| 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 | #include "OGLExtensions.h" |
| 20 | |
| 21 | #include "OGLCombinerNV.h" |
| 22 | #include "OGLRender.h" |
| 23 | #include "OGLGraphicsContext.h" |
| 24 | |
| 25 | //======================================================================== |
| 26 | #define MUX_E_F (MUX_PRIMLODFRAC+1) |
| 27 | #define MUX_SPARE1 (MUX_E_F+1) |
| 28 | #define MUX_SECONDARY_COLOR (MUX_SPARE1+1) |
| 29 | #define MUX_NOT_USED MUX_ERR |
| 30 | #define MUX_COMBINED_SIGNED (MUX_SECONDARY_COLOR+1) //Use only by Nvidia register combiner |
| 31 | |
| 32 | |
| 33 | typedef struct { |
| 34 | GLenum input; |
| 35 | GLenum mapping; |
| 36 | GLenum componentUsage; |
| 37 | }RGBMapType; |
| 38 | |
| 39 | RGBMapType RGBmap1[] = |
| 40 | { |
| 41 | {GL_ZERO, GL_UNSIGNED_IDENTITY_NV, GL_RGB}, //MUX_0 = 0, |
| 42 | {GL_ZERO, GL_UNSIGNED_INVERT_NV, GL_RGB}, //MUX_1, = ZERO NEG |
| 43 | {GL_SPARE0_NV, GL_UNSIGNED_IDENTITY_NV, GL_RGB}, //MUX_COMBINED, |
| 44 | {GL_TEXTURE0_ARB, GL_UNSIGNED_IDENTITY_NV, GL_RGB}, //MUX_TEXEL0, |
| 45 | {GL_TEXTURE1_ARB, GL_UNSIGNED_IDENTITY_NV, GL_RGB}, //MUX_TEXEL1, |
| 46 | {GL_CONSTANT_COLOR0_NV, GL_UNSIGNED_IDENTITY_NV, GL_RGB}, //MUX_PRIM, |
| 47 | {GL_PRIMARY_COLOR_NV, GL_UNSIGNED_IDENTITY_NV, GL_RGB}, //MUX_SHADE, |
| 48 | {GL_CONSTANT_COLOR1_NV, GL_UNSIGNED_IDENTITY_NV, GL_RGB}, //MUX_ENV, |
| 49 | {GL_SPARE0_NV, GL_UNSIGNED_IDENTITY_NV, GL_ALPHA}, //MUX_COMBALPHA, |
| 50 | {GL_TEXTURE0_ARB, GL_UNSIGNED_IDENTITY_NV, GL_ALPHA}, //MUX_T0_ALPHA, |
| 51 | {GL_TEXTURE1_ARB, GL_UNSIGNED_IDENTITY_NV, GL_ALPHA}, //MUX_T1_ALPHA, |
| 52 | {GL_CONSTANT_COLOR0_NV, GL_UNSIGNED_IDENTITY_NV, GL_ALPHA}, //MUX_PRIM_ALPHA, |
| 53 | {GL_PRIMARY_COLOR_NV, GL_UNSIGNED_IDENTITY_NV, GL_ALPHA}, //MUX_SHADE_ALPHA, |
| 54 | {GL_CONSTANT_COLOR1_NV, GL_UNSIGNED_IDENTITY_NV, GL_ALPHA}, //MUX_ENV_ALPHA, |
| 55 | {GL_CONSTANT_COLOR1_NV, GL_UNSIGNED_IDENTITY_NV, GL_RGB}, //MUX_LODFRAC, |
| 56 | {GL_CONSTANT_COLOR1_NV, GL_UNSIGNED_IDENTITY_NV, GL_RGB}, //MUX_PRIMLODFRAC, |
| 57 | {GL_E_TIMES_F_NV, GL_UNSIGNED_IDENTITY_NV, GL_RGB}, //MUX_E_F, |
| 58 | {GL_SPARE1_NV, GL_UNSIGNED_IDENTITY_NV, GL_RGB}, //MUX_SPARE1, |
| 59 | {GL_SECONDARY_COLOR_NV, GL_UNSIGNED_IDENTITY_NV, GL_RGB}, //MUX_SECONDARY_COLOR, |
| 60 | {GL_SPARE0_NV, GL_SIGNED_IDENTITY_NV, GL_RGB}, //MUX_COMBINED_SIGNED, |
| 61 | }; |
| 62 | |
| 63 | |
| 64 | //======================================================================== |
| 65 | COGLColorCombinerNvidia::COGLColorCombinerNvidia(CRender *pRender) : |
| 66 | COGLColorCombiner4(pRender) |
| 67 | { |
| 68 | m_bNVSupported = false; |
| 69 | delete m_pDecodedMux; |
| 70 | m_pDecodedMux = new COGLDecodedMux; |
| 71 | m_pDecodedMux->m_maxConstants=2; |
| 72 | } |
| 73 | |
| 74 | COGLColorCombinerNvidia::~COGLColorCombinerNvidia() |
| 75 | { |
| 76 | m_vCompiledSettings.clear(); |
| 77 | } |
| 78 | |
| 79 | |
| 80 | bool COGLColorCombinerNvidia::Initialize(void) |
| 81 | { |
| 82 | m_bNVSupported = false; |
| 83 | |
| 84 | if( COGLColorCombiner4::Initialize() ) |
| 85 | { |
| 86 | m_bSupportMultiTexture = true; |
| 87 | |
| 88 | COGLGraphicsContext *pcontext = (COGLGraphicsContext *)(CGraphicsContext::g_pGraphicsContext); |
| 89 | if( pcontext->IsExtensionSupported("GL_NV_texture_env_combine4") || pcontext->IsExtensionSupported("GL_NV_register_combiners") ) |
| 90 | { |
| 91 | m_bNVSupported = true; |
| 92 | glEnable(GL_REGISTER_COMBINERS_NV); |
| 93 | return true; |
| 94 | } |
| 95 | else |
| 96 | { |
| 97 | DebugMessage(M64MSG_ERROR, "Your video card does not support Nvidia OpenGL extension combiner"); |
| 98 | glDisable(GL_REGISTER_COMBINERS_NV); |
| 99 | return false; |
| 100 | } |
| 101 | } |
| 102 | |
| 103 | glDisable(GL_REGISTER_COMBINERS_NV); |
| 104 | return false; |
| 105 | } |
| 106 | |
| 107 | void COGLColorCombinerNvidia::InitCombinerCycle12(void) |
| 108 | { |
| 109 | if( !m_bNVSupported ) {COGLColorCombiner4::InitCombinerCycle12(); return;} |
| 110 | |
| 111 | glEnable(GL_REGISTER_COMBINERS_NV); |
| 112 | |
| 113 | #ifdef DEBUGGER |
| 114 | if( debuggerDropCombiners ) |
| 115 | { |
| 116 | m_vCompiledSettings.clear(); |
| 117 | m_dwLastMux0 = m_dwLastMux1 = 0; |
| 118 | debuggerDropCombiners = false; |
| 119 | } |
| 120 | #endif |
| 121 | |
| 122 | m_pOGLRender->EnableMultiTexture(); |
| 123 | bool combinerIsChanged = false; |
| 124 | |
| 125 | if( m_pDecodedMux->m_dwMux0 != m_dwLastMux0 || m_pDecodedMux->m_dwMux1 != m_dwLastMux1 || m_lastIndex < 0 ) |
| 126 | { |
| 127 | combinerIsChanged = true; |
| 128 | m_lastIndex = FindCompiledMux(); |
| 129 | if( m_lastIndex < 0 ) // Can not found |
| 130 | { |
| 131 | NVRegisterCombinerParserType result; |
| 132 | ParseDecodedMux(result); |
| 133 | m_lastIndex= SaveParserResult(result); |
| 134 | } |
| 135 | |
| 136 | m_dwLastMux0 = m_pDecodedMux->m_dwMux0; |
| 137 | m_dwLastMux1 = m_pDecodedMux->m_dwMux1; |
| 138 | GenerateNVRegisterCombinerSetting(m_lastIndex); |
| 139 | } |
| 140 | |
| 141 | m_pOGLRender->SetAllTexelRepeatFlag(); |
| 142 | |
| 143 | if( m_bCycleChanged || combinerIsChanged || gRDP.texturesAreReloaded || gRDP.colorsAreReloaded ) |
| 144 | { |
| 145 | gRDP.texturesAreReloaded = false; |
| 146 | if( m_bCycleChanged || combinerIsChanged ) |
| 147 | { |
