X-Git-Url: https://de.git.xonotic.org/?a=blobdiff_plain;f=mathlib.c;h=8f73d5eab2476072ebb21ede62645bfab358a93d;hb=44c9a08f2c4035bf5798e663bd4fbad506fd152e;hp=f1bbacff8f25cdb9009bdab5d02b842d27d33542;hpb=8dcce44300385b12c46d494c06aadcfa35a8bc14;p=xonotic%2Fdarkplaces.git diff --git a/mathlib.c b/mathlib.c index f1bbacff..8f73d5ea 100644 --- a/mathlib.c +++ b/mathlib.c @@ -8,7 +8,7 @@ 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. +MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. @@ -25,14 +25,155 @@ Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. void Sys_Error (char *error, ...); vec3_t vec3_origin = {0,0,0}; -int nanmask = 255<<23; +float ixtable[4096]; /*-----------------------------------------------------------------*/ -#define DEG2RAD( a ) ( a * M_PI ) / 180.0F +float m_bytenormals[NUMVERTEXNORMALS][3] = +{ +{-0.525731, 0.000000, 0.850651}, {-0.442863, 0.238856, 0.864188}, +{-0.295242, 0.000000, 0.955423}, {-0.309017, 0.500000, 0.809017}, +{-0.162460, 0.262866, 0.951056}, {0.000000, 0.000000, 1.000000}, +{0.000000, 0.850651, 0.525731}, {-0.147621, 0.716567, 0.681718}, +{0.147621, 0.716567, 0.681718}, {0.000000, 0.525731, 0.850651}, +{0.309017, 0.500000, 0.809017}, {0.525731, 0.000000, 0.850651}, +{0.295242, 0.000000, 0.955423}, {0.442863, 0.238856, 0.864188}, +{0.162460, 0.262866, 0.951056}, {-0.681718, 0.147621, 0.716567}, +{-0.809017, 0.309017, 0.500000}, {-0.587785, 0.425325, 0.688191}, +{-0.850651, 0.525731, 0.000000}, {-0.864188, 0.442863, 0.238856}, +{-0.716567, 0.681718, 0.147621}, {-0.688191, 0.587785, 0.425325}, +{-0.500000, 0.809017, 0.309017}, {-0.238856, 0.864188, 0.442863}, +{-0.425325, 0.688191, 0.587785}, {-0.716567, 0.681718, -0.147621}, +{-0.500000, 0.809017, -0.309017}, {-0.525731, 0.850651, 0.000000}, +{0.000000, 0.850651, -0.525731}, {-0.238856, 0.864188, -0.442863}, +{0.000000, 0.955423, -0.295242}, {-0.262866, 0.951056, -0.162460}, +{0.000000, 1.000000, 0.000000}, {0.000000, 0.955423, 0.295242}, +{-0.262866, 0.951056, 0.162460}, {0.238856, 0.864188, 0.442863}, +{0.262866, 0.951056, 0.162460}, {0.500000, 0.809017, 0.309017}, +{0.238856, 0.864188, -0.442863}, {0.262866, 0.951056, -0.162460}, +{0.500000, 0.809017, -0.309017}, {0.850651, 0.525731, 0.000000}, +{0.716567, 0.681718, 0.147621}, {0.716567, 0.681718, -0.147621}, +{0.525731, 0.850651, 0.000000}, {0.425325, 0.688191, 0.587785}, +{0.864188, 0.442863, 0.238856}, {0.688191, 0.587785, 0.425325}, +{0.809017, 0.309017, 0.500000}, {0.681718, 0.147621, 0.716567}, +{0.587785, 0.425325, 0.688191}, {0.955423, 0.295242, 0.000000}, +{1.000000, 0.000000, 0.000000}, {0.951056, 0.162460, 0.262866}, +{0.850651, -0.525731, 0.000000}, {0.955423, -0.295242, 0.000000}, +{0.864188, -0.442863, 0.238856}, {0.951056, -0.162460, 0.262866}, +{0.809017, -0.309017, 0.500000}, {0.681718, -0.147621, 0.716567}, +{0.850651, 0.000000, 0.525731}, {0.864188, 0.442863, -0.238856}, +{0.809017, 0.309017, -0.500000}, {0.951056, 0.162460, -0.262866}, +{0.525731, 0.000000, -0.850651}, {0.681718, 0.147621, -0.716567}, +{0.681718, -0.147621, -0.716567}, {0.850651, 0.000000, -0.525731}, +{0.809017, -0.309017, -0.500000}, {0.864188, -0.442863, -0.238856}, +{0.951056, -0.162460, -0.262866}, {0.147621, 0.716567, -0.681718}, +{0.309017, 0.500000, -0.809017}, {0.425325, 0.688191, -0.587785}, +{0.442863, 0.238856, -0.864188}, {0.587785, 0.425325, -0.688191}, +{0.688191, 0.587785, -0.425325}, {-0.147621, 0.716567, -0.681718}, +{-0.309017, 0.500000, -0.809017}, {0.000000, 0.525731, -0.850651}, +{-0.525731, 0.000000, -0.850651}, {-0.442863, 0.238856, -0.864188}, +{-0.295242, 0.000000, -0.955423}, {-0.162460, 0.262866, -0.951056}, +{0.000000, 0.000000, -1.000000}, {0.295242, 0.000000, -0.955423}, +{0.162460, 