| 1 | /* |
| 2 | Copyright (C) 2003 Rice1964 |
| 3 | |
| 4 | This program is free software; you can redistribute it and/or |
| 5 | modify it under the terms of the GNU General Public License |
| 6 | as published by the Free Software Foundation; either version 2 |
| 7 | of the License, or (at your option) any later version. |
| 8 | |
| 9 | This program is distributed in the hope that it will be useful, |
| 10 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 11 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 12 | GNU General Public License for more details. |
| 13 | |
| 14 | You should have received a copy of the GNU General Public License |
| 15 | along with this program; if not, write to the Free Software |
| 16 | Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. |
| 17 | |
| 18 | */ |
| 19 | |
| 20 | #include <cmath> |
| 21 | #include <vector> |
| 22 | |
| 23 | #include "osal_preproc.h" |
| 24 | #include "float.h" |
| 25 | #include "DeviceBuilder.h" |
| 26 | #include "VertexShaderConstantDef.h" |
| 27 | #include "Render.h" |
| 28 | #include "Timing.h" |
| 29 | |
| 30 | extern FiddledVtx * g_pVtxBase; |
| 31 | |
| 32 | #define ENABLE_CLIP_TRI |
| 33 | #define X_CLIP_MAX 0x1 |
| 34 | #define X_CLIP_MIN 0x2 |
| 35 | #define Y_CLIP_MAX 0x4 |
| 36 | #define Y_CLIP_MIN 0x8 |
| 37 | #define Z_CLIP_MAX 0x10 |
| 38 | #define Z_CLIP_MIN 0x20 |
| 39 | |
| 40 | #ifdef ENABLE_CLIP_TRI |
| 41 | |
| 42 | inline void RSP_Vtx_Clipping(int i) |
| 43 | { |
| 44 | g_clipFlag[i] = 0; |
| 45 | g_clipFlag2[i] = 0; |
| 46 | if( g_vecProjected[i].w > 0 ) |
| 47 | { |
| 48 | /* |
| 49 | if( gRSP.bRejectVtx ) |
| 50 | { |
| 51 | if( g_vecProjected[i].x > 1 ) |
| 52 | { |
| 53 | g_clipFlag2[i] |= X_CLIP_MAX; |
| 54 | if( g_vecProjected[i].x > gRSP.real_clip_ratio_posx ) |
| 55 | g_clipFlag[i] |= X_CLIP_MAX; |
| 56 | } |
| 57 | |
| 58 | if( g_vecProjected[i].x < -1 ) |
| 59 | { |
| 60 | g_clipFlag2[i] |= X_CLIP_MIN; |
| 61 | if( g_vecProjected[i].x < gRSP.real_clip_ratio_negx ) |
| 62 | g_clipFlag[i] |= X_CLIP_MIN; |
| 63 | } |
| 64 | |
| 65 | if( g_vecProjected[i].y > 1 ) |
| 66 | { |
| 67 | g_clipFlag2[i] |= Y_CLIP_MAX; |
| 68 | if( g_vecProjected[i].y > gRSP.real_clip_ratio_posy ) |
| 69 | g_clipFlag[i] |= Y_CLIP_MAX; |
| 70 | } |
| 71 | |
| 72 | if( g_vecProjected[i].y < -1 ) |
| 73 | { |
| 74 | g_clipFlag2[i] |= Y_CLIP_MIN; |
| 75 | if( g_vecProjected[i].y < gRSP.real_clip_ratio_negy ) |
| 76 | g_clipFlag[i] |= Y_CLIP_MIN; |
| 77 | } |
| 78 | |
| 79 | //if( g_vecProjected[i].z > 1.0f ) |
| 80 | //{ |
| 81 | // g_clipFlag2[i] |= Z_CLIP_MAX; |
| 82 | // g_clipFlag[i] |= Z_CLIP_MAX; |
| 83 | //} |
| 84 | |
| 85 | //if( gRSP.bNearClip && g_vecProjected[i].z < -1.0f ) |
| 86 | //{ |
| 87 | // g_clipFlag2[i] |= Z_CLIP_MIN; |
| 88 | // g_clipFlag[i] |= Z_CLIP_MIN; |
| 89 | //} |
| 90 | } |
| 91 | else |
| 92 | */ |
| 93 | { |
| 94 | if( g_vecProjected[i].x > 1 ) g_clipFlag2[i] |= X_CLIP_MAX; |
| 95 | if( g_vecProjected[i].x < -1 ) g_clipFlag2[i] |= X_CLIP_MIN; |
| 96 | if( g_vecProjected[i].y > 1 ) g_clipFlag2[i] |= Y_CLIP_MAX; |
| 97 | if( g_vecProjected[i].y < -1 ) g_clipFlag2[i] |= Y_CLIP_MIN; |
| 98 | //if( g_vecProjected[i].z > 1.0f ) g_clipFlag2[i] |= Z_CLIP_MAX; |
| 99 | //if( gRSP.bNearClip && g_vecProjected[i].z < -1.0f ) g_clipFlag2[i] |= Z_CLIP_MIN; |
| 100 | } |
| 101 | |
| 102 | } |
| 103 | } |
| 104 | |
| 105 | #else |
| 106 | inline void RSP_Vtx_Clipping(int i) {} |
| 107 | #endif |
| 108 | |
| 109 | /* |
| 110 | * Global variables |
| 111 | */ |
| 112 | ALIGN(16,RSP_Options gRSP) |
| 113 | ALIGN(16,RDP_Options gRDP) |
| 114 | |
| 115 | static ALIGN(16,XVECTOR4 g_normal) |
| 116 | //static int norms[3]; |
| 117 | |
| 118 | ALIGN(16,XVECTOR4 g_vtxNonTransformed[MAX_VERTS]) |
| 119 | ALIGN(16,XVECTOR4 g_vecProjected[MAX_VERTS]) |
| 120 | ALIGN(16,XVECTOR4 g_vtxTransformed[MAX_VERTS]) |
| 121 | |
| 122 | float g_vtxProjected5[1000][5]; |
| 123 | float g_vtxProjected5Clipped[2000][5]; |
| 124 | |
| 125 | //uint32 g_dwVtxFlags[MAX_VERTS]; // Z_POS Z_NEG etc |
| 126 | VECTOR2 g_fVtxTxtCoords[MAX_VERTS]; |
| 127 | uint32 g_dwVtxDifColor[MAX_VERTS]; |
| 128 | uint32 g_clipFlag[MAX_VERTS]; |
| 129 | uint32 g_clipFlag2[MAX_VERTS]; |
| 130 | RenderTexture g_textures[MAX_TEXTURES]; |
| 131 | float g_fFogCoord[MAX_VERTS]; |
| 132 | |
| 133 | EXTERNAL_VERTEX g_vtxForExternal[MAX_VERTS]; |
| 134 | |
| 135 | TLITVERTEX g_vtxBuffer[1000]; |
| 136 | TLITVERTEX g_clippedVtxBuffer[2000]; |
| 137 | uint8 g_oglVtxColors[1000][4]; |
| 138 | int g_clippedVtxCount=0; |
| 139 | TLITVERTEX g_texRectTVtx[4]; |
| 140 | unsigned short g_vtxIndex[1000]; |
| 141 | unsigned int g_minIndex, g_maxIndex; |
| 142 | |
| 143 | float gRSPfFogMin; |
| 144 | float gRSPfFogMax; |
| 145 | float gRSPfFogDivider; |
| 146 | |
| 147 | uint32 gRSPnumLights; |
| 148 | Light gRSPlights[16]; |
| 149 | |
| 150 | ALIGN(16,Matrix gRSPworldProjectTransported) |
| 151 | ALIGN(16,Matrix gRSPworldProject) |
| 152 | ALIGN(16,Matrix gRSPmodelViewTop) |
| 153 | ALIGN(16,Matrix gRSPmodelViewTopTranspose) |
| 154 | ALIGN(16,Matrix dkrMatrixTransposed) |
| 155 | |
| 156 | N64Light gRSPn64lights[16]; |
| 157 | |
| 158 | |
| 159 | void (*ProcessVertexData)(uint32 dwAddr, uint32 dwV0, uint32 dwNum)=NULL; |
| 160 | |
| 161 | /* |
| 162 | * |
| 163 | */ |
| 164 | |
| 165 | |
| 166 | /*n.x = (g_normal.x * matWorld.m00) + (g_normal.y * matWorld.m10) + (g_normal.z * matWorld.m20); |
| 167 | n.y = (g_normal.x * matWorld.m01) + (g_normal.y * matWorld.m11) + (g_normal.z * matWorld.m21); |
| 168 | n.z = (g_normal.x * matWorld.m02) + (g_normal.y * matWorld.m12) + (g_normal.z * matWorld.m22);*/ |
| 169 | |
| 170 | // Multiply (x,y,z,0) by matrix m, then normalize |
| 171 | #if defined(__INTEL_COMPILER) && !defined(NO_ASM) |
| 172 | #define Vec3TransformNormal(vec, m) __asm \ |
| 173 | { \ |
| 174 | __asm fld dword ptr [vec + 0] \ |
| 175 | __asm fmul dword ptr [m + 0] \ /* x m00*/ |
| 176 | __asm fld dword ptr [vec + 0] \ |
| 177 | __asm fmul dword ptr [m + 4] \ /* x m01 x m00*/ |
| 178 | __asm fld dword ptr [vec + 0] \ |
| 179 | __asm fmul dword ptr [m + 8] \ /* x m02 x m01 x m00*/ |
| 180 | \ |
| 181 | __asm fld dword ptr [vec + 4] \ |
| 182 | __asm fmul dword ptr [m + 16] \ /* y m10 x m02 x m01 x m00*/ |
| 183 | __asm fld dword ptr [vec + 4] \ |
| 184 | __asm fmul dword ptr [m + 20] \ /* y m11 y m10 x m02 x m01 x m00*/ |
| 185 | __asm fld dword ptr [vec + 4] \ |
| 186 | __asm fmul dword ptr [m + 24] \ /* y m12 y m11 y m10 x m02 x m01 x m00*/ |
| 187 | \ |
| 188 | __asm fxch st(2) \ /* y m10 y m11 y m12 x m02 x m01 x m00*/ |
| 189 | __asm faddp st(5), st(0) \ /* y m11 y m12 x m02 x m01 (x m00 + y m10)*/ |
| 190 | __asm faddp st(3), st(0) \ /* y m12 x m02 (x m01 + ym11) (x m00 + y m10)*/ |
| 191 | __asm faddp st(1), st(0) \ /* (x m02 + y m12) (x m01 + ym11) (x m00 + y m10)*/ |
| 192 | \ |
| 193 | __asm fld dword ptr [vec + 8] \ |
| 194 | __asm fmul dword ptr [m + 32] \ /* z m20 (x m02 + y m12) (x m01 + ym11) (x m00 + y m10)*/ |
| 195 | __asm fld dword ptr [vec + 8] \ |
| 196 | __asm fmul dword ptr [m + 36] \ /* z m21 z m20 (x m02 + y m12) (x m01 + ym11) (x m00 + y m10)*/ |
| 197 | __asm fld dword ptr [vec + 8] \ |
| 198 | __asm fmul dword ptr [m + 40] \ /* z m22 z m21 z m20 (x m02 + y m12) (x m01 + ym11) (x m00 + y m10)*/ |
| 199 | \ |
| 200 | __asm fxch st(2) \ /* z m20 z m21 z m22 (x m02 + y m12) (x m01 + ym11) (x m00 + y m10)*/ |
| 201 | __asm faddp st(5), st(0) \ /* z m21 z m22 (x m02 + y m12) (x m01 + ym11) (x m00 + y m10 + z m20)*/ |
| 202 | __asm faddp st(3), st(0) \ /* z m22 (x m02 + y m12) (x m01 + ym11 + z m21) (x m00 + y m10 + z m20)*/ |
| 203 | __asm faddp st(1), st(0) \ /* (x m02 + y m12 + z m 22) (x m01 + ym11 + z m21) (x m00 + y m10 + z m20)*/ |
| 204 | \ |
| 205 | __asm fxch st(2) \ /* (x m00 + y m10 + z m20) (x m01 + ym11 + z m21) (x m02 + y m12 + z m 22) */ |
| 206 | \ |
| 207 | __asm fld1 \ /* 1 x y z */ |
| 208 | __asm fld st(1) \ /* x 1 x y z */ |
| 209 | __asm fmul st(0),st(0) \ /* xx 1 x y z */ |
| 210 | __asm fld st(3) \ /* y xx 1 x y z */ |
| 211 | __asm fmul st(0),st(0) \ /* yy xx 1 x y z */ |
| 212 | __asm fld st(5) \ /* z yy xx 1 x y z */ |
| 213 | __asm fmul st(0),st(0) \ /* zz yy xx 1 x y z */ |
| 214 | \ |
| 215 | __asm fxch st(2) \ /* xx yy zz 1 x y z */ |
| 216 | \ |
| 217 | __asm faddp st(1),st(0) \ /* (xx+yy) zz 1 x y z */ |
| 218 | __asm faddp st(1),st(0) \ /* (xx+yy+zz) 1 x y z */ |
| 219 | \ |
| 220 | __asm ftst \ /* Compare ST to 0 */ |
| 221 | __asm fstsw ax \ /* Store FPU status word in a */ |
| 222 | __asm sahf \ /* Transfer ax to flags register */ |
| 223 | __asm jz l2 \ /* Skip if length is zero */ |
| 224 | \ |
| 225 | __asm fsqrt \ /* l 1 x y z */ |
| 226 | \ |
| 227 | __asm fdivp st(1),st(0) \ /* (1/l) x y z */ |
| 228 | \ |
| 229 | __asm fmul st(3),st(0) \ /* f x y fz */ |
| 230 | __asm fmul st(2),st(0) \ /* f x fy fz */ |
| 231 | __asm fmulp st(1),st(0) \ /* fx fy fz */ |
| 232 | \ |
| 233 | __asm fstp dword ptr [vec + 0] \ /* fy fz*/ |
| 234 | __asm fstp dword ptr [vec + 4] \ /* fz */ |
| 235 | __asm fstp dword ptr [vec + 8] \ /* done */ |
| 236 | __asm jmp l3 \ |
| 237 | __asm l2: \ |
| 238 | __asm mov dword ptr [vec + 0], 0 \ |
| 239 | __asm mov dword ptr [vec + 4], 0 \ |
| 240 | __asm mov dword ptr [vec + 8], 0 \ |
| 241 | __asm l3: \ |
| 242 | } \ |
| 243 | |
| 244 | //#else // use C code in other cases, this is probably faster anyway |
| 245 | #elif defined(__ARM_NEON__0) |
| 246 | #define Vec3TransformNormal(vec, mtx) \ |
| 247 | { \ |
| 248 | asm volatile ( \ |
| 249 | "vld1.32 {d0}, [%0] \n\t" \ |
| 250 | "flds s2, [%0, #8] \n\t" \ |
| 251 | "vld1.32 {d18, d19}, [%1] \n\t" \ |
| 252 | "vld1.32 {d20, d21}, [%1, #16] \n\t" \ |
| 253 | "vld1.32 {d22, d23}, [%1, #32] \n\t" \ |
| 254 | \ |
| 255 | "vmul.f32 q2, q9, d0[0] \n\t" \ |
| 256 | "vmla.f32 q2, q10, d0[1] \n\t" \ |
| 257 | "vmla.f32 q2, q11, d1[0] \n\t" \ |
| 258 | \ |
| 259 | "vmul.f32 d0, d4, d4 \n\t" \ |
| 260 | "vpadd.f32 d0, d0, d0 \n\t" \ |
| 261 | "vmla.f32 d0, d5, d5 \n\t" \ |
| 262 | \ |
| 263 | "vmov.f32 d1, d0 \n\t" \ |
| 264 | "vrsqrte.f32 d0, d0 \n\t" \ |
| 265 | "vmul.f32 d2, d0, d1 \n\t" \ |
| 266 | "vrsqrts.f32 d3, d2, d0 \n\t" \ |
| 267 | "vmul.f32 d0, d0, d3 \n\t" \ |
| 268 | "vmul.f32 d2, d0, d1 \n\t" \ |
| 269 | "vrsqrts.f32 d3, d2, d0 \n\t" \ |
| 270 | "vmul.f32 d0, d0, d3 \n\t" \ |
| 271 | \ |
| 272 | "vmul.f32 q2, q2, d0[0] \n\t" \ |
| 273 | \ |
| 274 | "vst1.32 {d4}, [%0] \n\t" \ |
| 275 | "fsts s10, [%0, #8] \n\t" \ |
| 276 | : :"r"(vec), "r"(&mtx._11) \ |
| 277 | : "d0","d1","d2","d3","d18","d19","d20","d21","d22", "d23", "memory" \ |
| 278 | ); \ |
| 279 | } |
| 280 | #else |
| 281 | #define Vec3TransformNormal(vec, m) \ |
| 282 | VECTOR3 temp; \ |
| 283 | temp.x = (vec.x * m._11) + (vec.y * m._21) + (vec.z * m._31); \ |
| 284 | temp.y = (vec.x * m._12) + (vec.y * m._22) + (vec.z * m._32); \ |
| 285 | temp.z = (vec.x * m._13) + (vec.y * m._23) + (vec.z * m._33); \ |
| 286 | float norm = sqrtf(temp.x*temp.x+temp.y*temp.y+temp.z*temp.z); \ |
| 287 | if (norm == 0.0) { vec.x = 0.0; vec.y = 0.0; vec.z = 0.0;} else \ |
| 288 | { float rep=1/norm;vec.x = temp.x*rep; vec.y = temp.y*rep; vec.z = temp.z*rep; } |
| 289 | #endif |
| 290 | |
| 291 | |
| 292 | #if !defined(__GNUC__) && !defined(NO_ASM) |
| 293 | __declspec( naked ) void __fastcall SSEVec3Transform(int i) |
| 294 | { |
| 295 | __asm |
| 296 | { |
| 297 | shl ecx,4; // ecx = i |