| 148 | GenerateNVRegisterCombinerSettingConstants(m_lastIndex); |
| 149 | GenerateNVRegisterCombinerSetting(m_lastIndex); |
| 150 | ApplyFogAtFinalStage(); |
| 151 | } |
| 152 | else if( gRDP.colorsAreReloaded ) |
| 153 | { |
| 154 | GenerateNVRegisterCombinerSettingConstants(m_lastIndex); |
| 155 | } |
| 156 | |
| 157 | gRDP.colorsAreReloaded = false; |
| 158 | } |
| 159 | } |
| 160 | |
| 161 | void COGLColorCombinerNvidia::ParseDecodedMux(NVRegisterCombinerParserType &result) // Compile the decodedMux into NV register combiner setting |
| 162 | { |
| 163 | //int stagesForRGB=0; |
| 164 | //int stagesForAlpha=0; |
| 165 | //int stages=0; |
| 166 | |
| 167 | COGLDecodedMux &mux = *(COGLDecodedMux*)m_pDecodedMux; |
| 168 | mux.To_AB_Add_CD_Format(); |
| 169 | |
| 170 | result.stagesUsed=0; |
| 171 | |
| 172 | if( StagesNeedToUse(mux, N64Cycle0RGB) == 0 ) |
| 173 | { |
| 174 | // Nothing to be done for RGB |
| 175 | ByPassGeneralStage(result.s1rgb); |
| 176 | ByPassGeneralStage(result.s2rgb); |
| 177 | ByPassFinalStage(result.finalrgb); |
| 178 | } |
| 179 | else if( StagesNeedToUse(mux, N64Cycle0RGB) == 1 ) |
| 180 | { |
| 181 | result.stagesUsed = 1; |
| 182 | Parse1Mux(mux, N64Cycle0RGB, result.s1rgb); |
| 183 | if( StagesNeedToUse(mux, N64Cycle1RGB) == 0 ) |
| 184 | { |
| 185 | ByPassGeneralStage(result.s2rgb); |
| 186 | ByPassFinalStage(result.finalrgb); |
| 187 | } |
| 188 | else |
| 189 | { |
| 190 | result.stagesUsed = 2; |
| 191 | Parse1MuxForStage2AndFinalStage(mux, N64Cycle1RGB, result.s2rgb, result.finalrgb); |
| 192 | } |
| 193 | } |
| 194 | else |
| 195 | { |
| 196 | result.stagesUsed = 2; |
| 197 | Parse1Mux2Stages(mux, N64Cycle0RGB, result.s1rgb, result.s2rgb); |
| 198 | Parse1MuxForFinalStage(mux, N64Cycle1RGB, result.finalrgb); |
| 199 | } |
| 200 | |
| 201 | // Debug texel1 |
| 202 | /* |
| 203 | if( m_pDecodedMux->m_bTexel0IsUsed && m_pDecodedMux->m_bTexel1IsUsed ) |
| 204 | { |
| 205 | result.finalrgb.a = MUX_TEXEL0; |
| 206 | result.finalrgb.b = MUX_TEXEL1; |
| 207 | result.finalrgb.c = MUX_0; |
| 208 | result.finalrgb.d = MUX_0; |
| 209 | } |
| 210 | */ |
| 211 | |
| 212 | if( StagesNeedToUse(mux, N64Cycle0Alpha) == 0 ) |
| 213 | { |
| 214 | // Nothing to be done for Alpha |
| 215 | ByPassGeneralStage(result.s1alpha); |
| 216 | ByPassGeneralStage(result.s2alpha); |
| 217 | ByPassFinalStage(result.finalalpha); |
| 218 | } |
| 219 | else if( Parse1Mux2Stages(mux, N64Cycle0Alpha, result.s1alpha, result.s2alpha) == 1 ) |
| 220 | { |
| 221 | // Only 1 NV stage is used |
| 222 | if( result.stagesUsed == 0 ) result.stagesUsed = 1; |
| 223 | if( StagesNeedToUse(mux, N64Cycle1Alpha) == 0 ) |
| 224 | { |
| 225 | ByPassGeneralStage(result.s2alpha); |
| 226 | } |
| 227 | else |
| 228 | { |
| 229 | Parse1Mux(mux, N64Cycle1Alpha, result.s2alpha); |
| 230 | result.stagesUsed = 2; |
| 231 | } |
| 232 | } |
| 233 | else |
| 234 | { |
| 235 | // The 1st is used 2 stages, skip the 2nd N64 alpha setting |
| 236 | result.stagesUsed = 2; |
| 237 | result.s2alpha.a=MUX_COMBINED; |
| 238 | result.s2alpha.b=MUX_1; |
| 239 | result.s2alpha.c=m_pDecodedMux->m_n64Combiners[N64Cycle0Alpha].d; |
| 240 | result.s2alpha.d=MUX_1; |
| 241 | } |
| 242 | |
| 243 | // Parse Alpha setting, alpha does not have a final stage |
| 244 | ByPassFinalStage(result.finalalpha); |
| 245 | ParseDecodedMuxForConstant(result); |
| 246 | } |
| 247 | |
| 248 | void COGLColorCombinerNvidia::ParseDecodedMuxForConstant(NVRegisterCombinerParserType &result) |
| 249 | { |
| 250 | result.constant0 = MUX_0; |
| 251 | result.constant1 = MUX_0; |
| 252 | bool const0Used=false; |
| 253 | bool const1Used=false; |
| 254 | if( m_pDecodedMux->isUsed(MUX_PRIM) ) |
| 255 | { |
| 256 | result.constant0 = MUX_PRIM; |
| 257 | const0Used = true; |
| 258 | } |
| 259 | if( m_pDecodedMux->isUsed(MUX_ENV) ) |
| 260 | { |
| 261 | if( const0Used ) |
| 262 | { |
| 263 | result.constant1 = MUX_ENV; |
| 264 | const1Used = true; |
| 265 | } |
| 266 | else |
| 267 | { |
| 268 | result.constant0 = MUX_ENV; |
| 269 | const0Used = true; |
| 270 | } |
| 271 | } |
| 272 | if( m_pDecodedMux->isUsed(MUX_LODFRAC) && !const1Used ) |
| 273 | { |
| 274 | if( !const1Used ) |
| 275 | { |
| 276 | result.constant1 = MUX_LODFRAC; |
| 277 | const1Used = true; |
| 278 | } |
| 279 | else if( !const0Used ) |
| 280 | { |
| 281 | result.constant0 = MUX_LODFRAC; |
| 282 | const0Used = true; |
| 283 | } |
| 284 | } |
| 285 | |
| 286 | if( m_pDecodedMux->isUsed(MUX_PRIMLODFRAC) && !const1Used ) |
| 287 | { |
| 288 | if( !const1Used ) |
| 289 | { |
| 290 | result.constant1 = MUX_PRIMLODFRAC; |
| 291 | const1Used = true; |
| 292 | } |
| 293 | else if( !const0Used ) |
| 294 | { |
| 295 | result.constant0 = MUX_PRIMLODFRAC; |
| 296 | const0Used = true; |
| 297 | } |
| 298 | } |
| 299 | } |
| 300 | |
| 301 | int COGLColorCombinerNvidia::StagesNeedToUse(COGLDecodedMux &mux, N64StageNumberType stage) |
| 302 | { |
| 303 | N64CombinerType &m = mux.m_n64Combiners[stage]; |
| 304 | |
| 305 | switch(mux.splitType[stage]) |
| 306 | { |
| 307 | case CM_FMT_TYPE_NOT_USED: |
| 308 | return 0; |
| 309 | case CM_FMT_TYPE_D: // = A ==> can be done in 1 NV stage |
| 310 | case CM_FMT_TYPE_A_ADD_D: // = A+D ==> can be done in 1 NV stage |
| 311 | case CM_FMT_TYPE_A_MOD_C: // = A*C ==> can be done in 1 NV stage |
