notaz.gp2x.de / mupen64plus-pandora.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
rice: C version of upcoming NEON func
[mupen64plus-pandora.git] / source / gles2rice / src / RenderBase.cpp
CommitLineData
292f9317 1/*
2Copyright (C) 2003 Rice1964
3
4This program is free software; you can redistribute it and/or
5modify it under the terms of the GNU General Public License
6as published by the Free Software Foundation; either version 2
7of the License, or (at your option) any later version.
8
9This program is distributed in the hope that it will be useful,
10but WITHOUT ANY WARRANTY; without even the implied warranty of
11MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12GNU General Public License for more details.
13
14You should have received a copy of the GNU General Public License
15along with this program; if not, write to the Free Software
16Foundation, 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
30extern 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
42inline 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
106inline void RSP_Vtx_Clipping(int i) {}
107#endif
108
109/*
110 * Global variables
111 */
112ALIGN(16,RSP_Options gRSP)
113ALIGN(16,RDP_Options gRDP)
114
115static ALIGN(16,XVECTOR4 g_normal)
116//static int norms[3];
117
118ALIGN(16,XVECTOR4 g_vtxNonTransformed[MAX_VERTS])
119ALIGN(16,XVECTOR4 g_vecProjected[MAX_VERTS])
120ALIGN(16,XVECTOR4 g_vtxTransformed[MAX_VERTS])
121
122float g_vtxProjected5[1000][5];
123float g_vtxProjected5Clipped[2000][5];
124
125//uint32 g_dwVtxFlags[MAX_VERTS]; // Z_POS Z_NEG etc
126VECTOR2 g_fVtxTxtCoords[MAX_VERTS];
127uint32 g_dwVtxDifColor[MAX_VERTS];
128uint32 g_clipFlag[MAX_VERTS];
129uint32 g_clipFlag2[MAX_VERTS];
130RenderTexture g_textures[MAX_TEXTURES];
131float g_fFogCoord[MAX_VERTS];
132
133EXTERNAL_VERTEX g_vtxForExternal[MAX_VERTS];
134
135TLITVERTEX g_vtxBuffer[1000];
136TLITVERTEX g_clippedVtxBuffer[2000];
137uint8 g_oglVtxColors[1000][4];
138int g_clippedVtxCount=0;
139TLITVERTEX g_texRectTVtx[4];
140unsigned short g_vtxIndex[1000];
141unsigned int g_minIndex, g_maxIndex;
142
143float gRSPfFogMin;
144float gRSPfFogMax;
145float gRSPfFogDivider;
146
147uint32 gRSPnumLights;
148Light gRSPlights[16];
149
150ALIGN(16,Matrix gRSPworldProjectTransported)
151ALIGN(16,Matrix gRSPworldProject)
152ALIGN(16,Matrix gRSPmodelViewTop)
153ALIGN(16,Matrix gRSPmodelViewTopTranspose)
154ALIGN(16,Matrix dkrMatrixTransposed)
155
156N64Light gRSPn64lights[16];
157
158
159void (*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);
167n.y = (g_normal.x * matWorld.m01) + (g_normal.y * matWorld.m11) + (g_normal.z * matWorld.m21);
168n.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#define Vec3TransformNormal(vec, m) \
246 VECTOR3 temp; \
247 temp.x = (vec.x * m._11) + (vec.y * m._21) + (vec.z * m._31); \
248 temp.y = (vec.x * m._12) + (vec.y * m._22) + (vec.z * m._32); \
249 temp.z = (vec.x * m._13) + (vec.y * m._23) + (vec.z * m._33); \
250 float norm = sqrt(temp.x*temp.x+temp.y*temp.y+temp.z*temp.z); \
251 if (norm == 0.0) { vec.x = 0.0; vec.y = 0.0; vec.z = 0.0;} else \
252 { vec.x = temp.x/norm; vec.y = temp.y/norm; vec.z = temp.z/norm; }
253#endif
254
255
256#if !defined(__GNUC__) && !defined(NO_ASM)
257__declspec( naked ) void __fastcall SSEVec3Transform(int i)
258{
259 __asm
260 {
261 shl ecx,4; // ecx = i
262
263 movaps xmm1, DWORD PTR g_vtxNonTransformed [ecx]; // xmm1 as original vector
264
265 movaps xmm4, DWORD PTR gRSPworldProjectTransported; // row1
266 movaps xmm5, DWORD PTR gRSPworldProjectTransported[0x10]; // row2
267 movaps xmm6, DWORD PTR gRSPworldProjectTransported[0x20]; // row3
268 movaps xmm7, DWORD PTR gRSPworldProjectTransported[0x30]; // row4
269
270 mulps xmm4, xmm1; // row 1
271 mulps xmm5, xmm1; // row 2
272 mulps xmm6, xmm1; // row 3
273 mulps xmm7, xmm1; // row 4
274
275 movhlps xmm0, xmm4; // xmm4 high to xmm0 low
276 movlhps xmm0, xmm5; // xmm5 low to xmm0 high
277
278 addps xmm4, xmm0; // result of add are in xmm4 low
279 addps xmm5, xmm0; // result of add are in xmm5 high
280
281 shufps xmm0, xmm4, 0x44; // move xmm4 low DWORDs to xmm0 high
282 shufps xmm4, xmm5, 0xe4; // move xmm5 high DWORS to xmm4
283 movhlps xmm5, xmm0; // xmm4, xmm5 are mirrored
284
285 shufps xmm4, xmm4, 0x08; // move xmm4's 3rd uint32 to its 2nd uint32
286 shufps xmm5, xmm5, 0x0d; // move xmm5's 4th uint32 to its 2nd uint32,
287 // and move its 2nd uint32 to its 1st uint32
288
289 addps xmm4, xmm5; // results are in 1st and 2nd uint32
290
291
292 movhlps xmm0, xmm6; // xmm6 high to xmm0 low
293 movlhps xmm0, xmm7; // xmm7 low to xmm0 high
294
295 addps xmm6, xmm0; // result of add are in xmm6 low
296 addps xmm7, xmm0; // result of add are in xmm7 high
297
298 shufps xmm0, xmm6, 0x44; // move xmm6 low DWORDs to xmm0 high
299 shufps xmm6, xmm7, 0xe4; // move xmm7 high DWORS to xmm6
300 movhlps xmm7, xmm0; // xmm6, xmm7 are mirrored
301
302 shufps xmm6, xmm6, 0x08; // move xmm6's 3rd uint32 to its 2nd uint32
303 shufps xmm7, xmm7, 0x0d; // move xmm7's 4th uint32 to its 2nd uint32,
304 // and move its 2nd uint32 to its 1st uint32
305
306 addps xmm6, xmm7; // results are in 1st and 2nd uint32
307
308 movlhps xmm4, xmm6; // final result is in xmm4
309 movaps DWORD PTR g_vtxTransformed [ecx], xmm4;
310
311 movaps xmm0,xmm4;
312 shufps xmm0,xmm0,0xff;
313 divps xmm4,xmm0;
314 rcpps xmm0,xmm0;
315 movhlps xmm0,xmm4;
316 shufps xmm0,xmm0,0xe8;
317 movlhps xmm4,xmm0;
318
319 movaps DWORD PTR g_vecProjected [ecx], xmm4;
320
321 emms;
322 ret;
323 }
324}
325
326// Only used by DKR
327__declspec( naked ) void __fastcall SSEVec3TransformDKR(XVECTOR4 &pOut, const XVECTOR4 &pV)
328{
329 __asm
330 {
331 movaps xmm1, DWORD PTR [edx]; // xmm1 as original vector
332
333 movaps xmm4, DWORD PTR dkrMatrixTransposed; // row1
334 movaps xmm5, DWORD PTR dkrMatrixTransposed[0x10]; // row2
335 movaps xmm6, DWORD PTR dkrMatrixTransposed[0x20]; // row3
336 movaps xmm7, DWORD PTR dkrMatrixTransposed[0x30]; // row4
337
338 mulps xmm4, xmm1; // row 1
339 mulps xmm5, xmm1; // row 2
340 mulps xmm6, xmm1; // row 3
341 mulps xmm7, xmm1; // row 4
342
343 movhlps xmm0, xmm4; // xmm4 high to xmm0 low
344 movlhps xmm0, xmm5; // xmm5 low to xmm0 high
345
346 addps xmm4, xmm0; // result of add are in xmm4 low
347 addps xmm5, xmm0; // result of add are in xmm5 high
348
349 shufps xmm0, xmm4, 0x44; // move xmm4 low DWORDs to xmm0 high
350 shufps xmm4, xmm5, 0xe4; // move xmm5 high DWORS to xmm4
351 movhlps xmm5, xmm0; // xmm4, xmm5 are mirrored
352