| 298 | |
| 299 | movaps xmm1, DWORD PTR g_vtxNonTransformed [ecx]; // xmm1 as original vector |
| 300 | |
| 301 | movaps xmm4, DWORD PTR gRSPworldProjectTransported; // row1 |
| 302 | movaps xmm5, DWORD PTR gRSPworldProjectTransported[0x10]; // row2 |
| 303 | movaps xmm6, DWORD PTR gRSPworldProjectTransported[0x20]; // row3 |
| 304 | movaps xmm7, DWORD PTR gRSPworldProjectTransported[0x30]; // row4 |
| 305 | |
| 306 | mulps xmm4, xmm1; // row 1 |
| 307 | mulps xmm5, xmm1; // row 2 |
| 308 | mulps xmm6, xmm1; // row 3 |
| 309 | mulps xmm7, xmm1; // row 4 |
| 310 | |
| 311 | movhlps xmm0, xmm4; // xmm4 high to xmm0 low |
| 312 | movlhps xmm0, xmm5; // xmm5 low to xmm0 high |
| 313 | |
| 314 | addps xmm4, xmm0; // result of add are in xmm4 low |
| 315 | addps xmm5, xmm0; // result of add are in xmm5 high |
| 316 | |
| 317 | shufps xmm0, xmm4, 0x44; // move xmm4 low DWORDs to xmm0 high |
| 318 | shufps xmm4, xmm5, 0xe4; // move xmm5 high DWORS to xmm4 |
| 319 | movhlps xmm5, xmm0; // xmm4, xmm5 are mirrored |
| 320 | |
| 321 | shufps xmm4, xmm4, 0x08; // move xmm4's 3rd uint32 to its 2nd uint32 |
| 322 | shufps xmm5, xmm5, 0x0d; // move xmm5's 4th uint32 to its 2nd uint32, |
| 323 | // and move its 2nd uint32 to its 1st uint32 |
| 324 | |
| 325 | addps xmm4, xmm5; // results are in 1st and 2nd uint32 |
| 326 | |
| 327 | |
| 328 | movhlps xmm0, xmm6; // xmm6 high to xmm0 low |
| 329 | movlhps xmm0, xmm7; // xmm7 low to xmm0 high |
| 330 | |
| 331 | addps xmm6, xmm0; // result of add are in xmm6 low |
| 332 | addps xmm7, xmm0; // result of add are in xmm7 high |
| 333 | |
| 334 | shufps xmm0, xmm6, 0x44; // move xmm6 low DWORDs to xmm0 high |
| 335 | shufps xmm6, xmm7, 0xe4; // move xmm7 high DWORS to xmm6 |
| 336 | movhlps xmm7, xmm0; // xmm6, xmm7 are mirrored |
| 337 | |
| 338 | shufps xmm6, xmm6, 0x08; // move xmm6's 3rd uint32 to its 2nd uint32 |
| 339 | shufps xmm7, xmm7, 0x0d; // move xmm7's 4th uint32 to its 2nd uint32, |
| 340 | // and move its 2nd uint32 to its 1st uint32 |
| 341 | |
| 342 | addps xmm6, xmm7; // results are in 1st and 2nd uint32 |
| 343 | |
| 344 | movlhps xmm4, xmm6; // final result is in xmm4 |
| 345 | movaps DWORD PTR g_vtxTransformed [ecx], xmm4; |
| 346 | |
| 347 | movaps xmm0,xmm4; |
| 348 | shufps xmm0,xmm0,0xff; |
| 349 | divps xmm4,xmm0; |
| 350 | rcpps xmm0,xmm0; |
| 351 | movhlps xmm0,xmm4; |
| 352 | shufps xmm0,xmm0,0xe8; |
| 353 | movlhps xmm4,xmm0; |
| 354 | |
| 355 | movaps DWORD PTR g_vecProjected [ecx], xmm4; |
| 356 | |
| 357 | emms; |
| 358 | ret; |
| 359 | } |
| 360 | } |
| 361 | |
| 362 | // Only used by DKR |
| 363 | __declspec( naked ) void __fastcall SSEVec3TransformDKR(XVECTOR4 &pOut, const XVECTOR4 &pV) |
| 364 | { |
| 365 | __asm |
| 366 | { |
| 367 | movaps xmm1, DWORD PTR [edx]; // xmm1 as original vector |
| 368 | |
| 369 | movaps xmm4, DWORD PTR dkrMatrixTransposed; // row1 |
| 370 | movaps xmm5, DWORD PTR dkrMatrixTransposed[0x10]; // row2 |
| 371 | movaps xmm6, DWORD PTR dkrMatrixTransposed[0x20]; // row3 |
| 372 | movaps xmm7, DWORD PTR dkrMatrixTransposed[0x30]; // row4 |
| 373 | |
| 374 | mulps xmm4, xmm1; // row 1 |
| 375 | mulps xmm5, xmm1; // row 2 |
| 376 | mulps xmm6, xmm1; // row 3 |
| 377 | mulps xmm7, xmm1; // row 4 |
| 378 | |
| 379 | movhlps xmm0, xmm4; // xmm4 high to xmm0 low |
| 380 | movlhps xmm0, xmm5; // xmm5 low to xmm0 high |
| 381 | |
| 382 | addps xmm4, xmm0; // result of add are in xmm4 low |
| 383 | addps xmm5, xmm0; // result of add are in xmm5 high |
| 384 | |
| 385 | shufps xmm0, xmm4, 0x44; // move xmm4 low DWORDs to xmm0 high |
| 386 | shufps xmm4, xmm5, 0xe4; // move xmm5 high DWORS to xmm4 |
| 387 | movhlps xmm5, xmm0; // xmm4, xmm5 are mirrored |
| 388 | |
| 389 | shufps xmm4, xmm4, 0x08; // move xmm4's 3rd uint32 to its 2nd uint32 |
| 390 | shufps xmm5, xmm5, 0x0d; // move xmm5's 4th uint32 to its 2nd uint32, |
| 391 | // and move its 2nd uint32 to its 1st uint32 |
| 392 | |
| 393 | addps xmm4, xmm5; // results are in 1st and 2nd uint32 |
| 394 | |
| 395 | |
| 396 | movhlps xmm0, xmm6; // xmm6 high to xmm0 low |
| 397 | movlhps xmm0, xmm7; // xmm7 low to xmm0 high |
| 398 | |
| 399 | addps xmm6, xmm0; // result of add are in xmm6 low |
| 400 | addps xmm7, xmm0; // result of add are in xmm7 high |
| 401 | |
| 402 | shufps xmm0, xmm6, 0x44; // move xmm6 low DWORDs to xmm0 high |
| 403 | shufps xmm6, xmm7, 0xe4; // move xmm7 high DWORS to xmm6 |
| 404 | movhlps xmm7, xmm0; // xmm6, xmm7 are mirrored |
| 405 | |
| 406 | shufps xmm6, xmm6, 0x08; // move xmm6's 3rd uint32 to its 2nd uint32 |
| 407 | shufps xmm7, xmm7, 0x0d; // move xmm7's 4th uint32 to its 2nd uint32, |
| 408 | // and move its 2nd uint32 to its 1st uint32 |
| 409 | |
| 410 | addps xmm6, xmm7; // results are in 1st and 2nd uint32 |
| 411 | |
| 412 | movlhps xmm4, xmm6; // final result is in xmm4 |
| 413 | movaps DWORD PTR [ecx], xmm4; |
| 414 | |
| 415 | emms; |
| 416 | ret; |
| 417 | } |
| 418 | } |
| 419 | #elif defined(__GNUC__) && defined(__x86_64__) && !defined(NO_ASM) |
| 420 | void SSEVec3Transform(int i) |
| 421 | { |
| 422 | asm volatile(" shl $4, %0 \n" |
| 423 | " movslq %k0, %q0 \n" |
| 424 | " movaps (%1,%q0), %%xmm1 \n" |
| 425 | " movaps 0(%2), %%xmm4 \n" |
| 426 | " movaps 16(%2), %%xmm5 \n" |
| 427 | " movaps 32(%2), %%xmm6 \n" |
| 428 | " movaps 48(%2), %%xmm7 \n" |
| 429 | " mulps %%xmm1, %%xmm4 \n" |
| 430 | " mulps %%xmm1, %%xmm5 \n" |
| 431 | " mulps %%xmm1, %%xmm6 \n" |
| 432 | " mulps %%xmm1, %%xmm7 \n" |
| 433 | " movhlps %%xmm4, %%xmm0 \n" |
| 434 | " movlhps %%xmm5, %%xmm0 \n" |
| 435 | " addps %%xmm0, %%xmm4 \n" |
| 436 | " addps %%xmm0, %%xmm5 \n" |
| 437 | " shufps $0x44, %%xmm4, %%xmm0 \n" |
| 438 | " shufps $0xe4, %%xmm5, %%xmm4 \n" |
| 439 | " movhlps %%xmm0, %%xmm5 \n" |
| 440 | " shufps $0x08, %%xmm4, %%xmm4 \n" |
| 441 | " shufps $0x0d, %%xmm5, %%xmm5 \n" |
| 442 | " addps %%xmm5, %%xmm4 \n" |
| 443 | " movhlps %%xmm6, %%xmm0 \n" |
| 444 | " movlhps %%xmm7, %%xmm0 \n" |
| 445 | " addps %%xmm0, %%xmm6 \n" |
| 446 | " addps %%xmm0, %%xmm7 \n" |
| 447 | " shufps $0x44, %%xmm6, %%xmm0 \n" |
| 448 | " shufps $0xe4, %%xmm7, %%xmm6 \n" |
| 449 | " movhlps %%xmm0, %%xmm7 \n" |
| 450 | " shufps $0x08, %%xmm6, %%xmm6 \n" |
| 451 | " shufps $0x0d, %%xmm7, %%xmm7 \n" |
| 452 | " addps %%xmm7, %%xmm6 \n" |
| 453 | " movlhps %%xmm6, %%xmm4 \n" |
| 454 | " movaps %%xmm4, (%3,%q0) \n" |
| 455 | " movaps %%xmm4, %%xmm0 \n" |
| 456 | " shufps $0xff, %%xmm0, %%xmm0 \n" |
| 457 | " divps %%xmm0, %%xmm4 \n" |
| 458 | " rcpps %%xmm0, %%xmm0 \n" |
| 459 | " movhlps %%xmm4, %%xmm0 \n" |
| 460 | " shufps $0xe8, %%xmm0, %%xmm0 \n" |
| 461 | " movlhps %%xmm0, %%xmm4 \n" |
| 462 | " movaps %%xmm4, (%4,%q0) \n" |
| 463 | : "+r"(i) |
| 464 | : "r"(g_vtxNonTransformed), "r"(&gRSPworldProjectTransported.m[0][0]), "r"(g_vtxTransformed), "r"(g_vecProjected) |
| 465 | : "memory", "%xmm0", "%xmm1", "%xmm4", "%xmm5", "%xmm6", "%xmm7" |
| 466 | ); |
| 467 | } |
| 468 | #elif !defined(NO_ASM) // 32-bit GCC assumed |
| 469 | void SSEVec3Transform(int i) |
| 470 | { |
| 471 | asm volatile(" shl $4, %0 \n" |
| 472 | " movaps (%1,%0), %%xmm1 \n" |
| 473 | " movaps 0(%2), %%xmm4 \n" |
| 474 | " movaps 16(%2), %%xmm5 \n" |
| 475 | " movaps 32(%2), %%xmm6 \n" |
| 476 | " movaps 48(%2), %%xmm7 \n" |
| 477 | " mulps %%xmm1, %%xmm4 \n" |
| 478 | " mulps %%xmm1, %%xmm5 \n" |
| 479 | " mulps %%xmm1, %%xmm6 \n" |
| 480 | " mulps %%xmm1, %%xmm7 \n" |
| 481 | " movhlps %%xmm4, %%xmm0 \n" |
| 482 | " movlhps %%xmm5, %%xmm0 \n" |
| 483 | " addps %%xmm0, %%xmm4 \n" |
| 484 | " addps %%xmm0, %%xmm5 \n" |
| 485 | " shufps $0x44, %%xmm4, %%xmm0 \n" |
| 486 | " shufps $0xe4, %%xmm5, %%xmm4 \n" |
| 487 | " movhlps %%xmm0, %%xmm5 \n" |
| 488 | " shufps $0x08, %%xmm4, %%xmm4 \n" |
| 489 | " shufps $0x0d, %%xmm5, %%xmm5 \n" |
| 490 | " addps %%xmm5, %%xmm4 \n" |
| 491 | " movhlps %%xmm6, %%xmm0 \n" |
| 492 | " movlhps %%xmm7, %%xmm0 \n" |
| 493 | " addps %%xmm0, %%xmm6 \n" |
| 494 | " addps %%xmm0, %%xmm7 \n" |
| 495 | " shufps $0x44, %%xmm6, %%xmm0 \n" |
| 496 | " shufps $0xe4, %%xmm7, %%xmm6 \n" |
| 497 | " movhlps %%xmm0, %%xmm7 \n" |
| 498 | " shufps $0x08, %%xmm6, %%xmm6 \n" |
| 499 | " shufps $0x0d, %%xmm7, %%xmm7 \n" |
| 500 | " addps %%xmm7, %%xmm6 \n" |
| 501 | " movlhps %%xmm6, %%xmm4 \n" |
| 502 | " movaps %%xmm4, (%3,%0) \n" |
| 503 | " movaps %%xmm4, %%xmm0 \n" |
| 504 | " shufps $0xff, %%xmm0, %%xmm0 \n" |
| 505 | " divps %%xmm0, %%xmm4 \n" |
| 506 | " rcpps %%xmm0, %%xmm0 \n" |
| 507 | " movhlps %%xmm4, %%xmm0 \n" |
| 508 | " shufps $0xe8, %%xmm0, %%xmm0 \n" |
| 509 | " movlhps %%xmm0, %%xmm4 \n" |
| 510 | " movaps %%xmm4, (%4,%0) \n" |
| 511 | : "+r"(i) |
| 512 | : "r"(g_vtxNonTransformed), "r"(&gRSPworldProjectTransported.m[0][0]), "r"(g_vtxTransformed), "r"(g_vecProjected) |
| 513 | : "memory", "%xmm0", "%xmm1", "%xmm4", "%xmm5", "%xmm6", "%xmm7" |
| 514 | ); |
| 515 | } |
| 516 | #endif |
| 517 | float real255 = 255.0f; |
| 518 | float real128 = 128.0f; |
| 519 | |
| 520 | #if !defined(__GNUC__) && !defined(NO_ASM) |
| 521 | __declspec( naked ) void __fastcall SSEVec3TransformNormal() |
| 522 | { |
| 523 | __asm |
| 524 | { |
| 525 | mov DWORD PTR [g_normal][12], 0; |
| 526 | |
| 527 | movaps xmm4, DWORD PTR gRSPmodelViewTopTranspose; // row1 |
| 528 | movaps xmm5, DWORD PTR gRSPmodelViewTopTranspose[0x10]; // row2 |
| 529 | movaps xmm1, DWORD PTR [g_normal]; // xmm1 as the normal vector |
| 530 | movaps xmm6, DWORD PTR gRSPmodelViewTopTranspose[0x20]; // row3 |
| 531 | |
| 532 | mulps xmm4, xmm1; // row 1 |
| 533 | mulps xmm5, xmm1; // row 2 |
| 534 | mulps xmm6, xmm1; // row 3 |
| 535 | |
| 536 | movhlps xmm0, xmm4; // xmm4 high to xmm0 low |
| 537 | movlhps xmm0, xmm5; // xmm5 low to xmm0 high |
| 538 | |
| 539 | addps xmm4, xmm0; // result of add are in xmm4 low |
| 540 | addps xmm5, xmm0; // result of add are in xmm5 high |
| 541 | |
| 542 | shufps xmm0, xmm4, 0x44; // move xmm4 low DWORDs to xmm0 high |
| 543 | shufps xmm4, xmm5, 0xe4; // move xmm5 high DWORS to xmm4 |
| 544 | movhlps xmm5, xmm0; // xmm4, xmm5 are mirrored |
| 545 | |
| 546 | shufps xmm4, xmm4, 0x08; // move xmm4's 3rd uint32 to its 2nd uint32 |
| 547 | shufps xmm5, xmm5, 0x0d; // move xmm5's 4th uint32 to its 2nd uint32, |
| 548 | |
| 549 | addps xmm4, xmm5; // results are in 1st and 2nd uint32 |
| 550 | |
| 551 | movaps xmm1,xmm4; |
| 552 | mulps xmm1,xmm1; //square |
| 553 | movlhps xmm7, xmm1; |
| 554 | shufps xmm7, xmm7,0x03; |
| 555 | addss xmm7, xmm1; |
| 556 | |
| 557 | movhlps xmm0, xmm6; // xmm6 high to xmm0 low |
| 558 | addps xmm6, xmm0; // result of add are in xmm6 low |
| 559 | |
| 560 | movlhps xmm0, xmm6; |
| 561 | shufps xmm0, xmm0, 0x03; |
| 562 | addss xmm0, xmm6; // result of add is at xmm0's 1st uint32 |
| 563 | |
| 564 | movlhps xmm4, xmm0; |
| 565 | |
| 566 | mulss xmm0,xmm0; |
| 567 | addss xmm7,xmm0; // xmm7 1st uint32 is the sum of squares |
| 568 | |
| 569 | #ifdef DEBUGGER |
| 570 | movaps DWORD PTR [g_normal], xmm4; |
| 571 | movss DWORD PTR [g_normal][12], xmm7; |
| 572 | #endif |
| 573 | xorps xmm0,xmm0; |
| 574 | ucomiss xmm0,xmm7; |
| 575 | jz l2 |
| 576 | |
| 577 | rsqrtss xmm7,xmm7; |
| 578 | shufps xmm7,xmm7,0; |
| 579 | #ifdef DEBUGGER |
| 580 | movss DWORD PTR [g_normal][12], xmm7; |
| 581 | #endif |
| 582 | mulps xmm4,xmm7; |
| 583 | |
| 584 | movaps DWORD PTR [g_normal], xmm4; // Normalized |