| 312 | case CM_FMT_TYPE_A_SUB_B: // = A-B ==> can be done in 1 NV stage |
| 313 | case CM_FMT_TYPE_A_MOD_C_ADD_D: // = A*C+D ==> can be done in 1 NV stage |
| 314 | case CM_FMT_TYPE_A_LERP_B_C: // = (A-B)*C+B ==> can be done in 1 NV stage |
| 315 | case CM_FMT_TYPE_A_SUB_B_MOD_C: // = (A-B)*C ==> can be done in 1 NV stage |
| 316 | case CM_FMT_TYPE_AB_ADD_CD: // = AB+CD |
| 317 | case CM_FMT_TYPE_AB_SUB_CD: // = AB-CD |
| 318 | return 1; |
| 319 | case CM_FMT_TYPE_A_SUB_B_ADD_D: // = A-B+D ==> can not be done in 1 stage |
| 320 | if( m.a == m.d ) // = 2A-B, simply it to A-B, in fact,we can do 2A-B with NV register combiner |
| 321 | return 1; |
| 322 | else // Need two NV stages for this N64 combiner |
| 323 | return 2; |
| 324 | case CM_FMT_TYPE_A_B_C_D: // = (A-B)*C+D ==> can not be done in 1 stage |
| 325 | default: |
| 326 | //if( m.a == m.d ) // = (A-B)*C+A = A(C+1)-B*C = A-B*C |
| 327 | // return 1; |
| 328 | //else |
| 329 | if( m.d == m.c ) // = (A-B)*C+C = A*C+(1-B)*C |
| 330 | return 1; |
| 331 | else // = (A-B)*C+D, need two NV stages |
| 332 | return 2; |
| 333 | } |
| 334 | } |
| 335 | |
| 336 | bool isTex(uint8 val) |
| 337 | { |
| 338 | if( (val&MUX_MASK) == MUX_TEXEL0 || (val&MUX_MASK) == MUX_TEXEL1 ) |
| 339 | return true; |
| 340 | else |
| 341 | return false; |
| 342 | } |
| 343 | int COGLColorCombinerNvidia::Parse1Mux(COGLDecodedMux &mux, N64StageNumberType stage, NVGeneralCombinerType &res) // Compile the decodedMux into NV register combiner setting |
| 344 | { |
| 345 | // Parse 1 N64 combiner, generate result and return how many NV stage is needed. |
| 346 | // result will be put into only 1 NV stage, not the 2nd one even if 2nd one is needed. |
| 347 | // The caller of this function will handle the 2nd NV stage if it is needed |
| 348 | |
| 349 | |
| 350 | // Up to here, the m_pDecodedMux is already simplied, N64 stage 1 and stage 2 have been |
| 351 | // adjusted so stage1 is almost always complicated than stage 2 |
| 352 | |
| 353 | // The stage type in decodedMux is still in (A-B)*C+D format |
| 354 | // we need to parser and translate it to A*B+C*D format for NV register general combiner |
| 355 | // and to A*D+(1-A)*C+D format for the NV final combiner |
| 356 | |
| 357 | // Remember that N64 has two stages, NV has two general combiner stages and 1 final combiner stage |
| 358 | // NV should be able to simulate exactly all possible N64 combiner settings |
| 359 | /* |
| 360 | CM_FMT_TYPE1_D, // = A ==> can be done in 1 NV stage |
| 361 | CM_FMT_TYPE2_A_ADD_D, // = A+D ==> can be done in 1 NV stage |
| 362 | CM_FMT_TYPE3_A_MOD_C, // = A*C ==> can be done in 1 NV stage |
| 363 | CM_FMT_TYPE4_A_SUB_B, // = A-B ==> can be done in 1 NV stage |
| 364 | CM_FMT_TYPE5_A_MOD_C_ADD_D, // = A*C+D ==> can be done in 1 NV stage |
| 365 | CM_FMT_TYPE6_A_LERP_B_C, // = (A-B)*C+B ==> can be done in 1 NV stage |
| 366 | CM_FMT_TYPE8_A_SUB_B_MOD_C, // = (A-B)*C ==> can be done in 1 NV stage |
| 367 | |
| 368 | CM_FMT_TYPE7_A_SUB_B_ADD_D, // = A-B+C ==> can not be done in 1 stage |
| 369 | CM_FMT_TYPE9_A_B_C_D, // = (A-B)*C+D ==> can not be done in 1 stage |
| 370 | |
| 371 | the last two ones, since we can neither do it in the final stage, if the 1st N64 stage |
| 372 | happen to be one of the two types and have used the two NV general combiners, and if the 2nd N64 |
| 373 | combiner happens to be one of the two types as well, then we have to simplify the N64 combiner so |
| 374 | to implement it. In such as case, the N64 combiners are too complicated, we just do what either as |
| 375 | we can to implement it. |
| 376 | |
| 377 | Use UNSIGNED_INVERT of ZERO ==> ONE |
| 378 | |
| 379 | // If the 1st N64 stage can not be done in 1 NV stage, then we will do 1st N64 stage |
| 380 | // by using 2 NV general combiner stages, and the 2nd N64 stage by using the NV final |
| 381 | // combiner stage. |
| 382 | |
| 383 | // RGB channel and alpha channel is the same in the general combiner, but different in |
| 384 | // the final combiner. In fact, final combiner does not do anything for alpha channel |
| 385 | // so alpha channel setting of both N64 combiner must be implemented by the two NV general |
| 386 | // combiner |
| 387 | |
| 388 | If we can not implement the two alpha setting in 2 NV combiner stages, we will do what either |
| 389 | as we can. |
| 390 | |
| 391 | */ |
| 392 | N64CombinerType &m = mux.m_n64Combiners[stage]; |
| 393 | |
| 394 | switch(mux.splitType[stage]) |
| 395 | { |
| 396 | case CM_FMT_TYPE_NOT_USED: |
| 397 | res.a=MUX_0; |
| 398 | res.b=MUX_0; |
| 399 | res.c=MUX_0; |
| 400 | res.d=MUX_0; |
| 401 | return 0; |
| 402 | break; |
| 403 | case CM_FMT_TYPE_D: // = A ==> can be done in 1 NV stage |
| 404 | res.a=m.d; |
| 405 | res.b=MUX_1; |
| 406 | res.c=MUX_0; |
| 407 | res.d=MUX_0; |
| 408 | return 1; |
| 409 | break; |
| 410 | case CM_FMT_TYPE_A_ADD_D: // = A+D ==> can be done in 1 NV stage |
| 411 | res.a=m.a; |
| 412 | res.b=MUX_1; |
| 413 | res.c=m.d; |
| 414 | res.d=MUX_1; |
| 415 | return 1; |
| 416 | break; |
| 417 | case CM_FMT_TYPE_A_MOD_C: // = A*C ==> can be done in 1 NV stage |
| 418 | res.a=m.a; |
| 419 | res.b=m.c; |
| 420 | res.c=MUX_0; |
| 421 | res.d=MUX_0; |
| 422 | return 1; |
| 423 | break; |