353 shufps xmm4, xmm4, 0x08; // move xmm4's 3rd uint32 to its 2nd uint32
354 shufps xmm5, xmm5, 0x0d; // move xmm5's 4th uint32 to its 2nd uint32,
355 // and move its 2nd uint32 to its 1st uint32
356
357 addps xmm4, xmm5; // results are in 1st and 2nd uint32
358
359
360 movhlps xmm0, xmm6; // xmm6 high to xmm0 low
361 movlhps xmm0, xmm7; // xmm7 low to xmm0 high
362
363 addps xmm6, xmm0; // result of add are in xmm6 low
364 addps xmm7, xmm0; // result of add are in xmm7 high
365
366 shufps xmm0, xmm6, 0x44; // move xmm6 low DWORDs to xmm0 high
367 shufps xmm6, xmm7, 0xe4; // move xmm7 high DWORS to xmm6
368 movhlps xmm7, xmm0; // xmm6, xmm7 are mirrored
369
370 shufps xmm6, xmm6, 0x08; // move xmm6's 3rd uint32 to its 2nd uint32
371 shufps xmm7, xmm7, 0x0d; // move xmm7's 4th uint32 to its 2nd uint32,
372 // and move its 2nd uint32 to its 1st uint32
373
374 addps xmm6, xmm7; // results are in 1st and 2nd uint32
375
376 movlhps xmm4, xmm6; // final result is in xmm4
377 movaps DWORD PTR [ecx], xmm4;
378
379 emms;
380 ret;
381 }
382}
383#elif defined(__GNUC__) && defined(__x86_64__) && !defined(NO_ASM)
384void SSEVec3Transform(int i)
385{
386 asm volatile(" shl $4, %0 \n"
387 " movslq %k0, %q0 \n"
388 " movaps (%1,%q0), %%xmm1 \n"
389 " movaps 0(%2), %%xmm4 \n"
390 " movaps 16(%2), %%xmm5 \n"
391 " movaps 32(%2), %%xmm6 \n"
392 " movaps 48(%2), %%xmm7 \n"
393 " mulps %%xmm1, %%xmm4 \n"
394 " mulps %%xmm1, %%xmm5 \n"
395 " mulps %%xmm1, %%xmm6 \n"
396 " mulps %%xmm1, %%xmm7 \n"
397 " movhlps %%xmm4, %%xmm0 \n"
398 " movlhps %%xmm5, %%xmm0 \n"
399 " addps %%xmm0, %%xmm4 \n"
400 " addps %%xmm0, %%xmm5 \n"
401 " shufps $0x44, %%xmm4, %%xmm0 \n"
402 " shufps $0xe4, %%xmm5, %%xmm4 \n"
403 " movhlps %%xmm0, %%xmm5 \n"
404 " shufps $0x08, %%xmm4, %%xmm4 \n"
405 " shufps $0x0d, %%xmm5, %%xmm5 \n"
406 " addps %%xmm5, %%xmm4 \n"
407 " movhlps %%xmm6, %%xmm0 \n"
408 " movlhps %%xmm7, %%xmm0 \n"
409 " addps %%xmm0, %%xmm6 \n"
410 " addps %%xmm0, %%xmm7 \n"
411 " shufps $0x44, %%xmm6, %%xmm0 \n"
412 " shufps $0xe4, %%xmm7, %%xmm6 \n"
413 " movhlps %%xmm0, %%xmm7 \n"
414 " shufps $0x08, %%xmm6, %%xmm6 \n"
415 " shufps $0x0d, %%xmm7, %%xmm7 \n"
416 " addps %%xmm7, %%xmm6 \n"
417 " movlhps %%xmm6, %%xmm4 \n"
418 " movaps %%xmm4, (%3,%q0) \n"
419 " movaps %%xmm4, %%xmm0 \n"
420 " shufps $0xff, %%xmm0, %%xmm0 \n"
421 " divps %%xmm0, %%xmm4 \n"
422 " rcpps %%xmm0, %%xmm0 \n"
423 " movhlps %%xmm4, %%xmm0 \n"
424 " shufps $0xe8, %%xmm0, %%xmm0 \n"
425 " movlhps %%xmm0, %%xmm4 \n"
426 " movaps %%xmm4, (%4,%q0) \n"
427 : "+r"(i)
428 : "r"(g_vtxNonTransformed), "r"(&gRSPworldProjectTransported.m[0][0]), "r"(g_vtxTransformed), "r"(g_vecProjected)
429 : "memory", "%xmm0", "%xmm1", "%xmm4", "%xmm5", "%xmm6", "%xmm7"
430 );
431}
432#elif !defined(NO_ASM) // 32-bit GCC assumed
433void SSEVec3Transform(int i)
434{
435 asm volatile(" shl $4, %0 \n"
436 " movaps (%1,%0), %%xmm1 \n"
437 " movaps 0(%2), %%xmm4 \n"
438 " movaps 16(%2), %%xmm5 \n"
439 " movaps 32(%2), %%xmm6 \n"
440 " movaps 48(%2), %%xmm7 \n"
441 " mulps %%xmm1, %%xmm4 \n"
442 " mulps %%xmm1, %%xmm5 \n"
443 " mulps %%xmm1, %%xmm6 \n"
444 " mulps %%xmm1, %%xmm7 \n"
445 " movhlps %%xmm4, %%xmm0 \n"
446 " movlhps %%xmm5, %%xmm0 \n"
447 " addps %%xmm0, %%xmm4 \n"
448 " addps %%xmm0, %%xmm5 \n"
449 " shufps $0x44, %%xmm4, %%xmm0 \n"
450 " shufps $0xe4, %%xmm5, %%xmm4 \n"
451 " movhlps %%xmm0, %%xmm5 \n"
452 " shufps $0x08, %%xmm4, %%xmm4 \n"
453 " shufps $0x0d, %%xmm5, %%xmm5 \n"
454 " addps %%xmm5, %%xmm4 \n"
455 " movhlps %%xmm6, %%xmm0 \n"
456 " movlhps %%xmm7, %%xmm0 \n"
457 " addps %%xmm0, %%xmm6 \n"
458 " addps %%xmm0, %%xmm7 \n"
459 " shufps $0x44, %%xmm6, %%xmm0 \n"
460 " shufps $0xe4, %%xmm7, %%xmm6 \n"
461 " movhlps %%xmm0, %%xmm7 \n"
462 " shufps $0x08, %%xmm6, %%xmm6 \n"
463 " shufps $0x0d, %%xmm7, %%xmm7 \n"
464 " addps %%xmm7, %%xmm6 \n"
465 " movlhps %%xmm6, %%xmm4 \n"
466 " movaps %%xmm4, (%3,%0) \n"
467 " movaps %%xmm4, %%xmm0 \n"
468 " shufps $0xff, %%xmm0, %%xmm0 \n"
469 " divps %%xmm0, %%xmm4 \n"
470 " rcpps %%xmm0, %%xmm0 \n"
471 " movhlps %%xmm4, %%xmm0 \n"
472 " shufps $0xe8, %%xmm0, %%xmm0 \n"
473 " movlhps %%xmm0, %%xmm4 \n"
474 " movaps %%xmm4, (%4,%0) \n"
475 : "+r"(i)
476 : "r"(g_vtxNonTransformed), "r"(&gRSPworldProjectTransported.m[0][0]), "r"(g_vtxTransformed), "r"(g_vecProjected)
477 : "memory", "%xmm0", "%xmm1", "%xmm4", "%xmm5", "%xmm6", "%xmm7"
478 );
479}
480#endif
481float real255 = 255.0f;
482float real128 = 128.0f;
483
484#if !defined(__GNUC__) && !defined(NO_ASM)
485__declspec( naked ) void __fastcall SSEVec3TransformNormal()
486{
487 __asm
488 {
489 mov DWORD PTR [g_normal][12], 0;
490
491 movaps xmm4, DWORD PTR gRSPmodelViewTopTranspose; // row1
492 movaps xmm5, DWORD PTR gRSPmodelViewTopTranspose[0x10]; // row2
493 movaps xmm1, DWORD PTR [g_normal]; // xmm1 as the normal vector
494 movaps xmm6, DWORD PTR gRSPmodelViewTopTranspose[0x20]; // row3
495
496 mulps xmm4, xmm1; // row 1
497 mulps xmm5, xmm1; // row 2
498 mulps xmm6, xmm1; // row 3
499
500 movhlps xmm0, xmm4; // xmm4 high to xmm0 low
501 movlhps xmm0, xmm5; // xmm5 low to xmm0 high
502
503 addps xmm4, xmm0; // result of add are in xmm4 low
504 addps xmm5, xmm0; // result of add are in xmm5 high
505
506 shufps xmm0, xmm4, 0x44; // move xmm4 low DWORDs to xmm0 high
507 shufps xmm4, xmm5, 0xe4; // move xmm5 high DWORS to xmm4
508 movhlps xmm5, xmm0; // xmm4, xmm5 are mirrored
509
510 shufps xmm4, xmm4, 0x08; // move xmm4's 3rd uint32 to its 2nd uint32
511 shufps xmm5, xmm5, 0x0d; // move xmm5's 4th uint32 to its 2nd uint32,
512
513 addps xmm4, xmm5; // results are in 1st and 2nd uint32
514
515 movaps xmm1,xmm4;
516 mulps xmm1,xmm1; //square
517 movlhps xmm7, xmm1;
518 shufps xmm7, xmm7,0x03;
519 addss xmm7, xmm1;
520
521 movhlps xmm0, xmm6; // xmm6 high to xmm0 low
522 addps xmm6, xmm0; // result of add are in xmm6 low
523
524 movlhps xmm0, xmm6;
525 shufps xmm0, xmm0, 0x03;
526 addss xmm0, xmm6; // result of add is at xmm0's 1st uint32
527
528 movlhps xmm4, xmm0;
529
530 mulss xmm0,xmm0;
531 addss xmm7,xmm0; // xmm7 1st uint32 is the sum of squares
532
533#ifdef DEBUGGER
534 movaps DWORD PTR [g_normal], xmm4;
535 movss DWORD PTR [g_normal][12], xmm7;
536#endif
537 xorps xmm0,xmm0;
538 ucomiss xmm0,xmm7;
539 jz l2
540
541 rsqrtss xmm7,xmm7;
542 shufps xmm7,xmm7,0;
543#ifdef DEBUGGER
544 movss DWORD PTR [g_normal][12], xmm7;
545#endif
546 mulps xmm4,xmm7;
547
548 movaps DWORD PTR [g_normal], xmm4; // Normalized
549 mov DWORD PTR [g_normal][12], 0;
550
551 emms;
552 ret;
553l2:
554 movss DWORD PTR [g_normal], xmm0;