| 585 | mov DWORD PTR [g_normal][12], 0; |
| 586 | |
| 587 | emms; |
| 588 | ret; |
| 589 | l2: |
| 590 | movss DWORD PTR [g_normal], xmm0; |
| 591 | movss DWORD PTR [g_normal][12], xmm0; |
| 592 | emms; |
| 593 | ret; |
| 594 | } |
| 595 | } |
| 596 | #elif defined(__GNUC__) && !defined(NO_ASM) // this code should compile for both 64-bit and 32-bit architectures |
| 597 | void SSEVec3TransformNormal(void) |
| 598 | { |
| 599 | asm volatile(" movl $0, 12(%0) \n" |
| 600 | " movaps (%1), %%xmm4 \n" |
| 601 | " movaps 16(%1), %%xmm5 \n" |
| 602 | " movaps (%0), %%xmm1 \n" |
| 603 | " movaps 32(%1), %%xmm6 \n" |
| 604 | " mulps %%xmm1, %%xmm4 \n" |
| 605 | " mulps %%xmm1, %%xmm5 \n" |
| 606 | " mulps %%xmm1, %%xmm6 \n" |
| 607 | " movhlps %%xmm4, %%xmm0 \n" |
| 608 | " movlhps %%xmm5, %%xmm0 \n" |
| 609 | " addps %%xmm0, %%xmm4 \n" |
| 610 | " addps %%xmm0, %%xmm5 \n" |
| 611 | " shufps $0x44, %%xmm4, %%xmm0 \n" |
| 612 | " shufps $0xe4, %%xmm5, %%xmm4 \n" |
| 613 | " movhlps %%xmm0, %%xmm5 \n" |
| 614 | " shufps $0x08, %%xmm4, %%xmm4 \n" |
| 615 | " shufps $0x0d, %%xmm5, %%xmm5 \n" |
| 616 | " addps %%xmm5, %%xmm4 \n" |
| 617 | " movaps %%xmm4, %%xmm1 \n" |
| 618 | " mulps %%xmm1, %%xmm1 \n" |
| 619 | " movlhps %%xmm1, %%xmm7 \n" |
| 620 | " shufps $0x03, %%xmm7, %%xmm7 \n" |
| 621 | " addss %%xmm1, %%xmm7 \n" |
| 622 | " movhlps %%xmm6, %%xmm0 \n" |
| 623 | " addps %%xmm0, %%xmm6 \n" |
| 624 | " movlhps %%xmm6, %%xmm0 \n" |
| 625 | " shufps $0x03, %%xmm0, %%xmm0 \n" |
| 626 | " addss %%xmm6, %%xmm0 \n" |
| 627 | " movlhps %%xmm0, %%xmm4 \n" |
| 628 | " mulss %%xmm0, %%xmm0 \n" |
| 629 | " addss %%xmm0, %%xmm7 \n" |
| 630 | #ifdef DEBUGGER |
| 631 | " movaps %%xmm4, (%0) \n" |
| 632 | " movss %%xmm7, 12(%0) \n" |
| 633 | #endif |
| 634 | " xorps %%xmm0, %%xmm0 \n" |
| 635 | " ucomiss %%xmm7, %%xmm0 \n" |
| 636 | " jz 0f \n" |
| 637 | " rsqrtss %%xmm7, %%xmm7 \n" |
| 638 | " shufps $0x00, %%xmm7, %%xmm7 \n" |
| 639 | #ifdef DEBUGGER |
| 640 | " movss %%xmm7, 12(%0) \n" |
| 641 | #endif |
| 642 | " mulps %%xmm7, %%xmm4 \n" |
| 643 | " movaps %%xmm4, (%0) \n" |
| 644 | " movl $0, 12(%0) \n" |
| 645 | " jmp 1f \n" |
| 646 | "0: \n" |
| 647 | " movss %%xmm0, (%0) \n" |
| 648 | " movss %%xmm0, 12(%0) \n" |
| 649 | "1: \n" |
| 650 | : |
| 651 | : "r"(&g_normal.x), "r"(&gRSPmodelViewTopTranspose.m[0][0]) |
| 652 | : "memory", "cc", "%xmm0", "%xmm1", "%xmm4", "%xmm5", "%xmm6", "%xmm7" |
| 653 | ); |
| 654 | } |
| 655 | #endif |
| 656 | |
| 657 | void NormalizeNormalVec() |
| 658 | #ifdef __ARM_NEON__0 |
| 659 | { |
| 660 | asm volatile ( |
| 661 | "vld1.32 {d4}, [%0] \n\t" //d4={x,y} |
| 662 | "flds s10, [%0, #8] \n\t" //d5[0] = z |
| 663 | // "sub %0, %0, #8 \n\t" //d5[0] = z |
| 664 | "vmul.f32 d0, d4, d4 \n\t" //d0= d4*d4 |
| 665 | "vpadd.f32 d0, d0, d0 \n\t" //d0 = d[0] + d[1] |
| 666 | "vmla.f32 d0, d5, d5 \n\t" //d0 = d0 + d5*d5 |
| 667 | |
| 668 | "vmov.f32 d1, d0 \n\t" //d1 = d0 |
| 669 | "vrsqrte.f32 d0, d0 \n\t" //d0 = ~ 1.0 / sqrt(d0) |
| 670 | "vmul.f32 d2, d0, d1 \n\t" //d2 = d0 * d1 |
| 671 | "vrsqrts.f32 d3, d2, d0 \n\t" //d3 = (3 - d0 * d2) / 2 |
| 672 | "vmul.f32 d0, d0, d3 \n\t" //d0 = d0 * d3 |
| 673 | "vmul.f32 d2, d0, d1 \n\t" //d2 = d0 * d1 |
| 674 | "vrsqrts.f32 d3, d2, d0 \n\t" //d3 = (3 - d0 * d3) / 2 |
| 675 | "vmul.f32 d0, d0, d3 \n\t" //d0 = d0 * d4 |
| 676 | |
| 677 | "vmul.f32 q2, q2, d0[0] \n\t" //d0= d2*d4 |
| 678 | "vst1.32 {d4}, [%0] \n\t" //d2={x0,y0}, d3={z0, w0} |
| 679 | "fsts s10, [%0, #8] \n\t" //d2={x0,y0}, d3={z0, w0} |
| 680 | |
| 681 | ::"r"(g_normal) |
| 682 | : "d0", "d1", "d2", "d3", "d4", "d5", "memory" |
| 683 | ); |
| 684 | } |
| 685 | #else |
| 686 | { |
| 687 | float w = 1/sqrtf(g_normal.x*g_normal.x + g_normal.y*g_normal.y + g_normal.z*g_normal.z); |
| 688 | g_normal.x *= w; |
| 689 | g_normal.y *= w; |
| 690 | g_normal.z *= w; |
| 691 | } |
| 692 | #endif |
| 693 | |
| 694 | void InitRenderBase() |
| 695 | { |
| 696 | #if !defined(NO_ASM) |
| 697 | if( status.isSSEEnabled && !g_curRomInfo.bPrimaryDepthHack && options.enableHackForGames != HACK_FOR_NASCAR) |
| 698 | { |
| 699 | ProcessVertexData = ProcessVertexDataSSE; |
| 700 | } |
| 701 | else |
| 702 | #endif |
| 703 | { |
| 704 | ProcessVertexData = ProcessVertexDataNoSSE; |
| 705 | } |
| 706 | |
| 707 | gRSPfFogMin = gRSPfFogMax = 0.0f; |
| 708 | windowSetting.fMultX = windowSetting.fMultY = 2.0f; |
| 709 | windowSetting.vpLeftW = windowSetting.vpTopW = 0; |
| 710 | windowSetting.vpRightW = windowSetting.vpWidthW = 640; |
| 711 | windowSetting.vpBottomW = windowSetting.vpHeightW = 480; |
| 712 | gRSP.maxZ = 0; |
| 713 | gRSP.nVPLeftN = gRSP.nVPTopN = 0; |
| 714 | gRSP.nVPRightN = 640; |
| 715 | gRSP.nVPBottomN = 640; |
| 716 | gRSP.nVPWidthN = 640; |
| 717 | gRSP.nVPHeightN = 640; |
| 718 | gRDP.scissor.left=gRDP.scissor.top=0; |
| 719 | gRDP.scissor.right=gRDP.scissor.bottom=640; |
| 720 | |
| 721 | gRSP.bLightingEnable = gRSP.bTextureGen = false; |
| 722 | gRSP.curTile=gRSPnumLights=gRSP.ambientLightColor=gRSP.ambientLightIndex= 0; |
| 723 | gRSP.fAmbientLightR=gRSP.fAmbientLightG=gRSP.fAmbientLightB=0; |
| 724 | gRSP.projectionMtxTop = gRSP.modelViewMtxTop = 0; |
| 725 | gRDP.fogColor = gRDP.primitiveColor = gRDP.envColor = gRDP.primitiveDepth = gRDP.primLODMin = gRDP.primLODFrac = gRDP.LODFrac = 0; |
| 726 | gRDP.fPrimitiveDepth = 0; |
| 727 | gRSP.numVertices = 0; |
| 728 | gRSP.maxVertexID = 0; |
| 729 | gRSP.bCullFront=false; |
| 730 | gRSP.bCullBack=true; |
| 731 | gRSP.bFogEnabled=gRDP.bFogEnableInBlender=false; |
| 732 | gRSP.bZBufferEnabled=true; |
| 733 | gRSP.shadeMode=SHADE_SMOOTH; |
| 734 | gRDP.keyR=gRDP.keyG=gRDP.keyB=gRDP.keyA=gRDP.keyRGB=gRDP.keyRGBA = 0; |
| 735 | gRDP.fKeyA = 0; |
| 736 | gRSP.DKRCMatrixIndex = gRSP.dwDKRVtxAddr = gRSP.dwDKRMatrixAddr = 0; |
| 737 | gRSP.DKRBillBoard = false; |
| 738 | |
| 739 | gRSP.fTexScaleX = 1/32.0f; |
| 740 | gRSP.fTexScaleY = 1/32.0f; |
| 741 | gRSP.bTextureEnabled = FALSE; |
| 742 | |
| 743 | gRSP.clip_ratio_left = 0; |
| 744 | gRSP.clip_ratio_top = 0; |
| 745 | gRSP.clip_ratio_right = 640; |
| 746 | gRSP.clip_ratio_bottom = 480; |
| 747 | gRSP.clip_ratio_negx = 1; |
| 748 | gRSP.clip_ratio_negy = 1; |
| 749 | gRSP.clip_ratio_posx = 1; |
| 750 | gRSP.clip_ratio_posy = 1; |
| 751 | gRSP.real_clip_scissor_left = 0; |
| 752 | gRSP.real_clip_scissor_top = 0; |
| 753 | gRSP.real_clip_scissor_right = 640; |
| 754 | gRSP.real_clip_scissor_bottom = 480; |
| 755 | windowSetting.clipping.left = 0; |
| 756 | windowSetting.clipping.top = 0; |
| 757 | windowSetting.clipping.right = 640; |
| 758 | windowSetting.clipping.bottom = 480; |
| 759 | windowSetting.clipping.width = 640; |
| 760 | windowSetting.clipping.height = 480; |
| 761 | windowSetting.clipping.needToClip = false; |
| 762 | gRSP.real_clip_ratio_negx = 1; |
| 763 | gRSP.real_clip_ratio_negy = 1; |
| 764 | gRSP.real_clip_ratio_posx = 1; |
| 765 | gRSP.real_clip_ratio_posy = 1; |
| 766 | |
| 767 | gRSP.DKRCMatrixIndex=0; |
| 768 | gRSP.DKRVtxCount=0; |
| 769 | gRSP.DKRBillBoard = false; |
| 770 | gRSP.dwDKRVtxAddr=0; |
| 771 | gRSP.dwDKRMatrixAddr=0; |
| 772 | |
| 773 | |
| 774 | gRDP.geometryMode = 0; |
| 775 | gRDP.otherModeL = 0; |
| 776 | gRDP.otherModeH = 0; |
| 777 | gRDP.fillColor = 0xFFFFFFFF; |
| 778 | gRDP.originalFillColor =0; |
| 779 | |
| 780 | gRSP.ucode = 1; |
| 781 | gRSP.vertexMult = 10; |
| 782 | gRSP.bNearClip = false; |
| 783 | gRSP.bRejectVtx = false; |
| 784 | |
| 785 | gRDP.texturesAreReloaded = false; |
| 786 | gRDP.textureIsChanged = false; |
| 787 | gRDP.colorsAreReloaded = false; |
| 788 | |
| 789 | memset(&gRDP.otherMode,0,sizeof(RDP_OtherMode)); |
| 790 | memset(&gRDP.tiles,0,sizeof(Tile)*8); |
| 791 | |
| 792 | for( int i=0; i<MAX_VERTS; i++ ) |
| 793 | { |
| 794 | g_clipFlag[i] = 0; |
| 795 | g_vtxNonTransformed[i].w = 1; |
| 796 | } |
| 797 | |
| 798 | memset(gRSPn64lights, 0, sizeof(N64Light)*16); |
| 799 | } |
| 800 | |
| 801 | void SetFogMinMax(float fMin, float fMax, float fMul, float fOffset) |
| 802 | { |
| 803 | if( fMin > fMax ) |
| 804 | { |
| 805 | float temp = fMin; |
| 806 | fMin = fMax; |
| 807 | fMax = temp; |
| 808 | } |
| 809 | |
| 810 | { |
| 811 | gRSPfFogMin = max(0,fMin/500-1); |
| 812 | gRSPfFogMax = fMax/500-1; |
| 813 | } |
| 814 | |
| 815 | gRSPfFogDivider = 255/(gRSPfFogMax-gRSPfFogMin); |
| 816 | CRender::g_pRender->SetFogMinMax(fMin, fMax); |
| 817 | } |
| 818 | |
| 819 | void InitVertexColors() |
| 820 | { |
| 821 | } |
| 822 | |
| 823 | void InitVertexTextureConstants() |
| 824 | { |
| 825 | float scaleX; |
| 826 | float scaleY; |
| 827 | |
| 828 | RenderTexture &tex0 = g_textures[gRSP.curTile]; |
| 829 | //CTexture *surf = tex0.m_pCTexture; |
| 830 | Tile &tile0 = gRDP.tiles[gRSP.curTile]; |
| 831 | |
| 832 | scaleX = gRSP.fTexScaleX; |
| 833 | scaleY = gRSP.fTexScaleY; |
| 834 | |
| 835 | gRSP.tex0scaleX = scaleX * tile0.fShiftScaleS/tex0.m_fTexWidth; |
| 836 | gRSP.tex0scaleY = scaleY * tile0.fShiftScaleT/tex0.m_fTexHeight; |
| 837 | |
| 838 | gRSP.tex0OffsetX = tile0.fhilite_sl/tex0.m_fTexWidth; |
| 839 | gRSP.tex0OffsetY = tile0.fhilite_tl/tex0.m_fTexHeight; |
| 840 | |
| 841 | if( CRender::g_pRender->IsTexel1Enable() ) |
| 842 | { |
| 843 | RenderTexture &tex1 = g_textures[(gRSP.curTile+1)&7]; |
| 844 | //CTexture *surf = tex1.m_pCTexture; |
| 845 | Tile &tile1 = gRDP.tiles[(gRSP.curTile+1)&7]; |
| 846 | |
| 847 | gRSP.tex1scaleX = scaleX * tile1.fShiftScaleS/tex1.m_fTexWidth; |
| 848 | gRSP.tex1scaleY = scaleY * tile1.fShiftScaleT/tex1.m_fTexHeight; |
| 849 | |
| 850 | gRSP.tex1OffsetX = tile1.fhilite_sl/tex1.m_fTexWidth; |
| 851 | gRSP.tex1OffsetY = tile1.fhilite_tl/tex1.m_fTexHeight; |
| 852 | } |
| 853 | |
| 854 | gRSP.texGenXRatio = tile0.fShiftScaleS; |
| 855 | gRSP.texGenYRatio = gRSP.fTexScaleX/gRSP.fTexScaleY*tex0.m_fTexWidth/tex0.m_fTexHeight*tile0.fShiftScaleT; |
| 856 | } |
| 857 | |
| 858 | void TexGen(float &s, float &t) |
| 859 | { |
| 860 | if (gRDP.geometryMode & G_TEXTURE_GEN_LINEAR) |
| 861 | { |
| 862 | s = acosf(g_normal.x) / 3.14159f; |
| 863 | t = acosf(g_normal.y) / 3.14159f; |
| 864 | } |
| 865 | else |
| 866 | { |
| 867 | s = 0.5f * ( 1.0f + g_normal.x); |
| 868 | t = 0.5f * ( 1.0f - g_normal.y); |
| 869 | } |
| 870 | } |
| 871 | |
| 872 | void ComputeLOD(bool openGL) |
| 873 | { |
| 874 | TLITVERTEX &v0 = g_vtxBuffer[0]; |
| 875 | TLITVERTEX &v1 = g_vtxBuffer[1]; |
| 876 | RenderTexture &tex0 = g_textures[gRSP.curTile]; |
| 877 | |
| 878 | float d,dt; |
| 879 | if( openGL ) |
| 880 | { |
| 881 | float x = g_vtxProjected5[0][0] / g_vtxProjected5[0][4] - g_vtxProjected5[1][0] / g_vtxProjected5[1][4]; |
| 882 | float y = g_vtxProjected5[0][1] / g_vtxProjected5[0][4] - g_vtxProjected5[1][1] / g_vtxProjected5[1][4]; |
| 883 | |
| 884 | x = windowSetting.vpWidthW*x/windowSetting.fMultX/2; |
| 885 | y = windowSetting.vpHeightW*y/windowSetting.fMultY/2; |
| 886 | d = sqrtf(x*x+y*y); |
| 887 | } |
| 888 | else |
| 889 | { |
| 890 | float x = (v0.x - v1.x)/ windowSetting.fMultX; |
| 891 | float y = (v0.y - v1.y)/ windowSetting.fMultY; |
| 892 | d = sqrtf(x*x+y*y); |
| 893 | } |
| 894 | |
| 895 | float s0 = v0.tcord[0].u * tex0.m_fTexWidth; |
| 896 | float t0 = v0.tcord[0].v * tex0.m_fTexHeight; |
| 897 | float s1 = v1.tcord[0].u * tex0.m_fTexWidth; |
| 898 | float t1 = v1.tcord[0].v * tex0.m_fTexHeight; |
| 899 | |
| 900 | dt = sqrtf((s0-s1)*(s0-s1)+(t0-t1)*(t0-t1)); |
| 901 | |
| 902 | float lod = dt/d; |