0.262866, -0.951056}, {-0.442863, -0.238856, -0.864188}, +{-0.309017, -0.500000, -0.809017}, {-0.162460, -0.262866, -0.951056}, +{0.000000, -0.850651, -0.525731}, {-0.147621, -0.716567, -0.681718}, +{0.147621, -0.716567, -0.681718}, {0.000000, -0.525731, -0.850651}, +{0.309017, -0.500000, -0.809017}, {0.442863, -0.238856, -0.864188}, +{0.162460, -0.262866, -0.951056}, {0.238856, -0.864188, -0.442863}, +{0.500000, -0.809017, -0.309017}, {0.425325, -0.688191, -0.587785}, +{0.716567, -0.681718, -0.147621}, {0.688191, -0.587785, -0.425325}, +{0.587785, -0.425325, -0.688191}, {0.000000, -0.955423, -0.295242}, +{0.000000, -1.000000, 0.000000}, {0.262866, -0.951056, -0.162460}, +{0.000000, -0.850651, 0.525731}, {0.000000, -0.955423, 0.295242}, +{0.238856, -0.864188, 0.442863}, {0.262866, -0.951056, 0.162460}, +{0.500000, -0.809017, 0.309017}, {0.716567, -0.681718, 0.147621}, +{0.525731, -0.850651, 0.000000}, {-0.238856, -0.864188, -0.442863}, +{-0.500000, -0.809017, -0.309017}, {-0.262866, -0.951056, -0.162460}, +{-0.850651, -0.525731, 0.000000}, {-0.716567, -0.681718, -0.147621}, +{-0.716567, -0.681718, 0.147621}, {-0.525731, -0.850651, 0.000000}, +{-0.500000, -0.809017, 0.309017}, {-0.238856, -0.864188, 0.442863}, +{-0.262866, -0.951056, 0.162460}, {-0.864188, -0.442863, 0.238856}, +{-0.809017, -0.309017, 0.500000}, {-0.688191, -0.587785, 0.425325}, +{-0.681718, -0.147621, 0.716567}, {-0.442863, -0.238856, 0.864188}, +{-0.587785, -0.425325, 0.688191}, {-0.309017, -0.500000, 0.809017}, +{-0.147621, -0.716567, 0.681718}, {-0.425325, -0.688191, 0.587785}, +{-0.162460, -0.262866, 0.951056}, {0.442863, -0.238856, 0.864188}, +{0.162460, -0.262866, 0.951056}, {0.309017, -0.500000, 0.809017}, +{0.147621, -0.716567, 0.681718}, {0.000000, -0.525731, 0.850651}, +{0.425325, -0.688191, 0.587785}, {0.587785, -0.425325, 0.688191}, +{0.688191, -0.587785, 0.425325}, {-0.955423, 0.295242, 0.000000}, +{-0.951056, 0.162460, 0.262866}, {-1.000000, 0.000000, 0.000000}, +{-0.850651, 0.000000, 0.525731}, {-0.955423, -0.295242, 0.000000}, +{-0.951056, -0.162460, 0.262866}, {-0.864188, 0.442863, -0.238856}, +{-0.951056, 0.162460, -0.262866}, {-0.809017, 0.309017, -0.500000}, +{-0.864188, -0.442863, -0.238856}, {-0.951056, -0.162460, -0.262866}, +{-0.809017, -0.309017, -0.500000}, {-0.681718, 0.147621, -0.716567}, +{-0.681718, -0.147621, -0.716567}, {-0.850651, 0.000000, -0.525731}, +{-0.688191, 0.587785, -0.425325}, {-0.587785, 0.425325, -0.688191}, +{-0.425325, 0.688191, -0.587785}, {-0.425325, -0.688191, -0.587785}, +{-0.587785, -0.425325, -0.688191}, {-0.688191, -0.587785, -0.425325}, +}; + +byte NormalToByte(vec3_t n) +{ + int i, best; + float bestdistance, distance; + + best = 0; + bestdistance = DotProduct (n, m_bytenormals[0]); + for (i = 1;i < NUMVERTEXNORMALS;i++) + { + distance = DotProduct (n, m_bytenormals[i]); + if (distance > bestdistance) + { + bestdistance = distance; + best = i; + } + } + return best; +} + +// note: uses byte partly to force unsigned for the validity check +void ByteToNormal(byte num, vec3_t n) +{ + if (num < NUMVERTEXNORMALS) + VectorCopy(m_bytenormals[num], n); + else + VectorClear(n); // FIXME: complain? +} + +float Q_RSqrt(float number) +{ + float y; + + if (number == 0.0f) + return 0.0f; + + *((long *)&y) = 0x5f3759df - ((* (long *) &number) >> 1); + return y * (1.5f - (number * 0.5f * y * y)); +} + +void _VectorNormalizeFast(vec3_t v) +{ + float y, number; + + number = DotProduct(v, v); + + if (number != 0.0) + { + *((long *)&y) = 0x5f3759df - ((* (long *) &number) >> 1); + y = y * (1.5f - (number * 0.5f * y * y)); + + VectorScale(v, y, v); + } +} +#if 0 +// LordHavoc: no longer used at all void ProjectPointOnPlane( vec3_t dst, const vec3_t p, const vec3_t normal ) { +#if 0 + // LordHavoc: the old way... float d; vec3_t n; float inv_denom; @@ -48,21 +189,32 @@ void ProjectPointOnPlane( vec3_t dst, const vec3_t p, const vec3_t normal ) dst[0] = p[0] - d * n[0]; dst[1] = p[1] - d * n[1]; dst[2] = p[2] - d * n[2]; +#else + // LordHavoc: optimized to death and beyond + float d; + + // LordHavoc: the normal is a unit vector by definition, + // therefore inv_denom was pointless. + d = DotProduct(normal, p); + dst[0] = p[0] - d * normal[0]; + dst[1] = p[1] - d * normal[1]; + dst[2] = p[2] - d * normal[2]; +#endif } +#endif -/* -** assumes "src" is normalized -*/ +// assumes "src" is normalized void PerpendicularVector( vec3_t dst, const vec3_t src ) { +#if 0 + // LordHavoc: the old way... int pos; int i; - float minelem = 1.0F; + float minelem, d; vec3_t tempvec; - /* - ** find the smallest magnitude axially aligned vector - */ + // find the smallest magnitude axially aligned vector + minelem = 1.0F; for ( pos = 0, i = 0; i < 3; i++ ) { if ( fabs( src[i] ) < minelem ) @@ -71,27 +223,102 @@ void PerpendicularVector( vec3_t dst, const vec3_t src ) minelem = fabs( src[i] ); } } - tempvec[0] = tempvec[1] = tempvec[2] = 0.0F; + VectorClear(tempvec); tempvec[pos] = 1.0F; - /* - ** project the point onto the plane defined by src - */ + // project the point onto the plane defined by src ProjectPointOnPlane( dst, tempvec, src ); - /* - ** normalize the result - */ - VectorNormalize( dst ); -} + // normalize the result + VectorNormalize(dst); +#else + // LordHavoc: optimized to death and beyond + int pos; + float minelem; -#ifdef _WIN32 -#pragma optimize( "", off ) + if (src[0]) + { + dst[0] = 0; + if (src[1]) + { + dst[1] = 0; + if (src[2]) + { + dst[2] = 0; + pos = 0; + minelem = fabs(src[0]); + if (fabs(src[1]) < minelem) + { + pos = 1; + minelem = fabs(src[1]); + } + if (fabs(src[2]) < minelem) + pos = 2; + + dst[pos] = 1; + dst[0] -= src[pos] * src[0]; + dst[1] -= src[pos] * src[1]; + dst[2] -= src[pos] * src[2]; + + // normalize the result + VectorNormalize(dst); + } + else + dst[2] = 1; + } + else + { + dst[1] = 1; + dst[2] = 0; + } + } + else + { + dst[0] = 1; + dst[1] = 0; + dst[2] = 0; + } #endif +} +// LordHavoc: like AngleVectors, but taking a forward vector instead of angles, useful! +void VectorVectors(const vec3_t forward, vec3_t right, vec3_t up) +{ + float d; + + right[0] = forward[2]; + right[1] = -forward[0]; + right[2] = forward[1]; + + d = DotProduct(forward, right); + right[0] -= d * forward[0]; + right[1] -= d * forward[1]; + right[2] -= d * forward[2]; + VectorNormalizeFast(right); + CrossProduct(right, forward, up); +} + +void VectorVectorsDouble(const double *forward, double *right, double *up) +{ + double d; + + right[0] = forward[2]; + right[1] = -forward[0]; + right[2] = forward[1]; + + d = DotProduct(forward, right); + right[0] -= d * forward[0]; + right[1] -= d * forward[1]; + right[2] -= d * forward[2]; + VectorNormalize(right); + CrossProduct(right, forward, up); +} + void RotatePointAroundVector( vec3_t dst, const vec3_t dir, const vec3_t point, float degrees ) { +#if 0 + // LordHavoc: the old way, cryptic brute force... float m[3][3]; float im[3][3]; float zrot[3][3]; @@ -140,210 +367,299 @@ void RotatePointAroundVector( vec3_t dst, const vec3_t dir, const vec3_t point, R_ConcatRotations( tmpmat, im, rot ); for ( i = 0; i < 3; i++ ) - { dst[i] = rot[i][0] * point[0] + rot[i][1] * point[1] + rot[i][2] * point[2]; - } -} +#elif 0 + // LordHavoc: on the path to unintelligible code... +// float m[3][3]; +// float im[3][3]; +// float zrot[3][3]; + float tmpmat[3][3]; +// float rot[3][3]; + float angle, c, s; +// int i; + vec3_t vr, vu, vf; -#ifdef _WIN32 -#pragma optimize( "", on ) -#endif + angle = DEG2RAD(degrees); -/*-----------------------------------------------------------------*/ + c = cos(angle); + s = sin(angle); + vf[0] = dir[0]; + vf[1] = dir[1]; + vf[2] = dir[2]; -float anglemod(float a) -{ -#if 0 - if (a >= 0) - a -= 360*(int)(a/360); - else - a += 360*( 1 + (int)(-a/360) ); -#endif - a = (360.0/65536) * ((int)(a*(65536/360.0)) & 65535); - return