555 movss DWORD PTR [g_normal][12], xmm0;
556 emms;
557 ret;
558 }
559}
560#elif defined(__GNUC__) && !defined(NO_ASM) // this code should compile for both 64-bit and 32-bit architectures
561void SSEVec3TransformNormal(void)
562{
563 asm volatile(" movl $0, 12(%0) \n"
564 " movaps (%1), %%xmm4 \n"
565 " movaps 16(%1), %%xmm5 \n"
566 " movaps (%0), %%xmm1 \n"
567 " movaps 32(%1), %%xmm6 \n"
568 " mulps %%xmm1, %%xmm4 \n"
569 " mulps %%xmm1, %%xmm5 \n"
570 " mulps %%xmm1, %%xmm6 \n"
571 " movhlps %%xmm4, %%xmm0 \n"
572 " movlhps %%xmm5, %%xmm0 \n"
573 " addps %%xmm0, %%xmm4 \n"
574 " addps %%xmm0, %%xmm5 \n"
575 " shufps $0x44, %%xmm4, %%xmm0 \n"
576 " shufps $0xe4, %%xmm5, %%xmm4 \n"
577 " movhlps %%xmm0, %%xmm5 \n"
578 " shufps $0x08, %%xmm4, %%xmm4 \n"
579 " shufps $0x0d, %%xmm5, %%xmm5 \n"
580 " addps %%xmm5, %%xmm4 \n"
581 " movaps %%xmm4, %%xmm1 \n"
582 " mulps %%xmm1, %%xmm1 \n"
583 " movlhps %%xmm1, %%xmm7 \n"
584 " shufps $0x03, %%xmm7, %%xmm7 \n"
585 " addss %%xmm1, %%xmm7 \n"
586 " movhlps %%xmm6, %%xmm0 \n"
587 " addps %%xmm0, %%xmm6 \n"
588 " movlhps %%xmm6, %%xmm0 \n"
589 " shufps $0x03, %%xmm0, %%xmm0 \n"
590 " addss %%xmm6, %%xmm0 \n"
591 " movlhps %%xmm0, %%xmm4 \n"
592 " mulss %%xmm0, %%xmm0 \n"
593 " addss %%xmm0, %%xmm7 \n"
594#ifdef DEBUGGER
595 " movaps %%xmm4, (%0) \n"
596 " movss %%xmm7, 12(%0) \n"
597#endif
598 " xorps %%xmm0, %%xmm0 \n"
599 " ucomiss %%xmm7, %%xmm0 \n"
600 " jz 0f \n"
601 " rsqrtss %%xmm7, %%xmm7 \n"
602 " shufps $0x00, %%xmm7, %%xmm7 \n"
603#ifdef DEBUGGER
604 " movss %%xmm7, 12(%0) \n"
605#endif
606 " mulps %%xmm7, %%xmm4 \n"
607 " movaps %%xmm4, (%0) \n"
608 " movl $0, 12(%0) \n"
609 " jmp 1f \n"
610 "0: \n"
611 " movss %%xmm0, (%0) \n"
612 " movss %%xmm0, 12(%0) \n"
613 "1: \n"
614 :
615 : "r"(&g_normal.x), "r"(&gRSPmodelViewTopTranspose.m[0][0])
616 : "memory", "cc", "%xmm0", "%xmm1", "%xmm4", "%xmm5", "%xmm6", "%xmm7"
617 );
618}
619#endif
620
621void NormalizeNormalVec()
622{
623 float w = 1/sqrtf(g_normal.x*g_normal.x + g_normal.y*g_normal.y + g_normal.z*g_normal.z);
624 g_normal.x *= w;
625 g_normal.y *= w;
626 g_normal.z *= w;
627}
628
629
630void InitRenderBase()
631{
632#if !defined(NO_ASM)
633 if( status.isSSEEnabled && !g_curRomInfo.bPrimaryDepthHack && options.enableHackForGames != HACK_FOR_NASCAR)
634 {
635 ProcessVertexData = ProcessVertexDataSSE;
636 }
637 else
0764a475 638#elif defined(__ARM_NEON__)
639 if( !g_curRomInfo.bPrimaryDepthHack && options.enableHackForGames != HACK_FOR_NASCAR && options.enableHackForGames != HACK_FOR_ZELDA_MM && !options.bWinFrameMode)
640 {
641 ProcessVertexData = ProcessVertexDataNEON;
642 }
643 else
292f9317 644#endif
645 {
646 ProcessVertexData = ProcessVertexDataNoSSE;
647 }
648
649 gRSPfFogMin = gRSPfFogMax = 0.0f;
650 windowSetting.fMultX = windowSetting.fMultY = 2.0f;
651 windowSetting.vpLeftW = windowSetting.vpTopW = 0;
652 windowSetting.vpRightW = windowSetting.vpWidthW = 640;
653 windowSetting.vpBottomW = windowSetting.vpHeightW = 480;
654 gRSP.maxZ = 0;
655 gRSP.nVPLeftN = gRSP.nVPTopN = 0;
656 gRSP.nVPRightN = 640;
657 gRSP.nVPBottomN = 640;
658 gRSP.nVPWidthN = 640;
659 gRSP.nVPHeightN = 640;
660 gRDP.scissor.left=gRDP.scissor.top=0;
661 gRDP.scissor.right=gRDP.scissor.bottom=640;
662
663 gRSP.bLightingEnable = gRSP.bTextureGen = false;
664 gRSP.curTile=gRSPnumLights=gRSP.ambientLightColor=gRSP.ambientLightIndex= 0;
665 gRSP.fAmbientLightR=gRSP.fAmbientLightG=gRSP.fAmbientLightB=0;
666 gRSP.projectionMtxTop = gRSP.modelViewMtxTop = 0;
667 gRDP.fogColor = gRDP.primitiveColor = gRDP.envColor = gRDP.primitiveDepth = gRDP.primLODMin = gRDP.primLODFrac = gRDP.LODFrac = 0;
668 gRDP.fPrimitiveDepth = 0;
669 gRSP.numVertices = 0;
670 gRSP.maxVertexID = 0;
671 gRSP.bCullFront=false;
672 gRSP.bCullBack=true;
673 gRSP.bFogEnabled=gRDP.bFogEnableInBlender=false;
674 gRSP.bZBufferEnabled=true;
675 gRSP.shadeMode=SHADE_SMOOTH;
676 gRDP.keyR=gRDP.keyG=gRDP.keyB=gRDP.keyA=gRDP.keyRGB=gRDP.keyRGBA = 0;
677 gRDP.fKeyA = 0;
678 gRSP.DKRCMatrixIndex = gRSP.dwDKRVtxAddr = gRSP.dwDKRMatrixAddr = 0;
679 gRSP.DKRBillBoard = false;
680
681 gRSP.fTexScaleX = 1/32.0f;
682 gRSP.fTexScaleY = 1/32.0f;
683 gRSP.bTextureEnabled = FALSE;
684
685 gRSP.clip_ratio_left = 0;
686 gRSP.clip_ratio_top = 0;
687 gRSP.clip_ratio_right = 640;
688 gRSP.clip_ratio_bottom = 480;
689 gRSP.clip_ratio_negx = 1;
690 gRSP.clip_ratio_negy = 1;
691 gRSP.clip_ratio_posx = 1;
692 gRSP.clip_ratio_posy = 1;
693 gRSP.real_clip_scissor_left = 0;
694 gRSP.real_clip_scissor_top = 0;
695 gRSP.real_clip_scissor_right = 640;
696 gRSP.real_clip_scissor_bottom = 480;
697 windowSetting.clipping.left = 0;
698 windowSetting.clipping.top = 0;
699 windowSetting.clipping.right = 640;
700 windowSetting.clipping.bottom = 480;
701 windowSetting.clipping.width = 640;
702 windowSetting.clipping.height = 480;
703 windowSetting.clipping.needToClip = false;
704 gRSP.real_clip_ratio_negx = 1;
705 gRSP.real_clip_ratio_negy = 1;
706 gRSP.real_clip_ratio_posx = 1;
707 gRSP.real_clip_ratio_posy = 1;
708
709 gRSP.DKRCMatrixIndex=0;
710 gRSP.DKRVtxCount=0;
711 gRSP.DKRBillBoard = false;
712 gRSP.dwDKRVtxAddr=0;
713 gRSP.dwDKRMatrixAddr=0;
714
715
716 gRDP.geometryMode = 0;
717 gRDP.otherModeL = 0;
718 gRDP.otherModeH = 0;
719 gRDP.fillColor = 0xFFFFFFFF;
720 gRDP.originalFillColor =0;
721
722 gRSP.ucode = 1;
723 gRSP.vertexMult = 10;
724 gRSP.bNearClip = false;
725 gRSP.bRejectVtx = false;
726
727 gRDP.texturesAreReloaded = false;
728 gRDP.textureIsChanged = false;
729 gRDP.colorsAreReloaded = false;
730
731 memset(&gRDP.otherMode,0,sizeof(RDP_OtherMode));
732 memset(&gRDP.tiles,0,sizeof(Tile)*8);
733
734 for( int i=0; i<MAX_VERTS; i++ )
735 {
736 g_clipFlag[i] = 0;
737 g_vtxNonTransformed[i].w = 1;
738 }
739
740 memset(gRSPn64lights, 0, sizeof(N64Light)*16);
741}
742
743void SetFogMinMax(float fMin, float fMax, float fMul, float fOffset)
744{
745 if( fMin > fMax )
746 {
747 float temp = fMin;
748 fMin = fMax;
749 fMax = temp;
750 }
751
752 {
753 gRSPfFogMin = max(0,fMin/500-1);
754 gRSPfFogMax = fMax/500-1;
755 }
756
757 gRSPfFogDivider = 255/(gRSPfFogMax-gRSPfFogMin);
758 CRender::g_pRender->SetFogMinMax(fMin, fMax);
759}
760
761void InitVertexColors()
762{
763}
764
765void InitVertexTextureConstants()
766{
767 float scaleX;
768 float scaleY;
769
770 RenderTexture &tex0 = g_textures[gRSP.curTile];
771 //CTexture *surf = tex0.m_pCTexture;
772 Tile &tile0 = gRDP.tiles[gRSP.curTile];
773
774 scaleX = gRSP.fTexScaleX;
775 scaleY = gRSP.fTexScaleY;
776
777 gRSP.tex0scaleX = scaleX * tile0.fShiftScaleS/tex0.m_fTexWidth;