| 903 | float frac = log10f(lod)/log10f(2.0f); |
| 904 | //DEBUGGER_IF_DUMP(pauseAtNext,{DebuggerAppendMsg("LOD frac = %f", frac);}); |
| 905 | frac = (lod / powf(2.0f,floorf(frac))); |
| 906 | frac = frac - floorf(frac); |
| 907 | //DEBUGGER_IF_DUMP(pauseAtNext,{DebuggerAppendMsg("LOD = %f, frac = %f", lod, frac);}); |
| 908 | gRDP.LODFrac = (uint32)(frac*255); |
| 909 | CRender::g_pRender->SetCombinerAndBlender(); |
| 910 | } |
| 911 | |
| 912 | bool bHalfTxtScale=false; |
| 913 | extern uint32 lastSetTile; |
| 914 | |
| 915 | void InitVertex(uint32 dwV, uint32 vtxIndex, bool bTexture, bool openGL) |
| 916 | { |
| 917 | VTX_DUMP(TRACE2("Init vertex (%d) to vtx buf[%d]:", dwV, vtxIndex)); |
| 918 | |
| 919 | TLITVERTEX &v = g_vtxBuffer[vtxIndex]; |
| 920 | VTX_DUMP(TRACE4(" Trans: x=%f, y=%f, z=%f, w=%f", g_vtxTransformed[dwV].x,g_vtxTransformed[dwV].y,g_vtxTransformed[dwV].z,g_vtxTransformed[dwV].w)); |
| 921 | if( openGL ) |
| 922 | { |
| 923 | g_vtxProjected5[vtxIndex][0] = g_vtxTransformed[dwV].x; |
| 924 | g_vtxProjected5[vtxIndex][1] = g_vtxTransformed[dwV].y; |
| 925 | g_vtxProjected5[vtxIndex][2] = g_vtxTransformed[dwV].z; |
| 926 | g_vtxProjected5[vtxIndex][3] = g_vtxTransformed[dwV].w; |
| 927 | g_vtxProjected5[vtxIndex][4] = g_vecProjected[dwV].z; |
| 928 | |
| 929 | if( g_vtxTransformed[dwV].w < 0 ) |
| 930 | g_vtxProjected5[vtxIndex][4] = 0; |
| 931 | |
| 932 | g_vtxIndex[vtxIndex] = vtxIndex; |
| 933 | } |
| 934 | |
| 935 | if( !openGL || options.bOGLVertexClipper == TRUE ) |
| 936 | { |
| 937 | v.x = g_vecProjected[dwV].x*gRSP.vtxXMul+gRSP.vtxXAdd; |
| 938 | v.y = g_vecProjected[dwV].y*gRSP.vtxYMul+gRSP.vtxYAdd; |
| 939 | v.z = (g_vecProjected[dwV].z + 1.0f) * 0.5f; // DirectX minZ=0, maxZ=1 |
| 940 | //v.z = g_vecProjected[dwV].z; // DirectX minZ=0, maxZ=1 |
| 941 | v.rhw = g_vecProjected[dwV].w; |
| 942 | VTX_DUMP(TRACE4(" Proj : x=%f, y=%f, z=%f, rhw=%f", v.x,v.y,v.z,v.rhw)); |
| 943 | |
| 944 | if( gRSP.bProcessSpecularColor ) |
| 945 | { |
| 946 | v.dcSpecular = CRender::g_pRender->PostProcessSpecularColor(); |
| 947 | if( gRSP.bFogEnabled ) |
| 948 | { |
| 949 | v.dcSpecular &= 0x00FFFFFF; |
| 950 | uint32 fogFct = 0xFF-(uint8)((g_fFogCoord[dwV]-gRSPfFogMin)*gRSPfFogDivider); |
| 951 | v.dcSpecular |= (fogFct<<24); |
| 952 | } |
| 953 | } |
| 954 | else if( gRSP.bFogEnabled ) |
| 955 | { |
| 956 | uint32 fogFct = 0xFF-(uint8)((g_fFogCoord[dwV]-gRSPfFogMin)*gRSPfFogDivider); |
| 957 | v.dcSpecular = (fogFct<<24); |
| 958 | } |
| 959 | } |
| 960 | VTX_DUMP(TRACE2(" (U,V): %f, %f", g_fVtxTxtCoords[dwV].x,g_fVtxTxtCoords[dwV].y)); |
| 961 | |
| 962 | v.dcDiffuse = g_dwVtxDifColor[dwV]; |
| 963 | if( gRDP.otherMode.key_en ) |
| 964 | { |
| 965 | v.dcDiffuse &= 0x00FFFFFF; |
| 966 | v.dcDiffuse |= (gRDP.keyA<<24); |
| 967 | } |
| 968 | else if( gRDP.otherMode.aa_en && gRDP.otherMode.clr_on_cvg==0 ) |
| 969 | { |
| 970 | v.dcDiffuse |= 0xFF000000; |
| 971 | } |
| 972 | |
| 973 | if( gRSP.bProcessDiffuseColor ) |
| 974 | { |
| 975 | v.dcDiffuse = CRender::g_pRender->PostProcessDiffuseColor(v.dcDiffuse); |
| 976 | } |
| 977 | if( options.bWinFrameMode ) |
| 978 | { |
| 979 | v.dcDiffuse = g_dwVtxDifColor[dwV]; |
| 980 | } |
| 981 | |
| 982 | if( openGL ) |
| 983 | { |
| 984 | g_oglVtxColors[vtxIndex][0] = v.r; |
| 985 | g_oglVtxColors[vtxIndex][1] = v.g; |
| 986 | g_oglVtxColors[vtxIndex][2] = v.b; |
| 987 | g_oglVtxColors[vtxIndex][3] = v.a; |
| 988 | } |
| 989 | |
| 990 | if( bTexture ) |
| 991 | { |
| 992 | // If the vert is already lit, then there is no normal (and hence we can't generate tex coord) |
| 993 | // Only scale if not generated automatically |
| 994 | if (gRSP.bTextureGen && gRSP.bLightingEnable) |
| 995 | { |
| 996 | // Correction for texGen result |
| 997 | float u0,u1,v0,v1; |
| 998 | RenderTexture &tex0 = g_textures[gRSP.curTile]; |
| 999 | u0 = g_fVtxTxtCoords[dwV].x * 32 * 1024 * gRSP.fTexScaleX / tex0.m_fTexWidth; |
| 1000 | v0 = g_fVtxTxtCoords[dwV].y * 32 * 1024 * gRSP.fTexScaleY / tex0.m_fTexHeight; |
| 1001 | u0 *= (gRDP.tiles[gRSP.curTile].fShiftScaleS); |
| 1002 | v0 *= (gRDP.tiles[gRSP.curTile].fShiftScaleT); |
| 1003 | |
| 1004 | if( CRender::g_pRender->IsTexel1Enable() ) |
| 1005 | { |
| 1006 | RenderTexture &tex1 = g_textures[(gRSP.curTile+1)&7]; |
| 1007 | u1 = g_fVtxTxtCoords[dwV].x * 32 * 1024 * gRSP.fTexScaleX / tex1.m_fTexWidth; |
| 1008 | v1 = g_fVtxTxtCoords[dwV].y * 32 * 1024 * gRSP.fTexScaleY / tex1.m_fTexHeight; |
| 1009 | u1 *= gRDP.tiles[(gRSP.curTile+1)&7].fShiftScaleS; |
| 1010 | v1 *= gRDP.tiles[(gRSP.curTile+1)&7].fShiftScaleT; |
| 1011 | CRender::g_pRender->SetVertexTextureUVCoord(v, u0, v0, u1, v1); |
| 1012 | } |
| 1013 | else |
| 1014 | { |
| 1015 | CRender::g_pRender->SetVertexTextureUVCoord(v, u0, v0); |
| 1016 | } |
| 1017 | } |
| 1018 | else |
| 1019 | { |
| 1020 | float tex0u = g_fVtxTxtCoords[dwV].x *gRSP.tex0scaleX - gRSP.tex0OffsetX ; |
| 1021 | float tex0v = g_fVtxTxtCoords[dwV].y *gRSP.tex0scaleY - gRSP.tex0OffsetY ; |
| 1022 | |
| 1023 | if( CRender::g_pRender->IsTexel1Enable() ) |
| 1024 | { |
| 1025 | float tex1u = g_fVtxTxtCoords[dwV].x *gRSP.tex1scaleX - gRSP.tex1OffsetX ; |
| 1026 | float tex1v = g_fVtxTxtCoords[dwV].y *gRSP.tex1scaleY - gRSP.tex1OffsetY ; |
| 1027 | |
| 1028 | CRender::g_pRender->SetVertexTextureUVCoord(v, tex0u, tex0v, tex1u, tex1v); |
| 1029 | VTX_DUMP(TRACE2(" (tex0): %f, %f", tex0u,tex0v)); |
| 1030 | VTX_DUMP(TRACE2(" (tex1): %f, %f", tex1u,tex1v)); |
| 1031 | } |
| 1032 | else |
| 1033 | { |
| 1034 | CRender::g_pRender->SetVertexTextureUVCoord(v, tex0u, tex0v); |
| 1035 | VTX_DUMP(TRACE2(" (tex0): %f, %f", tex0u,tex0v)); |
| 1036 | } |
| 1037 | } |
| 1038 | |
| 1039 | // Check for txt scale hack |
| 1040 | if( !bHalfTxtScale && g_curRomInfo.bTextureScaleHack && |
| 1041 | (gRDP.tiles[lastSetTile].dwSize == TXT_SIZE_32b || gRDP.tiles[lastSetTile].dwSize == TXT_SIZE_4b ) ) |
| 1042 | { |
| 1043 | int width = ((gRDP.tiles[lastSetTile].sh-gRDP.tiles[lastSetTile].sl+1)<<1); |
| 1044 | int height = ((gRDP.tiles[lastSetTile].th-gRDP.tiles[lastSetTile].tl+1)<<1); |
| 1045 | if( g_fVtxTxtCoords[dwV].x*gRSP.fTexScaleX == width || g_fVtxTxtCoords[dwV].y*gRSP.fTexScaleY == height ) |
| 1046 | { |
| 1047 | bHalfTxtScale=true; |
| 1048 | } |
| 1049 | } |
| 1050 | } |
| 1051 | |
| 1052 | if( g_curRomInfo.bEnableTxtLOD && vtxIndex == 1 && gRDP.otherMode.text_lod ) |
| 1053 | { |
| 1054 | if( CRender::g_pRender->IsTexel1Enable() && CRender::g_pRender->m_pColorCombiner->m_pDecodedMux->isUsed(MUX_LODFRAC) ) |
| 1055 | { |
| 1056 | ComputeLOD(openGL); |
| 1057 | } |
| 1058 | else |
| 1059 | { |
| 1060 | gRDP.LODFrac = 0; |
| 1061 | } |
| 1062 | } |
| 1063 | |
| 1064 | VTX_DUMP(TRACE2(" DIF(%08X), SPE(%08X)", v.dcDiffuse, v.dcSpecular)); |
| 1065 | VTX_DUMP(TRACE0("")); |
| 1066 | } |
| 1067 | |
| 1068 | uint32 LightVert(XVECTOR4 & norm, int vidx) |
| 1069 | { |
| 1070 | float fCosT; |
| 1071 | |
| 1072 | // Do ambient |
| 1073 | register float r = gRSP.fAmbientLightR; |
| 1074 | register float g = gRSP.fAmbientLightG; |
| 1075 | register float b = gRSP.fAmbientLightB; |
| 1076 | |
| 1077 | if( options.enableHackForGames != HACK_FOR_ZELDA_MM ) |
| 1078 | { |
| 1079 | for (register unsigned int l=0; l < gRSPnumLights; l++) |
| 1080 | { |
| 1081 | fCosT = norm.x*gRSPlights[l].x + norm.y*gRSPlights[l].y + norm.z*gRSPlights[l].z; |
| 1082 | |
| 1083 | if (fCosT > 0 ) |
| 1084 | { |
| 1085 | r += gRSPlights[l].fr * fCosT; |
| 1086 | g += gRSPlights[l].fg * fCosT; |
| 1087 | b += gRSPlights[l].fb * fCosT; |
| 1088 | } |
| 1089 | } |
| 1090 | } |
| 1091 | else |
| 1092 | { |
| 1093 | XVECTOR4 v; |
| 1094 | bool transformed = false; |
| 1095 | |
| 1096 | for (register unsigned int l=0; l < gRSPnumLights; l++) |
| 1097 | { |
| 1098 | if( gRSPlights[l].range == 0 ) |
| 1099 | { |
| 1100 | // Regular directional light |
| 1101 | fCosT = norm.x*gRSPlights[l].x + norm.y*gRSPlights[l].y + norm.z*gRSPlights[l].z; |
| 1102 | |
| 1103 | if (fCosT > 0 ) |
| 1104 | { |
| 1105 | r += gRSPlights[l].fr * fCosT; |
| 1106 | g += gRSPlights[l].fg * fCosT; |
| 1107 | b += gRSPlights[l].fb * fCosT; |
| 1108 | } |
| 1109 | } |
| 1110 | else //if( (gRSPlights[l].col&0x00FFFFFF) != 0x00FFFFFF ) |
| 1111 | { |
| 1112 | // Point light |
| 1113 | if( !transformed ) |
| 1114 | { |
| 1115 | Vec3Transform(&v, (XVECTOR3*)&g_vtxNonTransformed[vidx], &gRSPmodelViewTop); // Convert to w=1 |
| 1116 | transformed = true; |
| 1117 | } |
| 1118 | |
| 1119 | XVECTOR3 dir(gRSPlights[l].x - v.x, gRSPlights[l].y - v.y, gRSPlights[l].z - v.z); |
| 1120 | //XVECTOR3 dir(v.x-gRSPlights[l].x, v.y-gRSPlights[l].y, v.z-gRSPlights[l].z); |
| 1121 | float d2 = sqrtf(dir.x*dir.x+dir.y*dir.y+dir.z*dir.z); |
| 1122 | dir.x /= d2; |
| 1123 | dir.y /= d2; |
| 1124 | dir.z /= d2; |
| 1125 | |
| 1126 | fCosT = norm.x*dir.x + norm.y*dir.y + norm.z*dir.z; |
| 1127 | |
| 1128 | if (fCosT > 0 ) |
| 1129 | { |
| 1130 | //float f = d2/gRSPlights[l].range*50; |
| 1131 | float f = d2/15000*50; |
| 1132 | f = 1 - min(f,1); |
| 1133 | fCosT *= f*f; |
| 1134 | |
| 1135 | r += gRSPlights[l].fr * fCosT; |
| 1136 | g += gRSPlights[l].fg * fCosT; |
| 1137 | b += gRSPlights[l].fb * fCosT; |
| 1138 | } |
| 1139 | } |
| 1140 | } |
| 1141 | } |
| 1142 | |
| 1143 | if (r > 255) r = 255; |
| 1144 | if (g > 255) g = 255; |
| 1145 | if (b > 255) b = 255; |
| 1146 | return ((0xff000000)|(((uint32)r)<<16)|(((uint32)g)<<8)|((uint32)b)); |
| 1147 | } |
| 1148 | |
| 1149 | uint32 LightVertNew(XVECTOR4 & norm) |
| 1150 | { |
| 1151 | float fCosT; |
| 1152 | |
| 1153 | // Do ambient |
| 1154 | register float r = gRSP.fAmbientLightR; |
| 1155 | register float g = gRSP.fAmbientLightG; |
| 1156 | register float b = gRSP.fAmbientLightB; |
| 1157 | |
| 1158 | |
| 1159 | for (register unsigned int l=0; l < gRSPnumLights; l++) |
| 1160 | { |
| 1161 | fCosT = norm.x*gRSPlights[l].tx + norm.y*gRSPlights[l].ty + norm.z*gRSPlights[l].tz; |
| 1162 | |
| 1163 | if (fCosT > 0 ) |
| 1164 | { |
| 1165 | r += gRSPlights[l].fr * fCosT; |
| 1166 | g += gRSPlights[l].fg * fCosT; |
| 1167 | b += gRSPlights[l].fb * fCosT; |
| 1168 | } |
| 1169 | } |
| 1170 | |
| 1171 | if (r > 255) r = 255; |
| 1172 | if (g > 255) g = 255; |
| 1173 | if (b > 255) b = 255; |
| 1174 | return ((0xff000000)|(((uint32)r)<<16)|(((uint32)g)<<8)|((uint32)b)); |
| 1175 | } |
| 1176 | |
| 1177 | |
| 1178 | float zero = 0.0f; |
| 1179 | float onef = 1.0f; |
| 1180 | float fcosT; |
| 1181 | |
| 1182 | #if !defined(__GNUC__) && !defined(NO_ASM) |
| 1183 | __declspec( naked ) uint32 __fastcall SSELightVert() |
| 1184 | { |
| 1185 | __asm |
| 1186 | { |
| 1187 | movaps xmm3, DWORD PTR gRSP; // loading Ambient colors, xmm3 is the result color |
| 1188 | movaps xmm4, DWORD PTR [g_normal]; // xmm4 is the normal |
| 1189 | |
| 1190 | mov ecx, 0; |
| 1191 | loopback: |
| 1192 | cmp ecx, DWORD PTR gRSPnumLights; |
| 1193 | jae breakout; |
| 1194 | mov eax,ecx; |
| 1195 | imul eax,0x44; |
| 1196 | movups xmm5, DWORD PTR gRSPlights[eax]; // Light Dir |
| 1197 | movups xmm1, DWORD PTR gRSPlights[0x14][eax]; // Light color |
| 1198 | mulps xmm5, xmm4; // Lightdir * normals |
| 1199 | |
| 1200 | movhlps xmm0,xmm5; |
| 1201 | addps xmm0,xmm5; |
| 1202 | shufps xmm5,xmm0,0x01; |
| 1203 | addps xmm0,xmm5; |
| 1204 | |
| 1205 | comiss xmm0,zero; |
| 1206 | jc endloop |
| 1207 | |
| 1208 | shufps xmm0,xmm0,0; // fcosT |
| 1209 | mulps xmm1,xmm0; |
