(c, s, 0x90); // [cp][cp][sy][sr]
__m128 m2 = _mm_shuffle_ps(c, c, 0x09); // [cy][cr][cp][cp]
__m128 cp_mults = _mm_mul_ps(m1, m2); // [cp * cy][cp * cr][cp * sy][cp * sr];
m1 = _mm_shuffle_ps(s, s, 0x50); // [sp][sp][sy][sy]
m2 = _mm_shuffle_ps(c, s, 0x05); // [cy][cy][sp][sp]
__m128 m3 = _mm_shuffle_ps(s, c, 0xAA); // [sr][sr][cr][cr]
m1 = _mm_mul_ps(m1, m2);
m3 = _mm_shuffle_ps(m3, m3, 0xD8); // [sr][cr][sr][cr]
m3 = _mm_mul_ps(m3, m1); // [sp*cy*sr][sp*cy*cr][sy*sp*sr][sy*sp*cr]
m2 = _mm_shuffle_ps(c, s, 0xA6); // [cr][cy][sr][sr]
m1 = _mm_shuffle_ps(s, c, 0x65); // [sy][sy][cr][cy]
m2 = _mm_shuffle_ps(m2, m2, 0xD8); // [cr][sr][cy][sr]
m1 = _mm_xor_ps(m1, _mm_load_ps((float *)&negmask_1001)); // [-cr][cy][sr][-sr]
m1 = _mm_mul_ps(m1, m2); // [-cr*sy][sr*sy][cy*cr][-sr*cy]
m3 = _mm_add_ps(m3, m1);
if (forward)
{
forward[0] = _mm_cvtss_f32(cp_mults);
_mm_storel_pi((__m64*)(forward + 1), m3); // (sr*sp*cy + cr*-sy);
}
if (right)
{
right[0] = _mm_cvtss_f32(_mm_shuffle_ps(cp_mults, cp_mults, 0x02));
_mm_storel_pi((__m64*)(right + 1), _mm_shuffle_ps(m3, m3, 0x0E));
}
if (up)
{
up[0] = -_mm_cvtss_f32(s);
_mm_storel_pi((__m64 *)&up[1], _mm_shuffle_ps(cp_mults, cp_mults, 0x07));
}
}
void AngleMatrix(const vec_t *angles, float(*matrix)[4])
{
__m128 s, c;
sincos_ps(_mm_mul_ps(_mm_loadu_ps(angles), _mm_load_ps(deg2rad)), &s, &c);
/*
matrix[0][1] = sr * sp * cy - cr * sy;
matrix[1][1] = sr * sp * sy + cr * cy;
matrix[0][2] = cr * sp * cy + sr * sy;
matrix[1][2] = cr * sp * sy - sr * cy;
*/
__m128 m1;
__m128 m2 = _mm_shuffle_ps(s, c, 0x00); // [sp][sp][cp][cp]
__m128 m3 = _mm_shuffle_ps(c, s, 0x55); // [cy][cy][sy][sy]
m1 = _mm_shuffle_ps(s, c, 0xAA); // [sr][sr][cr][cr]
m2 = _mm_shuffle_ps(m2, m2, 0x00); // [sp][sp][sp][sp]
m3 = _mm_shuffle_ps(m3, m3, 0xD8); // [cy][sy][cy][sy]
m2 = _mm_mul_ps(m2, _mm_mul_ps(m1, m3)); // m1*m2*m3
m1 = _mm_shuffle_ps(m1, m1, 0x1B); // [cr][cr][sr][sr]
m3 = _mm_shuffle_ps(m3, m3, 0xB1); // [sy][cy][sy][cy]
m3 = _mm_xor_ps(m3, _mm_load_ps((float *)&negmask_1001));
m3 = _mm_mul_ps(m3, m1);
m2 = _mm_add_ps(m2, m3);
/*
matrix[0][0] = cp * cy;
matrix[1][0] = cp * sy;
matrix[2][1] = sr * cp;
matrix[2][2] = cr * cp;
*/
m1 = _mm_shuffle_ps(s, c, 0x29); // [sy][sr][cr][cp]
c = _mm_shuffle_ps(c, c, 0x40); // [cp][cp][cp][cy]
m1 = _mm_mul_ps(m1, c);
// matrix[0]
m3 = _mm_shuffle_ps(m2, m2, 0xE1);
_mm_storeu_ps(&matrix[0][0], m3);
matrix[0][0] = _mm_cvtss_f32(_mm_shuffle_ps(m1, m1, 0x03));
*(int *)&matrix[0][3] = 0;
// matrix[1]
m2 = _mm_shuffle_ps(m2, m2, 0xB4);
_mm_storeu_ps(&matrix[1][0], m2);
matrix[1][0] = _mm_cvtss_f32(m1);
*(int *)&matrix[1][3] = 0;
// matrix[2]
_mm_storeu_ps(&matrix[2][0], m1);
matrix[2][0] = -_mm_cvtss_f32(s);
*(int *)&matrix[2][3] = 0;
}
void VectorMA(const vec_t *veca, float scale, const vec_t *vecm, vec_t *out)
{
xmm2vec(out, _mm_add_ps(_mm_mul_ps(_mm_set_ps1(scale), _mm_loadu_ps(vecm)), _mm_loadu_ps(veca)));
}
float _DotProduct(const vec_t *v1, const vec_t *v2)
{
// _mm_loadu_ps - load xmm from unaligned address
// _mm_cvtss_f32 - return low float value of xmm
// _mm_dp_ps - dot product
// 0x71 = 0b01110001 - mask for multiplying operands and result
// dpps isn't binary compatible with separate sse2 instructions (max difference is about 0.0002f, but usually < 0.00001f)
return _mm_cvtss_f32(dotProduct3D(_mm_loadu_ps(v1), _mm_loadu_ps(v2)));
}
float Length(const vec_t *v)
{
return _mm_cvtss_f32(length3D(_mm_loadu_ps(v))); // rsqrt is very inaccurate :(
}
float Length2D(const vec_t *v)
{
return _mm_cvtss_f32(length2D(_mm_loadu_ps(v)));
}
void CrossProduct(const vec_t *v1, const vec_t *v2, vec_t *cross)
{
xmm2vec(cross, crossProduct3D(_mm_loadu_ps(v1), _mm_loadu_ps(v2)));
}
void R_ConcatTransforms(float in1[3][4], float in2[3][4], float out[3][4])
{
for (size_t i = 0; i < 3; i++)
{
__m128 a1 = _mm_mul_ps(_mm_set_ps1(in1[i][0]), _mm_loadu_ps(in2[0]));
__m128 a2 = _mm_mul_ps(_mm_set_ps1(in1[i][1]), _mm_loadu_ps(in2[1]));
__m128 a3 = _mm_mul_ps(_mm_set_ps1(in1[i][2]), _mm_loadu_ps(in2[2]));
_mm_storeu_ps(out[i], _mm_add_ps(a1, _mm_add_ps(a2, a3)));
out[i][3] += in1[i][3];
}
}
#endif // #if defined(REHLDS_SSE)