778 gRSP.tex0scaleY = scaleY * tile0.fShiftScaleT/tex0.m_fTexHeight;
779
780 gRSP.tex0OffsetX = tile0.fhilite_sl/tex0.m_fTexWidth;
781 gRSP.tex0OffsetY = tile0.fhilite_tl/tex0.m_fTexHeight;
782
783 if( CRender::g_pRender->IsTexel1Enable() )
784 {
785 RenderTexture &tex1 = g_textures[(gRSP.curTile+1)&7];
786 //CTexture *surf = tex1.m_pCTexture;
787 Tile &tile1 = gRDP.tiles[(gRSP.curTile+1)&7];
788
789 gRSP.tex1scaleX = scaleX * tile1.fShiftScaleS/tex1.m_fTexWidth;
790 gRSP.tex1scaleY = scaleY * tile1.fShiftScaleT/tex1.m_fTexHeight;
791
792 gRSP.tex1OffsetX = tile1.fhilite_sl/tex1.m_fTexWidth;
793 gRSP.tex1OffsetY = tile1.fhilite_tl/tex1.m_fTexHeight;
794 }
795
796 gRSP.texGenXRatio = tile0.fShiftScaleS;
797 gRSP.texGenYRatio = gRSP.fTexScaleX/gRSP.fTexScaleY*tex0.m_fTexWidth/tex0.m_fTexHeight*tile0.fShiftScaleT;
798}
799
800void TexGen(float &s, float &t)
801{
802 if (gRDP.geometryMode & G_TEXTURE_GEN_LINEAR)
803 {
804 s = acosf(g_normal.x) / 3.14159f;
805 t = acosf(g_normal.y) / 3.14159f;
806 }
807 else
808 {
809 s = 0.5f * ( 1.0f + g_normal.x);
810 t = 0.5f * ( 1.0f - g_normal.y);
811 }
812}
813
814void ComputeLOD(bool openGL)
815{
816 TLITVERTEX &v0 = g_vtxBuffer[0];
817 TLITVERTEX &v1 = g_vtxBuffer[1];
818 RenderTexture &tex0 = g_textures[gRSP.curTile];
819
820 float d,dt;
821 if( openGL )
822 {
823 float x = g_vtxProjected5[0][0] / g_vtxProjected5[0][4] - g_vtxProjected5[1][0] / g_vtxProjected5[1][4];
824 float y = g_vtxProjected5[0][1] / g_vtxProjected5[0][4] - g_vtxProjected5[1][1] / g_vtxProjected5[1][4];
825
826 x = windowSetting.vpWidthW*x/windowSetting.fMultX/2;
827 y = windowSetting.vpHeightW*y/windowSetting.fMultY/2;
828 d = sqrtf(x*x+y*y);
829 }
830 else
831 {
832 float x = (v0.x - v1.x)/ windowSetting.fMultX;
833 float y = (v0.y - v1.y)/ windowSetting.fMultY;
834 d = sqrtf(x*x+y*y);
835 }
836
837 float s0 = v0.tcord[0].u * tex0.m_fTexWidth;
838 float t0 = v0.tcord[0].v * tex0.m_fTexHeight;
839 float s1 = v1.tcord[0].u * tex0.m_fTexWidth;
840 float t1 = v1.tcord[0].v * tex0.m_fTexHeight;
841
842 dt = sqrtf((s0-s1)*(s0-s1)+(t0-t1)*(t0-t1));
843
844 float lod = dt/d;
845 float frac = log10f(lod)/log10f(2.0f);
846 //DEBUGGER_IF_DUMP(pauseAtNext,{DebuggerAppendMsg("LOD frac = %f", frac);});
847 frac = (lod / powf(2.0f,floorf(frac)));
848 frac = frac - floorf(frac);
849 //DEBUGGER_IF_DUMP(pauseAtNext,{DebuggerAppendMsg("LOD = %f, frac = %f", lod, frac);});
850 gRDP.LODFrac = (uint32)(frac*255);
851 CRender::g_pRender->SetCombinerAndBlender();
852}
853
854bool bHalfTxtScale=false;
855extern uint32 lastSetTile;
856
857void InitVertex(uint32 dwV, uint32 vtxIndex, bool bTexture, bool openGL)
858{
859 VTX_DUMP(TRACE2("Init vertex (%d) to vtx buf[%d]:", dwV, vtxIndex));
860
861 TLITVERTEX &v = g_vtxBuffer[vtxIndex];
862 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));
863 if( openGL )
864 {
865 g_vtxProjected5[vtxIndex][0] = g_vtxTransformed[dwV].x;
866 g_vtxProjected5[vtxIndex][1] = g_vtxTransformed[dwV].y;
867 g_vtxProjected5[vtxIndex][2] = g_vtxTransformed[dwV].z;
868 g_vtxProjected5[vtxIndex][3] = g_vtxTransformed[dwV].w;
869 g_vtxProjected5[vtxIndex][4] = g_vecProjected[dwV].z;
870
871 if( g_vtxTransformed[dwV].w < 0 )
872 g_vtxProjected5[vtxIndex][4] = 0;
873
874 g_vtxIndex[vtxIndex] = vtxIndex;
875 }
876
877 if( !openGL || options.bOGLVertexClipper == TRUE )
878 {
879 v.x = g_vecProjected[dwV].x*gRSP.vtxXMul+gRSP.vtxXAdd;
880 v.y = g_vecProjected[dwV].y*gRSP.vtxYMul+gRSP.vtxYAdd;
881 v.z = (g_vecProjected[dwV].z + 1.0f) * 0.5f; // DirectX minZ=0, maxZ=1
882 //v.z = g_vecProjected[dwV].z; // DirectX minZ=0, maxZ=1
883 v.rhw = g_vecProjected[dwV].w;
884 VTX_DUMP(TRACE4(" Proj : x=%f, y=%f, z=%f, rhw=%f", v.x,v.y,v.z,v.rhw));
885
886 if( gRSP.bProcessSpecularColor )
887 {
888 v.dcSpecular = CRender::g_pRender->PostProcessSpecularColor();
889 if( gRSP.bFogEnabled )
890 {
891 v.dcSpecular &= 0x00FFFFFF;
892 uint32 fogFct = 0xFF-(uint8)((g_fFogCoord[dwV]-gRSPfFogMin)*gRSPfFogDivider);
893 v.dcSpecular |= (fogFct<<24);
894 }
895 }
896 else if( gRSP.bFogEnabled )
897 {
898 uint32 fogFct = 0xFF-(uint8)((g_fFogCoord[dwV]-gRSPfFogMin)*gRSPfFogDivider);
899 v.dcSpecular = (fogFct<<24);
900 }
901 }
902 VTX_DUMP(TRACE2(" (U,V): %f, %f", g_fVtxTxtCoords[dwV].x,g_fVtxTxtCoords[dwV].y));
903
904 v.dcDiffuse = g_dwVtxDifColor[dwV];
905 if( gRDP.otherMode.key_en )
906 {
907 v.dcDiffuse &= 0x00FFFFFF;
908 v.dcDiffuse |= (gRDP.keyA<<24);
909 }
910 else if( gRDP.otherMode.aa_en && gRDP.otherMode.clr_on_cvg==0 )
911 {
912 v.dcDiffuse |= 0xFF000000;
913 }
914
915 if( gRSP.bProcessDiffuseColor )
916 {
917 v.dcDiffuse = CRender::g_pRender->PostProcessDiffuseColor(v.dcDiffuse);
918 }
919 if( options.bWinFrameMode )
920 {
921 v.dcDiffuse = g_dwVtxDifColor[dwV];
922 }
923
924 if( openGL )
925 {
926 g_oglVtxColors[vtxIndex][0] = v.r;
927 g_oglVtxColors[vtxIndex][1] = v.g;
928 g_oglVtxColors[vtxIndex][2] = v.b;
929 g_oglVtxColors[vtxIndex][3] = v.a;
930 }
931
932 if( bTexture )
933 {
934 // If the vert is already lit, then there is no normal (and hence we can't generate tex coord)
935 // Only scale if not generated automatically
936 if (gRSP.bTextureGen && gRSP.bLightingEnable)
937 {
938 // Correction for texGen result
939 float u0,u1,v0,v1;
940 RenderTexture &tex0 = g_textures[gRSP.curTile];
941 u0 = g_fVtxTxtCoords[dwV].x * 32 * 1024 * gRSP.fTexScaleX / tex0.m_fTexWidth;
942 v0 = g_fVtxTxtCoords[dwV].y * 32 * 1024 * gRSP.fTexScaleY / tex0.m_fTexHeight;
943 u0 *= (gRDP.tiles[gRSP.curTile].fShiftScaleS);
944 v0 *= (gRDP.tiles[gRSP.curTile].fShiftScaleT);
945
946 if( CRender::g_pRender->IsTexel1Enable() )
947 {
948 RenderTexture &tex1 = g_textures[(gRSP.curTile+1)&7];
949 u1 = g_fVtxTxtCoords[dwV].x * 32 * 1024 * gRSP.fTexScaleX / tex1.m_fTexWidth;
950 v1 = g_fVtxTxtCoords[dwV].y * 32 * 1024 * gRSP.fTexScaleY / tex1.m_fTexHeight;
951 u1 *= gRDP.tiles[(gRSP.curTile+1)&7].fShiftScaleS;
952 v1 *= gRDP.tiles[(gRSP.curTile+1)&7].fShiftScaleT;
953 CRender::g_pRender->SetVertexTextureUVCoord(v, u0, v0, u1, v1);
954 }
955 else
956 {
957 CRender::g_pRender->SetVertexTextureUVCoord(v, u0, v0);
958 }
959 }
960 else
961 {
962 float tex0u = g_fVtxTxtCoords[dwV].x *gRSP.tex0scaleX - gRSP.tex0OffsetX ;
963 float tex0v = g_fVtxTxtCoords[dwV].y *gRSP.tex0scaleY - gRSP.tex0OffsetY ;
964
965 if( CRender::g_pRender->IsTexel1Enable() )
966 {
967 float tex1u = g_fVtxTxtCoords[dwV].x *gRSP.tex1scaleX - gRSP.tex1OffsetX ;
968 float tex1v = g_fVtxTxtCoords[dwV].y *gRSP.tex1scaleY - gRSP.tex1OffsetY ;
969
970 CRender::g_pRender->SetVertexTextureUVCoord(v, tex0u, tex0v, tex1u, tex1v);