| 1210 | addps xmm3,xmm1; |
| 1211 | endloop: |
| 1212 | inc ecx; |
| 1213 | jmp loopback; |
| 1214 | breakout: |
| 1215 | |
| 1216 | movss xmm0,DWORD PTR real255; |
| 1217 | shufps xmm0,xmm0,0; |
| 1218 | minps xmm0,xmm3; |
| 1219 | |
| 1220 | // Without using a memory |
| 1221 | cvtss2si eax,xmm0; // move the 1st uint32 to eax |
| 1222 | shl eax,10h; |
| 1223 | or eax,0FF000000h; |
| 1224 | shufps xmm0,xmm0,0E5h; // move the 2nd uint32 to the 1st uint32 |
| 1225 | cvtss2si ecx,xmm0; // move the 1st uint32 to ecx |
| 1226 | shl ecx,8; |
| 1227 | or eax,ecx; |
| 1228 | shufps xmm0,xmm0,0E6h; // Move the 3rd uint32 to the 1st uint32 |
| 1229 | cvtss2si ecx,xmm0; |
| 1230 | or eax,ecx; |
| 1231 | |
| 1232 | ret; |
| 1233 | } |
| 1234 | } |
| 1235 | #elif defined(__GNUC__) && defined(__x86_64__) && !defined(NO_ASM) |
| 1236 | uint32 SSELightVert(void) |
| 1237 | { |
| 1238 | uint32 rval; |
| 1239 | float f255 = 255.0, fZero = 0.0; |
| 1240 | |
| 1241 | asm volatile(" movaps %1, %%xmm3 \n" // xmm3 == gRSP.fAmbientLight{RGBA} |
| 1242 | " movaps %2, %%xmm4 \n" // xmm4 == g_normal.{xyz} |
| 1243 | " xor %%rcx, %%rcx \n" |
| 1244 | "0: \n" |
| 1245 | " cmpl %3, %%ecx \n" |
| 1246 | " jae 2f \n" |
| 1247 | " mov %%rcx, %%rax \n" |
| 1248 | " imul $0x44, %%rax, %%rax \n" |
| 1249 | " movups (%4,%%rax,), %%xmm5 \n" // xmm5 == gRSPlights[l].{xyzr} |
| 1250 | " movups 20(%4,%%rax,), %%xmm1 \n" // xmm1 == gRSPlights[l].{frfgfbfa} |
| 1251 | " mulps %%xmm4, %%xmm5 \n" |
| 1252 | " movhlps %%xmm5, %%xmm0 \n" |
| 1253 | " addps %%xmm5, %%xmm0 \n" |
| 1254 | " shufps $0x01, %%xmm0, %%xmm5 \n" |
| 1255 | " addps %%xmm5, %%xmm0 \n" |
| 1256 | " comiss %6, %%xmm0 \n" |
| 1257 | " jc 1f \n" |
| 1258 | " shufps $0x00, %%xmm0, %%xmm0 \n" |
| 1259 | " mulps %%xmm0, %%xmm1 \n" |
| 1260 | " addps %%xmm1, %%xmm3 \n" |
| 1261 | "1: \n" |
| 1262 | " inc %%rcx \n" |
| 1263 | " jmp 0b \n" |
| 1264 | "2: \n" |
| 1265 | " movss %5, %%xmm0 \n" |
| 1266 | " shufps $0x00, %%xmm0, %%xmm0 \n" |
| 1267 | " minps %%xmm3, %%xmm0 \n" |
| 1268 | " cvtss2si %%xmm0, %%eax \n" |
| 1269 | " shll $0x10, %%eax \n" |
| 1270 | " orl $0xff000000, %%eax \n" |
| 1271 | " shufps $0xe5, %%xmm0, %%xmm0 \n" |
| 1272 | " cvtss2si %%xmm0, %%ecx \n" |
| 1273 | " shll $8, %%ecx \n" |
| 1274 | " orl %%ecx, %%eax \n" |
| 1275 | " shufps $0xe6, %%xmm0, %%xmm0 \n" |
| 1276 | " cvtss2si %%xmm0, %%ecx \n" |
| 1277 | " orl %%ecx, %%eax \n" |
| 1278 | : "=&a"(rval) |
| 1279 | : "m"(gRSP), "m"(g_normal), "m"(gRSPnumLights), "r"(gRSPlights), "m"(f255), "m"(fZero) |
| 1280 | : "%rcx", "memory", "cc", "%xmm0", "%xmm1", "%xmm3", "%xmm4", "%xmm5" |
| 1281 | ); |
| 1282 | return rval; |
| 1283 | } |
| 1284 | #elif !defined(NO_ASM) // 32-bit GCC assumed |
| 1285 | uint32 SSELightVert(void) |
| 1286 | { |
| 1287 | uint32 rval; |
| 1288 | float f255 = 255.0, fZero = 0.0; |
| 1289 | |
| 1290 | asm volatile(" movaps %1, %%xmm3 \n" |
| 1291 | " movaps %2, %%xmm4 \n" |
| 1292 | " xor %%ecx, %%ecx \n" |
| 1293 | "0: \n" |
| 1294 | " cmpl %3, %%ecx \n" |
| 1295 | " jae 2f \n" |
| 1296 | " mov %%ecx, %%eax \n" |
| 1297 | " imul $0x44, %%eax, %%eax \n" |
| 1298 | " movups (%4,%%eax,), %%xmm5 \n" |
| 1299 | " movups 20(%4,%%eax,), %%xmm1 \n" |
| 1300 | " mulps %%xmm4, %%xmm5 \n" |
| 1301 | " movhlps %%xmm5, %%xmm0 \n" |
| 1302 | " addps %%xmm5, %%xmm0 \n" |
| 1303 | " shufps $0x01, %%xmm0, %%xmm5 \n" |
| 1304 | " addps %%xmm5, %%xmm0 \n" |
| 1305 | " comiss %6, %%xmm0 \n" |
| 1306 | " jc 1f \n" |
| 1307 | " shufps $0x00, %%xmm0, %%xmm0 \n" |
| 1308 | " mulps %%xmm0, %%xmm1 \n" |
| 1309 | " addps %%xmm1, %%xmm3 \n" |
| 1310 | "1: \n" |
| 1311 | " inc %%ecx \n" |
| 1312 | " jmp 0b \n" |
| 1313 | "2: \n" |
| 1314 | " movss %5, %%xmm0 \n" |
| 1315 | " shufps $0x00, %%xmm0, %%xmm0 \n" |
| 1316 | " minps %%xmm3, %%xmm0 \n" |
| 1317 | " cvtss2si %%xmm0, %%eax \n" |
| 1318 | " shll $0x10, %%eax \n" |
| 1319 | " orl $0xff000000, %%eax \n" |
| 1320 | " shufps $0xe5, %%xmm0, %%xmm0 \n" |
| 1321 | " cvtss2si %%xmm0, %%ecx \n" |
| 1322 | " shll $8, %%ecx \n" |
| 1323 | " orl %%ecx, %%eax \n" |
| 1324 | " shufps $0xe6, %%xmm0, %%xmm0 \n" |
| 1325 | " cvtss2si %%xmm0, %%ecx \n" |
| 1326 | " orl %%ecx, %%eax \n" |
| 1327 | : "=&a"(rval) |
| 1328 | : "m"(gRSP), "m"(g_normal), "m"(gRSPnumLights), "r"(gRSPlights), "m"(f255), "m"(fZero) |
| 1329 | : "%rcx", "memory", "cc", "%xmm0", "%xmm1", "%xmm3", "%xmm4", "%xmm5" |
| 1330 | ); |
| 1331 | return rval; |
| 1332 | } |
| 1333 | #endif |
| 1334 | |
| 1335 | inline void ReplaceAlphaWithFogFactor(int i) |
| 1336 | { |
| 1337 | if( gRDP.geometryMode & G_FOG ) |
| 1338 | { |
| 1339 | // Use fog factor to replace vertex alpha |
| 1340 | if( g_vecProjected[i].z > 1 ) |
| 1341 | *(((uint8*)&(g_dwVtxDifColor[i]))+3) = 0xFF; |
| 1342 | if( g_vecProjected[i].z < 0 ) |
| 1343 | *(((uint8*)&(g_dwVtxDifColor[i]))+3) = 0; |
| 1344 | else |
| 1345 | *(((uint8*)&(g_dwVtxDifColor[i]))+3) = (uint8)(g_vecProjected[i].z*255); |
| 1346 | } |
| 1347 | } |
| 1348 | |
| 1349 | |
| 1350 | // Bits |
| 1351 | // +-+-+- |
| 1352 | // xxyyzz |
| 1353 | #define Z_NEG 0x01 |
| 1354 | #define Z_POS 0x02 |
| 1355 | #define Y_NEG 0x04 |
| 1356 | #define Y_POS 0x08 |
| 1357 | #define X_NEG 0x10 |
| 1358 | #define X_POS 0x20 |
| 1359 | |
| 1360 | // Assumes dwAddr has already been checked! |
| 1361 | // Don't inline - it's too big with the transform macros |
| 1362 | |
| 1363 | #if !defined(NO_ASM) |
| 1364 | void ProcessVertexDataSSE(uint32 dwAddr, uint32 dwV0, uint32 dwNum) |
| 1365 | { |
| 1366 | UpdateCombinedMatrix(); |
| 1367 | |
| 1368 | // This function is called upon SPvertex |
| 1369 | // - do vertex matrix transform |
| 1370 | // - do vertex lighting |
| 1371 | // - do texture cooridinate transform if needed |
| 1372 | // - calculate normal vector |
| 1373 | |
| 1374 | // Output: - g_vecProjected[i] -> transformed vertex x,y,z |
| 1375 | // - g_vecProjected[i].w -> saved vertex 1/w |
| 1376 | // - g_dwVtxFlags[i] -> flags |
| 1377 | // - g_dwVtxDifColor[i] -> vertex color |
| 1378 | // - g_fVtxTxtCoords[i] -> vertex texture cooridinates |
| 1379 | |
| 1380 | FiddledVtx * pVtxBase = (FiddledVtx*)(g_pRDRAMu8 + dwAddr); |
| 1381 | g_pVtxBase = pVtxBase; |
| 1382 | |
| 1383 | for (uint32 i = dwV0; i < dwV0 + dwNum; i++) |
| 1384 | { |
| 1385 | SP_Timing(RSP_GBI0_Vtx); |
| 1386 | |
| 1387 | FiddledVtx & vert = pVtxBase[i - dwV0]; |
| 1388 | |
| 1389 | g_vtxNonTransformed[i].x = (float)vert.x; |
| 1390 | g_vtxNonTransformed[i].y = (float)vert.y; |
| 1391 | g_vtxNonTransformed[i].z = (float)vert.z; |
| 1392 | |
| 1393 | SSEVec3Transform(i); |
| 1394 | |
| 1395 | if( gRSP.bFogEnabled ) |
| 1396 | { |
| 1397 | g_fFogCoord[i] = g_vecProjected[i].z; |
| 1398 | if( g_vecProjected[i].w < 0 || g_vecProjected[i].z < 0 || g_fFogCoord[i] < gRSPfFogMin ) |
| 1399 | g_fFogCoord[i] = gRSPfFogMin; |
| 1400 | } |
| 1401 | |
| 1402 | ReplaceAlphaWithFogFactor(i); |
| 1403 | |
| 1404 | |
| 1405 | VTX_DUMP( |
| 1406 | { |
| 1407 | uint32 *dat = (uint32*)(&vert); |
| 1408 | DebuggerAppendMsg("vtx %d: %08X %08X %08X %08X", i, dat[0],dat[1],dat[2],dat[3]); |
| 1409 | DebuggerAppendMsg(" : %f, %f, %f, %f", |
| 1410 | g_vtxTransformed[i].x,g_vtxTransformed[i].y,g_vtxTransformed[i].z,g_vtxTransformed[i].w); |
| 1411 | DebuggerAppendMsg(" : %f, %f, %f, %f", |
| 1412 | g_vecProjected[i].x,g_vecProjected[i].y,g_vecProjected[i].z,g_vecProjected[i].w); |
| 1413 | }); |
| 1414 | |
| 1415 | RSP_Vtx_Clipping(i); |
| 1416 | |
| 1417 | if( gRSP.bLightingEnable ) |
| 1418 | { |
| 1419 | g_normal.x = (float)vert.norma.nx; |
| 1420 | g_normal.y = (float)vert.norma.ny; |
| 1421 | g_normal.z = (float)vert.norma.nz; |
| 1422 | |
| 1423 | SSEVec3TransformNormal(); |
| 1424 | if( options.enableHackForGames != HACK_FOR_ZELDA_MM ) |
| 1425 | g_dwVtxDifColor[i] = SSELightVert(); |
| 1426 | else |
| 1427 | g_dwVtxDifColor[i] = LightVert(g_normal, i); |
| 1428 | *(((uint8*)&(g_dwVtxDifColor[i]))+3) = vert.rgba.a; // still use alpha from the vertex |
| 1429 | } |
| 1430 | else |
| 1431 | { |
| 1432 | if( (gRDP.geometryMode & G_SHADE) == 0 && gRSP.ucode < 5 ) //Shade is disabled |
| 1433 | { |
| 1434 | //FLAT shade |
| 1435 | g_dwVtxDifColor[i] = gRDP.primitiveColor; |
| 1436 | } |
| 1437 | else |
| 1438 | { |
| 1439 | register IColor &color = *(IColor*)&g_dwVtxDifColor[i]; |
| 1440 | color.b = vert.rgba.r; |
| 1441 | color.g = vert.rgba.g; |
| 1442 | color.r = vert.rgba.b; |
| 1443 | color.a = vert.rgba.a; |
| 1444 | } |
| 1445 | } |
| 1446 | |
| 1447 | if( options.bWinFrameMode ) |
| 1448 | { |
| 1449 | g_dwVtxDifColor[i] = COLOR_RGBA(vert.rgba.r, vert.rgba.g, vert.rgba.b, vert.rgba.a); |
| 1450 | } |
| 1451 | |
| 1452 | // Update texture coords n.b. need to divide tu/tv by bogus scale on addition to buffer |
| 1453 | |
| 1454 | // If the vert is already lit, then there is no normal (and hence we |
| 1455 | // can't generate tex coord) |
| 1456 | if (gRSP.bTextureGen && gRSP.bLightingEnable ) |
| 1457 | { |
| 1458 | TexGen(g_fVtxTxtCoords[i].x, g_fVtxTxtCoords[i].y); |
| 1459 | } |
| 1460 | else |
| 1461 | { |
| 1462 | g_fVtxTxtCoords[i].x = (float)vert.tu; |
| 1463 | g_fVtxTxtCoords[i].y = (float)vert.tv; |
| 1464 | } |
| 1465 | } |
| 1466 | |
| 1467 | VTX_DUMP(TRACE2("Setting Vertexes: %d - %d\n", dwV0, dwV0+dwNum-1)); |
| 1468 | DEBUGGER_PAUSE_AND_DUMP(NEXT_VERTEX_CMD,{TRACE0("Paused at Vertex Cmd");}); |
| 1469 | } |
| 1470 | #endif |
| 1471 | |
| 1472 | void ProcessVertexDataNoSSE(uint32 dwAddr, uint32 dwV0, uint32 dwNum) |
| 1473 | { |
| 1474 | |
| 1475 | UpdateCombinedMatrix(); |
| 1476 | |
| 1477 | // This function is called upon SPvertex |
| 1478 | // - do vertex matrix transform |
| 1479 | // - do vertex lighting |
| 1480 | // - do texture cooridinate transform if needed |
| 1481 | // - calculate normal vector |
| 1482 | |
| 1483 | // Output: - g_vecProjected[i] -> transformed vertex x,y,z |
| 1484 | // - g_vecProjected[i].w -> saved vertex 1/w |
| 1485 | // - g_dwVtxFlags[i] -> flags |
| 1486 | // - g_dwVtxDifColor[i] -> vertex color |
| 1487 | // - g_fVtxTxtCoords[i] -> vertex texture cooridinates |
| 1488 | |
| 1489 | FiddledVtx * pVtxBase = (FiddledVtx*)(g_pRDRAMu8 + dwAddr); |
| 1490 | g_pVtxBase = pVtxBase; |
| 1491 | |
| 1492 | for (uint32 i = dwV0; i < dwV0 + dwNum; i++) |
| 1493 | { |
| 1494 | SP_Timing(RSP_GBI0_Vtx); |
| 1495 | |
| 1496 | FiddledVtx & vert = pVtxBase[i - dwV0]; |
| 1497 | |
| 1498 | g_vtxNonTransformed[i].x = (float)vert.x; |
| 1499 | g_vtxNonTransformed[i].y = (float)vert.y; |
| 1500 | g_vtxNonTransformed[i].z = (float)vert.z; |
| 1501 | |
| 1502 | Vec3Transform(&g_vtxTransformed[i], (XVECTOR3*)&g_vtxNonTransformed[i], &gRSPworldProject); // Convert to w=1 |
| 1503 | |
| 1504 | g_vecProjected[i].w = 1.0f / g_vtxTransformed[i].w; |
| 1505 | g_vecProjected[i].x = g_vtxTransformed[i].x * g_vecProjected[i].w; |
| 1506 | g_vecProjected[i].y = g_vtxTransformed[i].y * g_vecProjected[i].w; |
| 1507 | if ((g_curRomInfo.bPrimaryDepthHack || options.enableHackForGames == HACK_FOR_NASCAR ) && gRDP.otherMode.depth_source ) |
| 1508 | { |
| 1509 | g_vecProjected[i].z = gRDP.fPrimitiveDepth; |
| 1510 | g_vtxTransformed[i].z = gRDP.fPrimitiveDepth*g_vtxTransformed[i].w; |
| 1511 | } |
| 1512 | else |
| 1513 | { |
| 1514 | g_vecProjected[i].z = g_vtxTransformed[i].z * g_vecProjected[i].w; |
| 1515 | } |
| 1516 | |
| 1517 | if( gRSP.bFogEnabled ) |
| 1518 | { |
| 1519 | g_fFogCoord[i] = g_vecProjected[i].z; |