| 424 | case CM_FMT_TYPE_A_SUB_B: // = A-B ==> can be done in 1 NV stage |
| 425 | res.a=m.a; |
| 426 | res.b=MUX_1; |
| 427 | res.c=m.b|MUX_NEG; |
| 428 | res.d=MUX_1; |
| 429 | return 1; |
| 430 | break; |
| 431 | case CM_FMT_TYPE_A_MOD_C_ADD_D: // = A*C+D ==> can be done in 1 NV stage |
| 432 | res.a=m.a; |
| 433 | res.b=m.c; |
| 434 | res.c=m.d; |
| 435 | res.d=MUX_1; |
| 436 | return 1; |
| 437 | break; |
| 438 | case CM_FMT_TYPE_A_LERP_B_C: // = (A-B)*C+B ==> can be done in 1 NV stage |
| 439 | // = AC+(1-C)B |
| 440 | res.a=m.a; |
| 441 | res.b=m.c; |
| 442 | res.c=m.c^MUX_COMPLEMENT; |
| 443 | res.d=m.b; |
| 444 | return 1; |
| 445 | break; |
| 446 | case CM_FMT_TYPE_A_SUB_B_MOD_C: // = (A-B)*C ==> can be done in 1 NV stage |
| 447 | res.a=m.a; |
| 448 | res.b=m.c; |
| 449 | res.c=m.b|MUX_NEG; |
| 450 | res.d=m.c; |
| 451 | return 1; |
| 452 | break; |
| 453 | case CM_FMT_TYPE_AB_ADD_CD: // = AB+CD |
| 454 | res.a = m.a; |
| 455 | res.b = m.b; |
| 456 | res.c = m.c; |
| 457 | res.d = m.d; |
| 458 | return 1; |
| 459 | break; |
| 460 | case CM_FMT_TYPE_AB_SUB_CD: // = AB-CD |
| 461 | res.a = m.a; |
| 462 | res.b = m.b; |
| 463 | res.c = m.c|MUX_NEG; |
| 464 | res.d = m.d; |
| 465 | return 1; |
| 466 | break; |
| 467 | case CM_FMT_TYPE_A_SUB_B_ADD_D: // = A-B+D ==> can not be done in 1 stage |
| 468 | if( m.a == m.d ) // = 2A-B, simply it to A-B, in fact,we can do 2A-B with NV register combiner |
| 469 | { |
| 470 | res.a=m.a; |
| 471 | res.b=MUX_1; |
| 472 | res.c=m.b|MUX_NEG; |
| 473 | res.d=MUX_1; |
| 474 | return 1; |
| 475 | } |
| 476 | else // Need two NV stages for this N64 combiner |
| 477 | { |
| 478 | // Stage 1: R1=A-B |
| 479 | res.a=m.a; |
| 480 | res.b=MUX_1; |
| 481 | |
| 482 | if( isTex(res.b) || !isTex(res.d) ) |
| 483 | { |
| 484 | res.c=m.b|MUX_NEG; |
| 485 | res.d=MUX_1; |
| 486 | } |
| 487 | else |
| 488 | { |
| 489 | res.c=m.d; |
| 490 | res.d=MUX_1; |
| 491 | } |
| 492 | return 2; |
| 493 | } |
| 494 | break; |
| 495 | case CM_FMT_TYPE_A_B_C_D: // = (A-B)*C+D ==> can not be done in 1 stage |
| 496 | default: |
| 497 | if( m.a == m.d ) // = (A-B)*C+A = A(C+1)-B*C = A-B*C |
| 498 | { |
| 499 | res.a=m.a; |
| 500 | res.b=m.c; |
| 501 | res.c=m.b|MUX_NEG; |
| 502 | res.d=m.c; |
| 503 | return 1; |
| 504 | } |
| 505 | else if( m.d == m.c ) // = (A-B)*C+C = A*C+(1-B)*C |
| 506 | { |
| 507 | res.a=m.a; |
| 508 | res.b=m.c; |
| 509 | res.c=m.b^MUX_COMPLEMENT; |
| 510 | res.d=m.c; |
| 511 | return 1; |
| 512 | } |
| 513 | else // = (A-B)*C+D, need two NV stages |
| 514 | { |
| 515 | // Stage 1: R1=(A-B)*C = AC-BC |
| 516 | if( isTex(m.d) ) |
| 517 | { |
| 518 | // = A*C+D |
| 519 | res.a=m.a; |
| 520 | res.b=m.c; |
| 521 | res.c=m.d; |
| 522 | res.d=MUX_1; |
| 523 | } |
| 524 | else |
| 525 | { |
| 526 | // = (A-B)*C = A*C - B*C |
| 527 | res.a=m.a; |
| 528 | res.b=m.c; |
| 529 | res.c=m.b|MUX_NEG; |
| 530 | res.d=m.c; |
| 531 | } |
| 532 | return 2; |
| 533 | } |
| 534 | break; |
| 535 | } |
| 536 | } |
| 537 | |
| 538 | int COGLColorCombinerNvidia::Parse1Mux2Stages(COGLDecodedMux &mux, N64StageNumberType stage, NVGeneralCombinerType &res, NVGeneralCombinerType &res2) |
| 539 | { |
| 540 | N64CombinerType &m = mux.m_n64Combiners[stage]; |
| 541 | switch(mux.splitType[stage]) |
| 542 | { |
| 543 | case CM_FMT_TYPE_A_SUB_B_ADD_D: // = A-B+D ==> can not be done in 1 stage |
| 544 | if( m.a != m.d ) // = 2A-B, simply it to A-B, in fact,we can do 2A-B with NV register combiner |
| 545 | { |
| 546 | // Stage 1: R1=A-B |
| 547 | res.a=m.a; |
| 548 | res.b=MUX_1; |
| 549 | res.c=m.b|MUX_NEG; |
| 550 | res.d=MUX_1; |
| 551 | |
| 552 | res2.a=MUX_COMBINED_SIGNED; |
| 553 | res2.b=MUX_1; |
| 554 | res2.c=m.d; |
| 555 | res2.d=MUX_1; |
| 556 | |
| 557 | return 2; |
| 558 | } |
| 559 | break; |
| 560 | case CM_FMT_TYPE_A_B_C_D: // = (A-B)*C+D ==> can not be done in 1 stage |
| 561 | case CM_FMT_TYPE_A_B_C_A: // = (A-B)*C+D ==> can not be done in 1 stage |
| 562 | //if( m.a != m.d && m.d != m.c ) |
| 563 | { |
| 564 | // Stage 1: R1=(A-B)*C = AC-BC |
| 565 | res.a=m.a; |
| 566 | res.b=m.c; |
| 567 | res.c=m.b|MUX_NEG; |
| 568 | res.d=m.c; |
| 569 | |
| 570 | res2.a=MUX_COMBINED_SIGNED; |
| 571 | res2.b=MUX_1; |
| 572 | res2.c=m.d; |
| 573 | res2.d=MUX_1; |
| 574 | |
| 575 | return 2; |
| 576 | } |
| 577 | break; |
| 578 | default: |
| 579 | break; |
| 580 | } |
| 581 | return Parse1Mux(mux, stage, res); |
| 582 | } |
| 583 | |
| 584 | |
| 585 | void COGLColorCombinerNvidia::Parse1MuxForFinalStage(COGLDecodedMux &mux, N64StageNumberType stage, NVFinalCombinerType &res) |
| 586 | { |
| 587 | N64CombinerType &m = mux.m_n64Combiners[stage]; |
| 588 | |
| 589 | // Final stage equation is: AB+(1-A)C+D |
| 590 | switch(mux.splitType[stage]) |
| 591 | { |
| 592 | case CM_FMT_TYPE_NOT_USED: |
| 593 | res.a=MUX_0; |
| 594 | res.b=MUX_0; |
| 595 | res.c=MUX_0; |
| 596 | res.d=MUX_0; |
| 597 | break; |
| 598 | case CM_FMT_TYPE_D: // = A ==> can be done in 1 NV stage |
| 599 | res.a=m.a; |
| 600 | res.b=MUX_1; |
| 601 | res.c=MUX_0; |
| 602 | res.d=MUX_0; |
| 603 | break; |
| 604 | case CM_FMT_TYPE_A_ADD_D: // = A+D ==> can be done in 1 NV stage |
| 605 | res.a=m.a; |
| 606 | res.b=MUX_1; |
| 607 | res.c=MUX_0; |
| 608 | res.d=m.d; |
| 609 | break; |