971 VTX_DUMP(TRACE2(" (tex0): %f, %f", tex0u,tex0v));
972 VTX_DUMP(TRACE2(" (tex1): %f, %f", tex1u,tex1v));
973 }
974 else
975 {
976 CRender::g_pRender->SetVertexTextureUVCoord(v, tex0u, tex0v);
977 VTX_DUMP(TRACE2(" (tex0): %f, %f", tex0u,tex0v));
978 }
979 }
980
981 // Check for txt scale hack
982 if( !bHalfTxtScale && g_curRomInfo.bTextureScaleHack &&
983 (gRDP.tiles[lastSetTile].dwSize == TXT_SIZE_32b || gRDP.tiles[lastSetTile].dwSize == TXT_SIZE_4b ) )
984 {
985 int width = ((gRDP.tiles[lastSetTile].sh-gRDP.tiles[lastSetTile].sl+1)<<1);
986 int height = ((gRDP.tiles[lastSetTile].th-gRDP.tiles[lastSetTile].tl+1)<<1);
987 if( g_fVtxTxtCoords[dwV].x*gRSP.fTexScaleX == width || g_fVtxTxtCoords[dwV].y*gRSP.fTexScaleY == height )
988 {
989 bHalfTxtScale=true;
990 }
991 }
992 }
993
994 if( g_curRomInfo.bEnableTxtLOD && vtxIndex == 1 && gRDP.otherMode.text_lod )
995 {
996 if( CRender::g_pRender->IsTexel1Enable() && CRender::g_pRender->m_pColorCombiner->m_pDecodedMux->isUsed(MUX_LODFRAC) )
997 {
998 ComputeLOD(openGL);
999 }
1000 else
1001 {
1002 gRDP.LODFrac = 0;
1003 }
1004 }
1005
1006 VTX_DUMP(TRACE2(" DIF(%08X), SPE(%08X)", v.dcDiffuse, v.dcSpecular));
1007 VTX_DUMP(TRACE0(""));
1008}
1009
1010uint32 LightVert(XVECTOR4 & norm, int vidx)
1011{
1012 float fCosT;
1013
1014 // Do ambient
1015 register float r = gRSP.fAmbientLightR;
1016 register float g = gRSP.fAmbientLightG;
1017 register float b = gRSP.fAmbientLightB;
1018
1019 if( options.enableHackForGames != HACK_FOR_ZELDA_MM )
1020 {
1021 for (register unsigned int l=0; l < gRSPnumLights; l++)
1022 {
1023 fCosT = norm.x*gRSPlights[l].x + norm.y*gRSPlights[l].y + norm.z*gRSPlights[l].z;
1024
1025 if (fCosT > 0 )
1026 {
1027 r += gRSPlights[l].fr * fCosT;
1028 g += gRSPlights[l].fg * fCosT;
1029 b += gRSPlights[l].fb * fCosT;
1030 }
1031 }
1032 }
1033 else
1034 {
1035 XVECTOR4 v;
1036 bool transformed = false;
1037
1038 for (register unsigned int l=0; l < gRSPnumLights; l++)
1039 {
1040 if( gRSPlights[l].range == 0 )
1041 {
1042 // Regular directional light
1043 fCosT = norm.x*gRSPlights[l].x + norm.y*gRSPlights[l].y + norm.z*gRSPlights[l].z;
1044
1045 if (fCosT > 0 )
1046 {
1047 r += gRSPlights[l].fr * fCosT;
1048 g += gRSPlights[l].fg * fCosT;
1049 b += gRSPlights[l].fb * fCosT;
1050 }
1051 }
1052 else //if( (gRSPlights[l].col&0x00FFFFFF) != 0x00FFFFFF )
1053 {
1054 // Point light
1055 if( !transformed )
1056 {
1057 Vec3Transform(&v, (XVECTOR3*)&g_vtxNonTransformed[vidx], &gRSPmodelViewTop); // Convert to w=1
1058 transformed = true;
1059 }
1060
1061 XVECTOR3 dir(gRSPlights[l].x - v.x, gRSPlights[l].y - v.y, gRSPlights[l].z - v.z);
1062 //XVECTOR3 dir(v.x-gRSPlights[l].x, v.y-gRSPlights[l].y, v.z-gRSPlights[l].z);
1063 float d2 = sqrtf(dir.x*dir.x+dir.y*dir.y+dir.z*dir.z);
1064 dir.x /= d2;
1065 dir.y /= d2;
1066 dir.z /= d2;
1067
1068 fCosT = norm.x*dir.x + norm.y*dir.y + norm.z*dir.z;
1069
1070 if (fCosT > 0 )
1071 {
1072 //float f = d2/gRSPlights[l].range*50;
1073 float f = d2/15000*50;
1074 f = 1 - min(f,1);
1075 fCosT *= f*f;
1076
1077 r += gRSPlights[l].fr * fCosT;
1078 g += gRSPlights[l].fg * fCosT;
1079 b += gRSPlights[l].fb * fCosT;
1080 }
1081 }
1082 }
1083 }
1084
1085 if (r > 255) r = 255;
1086 if (g > 255) g = 255;
1087 if (b > 255) b = 255;
1088 return ((0xff000000)|(((uint32)r)<<16)|(((uint32)g)<<8)|((uint32)b));
1089}
1090
1091uint32 LightVertNew(XVECTOR4 & norm)
1092{
1093 float fCosT;
1094
1095 // Do ambient
1096 register float r = gRSP.fAmbientLightR;
1097 register float g = gRSP.fAmbientLightG;
1098 register float b = gRSP.fAmbientLightB;
1099
1100
1101 for (register unsigned int l=0; l < gRSPnumLights; l++)
1102 {
1103 fCosT = norm.x*gRSPlights[l].tx + norm.y*gRSPlights[l].ty + norm.z*gRSPlights[l].tz;
1104
1105 if (fCosT > 0 )
1106 {
1107 r += gRSPlights[l].fr * fCosT;
1108 g += gRSPlights[l].fg * fCosT;
1109 b += gRSPlights[l].fb * fCosT;
1110 }
1111 }
1112
1113 if (r > 255) r = 255;
1114 if (g > 255) g = 255;
1115 if (b > 255) b = 255;
1116 return ((0xff000000)|(((uint32)r)<<16)|(((uint32)g)<<8)|((uint32)b));
1117}
1118
1119
1120float zero = 0.0f;
1121float onef = 1.0f;
1122float fcosT;
1123
1124#if !defined(__GNUC__) && !defined(NO_ASM)
1125__declspec( naked ) uint32 __fastcall SSELightVert()
1126{
1127 __asm
1128 {
1129 movaps xmm3, DWORD PTR gRSP; // loading Ambient colors, xmm3 is the result color
1130 movaps xmm4, DWORD PTR [g_normal]; // xmm4 is the normal
1131
1132 mov ecx, 0;
1133loopback:
1134 cmp ecx, DWORD PTR gRSPnumLights;
1135 jae breakout;
1136 mov eax,ecx;
1137 imul eax,0x44;
1138 movups xmm5, DWORD PTR gRSPlights[eax]; // Light Dir
1139 movups xmm1, DWORD PTR gRSPlights[0x14][eax]; // Light color
1140 mulps xmm5, xmm4; // Lightdir * normals
1141
1142 movhlps xmm0,xmm5;
1143 addps xmm0,xmm5;
1144 shufps xmm5,xmm0,0x01;
1145 addps xmm0,xmm5;
1146
1147 comiss xmm0,zero;
1148 jc endloop
1149
1150 shufps xmm0,xmm0,0; // fcosT
1151 mulps xmm1,xmm0;
1152 addps xmm3,xmm1;
1153endloop:
1154 inc ecx;
1155 jmp loopback;
1156breakout:
1157
1158 movss xmm0,DWORD PTR real255;
1159 shufps xmm0,xmm0,0;
1160 minps xmm0,xmm3;
1161
1162 // Without using a memory
1163 cvtss2si eax,xmm0; // move the 1st uint32 to eax
1164 shl eax,10h;
1165 or eax,0FF000000h;
1166 shufps xmm0,xmm0,0E5h; // move the 2nd uint32 to the 1st uint32
1167 cvtss2si ecx,xmm0; // move the 1st uint32 to ecx
1168 shl ecx,8;
1169 or eax,ecx;
1170 shufps xmm0,xmm0,0E6h; // Move the 3rd uint32 to the 1st uint32
1171 cvtss2si ecx,xmm0;
1172 or eax,ecx;
1173
1174 ret;
1175 }
1176}
1177#elif defined(__GNUC__) && defined(__x86_64__) && !defined(NO_ASM)
1178uint32 SSELightVert(void)
1179{
1180 uint32 rval;
1181 float f255 = 255.0, fZero = 0.0;
1182
1183 asm volatile(" movaps %1, %%xmm3 \n" // xmm3 == gRSP.fAmbientLight{RGBA}
1184 " movaps %2, %%xmm4 \n" // xmm4 == g_normal.{xyz}
1185 " xor %%rcx, %%rcx \n"
1186 "0: \n"
1187 " cmpl %3, %%ecx \n"
1188 " jae 2f \n"
1189 " mov %%rcx, %%rax \n"
1190 " imul $0x44, %%rax, %%rax \n"
1191 " movups (%4,%%rax,), %%xmm5 \n" // xmm5 == gRSPlights[l].{xyzr}
1192 " movups 20(%4,%%rax,), %%xmm1 \n" // xmm1 == gRSPlights[l].{frfgfbfa}
1193 " mulps %%xmm4, %%xmm5 \n"
1194 " movhlps %%xmm5, %%xmm0 \n"
1195 " addps %%xmm5, %%xmm0 \n"
1196 " shufps $0x01, %%xmm0, %%xmm5 \n"
1197 " addps %%xmm5, %%xmm0 \n"
1198 " comiss %6, %%xmm0 \n"
1199 " jc 1f \n"
1200 " shufps $0x00, %%xmm0, %%xmm0 \n"
1201 " mulps %%xmm0, %%xmm1 \n"
1202 " addps %%xmm1, %%xmm3 \n"
1203 "1: \n"
1204 " inc %%rcx \n"
1205 " jmp 0b \n"
1206 "2: \n"
1207 " movss %5, %%xmm0 \n"