| 1520 | if( g_vecProjected[i].w < 0 || g_vecProjected[i].z < 0 || g_fFogCoord[i] < gRSPfFogMin ) |
| 1521 | g_fFogCoord[i] = gRSPfFogMin; |
| 1522 | } |
| 1523 | |
| 1524 | VTX_DUMP( |
| 1525 | { |
| 1526 | uint32 *dat = (uint32*)(&vert); |
| 1527 | DebuggerAppendMsg("vtx %d: %08X %08X %08X %08X", i, dat[0],dat[1],dat[2],dat[3]); |
| 1528 | DebuggerAppendMsg(" : %f, %f, %f, %f", |
| 1529 | g_vtxTransformed[i].x,g_vtxTransformed[i].y,g_vtxTransformed[i].z,g_vtxTransformed[i].w); |
| 1530 | DebuggerAppendMsg(" : %f, %f, %f, %f", |
| 1531 | g_vecProjected[i].x,g_vecProjected[i].y,g_vecProjected[i].z,g_vecProjected[i].w); |
| 1532 | }); |
| 1533 | |
| 1534 | RSP_Vtx_Clipping(i); |
| 1535 | |
| 1536 | if( gRSP.bLightingEnable ) |
| 1537 | { |
| 1538 | g_normal.x = (float)vert.norma.nx; |
| 1539 | g_normal.y = (float)vert.norma.ny; |
| 1540 | g_normal.z = (float)vert.norma.nz; |
| 1541 | |
| 1542 | Vec3TransformNormal(g_normal, gRSPmodelViewTop); |
| 1543 | g_dwVtxDifColor[i] = LightVert(g_normal, i); |
| 1544 | *(((uint8*)&(g_dwVtxDifColor[i]))+3) = vert.rgba.a; // still use alpha from the vertex |
| 1545 | } |
| 1546 | else |
| 1547 | { |
| 1548 | if( (gRDP.geometryMode & G_SHADE) == 0 && gRSP.ucode < 5 ) //Shade is disabled |
| 1549 | { |
| 1550 | //FLAT shade |
| 1551 | g_dwVtxDifColor[i] = gRDP.primitiveColor; |
| 1552 | } |
| 1553 | else |
| 1554 | { |
| 1555 | register IColor &color = *(IColor*)&g_dwVtxDifColor[i]; |
| 1556 | color.b = vert.rgba.r; |
| 1557 | color.g = vert.rgba.g; |
| 1558 | color.r = vert.rgba.b; |
| 1559 | color.a = vert.rgba.a; |
| 1560 | } |
| 1561 | } |
| 1562 | |
| 1563 | if( options.bWinFrameMode ) |
| 1564 | { |
| 1565 | g_dwVtxDifColor[i] = COLOR_RGBA(vert.rgba.r, vert.rgba.g, vert.rgba.b, vert.rgba.a); |
| 1566 | } |
| 1567 | |
| 1568 | ReplaceAlphaWithFogFactor(i); |
| 1569 | |
| 1570 | // Update texture coords n.b. need to divide tu/tv by bogus scale on addition to buffer |
| 1571 | |
| 1572 | // If the vert is already lit, then there is no normal (and hence we |
| 1573 | // can't generate tex coord) |
| 1574 | if (gRSP.bTextureGen && gRSP.bLightingEnable ) |
| 1575 | { |
| 1576 | TexGen(g_fVtxTxtCoords[i].x, g_fVtxTxtCoords[i].y); |
| 1577 | } |
| 1578 | else |
| 1579 | { |
| 1580 | g_fVtxTxtCoords[i].x = (float)vert.tu; |
| 1581 | g_fVtxTxtCoords[i].y = (float)vert.tv; |
| 1582 | } |
| 1583 | } |
| 1584 | |
| 1585 | VTX_DUMP(TRACE2("Setting Vertexes: %d - %d\n", dwV0, dwV0+dwNum-1)); |
| 1586 | DEBUGGER_PAUSE_AND_DUMP(NEXT_VERTEX_CMD,{TRACE0("Paused at Vertex Cmd");}); |
| 1587 | } |
| 1588 | |
| 1589 | bool PrepareTriangle(uint32 dwV0, uint32 dwV1, uint32 dwV2) |
| 1590 | { |
| 1591 | if( status.isVertexShaderEnabled || status.bUseHW_T_L ) |
| 1592 | { |
| 1593 | g_vtxIndex[gRSP.numVertices++] = dwV0; |
| 1594 | g_vtxIndex[gRSP.numVertices++] = dwV1; |
| 1595 | g_vtxIndex[gRSP.numVertices++] = dwV2; |
| 1596 | status.dwNumTrisRendered++; |
| 1597 | gRSP.maxVertexID = max(gRSP.maxVertexID,max(dwV0,max(dwV1,dwV2))); |
| 1598 | } |
| 1599 | else |
| 1600 | { |
| 1601 | SP_Timing(SP_Each_Triangle); |
| 1602 | |
| 1603 | bool textureFlag = (CRender::g_pRender->IsTextureEnabled() || gRSP.ucode == 6 ); |
| 1604 | bool openGL = CDeviceBuilder::m_deviceGeneralType == OGL_DEVICE; |
| 1605 | |
| 1606 | InitVertex(dwV0, gRSP.numVertices, textureFlag, openGL); |
| 1607 | InitVertex(dwV1, gRSP.numVertices+1, textureFlag, openGL); |
| 1608 | InitVertex(dwV2, gRSP.numVertices+2, textureFlag, openGL); |
| 1609 | |
| 1610 | gRSP.numVertices += 3; |
| 1611 | status.dwNumTrisRendered++; |
| 1612 | } |
| 1613 | |
| 1614 | return true; |
| 1615 | } |
| 1616 | |
| 1617 | |
| 1618 | |
| 1619 | // Returns TRUE if it thinks the triangle is visible |
| 1620 | // Returns FALSE if it is clipped |
| 1621 | bool IsTriangleVisible(uint32 dwV0, uint32 dwV1, uint32 dwV2) |
| 1622 | { |
| 1623 | //return true; //fix me |
| 1624 | |
| 1625 | if( status.isVertexShaderEnabled || status.bUseHW_T_L ) return true; // We won't have access to transformed vertex data |
| 1626 | |
| 1627 | DEBUGGER_ONLY_IF( (!debuggerEnableTestTris || !debuggerEnableCullFace), {return TRUE;}); |
| 1628 | |
| 1629 | #ifdef DEBUGGER |
| 1630 | // Check vertices are valid! |
| 1631 | if (dwV0 >= MAX_VERTS || dwV1 >= MAX_VERTS || dwV2 >= MAX_VERTS) |
| 1632 | return false; |
| 1633 | #endif |
| 1634 | |
| 1635 | // Here we AND all the flags. If any of the bits is set for all |
| 1636 | // 3 vertices, it means that all three x, y or z lie outside of |
| 1637 | // the current viewing volume. |
| 1638 | // Currently disabled - still seems a bit dodgy |
| 1639 | if ((gRSP.bCullFront || gRSP.bCullBack) && gRDP.otherMode.zmode != 3) |
| 1640 | { |
| 1641 | XVECTOR4 & v0 = g_vecProjected[dwV0]; |
| 1642 | XVECTOR4 & v1 = g_vecProjected[dwV1]; |
| 1643 | XVECTOR4 & v2 = g_vecProjected[dwV2]; |
| 1644 | |
| 1645 | // Only try to clip if the tri is onscreen. For some reason, this |
| 1646 | // method doesnt' work well when the z value is outside of screenspace |
| 1647 | //if (v0.z < 1 && v1.z < 1 && v2.z < 1) |
| 1648 | { |
| 1649 | float V1 = v2.x - v0.x; |
| 1650 | float V2 = v2.y - v0.y; |
| 1651 | |
| 1652 | float W1 = v2.x - v1.x; |
| 1653 | float W2 = v2.y - v1.y; |
| 1654 | |
| 1655 | float fDirection = (V1 * W2) - (V2 * W1); |
| 1656 | fDirection = fDirection * v1.w * v2.w * v0.w; |
| 1657 | //float fDirection = v0.x*v1.y-v1.x*v0.y+v1.x*v2.y-v2.x*v1.y+v2.x*v0.y-v0.x*v2.y; |
| 1658 | |
| 1659 | if (fDirection < 0 && gRSP.bCullBack) |
| 1660 | { |
| 1661 | status.dwNumTrisClipped++; |
| 1662 | return false; |
| 1663 | } |
| 1664 | else if (fDirection > 0 && gRSP.bCullFront) |
| 1665 | { |
| 1666 | status.dwNumTrisClipped++; |
| 1667 | return false; |
| 1668 | } |
| 1669 | } |
| 1670 | } |
| 1671 | |
| 1672 | #ifdef ENABLE_CLIP_TRI |
| 1673 | //if( gRSP.bRejectVtx && (g_clipFlag[dwV0]|g_clipFlag[dwV1]|g_clipFlag[dwV2]) ) |
| 1674 | // return; |
| 1675 | if( g_clipFlag2[dwV0]&g_clipFlag2[dwV1]&g_clipFlag2[dwV2] ) |
| 1676 | { |
| 1677 | //DebuggerAppendMsg("Clipped"); |
| 1678 | return false; |
| 1679 | } |
| 1680 | #endif |
| 1681 | |
| 1682 | return true; |
| 1683 | } |
| 1684 | |
| 1685 | |
| 1686 | void SetPrimitiveColor(uint32 dwCol, uint32 LODMin, uint32 LODFrac) |
| 1687 | { |
| 1688 | gRDP.colorsAreReloaded = true; |
| 1689 | gRDP.primitiveColor = dwCol; |
| 1690 | gRDP.primLODMin = LODMin; |
| 1691 | gRDP.primLODFrac = LODFrac; |
| 1692 | if( gRDP.primLODFrac < gRDP.primLODMin ) |
| 1693 | { |
| 1694 | gRDP.primLODFrac = gRDP.primLODMin; |
| 1695 | } |
| 1696 | |
| 1697 | gRDP.fvPrimitiveColor[0] = ((dwCol>>16)&0xFF)/255.0f; //r |
| 1698 | gRDP.fvPrimitiveColor[1] = ((dwCol>>8)&0xFF)/255.0f; //g |
| 1699 | gRDP.fvPrimitiveColor[2] = ((dwCol)&0xFF)/255.0f; //b |
| 1700 | gRDP.fvPrimitiveColor[3] = ((dwCol>>24)&0xFF)/255.0f; //a |
| 1701 | } |
| 1702 | |
| 1703 | void SetPrimitiveDepth(uint32 z, uint32 dwDZ) |
| 1704 | { |
| 1705 | gRDP.primitiveDepth = z & 0x7FFF; |
| 1706 | gRDP.fPrimitiveDepth = (float)(gRDP.primitiveDepth)/(float)0x8000; |
| 1707 | |
| 1708 | //gRDP.fPrimitiveDepth = gRDP.fPrimitiveDepth*2-1; |
| 1709 | /* |
| 1710 | z=0xFFFF -> 1 the farest |
| 1711 | z=0 -> -1 the nearest |
| 1712 | */ |
| 1713 | |
| 1714 | //how to use dwDZ? |
| 1715 | |
| 1716 | #ifdef DEBUGGER |
| 1717 | if( (pauseAtNext && (eventToPause == NEXT_VERTEX_CMD || eventToPause == NEXT_FLUSH_TRI )) )//&& logTriangles ) |
| 1718 | { |
| 1719 | DebuggerAppendMsg("Set prim Depth: %f, (%08X, %08X)", gRDP.fPrimitiveDepth, z, dwDZ); |
| 1720 | } |
| 1721 | #endif |
| 1722 | } |
| 1723 | |
| 1724 | void SetVertexXYZ(uint32 vertex, float x, float y, float z) |
| 1725 | { |
| 1726 | g_vecProjected[vertex].x = x; |
| 1727 | g_vecProjected[vertex].y = y; |
| 1728 | g_vecProjected[vertex].z = z; |
| 1729 | |
| 1730 | g_vtxTransformed[vertex].x = x*g_vtxTransformed[vertex].w; |
| 1731 | g_vtxTransformed[vertex].y = y*g_vtxTransformed[vertex].w; |
| 1732 | g_vtxTransformed[vertex].z = z*g_vtxTransformed[vertex].w; |
| 1733 | } |
| 1734 | |
| 1735 | void ModifyVertexInfo(uint32 where, uint32 vertex, uint32 val) |
| 1736 | { |
| 1737 | switch (where) |
| 1738 | { |
| 1739 | case RSP_MV_WORD_OFFSET_POINT_RGBA: // Modify RGBA |
| 1740 | { |
| 1741 | uint32 r = (val>>24)&0xFF; |
| 1742 | uint32 g = (val>>16)&0xFF; |
| 1743 | uint32 b = (val>>8)&0xFF; |
| 1744 | uint32 a = val&0xFF; |
| 1745 | g_dwVtxDifColor[vertex] = COLOR_RGBA(r, g, b, a); |
| 1746 | LOG_UCODE("Modify vert %d color, 0x%08x", vertex, g_dwVtxDifColor[vertex]); |
| 1747 | } |
| 1748 | break; |
| 1749 | case RSP_MV_WORD_OFFSET_POINT_XYSCREEN: // Modify X,Y |
| 1750 | { |
| 1751 | uint16 nX = (uint16)(val>>16); |
| 1752 | short x = *((short*)&nX); |
| 1753 | x /= 4; |
| 1754 | |
| 1755 | uint16 nY = (uint16)(val&0xFFFF); |
| 1756 | short y = *((short*)&nY); |
| 1757 | y /= 4; |
| 1758 | |
| 1759 | // Should do viewport transform. |
| 1760 | |
| 1761 | |
| 1762 | x -= windowSetting.uViWidth/2; |
| 1763 | y = windowSetting.uViHeight/2-y; |
| 1764 | |
| 1765 | if( options.bEnableHacks && ((*g_GraphicsInfo.VI_X_SCALE_REG)&0xF) != 0 ) |
| 1766 | { |
| 1767 | // Tarzan |
| 1768 | // I don't know why Tarzan is different |
| 1769 | SetVertexXYZ(vertex, x/windowSetting.fViWidth, y/windowSetting.fViHeight, g_vecProjected[vertex].z); |
| 1770 | } |
| 1771 | else |
| 1772 | { |
| 1773 | // Toy Story 2 and other games |
| 1774 | SetVertexXYZ(vertex, x*2/windowSetting.fViWidth, y*2/windowSetting.fViHeight, g_vecProjected[vertex].z); |
| 1775 | } |
| 1776 | |
| 1777 | LOG_UCODE("Modify vert %d: x=%d, y=%d", vertex, x, y); |
| 1778 | VTX_DUMP(TRACE3("Modify vert %d: (%d,%d)", vertex, x, y)); |
| 1779 | } |
| 1780 | break; |
| 1781 | case RSP_MV_WORD_OFFSET_POINT_ZSCREEN: // Modify C |
| 1782 | { |
| 1783 | int z = val>>16; |
| 1784 | |
| 1785 | SetVertexXYZ(vertex, g_vecProjected[vertex].x, g_vecProjected[vertex].y, (((float)z/0x03FF)+0.5f)/2.0f ); |
| 1786 | LOG_UCODE("Modify vert %d: z=%d", vertex, z); |
| 1787 | VTX_DUMP(TRACE2("Modify vert %d: z=%d", vertex, z)); |
| 1788 | } |
| 1789 | break; |
| 1790 | case RSP_MV_WORD_OFFSET_POINT_ST: // Texture |
| 1791 | { |
| 1792 | short tu = short(val>>16); |
| 1793 | short tv = short(val & 0xFFFF); |
| 1794 | float ftu = tu / 32.0f; |
| 1795 | float ftv = tv / 32.0f; |
| 1796 | LOG_UCODE(" Setting vertex %d tu/tv to %f, %f", vertex, (float)tu, (float)tv); |
| 1797 | CRender::g_pRender->SetVtxTextureCoord(vertex, ftu/gRSP.fTexScaleX, ftv/gRSP.fTexScaleY); |
| 1798 | } |
| 1799 | break; |
| 1800 | } |
| 1801 | DEBUGGER_PAUSE_AND_DUMP(NEXT_VERTEX_CMD,{TRACE0("Paused at ModVertex Cmd");}); |
| 1802 | } |
| 1803 | |
| 1804 | void ProcessVertexDataDKR(uint32 dwAddr, uint32 dwV0, uint32 dwNum) |
| 1805 | { |
| 1806 | UpdateCombinedMatrix(); |
| 1807 | |
| 1808 | long long pVtxBase = (long long) (g_pRDRAMu8 + dwAddr); |
| 1809 | g_pVtxBase = (FiddledVtx*)pVtxBase; |
| 1810 | |
| 1811 | Matrix &matWorldProject = gRSP.DKRMatrixes[gRSP.DKRCMatrixIndex]; |
| 1812 | |
| 1813 | int nOff; |
| 1814 | |
| 1815 | bool addbase=false; |
| 1816 | if ((!gRSP.DKRBillBoard) || (gRSP.DKRCMatrixIndex != 2) ) |
| 1817 | addbase = false; |
| 1818 | else |
| 1819 | addbase = true; |
| 1820 | |
| 1821 | if( addbase && gRSP.DKRVtxCount == 0 && dwNum > 1 ) |
| 1822 | { |
| 1823 | gRSP.DKRVtxCount++; |
| 1824 | } |
| 1825 | |