| 610 | case CM_FMT_TYPE_A_MOD_C: // = A*C ==> can be done in 1 NV stage |
| 611 | res.a=m.a; |
| 612 | res.b=m.c; |
| 613 | res.c=MUX_0; |
| 614 | res.d=MUX_0; |
| 615 | break; |
| 616 | case CM_FMT_TYPE_A_SUB_B: // = A-B ==> can be done in 1 NV stage |
| 617 | res.a=m.a; |
| 618 | res.b=MUX_1; |
| 619 | res.c=MUX_0; |
| 620 | res.d=m.b|MUX_NEG; |
| 621 | break; |
| 622 | case CM_FMT_TYPE_A_MOD_C_ADD_D: // = A*C+D ==> can be done in 1 NV stage |
| 623 | res.a=m.a; |
| 624 | res.b=m.c; |
| 625 | res.c=MUX_0; |
| 626 | res.d=m.d; |
| 627 | break; |
| 628 | case CM_FMT_TYPE_A_LERP_B_C: // = (A-B)*C+B ==> can be done in 1 NV stage |
| 629 | // = AC+(1-B)C |
| 630 | res.a = m.c; |
| 631 | res.b = MUX_0; |
| 632 | res.c = m.b; |
| 633 | res.d = MUX_E_F; |
| 634 | res.e = m.a; |
| 635 | res.f = m.c; |
| 636 | break; |
| 637 | case CM_FMT_TYPE_A_SUB_B_MOD_C: // = (A-B)*C ==> can be done in 1 NV stage |
| 638 | res.a=m.c; |
| 639 | res.b=m.a; |
| 640 | res.c=m.b; |
| 641 | res.d=m.b|MUX_NEG; |
| 642 | break; |
| 643 | case CM_FMT_TYPE_AB_ADD_CD: // = AB+CD |
| 644 | res.a = m.a; |
| 645 | res.b = m.b; |
| 646 | res.e = m.c; |
| 647 | res.f = m.d; |
| 648 | res.c = MUX_0; |
| 649 | res.d = MUX_E_F; |
| 650 | break; |
| 651 | case CM_FMT_TYPE_AB_SUB_CD: // = AB-CD |
| 652 | res.a = m.a; |
| 653 | res.b = m.b; |
| 654 | res.e = m.c|MUX_NEG; |
| 655 | res.f = m.d; |
| 656 | res.c = MUX_0; |
| 657 | res.d = MUX_E_F; |
| 658 | break; |
| 659 | case CM_FMT_TYPE_A_SUB_B_ADD_D: // = A-B+D ==> can not be done in 1 stage |
| 660 | if( m.a == m.d ) // = 2A-B, simply it to A-B, in fact,we can do 2A-B with NV register combiner |
| 661 | { |
| 662 | res.a=m.a; |
| 663 | res.b=MUX_1; |
| 664 | res.c=MUX_0; |
| 665 | res.d=m.b|MUX_NEG; |
| 666 | } |
| 667 | else // Need two NV stages for this N64 combiner |
| 668 | { |
| 669 | TRACE0("NV Combiner parse, check me, not fully support this combiner"); |
| 670 | // final combiner can not fully support this combiner setting |
| 671 | // Stage 1: R1=A-B |
| 672 | res.a=m.a; |
| 673 | res.b=MUX_1; |
| 674 | res.c=MUX_0; |
| 675 | res.d=m.b|MUX_NEG; |
| 676 | } |
| 677 | break; |
| 678 | case CM_FMT_TYPE_A_B_C_D: // = (A-B)*C+D ==> can not be done in 1 stage |
| 679 | default: |
| 680 | if( m.a == m.d ) // = (A-B)*C+A = A(C+1)-B*C = A-B*C |
| 681 | { |
| 682 | /* |
| 683 | res.a=m.c; |
| 684 | res.b=m.b|MUX_NEG; |
| 685 | res.c=MUX_0; |
| 686 | res.d=m.a; |
| 687 | */ |
| 688 | res.a=m.c; |
| 689 | res.b=m.a; |
| 690 | res.c=m.b; |
| 691 | res.d=MUX_0; |
| 692 | } |
| 693 | else if( m.d == m.c ) // = (A-B)*C+C = A*C+(1-B)*C |
| 694 | { |
| 695 | res.a=m.b; |
| 696 | res.b=MUX_0; |
| 697 | res.c=m.c; |
| 698 | res.d=MUX_E_F; |
| 699 | res.e=m.a; |
| 700 | res.f=m.c; |
| 701 | } |
| 702 | else // = (A-B)*C+D, need two NV stages |
| 703 | { |
| 704 | TRACE0("NV Combiner parse, check me, not fully support this combiner"); |
| 705 | // final combiner can not fully support this combiner setting |
| 706 | // Stage 1: R1=(A-B)*C = AC-BC |
| 707 | res.a=m.c; |
| 708 | res.b=m.a; |
| 709 | res.c=m.b; |
| 710 | res.d=m.b|MUX_NEG; |
| 711 | } |
| 712 | break; |
| 713 | } |
| 714 | res.g=MUX_COMBINED; |
| 715 | } |
| 716 | |
| 717 | int COGLColorCombinerNvidia::Parse1MuxForStage2AndFinalStage(COGLDecodedMux &mux, N64StageNumberType stage, NVGeneralCombinerType &res, NVFinalCombinerType &fres) |
| 718 | { |
| 719 | if( Parse1Mux(mux, stage, res) == 1 ) |
| 720 | { |
| 721 | ByPassFinalStage(fres); |
| 722 | return 1; |
| 723 | } |
| 724 | else |
| 725 | { |
| 726 | ByPassFinalStage(fres); |
| 727 | fres.a=MUX_COMBINED; |
| 728 | fres.b=MUX_1; |
| 729 | fres.d = mux.m_n64Combiners[stage].d; |
| 730 | fres.g=MUX_COMBINED; |
| 731 | return 2; |
| 732 | } |
| 733 | } |
| 734 | |
| 735 | void COGLColorCombinerNvidia::ByPassFinalStage(NVFinalCombinerType &fres) |
| 736 | { |
| 737 | fres.a=MUX_0; |
| 738 | fres.b=MUX_0; |
| 739 | fres.c=MUX_0; |
| 740 | fres.d=MUX_COMBINED; |
| 741 | fres.e=MUX_0; |
| 742 | fres.f=MUX_0; |
| 743 | fres.g=MUX_COMBINED; |
| 744 | } |
| 745 | |
| 746 | void COGLColorCombinerNvidia::ByPassGeneralStage(NVGeneralCombinerType &res) |
| 747 | { |
| 748 | res.a=MUX_1; |
| 749 | res.b=MUX_COMBINED; |
| 750 | res.c=MUX_0; |
| 751 | res.d=MUX_0; |
| 752 | } |
| 753 | |
| 754 | int COGLColorCombinerNvidia::FindCompiledMux(void) |
| 755 | { |
| 756 | for( uint32 i=0; i<m_vCompiledSettings.size(); i++ ) |
| 757 | { |
| 758 | if( m_vCompiledSettings[i].dwMux0 == m_pDecodedMux->m_dwMux0 && m_vCompiledSettings[i].dwMux1 == m_pDecodedMux->m_dwMux1 ) |
| 759 | return i; |
| 760 | } |
| 761 | |
| 762 | return -1; |
| 763 | } |
| 764 | void COGLColorCombinerNvidia::GenerateNVRegisterCombinerSettingConstants(int index) |
| 765 | { |
| 766 | NVRegisterCombinerSettingType &info = m_vCompiledSettings[index]; |
| 767 | uint8 consts[2] = {info.constant0,info.constant1}; |
| 768 | |
| 769 | float *pf; |
| 770 | |
| 771 | for( int i=0; i<2; i++ ) |
| 772 | { |
| 773 | switch( consts[i] ) |
| 774 | { |
| 775 | case MUX_PRIM: |
| 776 | pf = GetPrimitiveColorfv(); |
| 777 | pglCombinerParameterfvNV(GL_CONSTANT_COLOR0_NV+i,pf); |
| 778 | break; |
| 779 | case MUX_ENV: |
| 780 | pf = GetEnvColorfv(); |
| 781 | pglCombinerParameterfvNV(GL_CONSTANT_COLOR0_NV+i,pf); |
| 782 | break; |
| 783 | case MUX_LODFRAC: |