1208 " shufps $0x00, %%xmm0, %%xmm0 \n"
1209 " minps %%xmm3, %%xmm0 \n"
1210 " cvtss2si %%xmm0, %%eax \n"
1211 " shll $0x10, %%eax \n"
1212 " orl $0xff000000, %%eax \n"
1213 " shufps $0xe5, %%xmm0, %%xmm0 \n"
1214 " cvtss2si %%xmm0, %%ecx \n"
1215 " shll $8, %%ecx \n"
1216 " orl %%ecx, %%eax \n"
1217 " shufps $0xe6, %%xmm0, %%xmm0 \n"
1218 " cvtss2si %%xmm0, %%ecx \n"
1219 " orl %%ecx, %%eax \n"
1220 : "=&a"(rval)
1221 : "m"(gRSP), "m"(g_normal), "m"(gRSPnumLights), "r"(gRSPlights), "m"(f255), "m"(fZero)
1222 : "%rcx", "memory", "cc", "%xmm0", "%xmm1", "%xmm3", "%xmm4", "%xmm5"
1223 );
1224 return rval;
1225}
1226#elif !defined(NO_ASM) // 32-bit GCC assumed
1227uint32 SSELightVert(void)
1228{
1229 uint32 rval;
1230 float f255 = 255.0, fZero = 0.0;
1231
1232 asm volatile(" movaps %1, %%xmm3 \n"
1233 " movaps %2, %%xmm4 \n"
1234 " xor %%ecx, %%ecx \n"
1235 "0: \n"
1236 " cmpl %3, %%ecx \n"
1237 " jae 2f \n"
1238 " mov %%ecx, %%eax \n"
1239 " imul $0x44, %%eax, %%eax \n"
1240 " movups (%4,%%eax,), %%xmm5 \n"
1241 " movups 20(%4,%%eax,), %%xmm1 \n"
1242 " mulps %%xmm4, %%xmm5 \n"
1243 " movhlps %%xmm5, %%xmm0 \n"
1244 " addps %%xmm5, %%xmm0 \n"
1245 " shufps $0x01, %%xmm0, %%xmm5 \n"
1246 " addps %%xmm5, %%xmm0 \n"
1247 " comiss %6, %%xmm0 \n"
1248 " jc 1f \n"
1249 " shufps $0x00, %%xmm0, %%xmm0 \n"
1250 " mulps %%xmm0, %%xmm1 \n"
1251 " addps %%xmm1, %%xmm3 \n"
1252 "1: \n"
1253 " inc %%ecx \n"
1254 " jmp 0b \n"
1255 "2: \n"
1256 " movss %5, %%xmm0 \n"
1257 " shufps $0x00, %%xmm0, %%xmm0 \n"
1258 " minps %%xmm3, %%xmm0 \n"
1259 " cvtss2si %%xmm0, %%eax \n"
1260 " shll $0x10, %%eax \n"
1261 " orl $0xff000000, %%eax \n"
1262 " shufps $0xe5, %%xmm0, %%xmm0 \n"
1263 " cvtss2si %%xmm0, %%ecx \n"
1264 " shll $8, %%ecx \n"
1265 " orl %%ecx, %%eax \n"
1266 " shufps $0xe6, %%xmm0, %%xmm0 \n"
1267 " cvtss2si %%xmm0, %%ecx \n"
1268 " orl %%ecx, %%eax \n"
1269 : "=&a"(rval)
1270 : "m"(gRSP), "m"(g_normal), "m"(gRSPnumLights), "r"(gRSPlights), "m"(f255), "m"(fZero)
1271 : "%rcx", "memory", "cc", "%xmm0", "%xmm1", "%xmm3", "%xmm4", "%xmm5"
1272 );
1273 return rval;
1274}
1275#endif
1276
1277inline void ReplaceAlphaWithFogFactor(int i)
1278{
1279 if( gRDP.geometryMode & G_FOG )
1280 {
1281 // Use fog factor to replace vertex alpha
1282 if( g_vecProjected[i].z > 1 )
1283 *(((uint8*)&(g_dwVtxDifColor[i]))+3) = 0xFF;
1284 if( g_vecProjected[i].z < 0 )
1285 *(((uint8*)&(g_dwVtxDifColor[i]))+3) = 0;
1286 else
1287 *(((uint8*)&(g_dwVtxDifColor[i]))+3) = (uint8)(g_vecProjected[i].z*255);
1288 }
1289}
1290
1291
1292// Bits
1293// +-+-+-
1294// xxyyzz
1295#define Z_NEG 0x01
1296#define Z_POS 0x02
1297#define Y_NEG 0x04
1298#define Y_POS 0x08
1299#define X_NEG 0x10
1300#define X_POS 0x20
1301
1302// Assumes dwAddr has already been checked!
1303// Don't inline - it's too big with the transform macros
1304
1305#if !defined(NO_ASM)
1306void ProcessVertexDataSSE(uint32 dwAddr, uint32 dwV0, uint32 dwNum)
1307{
1308 UpdateCombinedMatrix();
1309
1310 // This function is called upon SPvertex
1311 // - do vertex matrix transform
1312 // - do vertex lighting
1313 // - do texture cooridinate transform if needed
1314 // - calculate normal vector
1315
1316 // Output: - g_vecProjected[i] -> transformed vertex x,y,z
1317 // - g_vecProjected[i].w -> saved vertex 1/w
1318 // - g_dwVtxFlags[i] -> flags
1319 // - g_dwVtxDifColor[i] -> vertex color
1320 // - g_fVtxTxtCoords[i] -> vertex texture cooridinates
1321
1322 FiddledVtx * pVtxBase = (FiddledVtx*)(g_pRDRAMu8 + dwAddr);
1323 g_pVtxBase = pVtxBase;
1324
1325 for (uint32 i = dwV0; i < dwV0 + dwNum; i++)
1326 {
1327 SP_Timing(RSP_GBI0_Vtx);
1328
1329 FiddledVtx & vert = pVtxBase[i - dwV0];
1330
1331 g_vtxNonTransformed[i].x = (float)vert.x;
1332 g_vtxNonTransformed[i].y = (float)vert.y;
1333 g_vtxNonTransformed[i].z = (float)vert.z;
1334
1335 SSEVec3Transform(i);
1336
1337 if( gRSP.bFogEnabled )
1338 {
1339 g_fFogCoord[i] = g_vecProjected[i].z;
1340 if( g_vecProjected[i].w < 0 || g_vecProjected[i].z < 0 || g_fFogCoord[i] < gRSPfFogMin )
1341 g_fFogCoord[i] = gRSPfFogMin;
1342 }
1343
1344 ReplaceAlphaWithFogFactor(i);
1345
1346
1347 VTX_DUMP(
1348 {
1349 uint32 *dat = (uint32*)(&vert);
1350 DebuggerAppendMsg("vtx %d: %08X %08X %08X %08X", i, dat[0],dat[1],dat[2],dat[3]);
1351 DebuggerAppendMsg(" : %f, %f, %f, %f",
1352 g_vtxTransformed[i].x,g_vtxTransformed[i].y,g_vtxTransformed[i].z,g_vtxTransformed[i].w);
1353 DebuggerAppendMsg(" : %f, %f, %f, %f",
1354 g_vecProjected[i].x,g_vecProjected[i].y,g_vecProjected[i].z,g_vecProjected[i].w);
1355 });
1356
1357 RSP_Vtx_Clipping(i);
1358
1359 if( gRSP.bLightingEnable )
1360 {
1361 g_normal.x = (float)vert.norma.nx;
1362 g_normal.y = (float)vert.norma.ny;
1363 g_normal.z = (float)vert.norma.nz;
1364
1365 SSEVec3TransformNormal();
1366 if( options.enableHackForGames != HACK_FOR_ZELDA_MM )
1367 g_dwVtxDifColor[i] = SSELightVert();
1368 else
1369 g_dwVtxDifColor[i] = LightVert(g_normal, i);
1370 *(((uint8*)&(g_dwVtxDifColor[i]))+3) = vert.rgba.a; // still use alpha from the vertex
1371 }
1372 else
1373 {
1374 if( (gRDP.geometryMode & G_SHADE) == 0 && gRSP.ucode < 5 ) //Shade is disabled
1375 {
1376 //FLAT shade
1377 g_dwVtxDifColor[i] = gRDP.primitiveColor;
1378 }
1379 else
1380 {
1381 register IColor &color = *(IColor*)&g_dwVtxDifColor[i];
1382 color.b = vert.rgba.r;
1383 color.g = vert.rgba.g;
1384 color.r = vert.rgba.b;
1385 color.a = vert.rgba.a;
1386 }
1387 }
1388
1389 if( options.bWinFrameMode )
1390 {
1391 g_dwVtxDifColor[i] = COLOR_RGBA(vert.rgba.r, vert.rgba.g, vert.rgba.b, vert.rgba.a);
1392 }
1393
1394 // Update texture coords n.b. need to divide tu/tv by bogus scale on addition to buffer
1395
1396 // If the vert is already lit, then there is no normal (and hence we
1397 // can't generate tex coord)
1398 if (gRSP.bTextureGen && gRSP.bLightingEnable )
1399 {
1400 TexGen(g_fVtxTxtCoords[i].x, g_fVtxTxtCoords[i].y);
1401 }
1402 else
1403 {
1404 g_fVtxTxtCoords[i].x = (float)vert.tu;
1405 g_fVtxTxtCoords[i].y = (float)vert.tv;
1406 }
1407 }
1408
1409 VTX_DUMP(TRACE2("Setting Vertexes: %d - %d\n", dwV0, dwV0+dwNum-1));
1410 DEBUGGER_PAUSE_AND_DUMP(NEXT_VERTEX_CMD,{TRACE0("Paused at Vertex Cmd");});
1411}
1412#endif
1413
1414void ProcessVertexDataNoSSE(uint32 dwAddr, uint32 dwV0, uint32 dwNum)
1415{
1416
1417 UpdateCombinedMatrix();
1418
1419 // This function is called upon SPvertex
1420 // - do vertex matrix transform
1421 // - do vertex lighting
1422 // - do texture cooridinate transform if needed
1423 // - calculate normal vector
1424