| 1826 | LOG_UCODE(" ProcessVertexDataDKR, CMatrix = %d, Add base=%s", gRSP.DKRCMatrixIndex, gRSP.DKRBillBoard?"true":"false"); |
| 1827 | VTX_DUMP(TRACE2("DKR Setting Vertexes\nCMatrix = %d, Add base=%s", gRSP.DKRCMatrixIndex, gRSP.DKRBillBoard?"true":"false")); |
| 1828 | |
| 1829 | nOff = 0; |
| 1830 | uint32 end = dwV0 + dwNum; |
| 1831 | for (uint32 i = dwV0; i < end; i++) |
| 1832 | { |
| 1833 | XVECTOR3 w; |
| 1834 | |
| 1835 | g_vtxNonTransformed[i].x = (float)*(short*)((pVtxBase+nOff + 0) ^ 2); |
| 1836 | g_vtxNonTransformed[i].y = (float)*(short*)((pVtxBase+nOff + 2) ^ 2); |
| 1837 | g_vtxNonTransformed[i].z = (float)*(short*)((pVtxBase+nOff + 4) ^ 2); |
| 1838 | |
| 1839 | //if( status.isSSEEnabled ) |
| 1840 | // SSEVec3TransformDKR(g_vtxTransformed[i], g_vtxNonTransformed[i]); |
| 1841 | //else |
| 1842 | Vec3Transform(&g_vtxTransformed[i], (XVECTOR3*)&g_vtxNonTransformed[i], &matWorldProject); // Convert to w=1 |
| 1843 | |
| 1844 | if( gRSP.DKRVtxCount == 0 && dwNum==1 ) |
| 1845 | { |
| 1846 | gRSP.DKRBaseVec.x = g_vtxTransformed[i].x; |
| 1847 | gRSP.DKRBaseVec.y = g_vtxTransformed[i].y; |
| 1848 | gRSP.DKRBaseVec.z = g_vtxTransformed[i].z; |
| 1849 | gRSP.DKRBaseVec.w = g_vtxTransformed[i].w; |
| 1850 | } |
| 1851 | else if( addbase ) |
| 1852 | { |
| 1853 | g_vtxTransformed[i].x += gRSP.DKRBaseVec.x; |
| 1854 | g_vtxTransformed[i].y += gRSP.DKRBaseVec.y; |
| 1855 | g_vtxTransformed[i].z += gRSP.DKRBaseVec.z; |
| 1856 | g_vtxTransformed[i].w = gRSP.DKRBaseVec.w; |
| 1857 | } |
| 1858 | |
| 1859 | g_vecProjected[i].w = 1.0f / g_vtxTransformed[i].w; |
| 1860 | g_vecProjected[i].x = g_vtxTransformed[i].x * g_vecProjected[i].w; |
| 1861 | g_vecProjected[i].y = g_vtxTransformed[i].y * g_vecProjected[i].w; |
| 1862 | g_vecProjected[i].z = g_vtxTransformed[i].z * g_vecProjected[i].w; |
| 1863 | |
| 1864 | gRSP.DKRVtxCount++; |
| 1865 | |
| 1866 | VTX_DUMP(TRACE5("vtx %d: %f, %f, %f, %f", i, |
| 1867 | g_vtxTransformed[i].x,g_vtxTransformed[i].y,g_vtxTransformed[i].z,g_vtxTransformed[i].w)); |
| 1868 | |
| 1869 | if( gRSP.bFogEnabled ) |
| 1870 | { |
| 1871 | g_fFogCoord[i] = g_vecProjected[i].z; |
| 1872 | if( g_vecProjected[i].w < 0 || g_vecProjected[i].z < 0 || g_fFogCoord[i] < gRSPfFogMin ) |
| 1873 | g_fFogCoord[i] = gRSPfFogMin; |
| 1874 | } |
| 1875 | |
| 1876 | RSP_Vtx_Clipping(i); |
| 1877 | |
| 1878 | short wA = *(short*)((pVtxBase+nOff + 6) ^ 2); |
| 1879 | short wB = *(short*)((pVtxBase+nOff + 8) ^ 2); |
| 1880 | |
| 1881 | s8 r = (s8)(wA >> 8); |
| 1882 | s8 g = (s8)(wA); |
| 1883 | s8 b = (s8)(wB >> 8); |
| 1884 | s8 a = (s8)(wB); |
| 1885 | |
| 1886 | if (gRSP.bLightingEnable) |
| 1887 | { |
| 1888 | g_normal.x = (char)r; //norma.nx; |
| 1889 | g_normal.y = (char)g; //norma.ny; |
| 1890 | g_normal.z = (char)b; //norma.nz; |
| 1891 | |
| 1892 | Vec3TransformNormal(g_normal, matWorldProject) |
| 1893 | #if !defined(NO_ASM) |
| 1894 | if( status.isSSEEnabled ) |
| 1895 | g_dwVtxDifColor[i] = SSELightVert(); |
| 1896 | else |
| 1897 | #endif |
| 1898 | g_dwVtxDifColor[i] = LightVert(g_normal, i); |
| 1899 | } |
| 1900 | else |
| 1901 | { |
| 1902 | int nR, nG, nB, nA; |
| 1903 | |
| 1904 | nR = r; |
| 1905 | nG = g; |
| 1906 | nB = b; |
| 1907 | nA = a; |
| 1908 | // Assign true vert colour after lighting/fogging |
| 1909 | g_dwVtxDifColor[i] = COLOR_RGBA(nR, nG, nB, nA); |
| 1910 | } |
| 1911 | |
| 1912 | ReplaceAlphaWithFogFactor(i); |
| 1913 | |
| 1914 | g_fVtxTxtCoords[i].x = g_fVtxTxtCoords[i].y = 1; |
| 1915 | |
| 1916 | nOff += 10; |
| 1917 | } |
| 1918 | |
| 1919 | |
| 1920 | DEBUGGER_PAUSE_AND_DUMP(NEXT_VERTEX_CMD,{DebuggerAppendMsg("Paused at DKR Vertex Cmd, v0=%d, vn=%d, addr=%08X", dwV0, dwNum, dwAddr);}); |
| 1921 | } |
| 1922 | |
| 1923 | |
| 1924 | extern uint32 dwPDCIAddr; |
| 1925 | void ProcessVertexDataPD(uint32 dwAddr, uint32 dwV0, uint32 dwNum) |
| 1926 | { |
| 1927 | UpdateCombinedMatrix(); |
| 1928 | |
| 1929 | N64VtxPD * pVtxBase = (N64VtxPD*)(g_pRDRAMu8 + dwAddr); |
| 1930 | g_pVtxBase = (FiddledVtx*)pVtxBase; // Fix me |
| 1931 | |
| 1932 | for (uint32 i = dwV0; i < dwV0 + dwNum; i++) |
| 1933 | { |
| 1934 | N64VtxPD &vert = pVtxBase[i - dwV0]; |
| 1935 | |
| 1936 | g_vtxNonTransformed[i].x = (float)vert.x; |
| 1937 | g_vtxNonTransformed[i].y = (float)vert.y; |
| 1938 | g_vtxNonTransformed[i].z = (float)vert.z; |
| 1939 | |
| 1940 | #if !defined(NO_ASM) |
| 1941 | if( status.isSSEEnabled ) |
| 1942 | SSEVec3Transform(i); |
| 1943 | else |
| 1944 | #endif |
| 1945 | { |
| 1946 | Vec3Transform(&g_vtxTransformed[i], (XVECTOR3*)&g_vtxNonTransformed[i], &gRSPworldProject); // Convert to w=1 |
| 1947 | g_vecProjected[i].w = 1.0f / g_vtxTransformed[i].w; |
| 1948 | g_vecProjected[i].x = g_vtxTransformed[i].x * g_vecProjected[i].w; |
| 1949 | g_vecProjected[i].y = g_vtxTransformed[i].y * g_vecProjected[i].w; |
| 1950 | g_vecProjected[i].z = g_vtxTransformed[i].z * g_vecProjected[i].w; |
| 1951 | } |
| 1952 | |
| 1953 | g_fFogCoord[i] = g_vecProjected[i].z; |
| 1954 | if( g_vecProjected[i].w < 0 || g_vecProjected[i].z < 0 || g_fFogCoord[i] < gRSPfFogMin ) |
| 1955 | g_fFogCoord[i] = gRSPfFogMin; |
| 1956 | |
| 1957 | RSP_Vtx_Clipping(i); |
| 1958 | |
| 1959 | uint8 *addr = g_pRDRAMu8+dwPDCIAddr+ (vert.cidx&0xFF); |
| 1960 | uint32 a = addr[0]; |
| 1961 | uint32 r = addr[3]; |
| 1962 | uint32 g = addr[2]; |
| 1963 | uint32 b = addr[1]; |
| 1964 | |
| 1965 | if( gRSP.bLightingEnable ) |
| 1966 | { |
| 1967 | g_normal.x = (char)r; |
| 1968 | g_normal.y = (char)g; |
| 1969 | g_normal.z = (char)b; |
| 1970 | #if !defined(NO_ASM) |
| 1971 | if( status.isSSEEnabled ) |
| 1972 | { |
| 1973 | SSEVec3TransformNormal(); |
| 1974 | g_dwVtxDifColor[i] = SSELightVert(); |
| 1975 | } |
| 1976 | else |
| 1977 | #endif |
| 1978 | { |
| 1979 | Vec3TransformNormal(g_normal, gRSPmodelViewTop); |
| 1980 | g_dwVtxDifColor[i] = LightVert(g_normal, i); |
| 1981 | } |
| 1982 | *(((uint8*)&(g_dwVtxDifColor[i]))+3) = (uint8)a; // still use alpha from the vertex |
| 1983 | } |
| 1984 | else |
| 1985 | { |
| 1986 | if( (gRDP.geometryMode & G_SHADE) == 0 && gRSP.ucode < 5 ) //Shade is disabled |
| 1987 | { |
| 1988 | g_dwVtxDifColor[i] = gRDP.primitiveColor; |
| 1989 | } |
| 1990 | else //FLAT shade |
| 1991 | { |
| 1992 | g_dwVtxDifColor[i] = COLOR_RGBA(r, g, b, a); |
| 1993 | } |
| 1994 | } |
| 1995 | |
| 1996 | if( options.bWinFrameMode ) |
| 1997 | { |
| 1998 | g_dwVtxDifColor[i] = COLOR_RGBA(r, g, b, a); |
| 1999 | } |
| 2000 | |
| 2001 | ReplaceAlphaWithFogFactor(i); |
| 2002 | |
| 2003 | VECTOR2 & t = g_fVtxTxtCoords[i]; |
| 2004 | if (gRSP.bTextureGen && gRSP.bLightingEnable ) |
| 2005 | { |
| 2006 | // Not sure if we should transform the normal here |
| 2007 | //Matrix & matWV = gRSP.projectionMtxs[gRSP.projectionMtxTop]; |
| 2008 | //Vec3TransformNormal(g_normal, matWV); |
| 2009 | |
| 2010 | TexGen(g_fVtxTxtCoords[i].x, g_fVtxTxtCoords[i].y); |
| 2011 | } |
| 2012 | else |
| 2013 | { |
| 2014 | t.x = vert.s; |
| 2015 | t.y = vert.t; |
| 2016 | } |
| 2017 | |
| 2018 | |
| 2019 | VTX_DUMP( |
| 2020 | { |
| 2021 | DebuggerAppendMsg("vtx %d: %d %d %d", i, vert.x,vert.y,vert.z); |
| 2022 | DebuggerAppendMsg(" : %f, %f, %f, %f", |
| 2023 | g_vtxTransformed[i].x,g_vtxTransformed[i].y,g_vtxTransformed[i].z,g_vtxTransformed[i].w); |
| 2024 | DebuggerAppendMsg(" : %X, %X, %X, %X", r,g,b,a); |
| 2025 | DebuggerAppendMsg(" : u=%f, v=%f", t.x, t.y); |
| 2026 | }); |
| 2027 | } |
| 2028 | |
| 2029 | VTX_DUMP(TRACE2("Setting Vertexes: %d - %d\n", dwV0, dwV0+dwNum-1)); |
| 2030 | DEBUGGER_PAUSE_AND_DUMP(NEXT_VERTEX_CMD,{TRACE0("Paused at Vertex Cmd");}); |
| 2031 | } |
| 2032 | |
| 2033 | extern uint32 dwConkerVtxZAddr; |
| 2034 | void ProcessVertexDataConker(uint32 dwAddr, uint32 dwV0, uint32 dwNum) |
| 2035 | { |
| 2036 | UpdateCombinedMatrix(); |
| 2037 | |
| 2038 | FiddledVtx * pVtxBase = (FiddledVtx*)(g_pRDRAMu8 + dwAddr); |
| 2039 | g_pVtxBase = pVtxBase; |
| 2040 | //short *vertexColoraddr = (short*)(g_pRDRAMu8+dwConkerVtxZAddr); |
| 2041 | |
| 2042 | for (uint32 i = dwV0; i < dwV0 + dwNum; i++) |
| 2043 | { |
| 2044 | SP_Timing(RSP_GBI0_Vtx); |
| 2045 | |
| 2046 | FiddledVtx & vert = pVtxBase[i - dwV0]; |
| 2047 | |
| 2048 | g_vtxNonTransformed[i].x = (float)vert.x; |
| 2049 | g_vtxNonTransformed[i].y = (float)vert.y; |
| 2050 | g_vtxNonTransformed[i].z = (float)vert.z; |
| 2051 | |
| 2052 | #if !defined(NO_ASM) |
| 2053 | if( status.isSSEEnabled ) |
| 2054 | SSEVec3Transform(i); |
| 2055 | else |
| 2056 | #endif |
| 2057 | { |
| 2058 | Vec3Transform(&g_vtxTransformed[i], (XVECTOR3*)&g_vtxNonTransformed[i], &gRSPworldProject); // Convert to w=1 |
| 2059 | g_vecProjected[i].w = 1.0f / g_vtxTransformed[i].w; |
| 2060 | g_vecProjected[i].x = g_vtxTransformed[i].x * g_vecProjected[i].w; |
| 2061 | g_vecProjected[i].y = g_vtxTransformed[i].y * g_vecProjected[i].w; |
| 2062 | g_vecProjected[i].z = g_vtxTransformed[i].z * g_vecProjected[i].w; |
| 2063 | } |
| 2064 | |
| 2065 | g_fFogCoord[i] = g_vecProjected[i].z; |
| 2066 | if( g_vecProjected[i].w < 0 || g_vecProjected[i].z < 0 || g_fFogCoord[i] < gRSPfFogMin ) |
| 2067 | g_fFogCoord[i] = gRSPfFogMin; |
| 2068 | |
| 2069 | VTX_DUMP( |
| 2070 | { |
| 2071 | uint32 *dat = (uint32*)(&vert); |
| 2072 | DebuggerAppendMsg("vtx %d: %08X %08X %08X %08X", i, dat[0],dat[1],dat[2],dat[3]); |
| 2073 | DebuggerAppendMsg(" : %f, %f, %f, %f", |
| 2074 | g_vtxTransformed[i].x,g_vtxTransformed[i].y,g_vtxTransformed[i].z,g_vtxTransformed[i].w); |
| 2075 | DebuggerAppendMsg(" : %f, %f, %f, %f", |
| 2076 | g_vecProjected[i].x,g_vecProjected[i].y,g_vecProjected[i].z,g_vecProjected[i].w); |
| 2077 | }); |
| 2078 | |
| 2079 | RSP_Vtx_Clipping(i); |
| 2080 | |
| 2081 | if( gRSP.bLightingEnable ) |
| 2082 | { |
| 2083 | { |
| 2084 | uint32 r= ((gRSP.ambientLightColor>>16)&0xFF); |
| 2085 | uint32 g= ((gRSP.ambientLightColor>> 8)&0xFF); |
| 2086 | uint32 b= ((gRSP.ambientLightColor )&0xFF); |
| 2087 | for( uint32 k=1; k<=gRSPnumLights; k++) |
| 2088 | { |
| 2089 | r += gRSPlights[k].r; |
| 2090 | g += gRSPlights[k].g; |
| 2091 | b += gRSPlights[k].b; |
| 2092 | } |
| 2093 | if( r>255 ) r=255; |
| 2094 | if( g>255 ) g=255; |
| 2095 | if( b>255 ) b=255; |
| 2096 | r *= vert.rgba.r ; |
| 2097 | g *= vert.rgba.g ; |
| 2098 | b *= vert.rgba.b ; |
| 2099 | r >>= 8; |
| 2100 | g >>= 8; |
| 2101 | b >>= 8; |
| 2102 | g_dwVtxDifColor[i] = 0xFF000000; |
| 2103 | g_dwVtxDifColor[i] |= (r<<16); |
| 2104 | g_dwVtxDifColor[i] |= (g<< 8); |
| 2105 | g_dwVtxDifColor[i] |= (b ); |
| 2106 | } |
| 2107 | |
| 2108 | *(((uint8*)&(g_dwVtxDifColor[i]))+3) = vert.rgba.a; // still use alpha from the vertex |
| 2109 | } |
| 2110 | else |
| 2111 | { |
| 2112 | if( (gRDP.geometryMode & G_SHADE) == 0 && gRSP.ucode < 5 ) //Shade is disabled |
| 2113 | { |
| 2114 | g_dwVtxDifColor[i] = gRDP.primitiveColor; |
| 2115 | } |
| 2116 | else //FLAT shade |
| 2117 | { |
| 2118 | g_dwVtxDifColor[i] = COLOR_RGBA(vert.rgba.r, vert.rgba.g, vert.rgba.b, vert.rgba.a); |
| 2119 | } |
| 2120 | } |
| 2121 | |
| 2122 | if( options.bWinFrameMode ) |
| 2123 | { |
| 2124 | //g_vecProjected[i].z = 0; |
| 2125 | g_dwVtxDifColor[i] = COLOR_RGBA(vert.rgba.r, vert.rgba.g, vert.rgba.b, vert.rgba.a); |
| 2126 | } |
| 2127 | |
| 2128 | ReplaceAlphaWithFogFactor(i); |
| 2129 | |
| 2130 | // Update texture coords n.b. need to divide tu/tv by bogus scale on addition to buffer |
| 2131 | //VECTOR2 & t = g_fVtxTxtCoords[i]; |
| 2132 | |