a; -} + PerpendicularVector(vr, dir); + CrossProduct(vr, vf, vu); + +// m [0][0] = vr[0];m [0][1] = vu[0];m [0][2] = vf[0]; +// m [1][0] = vr[1];m [1][1] = vu[1];m [1][2] = vf[1]; +// m [2][0] = vr[2];m [2][1] = vu[2];m [2][2] = vf[2]; +// im [0][0] = vr[0];im [0][1] = vr[1];im [0][2] = vr[2]; +// im [1][0] = vu[0];im [1][1] = vu[1];im [1][2] = vu[2]; +// im [2][0] = vf[0];im [2][1] = vf[1];im [2][2] = vf[2]; +// zrot[0][0] = c;zrot[0][1] = s;zrot[0][2] = 0; +// zrot[1][0] = -s;zrot[1][1] = c;zrot[1][2] = 0; +// zrot[2][0] = 0;zrot[2][1] = 0;zrot[2][2] = 1; + +// tmpmat[0][0] = m[0][0] * zrot[0][0] + m[0][1] * zrot[1][0] + m[0][2] * zrot[2][0]; +// tmpmat[0][1] = m[0][0] * zrot[0][1] + m[0][1] * zrot[1][1] + m[0][2] * zrot[2][1]; +// tmpmat[0][2] = m[0][0] * zrot[0][2] + m[0][1] * zrot[1][2] + m[0][2] * zrot[2][2]; +// tmpmat[1][0] = m[1][0] * zrot[0][0] + m[1][1] * zrot[1][0] + m[1][2] * zrot[2][0]; +// tmpmat[1][1] = m[1][0] * zrot[0][1] + m[1][1] * zrot[1][1] + m[1][2] * zrot[2][1]; +// tmpmat[1][2] = m[1][0] * zrot[0][2] + m[1][1] * zrot[1][2] + m[1][2] * zrot[2][2]; +// tmpmat[2][0] = m[2][0] * zrot[0][0] + m[2][1] * zrot[1][0] + m[2][2] * zrot[2][0]; +// tmpmat[2][1] = m[2][0] * zrot[0][1] + m[2][1] * zrot[1][1] + m[2][2] * zrot[2][1]; +// tmpmat[2][2] = m[2][0] * zrot[0][2] + m[2][1] * zrot[1][2] + m[2][2] * zrot[2][2]; + + tmpmat[0][0] = vr[0] * c + vu[0] * -s; + tmpmat[0][1] = vr[0] * s + vu[0] * c; +// tmpmat[0][2] = vf[0]; + tmpmat[1][0] = vr[1] * c + vu[1] * -s; + tmpmat[1][1] = vr[1] * s + vu[1] * c; +// tmpmat[1][2] = vf[1]; + tmpmat[2][0] = vr[2] * c + vu[2] * -s; + tmpmat[2][1] = vr[2] * s + vu[2] * c; +// tmpmat[2][2] = vf[2]; + +// rot[0][0] = tmpmat[0][0] * vr[0] + tmpmat[0][1] * vu[0] + tmpmat[0][2] * vf[0]; +// rot[0][1] = tmpmat[0][0] * vr[1] + tmpmat[0][1] * vu[1] + tmpmat[0][2] * vf[1]; +// rot[0][2] = tmpmat[0][0] * vr[2] + tmpmat[0][1] * vu[2] + tmpmat[0][2] * vf[2]; +// rot[1][0] = tmpmat[1][0] * vr[0] + tmpmat[1][1] * vu[0] + tmpmat[1][2] * vf[0]; +// rot[1][1] = tmpmat[1][0] * vr[1] + tmpmat[1][1] * vu[1] + tmpmat[1][2] * vf[1]; +// rot[1][2] = tmpmat[1][0] * vr[2] + tmpmat[1][1] * vu[2] + tmpmat[1][2] * vf[2]; +// rot[2][0] = tmpmat[2][0] * vr[0] + tmpmat[2][1] * vu[0] + tmpmat[2][2] * vf[0]; +// rot[2][1] = tmpmat[2][0] * vr[1] + tmpmat[2][1] * vu[1] + tmpmat[2][2] * vf[1]; +// rot[2][2] = tmpmat[2][0] * vr[2] + tmpmat[2][1] * vu[2] + tmpmat[2][2] * vf[2]; + +// rot[0][0] = tmpmat[0][0] * vr[0] + tmpmat[0][1] * vu[0] + vf[0] * vf[0]; +// rot[0][1] = tmpmat[0][0] * vr[1] + tmpmat[0][1] * vu[1] + vf[0] * vf[1]; +// rot[0][2] = tmpmat[0][0] * vr[2] + tmpmat[0][1] * vu[2] + vf[0] * vf[2]; +// rot[1][0] = tmpmat[1][0] * vr[0] + tmpmat[1][1] * vu[0] + vf[1] * vf[0]; +// rot[1][1] = tmpmat[1][0] * vr[1] + tmpmat[1][1] * vu[1] + vf[1] * vf[1]; +// rot[1][2] = tmpmat[1][0] * vr[2] + tmpmat[1][1] * vu[2] + vf[1] * vf[2]; +// rot[2][0] = tmpmat[2][0] * vr[0] + tmpmat[2][1] * vu[0] + vf[2] * vf[0]; +// rot[2][1] = tmpmat[2][0] * vr[1] + tmpmat[2][1] * vu[1] + vf[2] * vf[1]; +// rot[2][2] = tmpmat[2][0] * vr[2] + tmpmat[2][1] * vu[2] + vf[2] * vf[2]; + +// dst[0] = rot[0][0] * point[0] + rot[0][1] * point[1] + rot[0][2] * point[2]; +// dst[1] = rot[1][0] * point[0] + rot[1][1] * point[1] + rot[1][2] * point[2]; +// dst[2] = rot[2][0] * point[0] + rot[2][1] * point[1] + rot[2][2] * point[2]; + + dst[0] = (tmpmat[0][0] * vr[0] + tmpmat[0][1] * vu[0] + vf[0] * vf[0]) * point[0] + + (tmpmat[0][0] * vr[1] + tmpmat[0][1] * vu[1] + vf[0] * vf[1]) * point[1] + + (tmpmat[0][0] * vr[2] + tmpmat[0][1] * vu[2] + vf[0] * vf[2]) * point[2]; + dst[1] = (tmpmat[1][0] * vr[0] + tmpmat[1][1] * vu[0] + vf[1] * vf[0]) * point[0] + + (tmpmat[1][0] * vr[1] + tmpmat[1][1] * vu[1] + vf[1] * vf[1]) * point[1] + + (tmpmat[1][0] * vr[2] + tmpmat[1][1] * vu[2] + vf[1] * vf[2]) * point[2]; + dst[2] = (tmpmat[2][0] * vr[0] + tmpmat[2][1] * vu[0] + vf[2] * vf[0]) * point[0] + + (tmpmat[2][0] * vr[1] + tmpmat[2][1] * vu[1] + vf[2] * vf[1]) * point[1] + + (tmpmat[2][0] * vr[2] + tmpmat[2][1] * vu[2] + vf[2] * vf[2]) * point[2]; +#else + // LordHavoc: optimized to death and beyond, cryptic in an entirely new way + float t0, t1; + float angle, c, s; + vec3_t vr, vu, vf; + + angle = DEG2RAD(degrees); + + c = cos(angle); + s = sin(angle); -/* -================== -BOPS_Error + vf[0] = dir[0]; + vf[1] = dir[1]; + vf[2] = dir[2]; -Split out like this for ASM to call. -================== -*/ -/* -void BOPS_Error (void) -{ - Sys_Error ("BoxOnPlaneSide: Bad signbits"); +// PerpendicularVector(vr, dir); +// CrossProduct(vr, vf, vu); + VectorVectors(vf, vr, vu); + + t0 = vr[0] * c + vu[0] * -s; + t1 = vr[0] * s + vu[0] * c; + dst[0] = (t0 * vr[0] + t1 * vu[0] + vf[0] * vf[0]) * point[0] + + (t0 * vr[1] + t1 * vu[1] + vf[0] * vf[1]) * point[1] + + (t0 * vr[2] + t1 * vu[2] + vf[0] * vf[2]) * point[2]; + + t0 = vr[1] * c + vu[1] * -s; + t1 = vr[1] * s + vu[1] * c; + dst[1] = (t0 * vr[0] + t1 * vu[0] + vf[1] * vf[0]) * point[0] + + (t0 * vr[1] + t1 * vu[1] + vf[1] * vf[1]) * point[1] + + (t0 * vr[2] + t1 * vu[2] + vf[1] * vf[2]) * point[2]; + + t0 = vr[2] * c + vu[2] * -s; + t1 = vr[2] * s + vu[2] * c; + dst[2] = (t0 * vr[0] + t1 * vu[0] + vf[2] * vf[0]) * point[0] + + (t0 * vr[1] + t1 * vu[1] + vf[2] * vf[1]) * point[1] + + (t0 * vr[2] + t1 * vu[2] + vf[2] * vf[2]) * point[2]; +#endif } +/*-----------------------------------------------------------------*/ -#if !id386 -*/ -/* -================== -BoxOnPlaneSide +// LordHavoc note 1: +// BoxOnPlaneSide did a switch on a 'signbits' value and had optimized +// assembly in an attempt to accelerate it further, very inefficient +// considering that signbits of the frustum planes only changed each +// frame, and the world planes changed only at load time. +// So, to optimize it further I took the obvious route of storing a function +// pointer in the plane struct itself, and shrunk each of the individual +// cases to a single return statement. +// LordHavoc note 2: +// realized axial cases would be a nice speedup for world geometry, although +// never useful for the frustum planes. +int BoxOnPlaneSideX (vec3_t emins, vec3_t emaxs, mplane_t *p) {return p->dist <= emins[0] ? 1 : (p->dist >= emaxs[0] ? 2 : 3);} +int BoxOnPlaneSideY (vec3_t emins, vec3_t emaxs, mplane_t *p) {return p->dist <= emins[1] ? 1 : (p->dist >= emaxs[1] ? 2 : 3);} +int BoxOnPlaneSideZ (vec3_t emins, vec3_t emaxs, mplane_t *p) {return p->dist <= emins[2] ? 1 : (p->dist >= emaxs[2] ? 2 : 3);} +int BoxOnPlaneSide0 (vec3_t emins, vec3_t emaxs, mplane_t *p) {return (((p->normal[0]*emaxs[0] + p->normal[1]*emaxs[1] + p->normal[2]*emaxs[2]) >= p->dist) | (((p->normal[0]*emins[0] + p->normal[1]*emins[1] + p->normal[2]*emins[2]) < p->dist) << 1));} +int BoxOnPlaneSide1 (vec3_t emins, vec3_t emaxs, mplane_t *p) {return (((p->normal[0]*emins[0] + p->normal[1]*emaxs[1] + p->normal[2]*emaxs[2]) >= p->dist) | (((p->normal[0]*emaxs[0] + p->normal[1]*emins[1] + p->normal[2]*emins[2]) < p->dist) << 1));} +int BoxOnPlaneSide2 (vec3_t emins, vec3_t emaxs, mplane_t *p) {return (((p->normal[0]*emaxs[0] + p->normal[1]*emins[1] + p->normal[2]*emaxs[2]) >= p->dist) | (((p->normal[0]*emins[0] + p->normal[1]*emaxs[1] + p->normal[2]*emins[2]) < p->dist) << 1));} +int BoxOnPlaneSide3 (vec3_t emins, vec3_t emaxs, mplane_t *p) {return (((p->normal[0]*emins[0] + p->normal[1]*emins[1] + p->normal[2]*emaxs[2]) >= p->dist) | (((p->normal[0]*emaxs[0] + p->normal[1]*emaxs[1] + p->normal[2]*emins[2]) < p->dist) << 1));} +int BoxOnPlaneSide4 (vec3_t emins, vec3_t emaxs, mplane_t *p) {return (((p->normal[0]*emaxs[0] + p->normal[1]*emaxs[1] + p->normal[2]*emins[2]) >= p->dist) | (((p->normal[0]*emins[0] + p->normal[1]*emins[1] + p->normal[2]*emaxs[2]) < p->dist) << 