1425 // Output: - g_vecProjected[i] -> transformed vertex x,y,z
1426 // - g_vecProjected[i].w -> saved vertex 1/w
1427 // - g_dwVtxFlags[i] -> flags
1428 // - g_dwVtxDifColor[i] -> vertex color
1429 // - g_fVtxTxtCoords[i] -> vertex texture cooridinates
1430
1431 FiddledVtx * pVtxBase = (FiddledVtx*)(g_pRDRAMu8 + dwAddr);
1432 g_pVtxBase = pVtxBase;
1433
1434 for (uint32 i = dwV0; i < dwV0 + dwNum; i++)
1435 {
1436 SP_Timing(RSP_GBI0_Vtx);
1437
1438 FiddledVtx & vert = pVtxBase[i - dwV0];
1439
1440 g_vtxNonTransformed[i].x = (float)vert.x;
1441 g_vtxNonTransformed[i].y = (float)vert.y;
1442 g_vtxNonTransformed[i].z = (float)vert.z;
1443
1444 Vec3Transform(&g_vtxTransformed[i], (XVECTOR3*)&g_vtxNonTransformed[i], &gRSPworldProject); // Convert to w=1
1445
1446 g_vecProjected[i].w = 1.0f / g_vtxTransformed[i].w;
1447 g_vecProjected[i].x = g_vtxTransformed[i].x * g_vecProjected[i].w;
1448 g_vecProjected[i].y = g_vtxTransformed[i].y * g_vecProjected[i].w;
1449 if ((g_curRomInfo.bPrimaryDepthHack || options.enableHackForGames == HACK_FOR_NASCAR ) && gRDP.otherMode.depth_source )
1450 {
1451 g_vecProjected[i].z = gRDP.fPrimitiveDepth;
1452 g_vtxTransformed[i].z = gRDP.fPrimitiveDepth*g_vtxTransformed[i].w;
1453 }
1454 else
1455 {
1456 g_vecProjected[i].z = g_vtxTransformed[i].z * g_vecProjected[i].w;
1457 }
1458
1459 if( gRSP.bFogEnabled )
1460 {
1461 g_fFogCoord[i] = g_vecProjected[i].z;
1462 if( g_vecProjected[i].w < 0 || g_vecProjected[i].z < 0 || g_fFogCoord[i] < gRSPfFogMin )
1463 g_fFogCoord[i] = gRSPfFogMin;
1464 }
1465
1466 VTX_DUMP(
1467 {
1468 uint32 *dat = (uint32*)(&vert);
1469 DebuggerAppendMsg("vtx %d: %08X %08X %08X %08X", i, dat[0],dat[1],dat[2],dat[3]);
1470 DebuggerAppendMsg(" : %f, %f, %f, %f",
1471 g_vtxTransformed[i].x,g_vtxTransformed[i].y,g_vtxTransformed[i].z,g_vtxTransformed[i].w);
1472 DebuggerAppendMsg(" : %f, %f, %f, %f",
1473 g_vecProjected[i].x,g_vecProjected[i].y,g_vecProjected[i].z,g_vecProjected[i].w);
1474 });
1475
1476 RSP_Vtx_Clipping(i);
1477
1478 if( gRSP.bLightingEnable )
1479 {
1480 g_normal.x = (float)vert.norma.nx;
1481 g_normal.y = (float)vert.norma.ny;
1482 g_normal.z = (float)vert.norma.nz;
1483
1484 Vec3TransformNormal(g_normal, gRSPmodelViewTop);
1485 g_dwVtxDifColor[i] = LightVert(g_normal, i);
1486 *(((uint8*)&(g_dwVtxDifColor[i]))+3) = vert.rgba.a; // still use alpha from the vertex
1487 }
1488 else
1489 {
1490 if( (gRDP.geometryMode & G_SHADE) == 0 && gRSP.ucode < 5 ) //Shade is disabled
1491 {
1492 //FLAT shade
1493 g_dwVtxDifColor[i] = gRDP.primitiveColor;
1494 }
1495 else
1496 {
1497 register IColor &color = *(IColor*)&g_dwVtxDifColor[i];
1498 color.b = vert.rgba.r;
1499 color.g = vert.rgba.g;
1500 color.r = vert.rgba.b;
1501 color.a = vert.rgba.a;
1502 }
1503 }
1504
1505 if( options.bWinFrameMode )
1506 {
1507 g_dwVtxDifColor[i] = COLOR_RGBA(vert.rgba.r, vert.rgba.g, vert.rgba.b, vert.rgba.a);
1508 }
1509
1510 ReplaceAlphaWithFogFactor(i);
1511
1512 // Update texture coords n.b. need to divide tu/tv by bogus scale on addition to buffer
1513
1514 // If the vert is already lit, then there is no normal (and hence we
1515 // can't generate tex coord)
1516 if (gRSP.bTextureGen && gRSP.bLightingEnable )
1517 {
1518 TexGen(g_fVtxTxtCoords[i].x, g_fVtxTxtCoords[i].y);
1519 }
1520 else
1521 {
1522 g_fVtxTxtCoords[i].x = (float)vert.tu;
1523 g_fVtxTxtCoords[i].y = (float)vert.tv;
1524 }
1525 }
1526
1527 VTX_DUMP(TRACE2("Setting Vertexes: %d - %d\n", dwV0, dwV0+dwNum-1));
1528 DEBUGGER_PAUSE_AND_DUMP(NEXT_VERTEX_CMD,{TRACE0("Paused at Vertex Cmd");});
1529}
1530
0764a475 1531extern "C" void pv_neon(XVECTOR4 *g_vtxTransformed, XVECTOR4 *g_vecProjected,
1532 uint32 *g_dwVtxDifColor, VECTOR2 *g_fVtxTxtCoords,
1533 float *g_fFogCoord, uint32 *g_clipFlag2,
1534 uint32 dwNum, const FiddledVtx *vtx,
1535 const Light *gRSPlights, const float *fRSPAmbientLightRGBA,
1536 const XMATRIX *gRSPworldProject, const XMATRIX *gRSPmodelViewTop,
1537 uint32 gRSPnumLights, float gRSPfFogMin);
1538
1539void ProcessVertexDataNEON(uint32 dwAddr, uint32 dwV0, uint32 dwNum)
1540{
1541 if (gRSP.bTextureGen && gRSP.bLightingEnable) {
1542 ProcessVertexDataNoSSE(dwAddr, dwV0,dwNum);
1543 return;
1544 }
1545
1546 // assumtions:
1547 // - g_clipFlag is not used at all
1548 // - g_vtxNonTransformed is not used after ProcessVertexData*() returns
1549 // - g_normal - same
1550
1551#define PV_NEON_ENABLE_LIGHT (1 << 0)
1552#define PV_NEON_ENABLE_SHADE (1 << 1)
1553#define PV_NEON_ENABLE_FOG (1 << 2)
1554#define PV_NEON_FOG_ALPHA (1 << 3)
1555
1556 int neon_state = 0;
1557 if ( gRSP.bLightingEnable )
1558 neon_state |= PV_NEON_ENABLE_LIGHT;
1559 if ( (gRDP.geometryMode & G_SHADE) || gRSP.ucode >= 5 )
1560 neon_state |= PV_NEON_ENABLE_SHADE;
1561 if ( gRSP.bFogEnabled )
1562 neon_state |= PV_NEON_ENABLE_FOG;
1563 if ( gRDP.geometryMode & G_FOG )
1564 neon_state |= PV_NEON_FOG_ALPHA;
1565
1566 uint32 i;
1567
1568 UpdateCombinedMatrix();
1569
1570 // This function is called upon SPvertex
1571 // - do vertex matrix transform
1572 // - do vertex lighting
1573 // - do texture cooridinate transform if needed
1574 // - calculate normal vector
1575
1576 // Output: - g_vecProjected[i] -> transformed vertex x,y,z
1577 // - g_vecProjected[i].w -> saved vertex 1/w
1578 // - g_vtxTransformed[i]
1579 // - g_dwVtxDifColor[i] -> vertex color
1580 // - g_fVtxTxtCoords[i] -> vertex texture cooridinates
1581 // - g_fFogCoord[i]
1582 // - g_clipFlag2[i]
1583
1584 const FiddledVtx * pVtxBase = (const FiddledVtx*)(g_pRDRAMu8 + dwAddr);
1585 g_pVtxBase = (FiddledVtx *)pVtxBase;
1586
1587 // SP_Timing(RSP_GBI0_Vtx);
1588 status.SPCycleCount += Timing_RSP_GBI0_Vtx * dwNum;
1589
1590 if (!(neon_state & (PV_NEON_ENABLE_LIGHT | PV_NEON_ENABLE_SHADE))) {
1591 for (i = dwV0; i < dwV0 + dwNum; i++)
1592 g_dwVtxDifColor[i] = gRDP.primitiveColor; // FLAT shade
1593 }
1594
1595 for (i = dwV0; i < dwV0 + dwNum; i++)
1596 {
1597 const FiddledVtx & vert = pVtxBase[i - dwV0];
1598 XVECTOR3 vtx_raw; // was g_vtxNonTransformed
1599
1600 vtx_raw.x = (float)vert.x;
1601 vtx_raw.y = (float)vert.y;
1602 vtx_raw.z = (float)vert.z;
1603
1604 Vec3Transform(&g_vtxTransformed[i], &vtx_raw, &gRSPworldProject); // Convert to w=1
1605
1606 g_vecProjected[i].w = 1.0f / g_vtxTransformed[i].w;
1607 g_vecProjected[i].x = g_vtxTransformed[i].x * g_vecProjected[i].w;
1608 g_vecProjected[i].y = g_vtxTransformed[i].y * g_vecProjected[i].w;
1609 g_vecProjected[i].z = g_vtxTransformed[i].z * g_vecProjected[i].w;
1610
1611 if( neon_state & PV_NEON_ENABLE_FOG )
1612 {
1613 g_fFogCoord[i] = g_vecProjected[i].z;