| 784 | case MUX_PRIMLODFRAC: |
| 785 | { |
| 786 | float frac = gRDP.primLODFrac / 255.0f; |
| 787 | float tempf[4] = {frac,frac,frac,frac}; |
| 788 | pglCombinerParameterfvNV(GL_CONSTANT_COLOR0_NV+i,tempf); |
| 789 | break; |
| 790 | } |
| 791 | } |
| 792 | } |
| 793 | } |
| 794 | |
| 795 | void COGLColorCombinerNvidia::GenerateNVRegisterCombinerSetting(int index) |
| 796 | { |
| 797 | if( index < 0 || index >= (int)m_vCompiledSettings.size() ) |
| 798 | { |
| 799 | TRACE0("NV Register combiner, vector index out of range"); |
| 800 | return; |
| 801 | } |
| 802 | |
| 803 | NVRegisterCombinerSettingType &info = m_vCompiledSettings[index]; |
| 804 | |
| 805 | pglCombinerParameteriNV(GL_NUM_GENERAL_COMBINERS_NV,info.numOfStages); |
| 806 | |
| 807 | uint32 i; |
| 808 | |
| 809 | if( info.numOfStages > 0 ) |
| 810 | { |
| 811 | for( i=0; i<4; i++ ) |
| 812 | { |
| 813 | pglCombinerInputNV(GL_COMBINER0_NV, GL_RGB, info.stage1RGB[i].variable, info.stage1RGB[i].input, |
| 814 | info.stage1RGB[i].mapping, info.stage1RGB[i].componentUsage ); |
| 815 | } |
| 816 | |
| 817 | for( i=0; i<4; i++ ) |
| 818 | { |
| 819 | pglCombinerInputNV(GL_COMBINER0_NV, GL_ALPHA, info.stage1Alpha[i].variable, info.stage1Alpha[i].input, |
| 820 | info.stage1Alpha[i].mapping, info.stage1Alpha[i].componentUsage ); |
| 821 | } |
| 822 | |
| 823 | pglCombinerOutputNV(GL_COMBINER0_NV, GL_RGB, info.stage1outputRGB.abOutput, info.stage1outputRGB.cdOutput, |
| 824 | info.stage1outputRGB.sumOutput, info.stage1outputRGB.scale, info.stage1outputRGB.bias, info.stage1outputRGB.abDotProduct, |
| 825 | info.stage1outputRGB.cdDotProduct, info.stage1outputRGB.muxSum); |
| 826 | |
| 827 | pglCombinerOutputNV(GL_COMBINER0_NV, GL_ALPHA, info.stage2outputAlpha.abOutput, info.stage2outputAlpha.cdOutput, |
| 828 | info.stage2outputAlpha.sumOutput, info.stage2outputAlpha.scale, info.stage2outputAlpha.bias, info.stage2outputAlpha.abDotProduct, |
| 829 | info.stage2outputAlpha.cdDotProduct, info.stage2outputAlpha.muxSum); |
| 830 | |
| 831 | if( info.numOfStages > 1 ) |
| 832 | { |
| 833 | for( i=0; i<4; i++ ) |
| 834 | { |
| 835 | pglCombinerInputNV(GL_COMBINER1_NV, GL_RGB, info.stage2RGB[i].variable, |
| 836 | info.stage2RGB[i].input, info.stage2RGB[i].mapping, info.stage2RGB[i].componentUsage ); |
| 837 | } |
| 838 | |
| 839 | for( i=0; i<4; i++ ) |
| 840 | { |
| 841 | pglCombinerInputNV(GL_COMBINER1_NV, GL_ALPHA, info.stage2Alpha[i].variable, info.stage2Alpha[i].input, |
| 842 | info.stage2Alpha[i].mapping, info.stage2Alpha[i].componentUsage ); |
| 843 | } |
| 844 | |
| 845 | pglCombinerOutputNV(GL_COMBINER1_NV, GL_RGB, info.stage2outputRGB.abOutput, info.stage2outputRGB.cdOutput, |
| 846 | info.stage2outputRGB.sumOutput, info.stage2outputRGB.scale, info.stage2outputRGB.bias, info.stage2outputRGB.abDotProduct, |
| 847 | info.stage2outputRGB.cdDotProduct, info.stage2outputRGB.muxSum); |
| 848 | |
| 849 | pglCombinerOutputNV(GL_COMBINER1_NV, GL_ALPHA, info.stage2outputAlpha.abOutput, info.stage2outputAlpha.cdOutput, |
| 850 | info.stage2outputAlpha.sumOutput, info.stage2outputAlpha.scale, info.stage2outputAlpha.bias, info.stage2outputAlpha.abDotProduct, |
| 851 | info.stage2outputAlpha.cdDotProduct, info.stage2outputAlpha.muxSum); |
| 852 | } |
| 853 | } |
| 854 | |
| 855 | for( i=0; i<7; i++ ) |
| 856 | { |
| 857 | pglFinalCombinerInputNV(info.finalStage[i].variable, info.finalStage[i].input, |
| 858 | info.finalStage[i].mapping, info.finalStage[i].componentUsage ); |
| 859 | } |
| 860 | } |
| 861 | |
| 862 | GLenum COGLColorCombinerNvidia::ConstMap(uint8 c) |
| 863 | { |
| 864 | switch(c) |
| 865 | { |
| 866 | case MUX_0: |
| 867 | return GL_ZERO; |
| 868 | case MUX_1: |
| 869 | return GL_ZERO; |
| 870 | case MUX_COMBINED: |
| 871 | case MUX_TEXEL0: |
| 872 | case MUX_TEXEL1: |
| 873 | case MUX_PRIM: |
| 874 | case MUX_SHADE: |
| 875 | case MUX_ENV: |
| 876 | case MUX_COMBALPHA: |
| 877 | case MUX_T0_ALPHA: |
| 878 | case MUX_T1_ALPHA: |
| 879 | case MUX_PRIM_ALPHA: |
| 880 | case MUX_SHADE_ALPHA: |
| 881 | case MUX_ENV_ALPHA: |
| 882 | case MUX_LODFRAC: |
| 883 | case MUX_PRIMLODFRAC: |
| 884 | break; |
| 885 | } |
| 886 | return GL_ZERO; |
| 887 | |
| 888 | } |
| 889 | |
| 890 | void Set1Variable(GLenum variable, uint8 val, NVCombinerInputType &record, const NVRegisterCombinerParserType &result, bool forRGB=true) |
| 891 | { |
| 892 | record.variable = variable; |
| 893 | record.componentUsage = RGBmap1[val&MUX_MASK].componentUsage; |
| 894 | record.input = RGBmap1[val&MUX_MASK].input; |
| 895 | record.mapping = RGBmap1[val&MUX_MASK].mapping; |
| 896 | |
| 897 | switch( val&MUX_MASK ) |
| 898 | { |
| 899 | case MUX_PRIM: |
| 900 | case MUX_ENV: |
| 901 | case MUX_PRIMLODFRAC: |
| 902 | case MUX_LODFRAC: |
| 903 | if( (val&MUX_MASK) == result.constant0 ) |
| 904 | { |
| 905 | record.input = GL_CONSTANT_COLOR0_NV; |
| 906 | } |
| 907 | else if( (val&MUX_MASK) == result.constant1 ) |
| 908 | { |
| 909 | record.input = GL_CONSTANT_COLOR1_NV; |
| 910 | } |
| 911 | else |
| 912 | { |
| 913 | record.input = GL_ZERO; |
| 914 | } |
| 915 | break; |
| 916 | } |
| 917 | |
| 918 | if( val&MUX_NEG ) |
| 919 | { |
| 920 | record.mapping = GL_SIGNED_NEGATE_NV; |
| 921 | } |
| 922 | else if( val == MUX_1 ) |
| 923 | { |