1));} +int BoxOnPlaneSide5 (vec3_t emins, vec3_t emaxs, mplane_t *p) {return (((p->normal[0]*emins[0] + p->normal[1]*emaxs[1] + p->normal[2]*emins[2]) >= p->dist) | (((p->normal[0]*emaxs[0] + p->normal[1]*emins[1] + p->normal[2]*emaxs[2]) < p->dist) << 1));} +int BoxOnPlaneSide6 (vec3_t emins, vec3_t emaxs, mplane_t *p) {return (((p->normal[0]*emaxs[0] + p->normal[1]*emins[1] + p->normal[2]*emins[2]) >= p->dist) | (((p->normal[0]*emins[0] + p->normal[1]*emaxs[1] + p->normal[2]*emaxs[2]) < p->dist) << 1));} +int BoxOnPlaneSide7 (vec3_t emins, vec3_t emaxs, mplane_t *p) {return (((p->normal[0]*emins[0] + p->normal[1]*emins[1] + p->normal[2]*emins[2]) >= p->dist) | (((p->normal[0]*emaxs[0] + p->normal[1]*emaxs[1] + p->normal[2]*emaxs[2]) < p->dist) << 1));} -Returns 1, 2, or 1 + 2 -================== -*/ -/* -int BoxOnPlaneSide (vec3_t emins, vec3_t emaxs, mplane_t *p) +void BoxOnPlaneSideClassify(mplane_t *p) { - float dist1, dist2; - int sides; - -#if 0 // this is done by the BOX_ON_PLANE_SIDE macro before calling this - // function -// fast axial cases - if (p->type < 3) + switch(p->type) { - if (p->dist <= emins[p->type]) - return 1; - if (p->dist >= emaxs[p->type]) - return 2; - return 3; - } -#endif - -// general case - switch (p->signbits) - { - case 0: -dist1 = p->normal[0]*emaxs[0] + p->normal[1]*emaxs[1] + p->normal[2]*emaxs[2]; -dist2 = p->normal[0]*emins[0] + p->normal[1]*emins[1] + p->normal[2]*emins[2]; - break; - case 1: -dist1 = p->normal[0]*emins[0] + p->normal[1]*emaxs[1] + p->normal[2]*emaxs[2]; -dist2 = p->normal[0]*emaxs[0] + p->normal[1]*emins[1] + p->normal[2]*emins[2]; - break; - case 2: -dist1 = p->normal[0]*emaxs[0] + p->normal[1]*emins[1] + p->normal[2]*emaxs[2]; -dist2 = p->normal[0]*emins[0] + p->normal[1]*emaxs[1] + p->normal[2]*emins[2]; - break; - case 3: -dist1 = p->normal[0]*emins[0] + p->normal[1]*emins[1] + p->normal[2]*emaxs[2]; -dist2 = p->normal[0]*emaxs[0] + p->normal[1]*emaxs[1] + p->normal[2]*emins[2]; + case 0: // x axis + p->BoxOnPlaneSideFunc = BoxOnPlaneSideX; break; - case 4: -dist1 = p->normal[0]*emaxs[0] + p->normal[1]*emaxs[1] + p->normal[2]*emins[2]; -dist2 = p->normal[0]*emins[0] + p->normal[1]*emins[1] + p->normal[2]*emaxs[2]; + case 1: // y axis + p->BoxOnPlaneSideFunc = BoxOnPlaneSideY; break; - case 5: -dist1 = p->normal[0]*emins[0] + p->normal[1]*emaxs[1] + p->normal[2]*emins[2]; -dist2 = p->normal[0]*emaxs[0] + p->normal[1]*emins[1] + p->normal[2]*emaxs[2]; - break; - case 6: -dist1 = p->normal[0]*emaxs[0] + p->normal[1]*emins[1] + p->normal[2]*emins[2]; -dist2 = p->normal[0]*emins[0] + p->normal[1]*emaxs[1] + p->normal[2]*emaxs[2]; - break; - case 7: -dist1 = p->normal[0]*emins[0] + p->normal[1]*emins[1] + p->normal[2]*emins[2]; -dist2 = p->normal[0]*emaxs[0] + p->normal[1]*emaxs[1] + p->normal[2]*emaxs[2]; + case 2: // z axis + p->BoxOnPlaneSideFunc = BoxOnPlaneSideZ; break; default: - dist1 = dist2 = 0; // shut up compiler - BOPS_Error (); - break; - } - -#if 0 - int i; - vec3_t corners[2]; - - for (i=0 ; i<3 ; i++) - { - if (plane->normal[i] < 0) + if (p->normal[2] < 0) // 4 { - corners[0][i] = emins[i]; - corners[1][i] = emaxs[i]; + if (p->normal[1] < 0) // 2 + { + if (p->normal[0] < 0) // 1 + p->BoxOnPlaneSideFunc = BoxOnPlaneSide7; + else + p->BoxOnPlaneSideFunc = BoxOnPlaneSide6; + } + else + { + if (p->normal[0] < 0) // 1 + p->BoxOnPlaneSideFunc = BoxOnPlaneSide5; + else + p->BoxOnPlaneSideFunc = BoxOnPlaneSide4; + } } else { - corners[1][i] = emins[i]; - corners[0][i] = emaxs[i]; + if (p->normal[1] < 0) // 2 + { + if (p->normal[0] < 0) // 1 + p->BoxOnPlaneSideFunc = BoxOnPlaneSide3; + else + p->BoxOnPlaneSideFunc = BoxOnPlaneSide2; + } + else + { + if (p->normal[0] < 0) // 1 + p->BoxOnPlaneSideFunc = BoxOnPlaneSide1; + else + p->BoxOnPlaneSideFunc = BoxOnPlaneSide0; + } } + break; } - dist = DotProduct (plane->normal, corners[0]) - plane->dist; - dist2 = DotProduct (plane->normal, corners[1]) - plane->dist; - sides = 0; - if (dist1 >= 0) - sides = 1; - if (dist2 < 0) - sides |= 2; - -#endif - - sides = 0; - if (dist1 >= p->dist) - sides = 1; - if (dist2 < p->dist) - sides |= 2; - -#ifdef PARANOID -if (sides == 0) - Sys_Error ("BoxOnPlaneSide: sides==0"); -#endif - - return sides; } -#endif -*/ - -int BoxOnPlaneSide0 (vec3_t emins, vec3_t emaxs, mplane_t *p) {return (((p->normal[0]*emaxs[0] + p->normal[1]*emaxs[1] + p->normal[2]*emaxs[2]) >= p->dist) | (((p->normal[0]*emins[0] + p->normal[1]*emins[1] + p->normal[2]*emins[2]) < p->dist) << 1));} -int BoxOnPlaneSide1 (vec3_t emins, vec3_t emaxs, mplane_t *p) {return (((p->normal[0]*emins[0] + p->normal[1]*emaxs[1] + p->normal[2]*emaxs[2]) >= p->dist) | (((p->normal[0]*emaxs[0] + p->normal[1]*emins[1] + p->normal[2]*emins[2]) < p->dist) << 1));} -int BoxOnPlaneSide2 (vec3_t emins, vec3_t emaxs, mplane_t *p) {return (((p->normal[0]*emaxs[0] + p->normal[1]*emins[1] + p->normal[2]*emaxs[2]) >= p->dist) | (((p->normal[0]*emins[0] + p->normal[1]*emaxs[1] + p->normal[2]*emins[2]) < p->dist) << 1));} -int BoxOnPlaneSide3 (vec3_t emins, vec3_t emaxs, mplane_t *p) {return (((p->normal[0]*emins[0] + p->normal[1]*emins[1] + p->normal[2]*emaxs[2]) >= p->dist) | (((p->normal[0]*emaxs[0] + p->normal[1]*emaxs[1] + p->normal[2]*emins[2]) < p->dist) << 1));} -int BoxOnPlaneSide4 (vec3_t emins, vec3_t emaxs, mplane_t *p) {return (((p->normal[0]*emaxs[0] + p->normal[1]*emaxs[1] + p->normal[2]*emins[2]) >= p->dist) | (((p->normal[0]*emins[0] + p->normal[1]*emins[1] + p->normal[2]*emaxs[2]) < p->dist) << 1));} -int BoxOnPlaneSide5 (vec3_t emins, vec3_t emaxs, mplane_t *p) {return (((p->normal[0]*emins[0] + p->normal[1]*emaxs[1] + p->normal[2]*emins[2]) >= p->dist) | (((p->normal[0]*emaxs[0] + p->normal[1]*emins[1] + p->normal[2]*emaxs[2]) < p->dist) << 1));} -int BoxOnPlaneSide6 (vec3_t emins, vec3_t emaxs, mplane_t *p) {return (((p->normal[0]*emaxs[0] + p->normal[1]*emins[1] + p->normal[2]*emins[2]) >= p->dist) | (((p->normal[0]*emins[0] + p->normal[1]*emaxs[1] + p->normal[2]*emaxs[2]) < p->dist) << 1));} -int BoxOnPlaneSide7 (vec3_t emins, vec3_t emaxs, mplane_t *p) {return (((p->normal[0]*emins[0] + p->normal[1]*emins[1] + p->normal[2]*emins[2]) >= p->dist) | (((p->normal[0]*emaxs[0] + p->normal[1]*emaxs[1] + p->normal[2]*emaxs[2]) < p->dist) << 1));} +void PlaneClassify(mplane_t *p) +{ + if (p->normal[0] == 1) + p->type = 0; + else if (p->normal[1] == 1) + p->type = 1; + else if (p->normal[2] == 1) + p->type = 2; + else + p->type = 3; + BoxOnPlaneSideClassify(p); +} -void BoxOnPlaneSideClassify(mplane_t *p) +void AngleVectors (vec3_t angles, vec3_t forward, vec3_t right, vec3_t up) { - if (p->normal[2] < 0) // 4 + double angle, sr, sp, sy, cr, cp, cy; + + angle = angles[YAW] * (M_PI*2 / 360); + sy = sin(angle); + cy = cos(angle); + angle = angles[PITCH] * (M_PI*2 / 360); + sp = sin(angle); + cp = cos(angle); + if (forward) { - if (p->normal[1] < 0) // 2 + forward[0] = cp*cy; + forward[1] = cp*sy; + forward[2] = -sp; + } + if (right || up) + { + angle = angles[ROLL] * (M_PI*2 / 360); + sr = sin(angle); + cr = cos(angle); + if (right) { - if (p->normal[0] < 0) // 1 - p->BoxOnPlaneSideFunc = BoxOnPlaneSide7; - else - p->BoxOnPlaneSideFunc = BoxOnPlaneSide6; + right[0] = -1*(sr*sp*cy+cr*-sy); + right[1] = -1*(sr*sp*sy+cr*cy); + right[2] = -1*(sr*cp); } - else + if (up) { - if (p->normal[0] < 0) // 1 - p->BoxOnPlaneSideFunc = BoxOnPlaneSide5; - else - p->BoxOnPlaneSideFunc = BoxOnPlaneSide4; + up[0] = (cr*sp*cy+-sr*-sy); + up[1] = (cr*sp*sy+-sr*cy); + up[2] = cr*cp; } } - else +} + +void AngleVectorsFLU (vec3_t angles, vec3_t forward, vec3_t left, vec3_t up) +{ + double angle, sr, sp, sy, cr, cp, cy; + + angle = angles[YAW] * (M_PI*2 / 360); + sy = sin(angle); + cy = cos(angle); + angle = angles[PITCH] * (M_PI*2 / 360); + sp = sin(angle); + cp = cos(angle); + if (forward) { - if (p->normal[1] < 0) // 2 + forward[0] = cp*cy; + forward[1] = cp*sy; + forward[2] = -sp; + } + if (left || up) + { + angle = angles[ROLL] * (M_PI*2 / 360); + sr = sin(angle); + cr = cos(angle); + if (left) { - if (p->normal[0] < 0) // 1 - p->BoxOnPlaneSideFunc = BoxOnPlaneSide3; - else - p->BoxOnPlaneSideFunc = BoxOnPlaneSide2; + left[0] = sr*sp*cy+cr*-sy; + left[1] = sr*sp*sy+cr*cy; + left[2] = sr*cp; } - else + if (up) { - if (p->normal[0] < 0) // 1 - p->BoxOnPlaneSideFunc = BoxOnPlaneSide1; - else - p->BoxOnPlaneSideFunc = BoxOnPlaneSide0; + up[0] = cr*sp*cy+-sr*-sy; + up[1] = cr*sp*sy+-sr*cy; + up[2] = cr*cp; } } } -void AngleVectors (vec3_t angles, vec3_t forward, vec3_t right, vec3_t up) +void AngleMatrix (vec3_t angles, vec3_t translate, vec_t matrix[][4]) { - float angle; - float sr, sp, sy, cr, cp, cy; - + double angle, sr, sp, sy, cr, cp, cy; + angle = angles[YAW] * (M_PI*2 / 360); sy = sin(angle); cy = cos(angle); @@ -353,30 +669,21 @@ void AngleVectors (vec3_t angles, vec3_t forward, vec3_t right, vec3_t up) angle = angles[ROLL] * (M_PI*2 / 360); sr = sin(angle); cr = cos(angle); - - forward[0] = cp*cy; - forward[1] = cp*sy; - forward[2] = -sp; - right[0] = (-1*sr*sp*cy+-1*cr*-sy); - right[1] = (-1*sr*sp*sy+-1*cr*cy); - right[2] = -1*sr*cp; - up[0] = (cr*sp*cy+-sr*-sy); - up[1] = (cr*sp*sy+-sr*cy); - up[2] = cr*cp; + matrix[0][0] = cp*cy; + matrix[0][1] = sr*sp*cy+cr*-sy; + matrix[0][2] = cr*sp*cy+-sr*-sy; + matrix[0][3] = translate[0]; + matrix[1][0] = cp*sy; + matrix[1][1] = sr*sp*sy+cr*cy; + matrix[1][2] = cr*sp*sy+-sr*cy; + matrix[1][3] = translate[1]; + matrix[2][0] = -sp; + matrix[2][1] = sr*cp; + matrix[2][2] = cr*cp; + matrix[2][3] = translate[2]; } -int VectorCompare (vec3_t v1, vec3_t v2) -{ - int i; - - for (i=0 ; i<3 ; i++) - if (v1[i] != v2[i]) - return 0; - - return 1; -} - -void VectorMA (vec3_t veca, float scale, vec3_t vecb, vec3_t vecc) +void VectorMASlow (vec3_t veca, float scale, vec3_t vecb, vec3_t vecc) { vecc[0] = veca[0] + scale*vecb[0]; vecc[1] = veca[1] + scale*vecb[1]; @@ -426,7 +733,7 @@ vec_t Length(vec3_t v) { int i; float length; - + length = 0; for (i=0 ; i< 3 ; i++) length += v[i]*v[i]; @@ -435,6 +742,27 @@ vec_t Length(vec3_t v) return length; } +/* +// LordHavoc: fixme: do more research on gcc assembly so that qftol_minushalf and result will not be considered unused +static double qftol_minushalf = -0.5; + +int qftol(double v) +{ + int result; +#ifdef _MSC_VER + __asm + { + fld v + fadd qftol_minushalf + fistp result + } +#else // gcc hopefully + asm("fldl v\n\tfaddl qftol_minushalf\n\tfistpl result"); +#endif + return result; +} +*/ + // LordHavoc: renamed these to Length, and made the normal ones #define float VectorNormalizeLength (vec3_t v) { @@ -476,14 +804,14 @@ float VectorNormalizeLength2 (vec3_t v, vec3_t dest) // LordHavoc: added to allo } -void VectorInverse (vec3_t v) +void _VectorInverse (vec3_t v) { v[0] = -v[0]; v[1] = -v[1]; v[2] = -v[2]; } -void VectorScale (vec3_t in, vec_t scale, vec3_t out) +void _VectorScale (vec3_t in, vec_t scale, vec3_t out) { out[0] = in[0]*scale; out[1] = in[1]*scale; @@ -615,25 +943,12 @@ int GreatestCommonDivisor (int i1, int i2) } -#if !id386 - -// TODO: move to nonintel.c - -/* -=================== -Invert24To16 - -Inverts an 8.24 value to a 16.16 value -==================== -*/ - -fixed16_t Invert24To16(fixed16_t val) +void Mathlib_Init(void) { - if (val < 256) - return (0xFFFFFFFF); + int a; - return (fixed16_t) - (((double)0x10000 * (double)0x1000000 / (double)val) + 0.5); + // LordHavoc: setup 1.0f / N table for quick recipricols of integers + ixtable[0] = 0; + for (a = 1;a < 4096;a++) + ixtable[a] = 1.0f / a; } - -#endif