| 2133 | // If the vert is already lit, then there is no normal (and hence we |
| 2134 | // can't generate tex coord) |
| 2135 | if (gRSP.bTextureGen && gRSP.bLightingEnable ) |
| 2136 | { |
| 2137 | g_normal.x = (float)*(char*)(g_pRDRAMu8+ (((i<<1)+0)^3)+dwConkerVtxZAddr); |
| 2138 | g_normal.y = (float)*(char*)(g_pRDRAMu8+ (((i<<1)+1)^3)+dwConkerVtxZAddr); |
| 2139 | g_normal.z = (float)*(char*)(g_pRDRAMu8+ (((i<<1)+2)^3)+dwConkerVtxZAddr); |
| 2140 | Vec3TransformNormal(g_normal, gRSPmodelViewTop); |
| 2141 | TexGen(g_fVtxTxtCoords[i].x, g_fVtxTxtCoords[i].y); |
| 2142 | } |
| 2143 | else |
| 2144 | { |
| 2145 | g_fVtxTxtCoords[i].x = (float)vert.tu; |
| 2146 | g_fVtxTxtCoords[i].y = (float)vert.tv; |
| 2147 | } |
| 2148 | } |
| 2149 | |
| 2150 | VTX_DUMP(TRACE2("Setting Vertexes: %d - %d\n", dwV0, dwV0+dwNum-1)); |
| 2151 | DEBUGGER_PAUSE_AND_DUMP(NEXT_VERTEX_CMD,{DebuggerAppendMsg("Paused at Vertex Cmd");}); |
| 2152 | } |
| 2153 | |
| 2154 | |
| 2155 | typedef struct{ |
| 2156 | short y; |
| 2157 | short x; |
| 2158 | short flag; |
| 2159 | short z; |
| 2160 | } RS_Vtx_XYZ; |
| 2161 | |
| 2162 | typedef union { |
| 2163 | struct { |
| 2164 | uint8 a; |
| 2165 | uint8 b; |
| 2166 | uint8 g; |
| 2167 | uint8 r; |
| 2168 | }; |
| 2169 | struct { |
| 2170 | char na; //a |
| 2171 | char nz; //b |
| 2172 | char ny; //g |
| 2173 | char nx; //r |
| 2174 | }; |
| 2175 | } RS_Vtx_Color; |
| 2176 | |
| 2177 | |
| 2178 | void ProcessVertexData_Rogue_Squadron(uint32 dwXYZAddr, uint32 dwColorAddr, uint32 dwXYZCmd, uint32 dwColorCmd) |
| 2179 | { |
| 2180 | UpdateCombinedMatrix(); |
| 2181 | |
| 2182 | uint32 dwV0 = 0; |
| 2183 | uint32 dwNum = (dwXYZCmd&0xFF00)>>10; |
| 2184 | |
| 2185 | RS_Vtx_XYZ * pVtxXYZBase = (RS_Vtx_XYZ*)(g_pRDRAMu8 + dwXYZAddr); |
| 2186 | RS_Vtx_Color * pVtxColorBase = (RS_Vtx_Color*)(g_pRDRAMu8 + dwColorAddr); |
| 2187 | |
| 2188 | uint32 i; |
| 2189 | for (i = dwV0; i < dwV0 + dwNum; i++) |
| 2190 | { |
| 2191 | RS_Vtx_XYZ & vertxyz = pVtxXYZBase[i - dwV0]; |
| 2192 | RS_Vtx_Color & vertcolors = pVtxColorBase[i - dwV0]; |
| 2193 | |
| 2194 | g_vtxNonTransformed[i].x = (float)vertxyz.x; |
| 2195 | g_vtxNonTransformed[i].y = (float)vertxyz.y; |
| 2196 | g_vtxNonTransformed[i].z = (float)vertxyz.z; |
| 2197 | |
| 2198 | #if !defined(NO_ASM) |
| 2199 | if( status.isSSEEnabled ) |
| 2200 | SSEVec3Transform(i); |
| 2201 | else |
| 2202 | #endif |
| 2203 | { |
| 2204 | Vec3Transform(&g_vtxTransformed[i], (XVECTOR3*)&g_vtxNonTransformed[i], &gRSPworldProject); // Convert to w=1 |
| 2205 | g_vecProjected[i].w = 1.0f / g_vtxTransformed[i].w; |
| 2206 | g_vecProjected[i].x = g_vtxTransformed[i].x * g_vecProjected[i].w; |
| 2207 | g_vecProjected[i].y = g_vtxTransformed[i].y * g_vecProjected[i].w; |
| 2208 | g_vecProjected[i].z = g_vtxTransformed[i].z * g_vecProjected[i].w; |
| 2209 | } |
| 2210 | |
| 2211 | VTX_DUMP( |
| 2212 | { |
| 2213 | DebuggerAppendMsg(" : %f, %f, %f, %f", |
| 2214 | g_vtxTransformed[i].x,g_vtxTransformed[i].y,g_vtxTransformed[i].z,g_vtxTransformed[i].w); |
| 2215 | DebuggerAppendMsg(" : %f, %f, %f, %f", |
| 2216 | g_vecProjected[i].x,g_vecProjected[i].y,g_vecProjected[i].z,g_vecProjected[i].w); |
| 2217 | }); |
| 2218 | |
| 2219 | g_fFogCoord[i] = g_vecProjected[i].z; |
| 2220 | if( g_vecProjected[i].w < 0 || g_vecProjected[i].z < 0 || g_fFogCoord[i] < gRSPfFogMin ) |
| 2221 | g_fFogCoord[i] = gRSPfFogMin; |
| 2222 | |
| 2223 | RSP_Vtx_Clipping(i); |
| 2224 | |
| 2225 | if( gRSP.bLightingEnable ) |
| 2226 | { |
| 2227 | g_normal.x = (float)vertcolors.nx; |
| 2228 | g_normal.y = (float)vertcolors.ny; |
| 2229 | g_normal.z = (float)vertcolors.nz; |
| 2230 | |
| 2231 | #if !defined(NO_ASM) |
| 2232 | if( status.isSSEEnabled ) |
| 2233 | { |
| 2234 | SSEVec3TransformNormal(); |
| 2235 | g_dwVtxDifColor[i] = SSELightVert(); |
| 2236 | } |
| 2237 | else |
| 2238 | #endif |
| 2239 | { |
| 2240 | Vec3TransformNormal(g_normal, gRSPmodelViewTop); |
| 2241 | g_dwVtxDifColor[i] = LightVert(g_normal, i); |
| 2242 | } |
| 2243 | *(((uint8*)&(g_dwVtxDifColor[i]))+3) = vertcolors.a; // still use alpha from the vertex |
| 2244 | } |
| 2245 | else |
| 2246 | { |
| 2247 | if( (gRDP.geometryMode & G_SHADE) == 0 && gRSP.ucode < 5 ) //Shade is disabled |
| 2248 | { |
| 2249 | g_dwVtxDifColor[i] = gRDP.primitiveColor; |
| 2250 | } |
| 2251 | else //FLAT shade |
| 2252 | { |
| 2253 | g_dwVtxDifColor[i] = COLOR_RGBA(vertcolors.r, vertcolors.g, vertcolors.b, vertcolors.a); |
| 2254 | } |
| 2255 | } |
| 2256 | |
| 2257 | if( options.bWinFrameMode ) |
| 2258 | { |
| 2259 | g_dwVtxDifColor[i] = COLOR_RGBA(vertcolors.r, vertcolors.g, vertcolors.b, vertcolors.a); |
| 2260 | } |
| 2261 | |
| 2262 | ReplaceAlphaWithFogFactor(i); |
| 2263 | |
| 2264 | /* |
| 2265 | // Update texture coords n.b. need to divide tu/tv by bogus scale on addition to buffer |
| 2266 | VECTOR2 & t = g_fVtxTxtCoords[i]; |
| 2267 | |
| 2268 | // If the vert is already lit, then there is no normal (and hence we |
| 2269 | // can't generate tex coord) |
| 2270 | if (gRSP.bTextureGen && gRSP.bLightingEnable && g_textures[gRSP.curTile].m_bTextureEnable ) |
| 2271 | { |
| 2272 | TexGen(g_fVtxTxtCoords[i].x, g_fVtxTxtCoords[i].y); |
| 2273 | } |
| 2274 | else |
| 2275 | { |
| 2276 | t.x = (float)vert.tu; |
| 2277 | t.y = (float)vert.tv; |
| 2278 | } |
| 2279 | */ |
| 2280 | } |
| 2281 | |
| 2282 | VTX_DUMP(TRACE2("Setting Vertexes: %d - %d\n", dwV0, dwV0+dwNum-1)); |
| 2283 | DEBUGGER_PAUSE_AND_DUMP(NEXT_VERTEX_CMD,{TRACE0("Paused at Vertex Cmd");}); |
| 2284 | } |
| 2285 | |
| 2286 | void SetLightCol(uint32 dwLight, uint32 dwCol) |
| 2287 | { |
| 2288 | gRSPlights[dwLight].r = (uint8)((dwCol >> 24)&0xFF); |
| 2289 | gRSPlights[dwLight].g = (uint8)((dwCol >> 16)&0xFF); |
| 2290 | gRSPlights[dwLight].b = (uint8)((dwCol >> 8)&0xFF); |
| 2291 | gRSPlights[dwLight].a = 255; // Ignore light alpha |
| 2292 | gRSPlights[dwLight].fr = (float)gRSPlights[dwLight].r; |
| 2293 | gRSPlights[dwLight].fg = (float)gRSPlights[dwLight].g; |
| 2294 | gRSPlights[dwLight].fb = (float)gRSPlights[dwLight].b; |
| 2295 | gRSPlights[dwLight].fa = 255; // Ignore light alpha |
| 2296 | |
| 2297 | //TRACE1("Set light %d color", dwLight); |
| 2298 | LIGHT_DUMP(TRACE2("Set Light %d color: %08X", dwLight, dwCol)); |
| 2299 | } |
| 2300 | |
| 2301 | void SetLightDirection(uint32 dwLight, float x, float y, float z, float range) |
| 2302 | { |
| 2303 | //gRSP.bLightIsUpdated = true; |
| 2304 | |
| 2305 | //gRSPlights[dwLight].ox = x; |
| 2306 | //gRSPlights[dwLight].oy = y; |
| 2307 | //gRSPlights[dwLight].oz = z; |
| 2308 | |
| 2309 | register float w = range == 0 ? (float)sqrt(x*x+y*y+z*z) : 1; |
| 2310 | |
| 2311 | gRSPlights[dwLight].x = x/w; |
| 2312 | gRSPlights[dwLight].y = y/w; |
| 2313 | gRSPlights[dwLight].z = z/w; |
| 2314 | gRSPlights[dwLight].range = range; |
| 2315 | DEBUGGER_PAUSE_AND_DUMP(NEXT_SET_LIGHT,TRACE5("Set Light %d dir: %.4f, %.4f, %.4f, %.4f", dwLight, x, y, z, range)); |
| 2316 | } |
| 2317 | |
| 2318 | static float maxS0, maxT0; |
| 2319 | static float maxS1, maxT1; |
| 2320 | static bool validS0, validT0; |
| 2321 | static bool validS1, validT1; |
| 2322 | |
| 2323 | void LogTextureCoords(float fTex0S, float fTex0T, float fTex1S, float fTex1T) |
| 2324 | { |
| 2325 | if( validS0 ) |
| 2326 | { |
| 2327 | if( fTex0S<0 || fTex0S>maxS0 ) validS0 = false; |
| 2328 | } |
| 2329 | if( validT0 ) |
| 2330 | { |
| 2331 | if( fTex0T<0 || fTex0T>maxT0 ) validT0 = false; |
| 2332 | } |
| 2333 | if( validS1 ) |
| 2334 | { |
| 2335 | if( fTex1S<0 || fTex1S>maxS1 ) validS1 = false; |
| 2336 | } |
| 2337 | if( validT1 ) |
| 2338 | { |
| 2339 | if( fTex1T<0 || fTex1T>maxT1 ) validT1 = false; |
| 2340 | } |
| 2341 | } |
| 2342 | |
| 2343 | bool CheckTextureCoords(int tex) |
| 2344 | { |
| 2345 | if( tex==0 ) |
| 2346 | { |
| 2347 | return validS0&&validT0; |
| 2348 | } |
| 2349 | else |
| 2350 | { |
| 2351 | return validS1&&validT1; |
| 2352 | } |
| 2353 | } |
| 2354 | |
| 2355 | void ResetTextureCoordsLog(float maxs0, float maxt0, float maxs1, float maxt1) |
| 2356 | { |
| 2357 | maxS0 = maxs0; |
| 2358 | maxT0 = maxt0; |
| 2359 | maxS1 = maxs1; |
| 2360 | maxT1 = maxt1; |
| 2361 | validS0 = validT0 = true; |
| 2362 | validS1 = validT1 = true; |
| 2363 | } |
| 2364 | |
| 2365 | void ForceMainTextureIndex(int dwTile) |
| 2366 | { |
| 2367 | if( dwTile == 1 && !(CRender::g_pRender->IsTexel0Enable()) && CRender::g_pRender->IsTexel1Enable() ) |
| 2368 | { |
| 2369 | // Hack |
| 2370 | gRSP.curTile = 0; |
| 2371 | } |
| 2372 | else |
| 2373 | { |
| 2374 | gRSP.curTile = dwTile; |
| 2375 | } |
| 2376 | } |
| 2377 | |
| 2378 | float HackZ2(float z) |
| 2379 | { |
| 2380 | z = (z+9)/10; |
| 2381 | return z; |
| 2382 | } |
| 2383 | |
| 2384 | float HackZ(float z) |
| 2385 | { |
| 2386 | return HackZ2(z); |
| 2387 | |
| 2388 | if( z < 0.1 && z >= 0 ) |
| 2389 | z = (.1f+z)/2; |
| 2390 | else if( z < 0 ) |
| 2391 | //return (10+z)/100; |
| 2392 | z = (expf(z)/20); |
| 2393 | return z; |
| 2394 | } |
| 2395 | |
| 2396 | void HackZ(std::vector<XVECTOR3>& points) |
| 2397 | { |
| 2398 | int size = points.size(); |
| 2399 | for( int i=0; i<size; i++) |
| 2400 | { |
| 2401 | XVECTOR3 &v = points[i]; |
| 2402 | v.z = (float)HackZ(v.z); |
| 2403 | } |
| 2404 | } |
| 2405 | |
| 2406 | void HackZAll() |
| 2407 | { |
| 2408 | if( CDeviceBuilder::m_deviceGeneralType == DIRECTX_DEVICE ) |
| 2409 | { |
| 2410 | for( uint32 i=0; i<gRSP.numVertices; i++) |
| 2411 | { |
| 2412 | g_vtxBuffer[i].z = HackZ(g_vtxBuffer[i].z); |
| 2413 | } |
| 2414 | } |
| 2415 | else |
| 2416 | { |
| 2417 | for( uint32 i=0; i<gRSP.numVertices; i++) |
| 2418 | { |
| 2419 | float w = g_vtxProjected5[i][3]; |
| 2420 | g_vtxProjected5[i][2] = HackZ(g_vtxProjected5[i][2]/w)*w; |
| 2421 | } |
| 2422 | } |
| 2423 | } |
| 2424 | |
| 2425 | |
| 2426 | extern XMATRIX reverseXY; |
| 2427 | extern XMATRIX reverseY; |
| 2428 | |
| 2429 | void UpdateCombinedMatrix() |
| 2430 | { |
| 2431 | if( gRSP.bMatrixIsUpdated ) |
| 2432 | { |
| 2433 | gRSPworldProject = gRSP.modelviewMtxs[gRSP.modelViewMtxTop] * gRSP.projectionMtxs[gRSP.projectionMtxTop]; |
| 2434 | gRSP.bMatrixIsUpdated = false; |
| 2435 | gRSP.bCombinedMatrixIsUpdated = true; |
| 2436 | } |
| 2437 | |
| 2438 | if( gRSP.bCombinedMatrixIsUpdated ) |
| 2439 | { |
| 2440 | if( options.enableHackForGames == HACK_REVERSE_XY_COOR ) |
| 2441 | { |
| 2442 | gRSPworldProject = gRSPworldProject * reverseXY; |
| 2443 | } |
| 2444 | if( options.enableHackForGames == HACK_REVERSE_Y_COOR ) |
| 2445 | { |
| 2446 | gRSPworldProject = gRSPworldProject * reverseY; |
| 2447 | } |
| 2448 | #if !defined(NO_ASM) |
| 2449 | if( status.isSSEEnabled ) |
| 2450 | { |
| 2451 | MatrixTranspose(&gRSPworldProjectTransported, &gRSPworldProject); |
| 2452 | } |
| 2453 | #endif |
| 2454 | gRSP.bCombinedMatrixIsUpdated = false; |
| 2455 | } |
| 2456 | |
| 2457 | //if( gRSP.bWorldMatrixIsUpdated || gRSP.bLightIsUpdated ) |
| 2458 | //{ |
| 2459 | // // Update lights with transported world matrix |
| 2460 | // for( unsigned int l=0; l<gRSPnumLights; l++) |
| 2461 | // { |
| 2462 | // Vec3TransformCoord(&gRSPlights[l].td, &gRSPlights[l].od, &gRSPmodelViewTopTranspose); |
| 2463 | // Vec3Normalize(&gRSPlights[l].td,&gRSPlights[l].td); |
| 2464 | // } |
| 2465 | |
| 2466 | // gRSP.bWorldMatrixIsUpdated = false; |
| 2467 | // gRSP.bLightIsUpdated = false; |
| 2468 | //} |
| 2469 | } |
| 2470 | |