| 924 | record.mapping = GL_UNSIGNED_INVERT_NV; |
| 925 | } |
| 926 | else if( val & MUX_COMPLEMENT ) |
| 927 | { |
| 928 | record.mapping = GL_UNSIGNED_INVERT_NV; |
| 929 | } |
| 930 | |
| 931 | if( val & MUX_ALPHAREPLICATE || !forRGB ) |
| 932 | { |
| 933 | record.componentUsage = GL_ALPHA; |
| 934 | } |
| 935 | } |
| 936 | |
| 937 | int COGLColorCombinerNvidia::SaveParserResult(const NVRegisterCombinerParserType &result) |
| 938 | { |
| 939 | NVRegisterCombinerSettingType save; |
| 940 | |
| 941 | // Stage 1 RGB |
| 942 | Set1Variable(GL_VARIABLE_A_NV, result.s1rgb.a, save.stage1RGB[0], result); |
| 943 | Set1Variable(GL_VARIABLE_B_NV, result.s1rgb.b, save.stage1RGB[1], result); |
| 944 | Set1Variable(GL_VARIABLE_C_NV, result.s1rgb.c, save.stage1RGB[2], result); |
| 945 | Set1Variable(GL_VARIABLE_D_NV, result.s1rgb.d, save.stage1RGB[3], result); |
| 946 | |
| 947 | // Stage 1 Alpha |
| 948 | Set1Variable(GL_VARIABLE_A_NV, result.s1alpha.a, save.stage1Alpha[0], result, false); |
| 949 | Set1Variable(GL_VARIABLE_B_NV, result.s1alpha.b, save.stage1Alpha[1], result, false); |
| 950 | Set1Variable(GL_VARIABLE_C_NV, result.s1alpha.c, save.stage1Alpha[2], result, false); |
| 951 | Set1Variable(GL_VARIABLE_D_NV, result.s1alpha.d, save.stage1Alpha[3], result, false); |
| 952 | |
| 953 | // Stage 2 RGB |
| 954 | Set1Variable(GL_VARIABLE_A_NV, result.s2rgb.a, save.stage2RGB[0], result); |
| 955 | Set1Variable(GL_VARIABLE_B_NV, result.s2rgb.b, save.stage2RGB[1], result); |
| 956 | Set1Variable(GL_VARIABLE_C_NV, result.s2rgb.c, save.stage2RGB[2], result); |
| 957 | Set1Variable(GL_VARIABLE_D_NV, result.s2rgb.d, save.stage2RGB[3], result); |
| 958 | |
| 959 | // Stage 2 Alpha |
| 960 | Set1Variable(GL_VARIABLE_A_NV, result.s2alpha.a, save.stage2Alpha[0], result, false); |
| 961 | Set1Variable(GL_VARIABLE_B_NV, result.s2alpha.b, save.stage2Alpha[1], result, false); |
| 962 | Set1Variable(GL_VARIABLE_C_NV, result.s2alpha.c, save.stage2Alpha[2], result, false); |
| 963 | Set1Variable(GL_VARIABLE_D_NV, result.s2alpha.d, save.stage2Alpha[3], result, false); |
| 964 | |
| 965 | // Final Stage RGB |
| 966 | Set1Variable(GL_VARIABLE_A_NV, result.finalrgb.a, save.finalStage[0], result); |
| 967 | Set1Variable(GL_VARIABLE_B_NV, result.finalrgb.b, save.finalStage[1], result); |
| 968 | Set1Variable(GL_VARIABLE_C_NV, result.finalrgb.c, save.finalStage[2], result); |
| 969 | Set1Variable(GL_VARIABLE_D_NV, result.finalrgb.d, save.finalStage[3], result); |
| 970 | Set1Variable(GL_VARIABLE_E_NV, result.finalrgb.e, save.finalStage[4], result); |
| 971 | //save.finalStage[4].componentUsage = GL_ALPHA; |
| 972 | Set1Variable(GL_VARIABLE_F_NV, result.finalrgb.f, save.finalStage[5], result); |
| 973 | //save.finalStage[5].componentUsage = GL_ALPHA; |
| 974 | Set1Variable(GL_VARIABLE_G_NV, result.finalrgb.g, save.finalStage[6], result); |
| 975 | save.finalStage[6].componentUsage = GL_ALPHA; |
| 976 | |
| 977 | save.numOfStages = result.stagesUsed; |
| 978 | save.dwMux0 = m_pDecodedMux->m_dwMux0; |
| 979 | save.dwMux1 = m_pDecodedMux->m_dwMux1; |
| 980 | |
| 981 | save.stage1outputRGB.scale = GL_NONE; |
| 982 | save.stage1outputRGB.sumOutput = GL_SPARE0_NV; |
| 983 | save.stage1outputRGB.abDotProduct = GL_FALSE; |
| 984 | save.stage1outputRGB.cdDotProduct = GL_FALSE; |
| 985 | save.stage1outputRGB.abOutput = GL_SPARE1_NV; |
| 986 | save.stage1outputRGB.cdOutput = GL_SECONDARY_COLOR_NV; |
| 987 | save.stage1outputRGB.bias = GL_NONE; |
| 988 | save.stage1outputRGB.muxSum = GL_FALSE; |
| 989 | |
| 990 | save.stage1outputAlpha.scale = GL_NONE; |
| 991 | save.stage1outputAlpha.sumOutput = GL_SPARE0_NV; |
| 992 | save.stage1outputAlpha.abDotProduct = GL_FALSE; |
| 993 | save.stage1outputAlpha.cdDotProduct = GL_FALSE; |
| 994 | save.stage1outputAlpha.abOutput = GL_SPARE1_NV; |
| 995 | save.stage1outputAlpha.cdOutput = GL_SECONDARY_COLOR_NV; |
| 996 | save.stage1outputAlpha.bias = GL_NONE; |
| 997 | save.stage1outputAlpha.muxSum = GL_FALSE; |
| 998 | |
| 999 | save.stage2outputRGB.scale = GL_NONE; |
| 1000 | save.stage2outputRGB.sumOutput = GL_SPARE0_NV; |
| 1001 | save.stage2outputRGB.abDotProduct = GL_FALSE; |
| 1002 | save.stage2outputRGB.cdDotProduct = GL_FALSE; |
| 1003 | save.stage2outputRGB.abOutput = GL_SPARE1_NV; |
| 1004 | save.stage2outputRGB.cdOutput = GL_SECONDARY_COLOR_NV; |
| 1005 | save.stage2outputRGB.bias = GL_NONE; |
| 1006 | save.stage2outputRGB.muxSum = GL_FALSE; |
| 1007 | |
| 1008 | save.stage2outputAlpha.scale = GL_NONE; |
| 1009 | save.stage2outputAlpha.sumOutput = GL_SPARE0_NV; |
| 1010 | save.stage2outputAlpha.abDotProduct = GL_FALSE; |
| 1011 | save.stage2outputAlpha.cdDotProduct = GL_FALSE; |
| 1012 | save.stage2outputAlpha.abOutput = GL_SPARE1_NV; |
| 1013 | save.stage2outputAlpha.cdOutput = GL_SECONDARY_COLOR_NV; |
| 1014 | save.stage2outputAlpha.bias = GL_NONE; |
| 1015 | save.stage2outputAlpha.muxSum = GL_FALSE; |
| 1016 | |
| 1017 | save.constant0 = result.constant0; |
| 1018 | save.constant1 = result.constant1; |
| 1019 | |
| 1020 | #ifdef DEBUGGER |
| 1021 | memcpy(&(save.parseResult),&result, sizeof(result)); |
| 1022 | if( logCombiners ) |
| 1023 | { |
| 1024 | TRACE0("\nNew Mux:\n"); |
| 1025 | DisplayMuxString(); |
| 1026 | COGLColorCombiner::DisplaySimpleMuxString(); |
| 1027 | DisplayNVCombinerString(save); |
| 1028 | } |
| 1029 | #endif |
| 1030 | |
| 1031 | m_vCompiledSettings.push_back(save); |
| 1032 | |
| 1033 | return m_vCompiledSettings.size()-1; // Return the index of the last element |
| 1034 | } |
| 1035 | |
| 1036 | |
| 1037 | void COGLColorCombinerNvidia::DisableCombiner(void) |
| 1038 | { |
| 1039 | glDisable(GL_REGISTER_COMBINERS_NV); |
| 1040 | COGLColorCombiner4::DisableCombiner(); |
| 1041 | } |
| 1042 | |
| 1043 | void COGLColorCombinerNvidia::InitCombinerCycleCopy(void) |
| 1044 | { |
| 1045 | glDisable(GL_REGISTER_COMBINERS_NV); |
| 1046 | COGLColorCombiner4::InitCombinerCycleCopy(); |
| 1047 | } |
| 1048 | |
| 1049 | void COGLColorCombinerNvidia::InitCombinerCycleFill(void) |
| 1050 | { |
| 1051 | glDisable(GL_REGISTER_COMBINERS_NV); |
| 1052 | COGLColorCombiner4::InitCombinerCycleFill(); |
| 1053 | } |
| 1054 | |
| 1055 | void COGLColorCombinerNvidia::InitCombinerBlenderForSimpleTextureDraw(uint32 tile) |
| 1056 | { |
| 1057 | glDisable(GL_REGISTER_COMBINERS_NV); |
| 1058 | COGLColorCombiner::InitCombinerBlenderForSimpleTextureDraw(tile); |
| 1059 | } |
| 1060 | |
| 1061 | void COGLColorCombinerNvidia::ApplyFogAtFinalStage() |
| 1062 | { |
| 1063 | // If we need to enable fog at final stage, the current flag stage setting |
| 1064 | // will be affect, which means correct combiner setting at final stage is lost |
| 1065 | // in order to use fog |
| 1066 | if( glIsEnabled(GL_FOG) ) |
| 1067 | { |
| 1068 | // Use final stage as: cmb*fogfactor+fog*(1-fogfactor) |
| 1069 | pglFinalCombinerInputNV(GL_VARIABLE_A_NV, GL_FOG, GL_UNSIGNED_IDENTITY_NV, GL_ALPHA ); |
| 1070 | pglFinalCombinerInputNV(GL_VARIABLE_B_NV, GL_SPARE0_NV, GL_UNSIGNED_IDENTITY_NV, GL_RGB ); |
| 1071 | pglFinalCombinerInputNV(GL_VARIABLE_C_NV, GL_FOG, GL_UNSIGNED_IDENTITY_NV, GL_RGB ); |
| 1072 | pglFinalCombinerInputNV(GL_VARIABLE_D_NV, GL_ZERO, GL_UNSIGNED_IDENTITY_NV, GL_RGB ); |
| 1073 | } |
| 1074 | } |
| 1075 | |
| 1076 | #ifdef DEBUGGER |
| 1077 | extern const char *translatedCombTypes[]; |
| 1078 | void COGLColorCombinerNvidia::DisplaySimpleMuxString(void) |
| 1079 | { |
| 1080 | COGLColorCombiner::DisplaySimpleMuxString(); |
| 1081 | TRACE0("\nNV Combiner setting\n"); |
| 1082 | uint32 index = FindCompiledMux(); |
| 1083 | if( index >= 0 ) |
| 1084 | { |
| 1085 | NVRegisterCombinerSettingType &record = m_vCompiledSettings[index]; |
| 1086 | DisplayNVCombinerString(record); |
| 1087 | } |
| 1088 | } |
| 1089 | |
| 1090 | char* FormatStrForFinalStage(uint8 val, char* buf) |
| 1091 | { |
| 1092 | if( (val&MUX_MASK) == MUX_E_F ) |
| 1093 | { |
| 1094 | strcpy(buf, "E_F"); |
| 1095 | return buf; |
| 1096 | } |
| 1097 | else |
| 1098 | return DecodedMux::FormatStr(val, buf); |
| 1099 | } |
| 1100 | |
| 1101 | void COGLColorCombinerNvidia::DisplayNVCombinerString(NVRegisterCombinerSettingType &record) |
| 1102 | { |
| 1103 | NVRegisterCombinerParserType &result = record.parseResult; |
| 1104 | |
| 1105 | char buf[2000]; |
| 1106 | char buf0[30]; |
| 1107 | char buf1[30]; |
| 1108 | char buf2[30]; |
| 1109 | char buf3[30]; |
| 1110 | char buf4[30]; |
| 1111 | char buf5[30]; |
| 1112 | char buf6[30]; |
| 1113 | buf[0]='\0'; |
| 1114 | |
| 1115 | TRACE0("\n\n"); |
| 1116 | TRACE0("\nNvidia combiner stages:\n"); |
| 1117 | |
| 1118 | DebuggerAppendMsg("//aRGB0:\t%s * %s + %s * %s\n", DecodedMux::FormatStr(result.s1rgb.a, buf0), |
| 1119 | DecodedMux::FormatStr(result.s1rgb.b, buf1), DecodedMux::FormatStr(result.s1rgb.c, buf2),DecodedMux::FormatStr(result.s1rgb.d, buf3)); |
| 1120 | DebuggerAppendMsg("//aA0:\t%s * %s + %s * %s\n", DecodedMux::FormatStr(result.s1alpha.a, buf0), |
| 1121 | DecodedMux::FormatStr(result.s1alpha.b, buf1), DecodedMux::FormatStr(result.s1alpha.c, buf2),DecodedMux::FormatStr(result.s1alpha.d, buf3)); |
| 1122 | if( record.numOfStages == 2 ) |
| 1123 | { |
| 1124 | DebuggerAppendMsg("//aRGB1:\t%s * %s + %s * %s\n", DecodedMux::FormatStr(result.s2rgb.a, buf0), |
| 1125 | DecodedMux::FormatStr(result.s2rgb.b, buf1), DecodedMux::FormatStr(result.s2rgb.c, buf2),DecodedMux::FormatStr(result.s2rgb.d, buf3)); |
| 1126 | DebuggerAppendMsg("//aA1:\t%s * %s + %s * %s\n", DecodedMux::FormatStr(result.s2alpha.a, buf0), |
| 1127 | DecodedMux::FormatStr(result.s2alpha.b, buf1), DecodedMux::FormatStr(result.s2alpha.c, buf2),DecodedMux::FormatStr(result.s2alpha.d, buf3)); |
| 1128 | } |
| 1129 | DebuggerAppendMsg("//Final:\t%s * %s + (1 - %s) * %s + %s\n\tE=%s, F=%s\n", FormatStrForFinalStage(result.finalrgb.a, buf0), |
| 1130 | FormatStrForFinalStage(result.finalrgb.b, buf1), FormatStrForFinalStage(result.finalrgb.a, buf2), |
| 1131 | FormatStrForFinalStage(result.finalrgb.c, buf3), FormatStrForFinalStage(result.finalrgb.d, buf4), |
| 1132 | FormatStrForFinalStage(result.finalrgb.e, buf5), FormatStrForFinalStage(result.finalrgb.f, buf6)); |
| 1133 | |
| 1134 | if( result.constant0 != MUX_0 ) |
| 1135 | { |
| 1136 | DebuggerAppendMsg("//Constant 0:\t%s\n", DecodedMux::FormatStr(result.constant0, buf0)); |
| 1137 | } |
| 1138 | if( result.constant1 != MUX_0 ) |
| 1139 | { |
| 1140 | DebuggerAppendMsg("//Constant 1:\t%s\n", DecodedMux::FormatStr(result.constant1, buf0)); |
| 1141 | } |
| 1142 | TRACE0("\n\n"); |
| 1143 | } |
| 1144 | |
| 1145 | #endif |
| 1146 | |