1614 if( g_vecProjected[i].w < 0 || g_vecProjected[i].z < 0 || g_fFogCoord[i] < gRSPfFogMin )
1615 g_fFogCoord[i] = gRSPfFogMin;
1616 }
1617
1618 // RSP_Vtx_Clipping(i);
1619 g_clipFlag2[i] = 0;
1620 if( g_vecProjected[i].w > 0 )
1621 {
1622 if( g_vecProjected[i].x > 1 ) g_clipFlag2[i] |= X_CLIP_MAX;
1623 if( g_vecProjected[i].x < -1 ) g_clipFlag2[i] |= X_CLIP_MIN;
1624 if( g_vecProjected[i].y > 1 ) g_clipFlag2[i] |= Y_CLIP_MAX;
1625 if( g_vecProjected[i].y < -1 ) g_clipFlag2[i] |= Y_CLIP_MIN;
1626 }
1627
1628 if( neon_state & PV_NEON_ENABLE_LIGHT )
1629 {
1630 XVECTOR3 normal; // was g_normal
1631 float r, g, b;
1632
1633 normal.x = (float)vert.norma.nx;
1634 normal.y = (float)vert.norma.ny;
1635 normal.z = (float)vert.norma.nz;
1636
1637 Vec3TransformNormal(normal, gRSPmodelViewTop);
1638
1639 r = gRSP.fAmbientLightR;
1640 g = gRSP.fAmbientLightG;
1641 b = gRSP.fAmbientLightB;
1642
1643 for (unsigned int l=0; l < gRSPnumLights; l++)
1644 {
1645 float fCosT = normal.x * gRSPlights[l].x + normal.y * gRSPlights[l].y + normal.z * gRSPlights[l].z;
1646
1647 if (fCosT > 0 )
1648 {
1649 r += gRSPlights[l].fr * fCosT;
1650 g += gRSPlights[l].fg * fCosT;
1651 b += gRSPlights[l].fb * fCosT;
1652 }
1653 }
1654 if (r > 255) r = 255;
1655 if (g > 255) g = 255;
1656 if (b > 255) b = 255;
1657 g_dwVtxDifColor[i] = ((vert.rgba.a<<24)|(((uint32)r)<<16)|(((uint32)g)<<8)|((uint32)b));
1658 }
1659 else if( neon_state & PV_NEON_ENABLE_SHADE )
1660 {
1661 IColor &color = *(IColor*)&g_dwVtxDifColor[i];
1662 color.b = vert.rgba.r;
1663 color.g = vert.rgba.g;
1664 color.r = vert.rgba.b;
1665 color.a = vert.rgba.a;
1666 }
1667
1668 // ReplaceAlphaWithFogFactor(i);
1669 if( neon_state & PV_NEON_FOG_ALPHA )
1670 {
1671 // Use fog factor to replace vertex alpha
1672 if( g_vecProjected[i].z > 1 )
1673 *(((uint8*)&(g_dwVtxDifColor[i]))+3) = 0xFF;
1674 if( g_vecProjected[i].z < 0 )
1675 *(((uint8*)&(g_dwVtxDifColor[i]))+3) = 0;
1676 else
1677 *(((uint8*)&(g_dwVtxDifColor[i]))+3) = (uint8)(g_vecProjected[i].z*255);
1678 }
1679
1680 g_fVtxTxtCoords[i].x = (float)vert.tu;
1681 g_fVtxTxtCoords[i].y = (float)vert.tv;
1682 }
1683}
1684
292f9317 1685bool PrepareTriangle(uint32 dwV0, uint32 dwV1, uint32 dwV2)
1686{
1687 if( status.isVertexShaderEnabled || status.bUseHW_T_L )
1688 {
1689 g_vtxIndex[gRSP.numVertices++] = dwV0;
1690 g_vtxIndex[gRSP.numVertices++] = dwV1;
1691 g_vtxIndex[gRSP.numVertices++] = dwV2;
1692 status.dwNumTrisRendered++;
1693 gRSP.maxVertexID = max(gRSP.maxVertexID,max(dwV0,max(dwV1,dwV2)));
1694 }
1695 else
1696 {
1697 SP_Timing(SP_Each_Triangle);
1698
1699 bool textureFlag = (CRender::g_pRender->IsTextureEnabled() || gRSP.ucode == 6 );
1700 bool openGL = CDeviceBuilder::m_deviceGeneralType == OGL_DEVICE;
1701
1702 InitVertex(dwV0, gRSP.numVertices, textureFlag, openGL);
1703 InitVertex(dwV1, gRSP.numVertices+1, textureFlag, openGL);
1704 InitVertex(dwV2, gRSP.numVertices+2, textureFlag, openGL);
1705
1706 gRSP.numVertices += 3;
1707 status.dwNumTrisRendered++;
1708 }
1709
1710 return true;
1711}
1712
1713
1714
1715// Returns TRUE if it thinks the triangle is visible
1716// Returns FALSE if it is clipped
1717bool IsTriangleVisible(uint32 dwV0, uint32 dwV1, uint32 dwV2)
1718{
1719 //return true; //fix me
1720
1721 if( status.isVertexShaderEnabled || status.bUseHW_T_L ) return true; // We won't have access to transformed vertex data
1722
1723 DEBUGGER_ONLY_IF( (!debuggerEnableTestTris || !debuggerEnableCullFace), {return TRUE;});
1724
1725#ifdef DEBUGGER
1726 // Check vertices are valid!
1727 if (dwV0 >= MAX_VERTS || dwV1 >= MAX_VERTS || dwV2 >= MAX_VERTS)
1728 return false;
1729#endif
1730
1731 // Here we AND all the flags. If any of the bits is set for all
1732 // 3 vertices, it means that all three x, y or z lie outside of
1733 // the current viewing volume.
1734 // Currently disabled - still seems a bit dodgy
1735 if ((gRSP.bCullFront || gRSP.bCullBack) && gRDP.otherMode.zmode != 3)
1736 {
1737 XVECTOR4 & v0 = g_vecProjected[dwV0];
1738 XVECTOR4 & v1 = g_vecProjected[dwV1];
1739 XVECTOR4 & v2 = g_vecProjected[dwV2];
1740
1741 // Only try to clip if the tri is onscreen. For some reason, this
1742 // method doesnt' work well when the z value is outside of screenspace
1743 //if (v0.z < 1 && v1.z < 1 && v2.z < 1)
1744 {
1745 float V1 = v2.x - v0.x;
1746 float V2 = v2.y - v0.y;
1747
1748 float W1 = v2.x - v1.x;
1749 float W2 = v2.y - v1.y;
1750
1751 float fDirection = (V1 * W2) - (V2 * W1);
1752 fDirection = fDirection * v1.w * v2.w * v0.w;
1753 //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;
1754
1755 if (fDirection < 0 && gRSP.bCullBack)
1756 {
1757 status.dwNumTrisClipped++;
1758 return false;
1759 }
1760 else if (fDirection > 0 && gRSP.bCullFront)
1761 {
1762 status.dwNumTrisClipped++;
1763 return false;
1764 }
1765 }
1766 }
1767
1768#ifdef ENABLE_CLIP_TRI
1769 //if( gRSP.bRejectVtx && (g_clipFlag[dwV0]|g_clipFlag[dwV1]|g_clipFlag[dwV2]) )
1770 // return;
1771 if( g_clipFlag2[dwV0]&g_clipFlag2[dwV1]&g_clipFlag2[dwV2] )
1772 {
1773 //DebuggerAppendMsg("Clipped");
1774 return false;
1775 }
1776#endif
1777
1778 return true;
1779}
1780
1781
1782void SetPrimitiveColor(uint32 dwCol, uint32 LODMin, uint32 LODFrac)
1783{
1784 gRDP.colorsAreReloaded = true;
1785 gRDP.primitiveColor = dwCol;
1786 gRDP.primLODMin = LODMin;
1787 gRDP.primLODFrac = LODFrac;
1788 if( gRDP.primLODFrac < gRDP.primLODMin )
1789 {
1790 gRDP.primLODFrac = gRDP.primLODMin;
1791 }
1792
1793 gRDP.fvPrimitiveColor[0] = ((dwCol>>16)&0xFF)/255.0f; //r
1794 gRDP.fvPrimitiveColor[1] = ((dwCol>>8)&0xFF)/255.0f; //g
1795 gRDP.fvPrimitiveColor[2] = ((dwCol)&0xFF)/255.0f; //b
1796 gRDP.fvPrimitiveColor[3] = ((dwCol>>24)&0xFF)/255.0f; //a
1797}
1798
1799void SetPrimitiveDepth(uint32 z, uint32 dwDZ)
1800{
1801 gRDP.primitiveDepth = z & 0x7FFF;
1802 gRDP.fPrimitiveDepth = (float)(gRDP.primitiveDepth)/(float)0x8000;
1803
1804 //gRDP.fPrimitiveDepth = gRDP.fPrimitiveDepth*2-1;
1805 /*
1806 z=0xFFFF -> 1 the farest
1807 z=0 -> -1 the nearest
1808 */
1809
1810 //how to use dwDZ?
1811
1812#ifdef DEBUGGER
1813 if( (pauseAtNext && (eventToPause == NEXT_VERTEX_CMD || eventToPause == NEXT_FLUSH_TRI )) )//&& logTriangles )
1814 {
1815 DebuggerAppendMsg("Set prim Depth: %f, (%08X, %08X)", gRDP.fPrimitiveDepth, z, dwDZ);
1816 }
1817#endif
1818}
1819
1820void SetVertexXYZ(uint32 vertex, float x, float y, float z)
1821{
1822 g_vecProjected[vertex].x = x;
1823 g_vecProjected[vertex].y = y;
1824 g_vecProjected[vertex].z = z;
1825
1826 g_vtxTransformed[vertex].x = x*g_vtxTransformed[vertex].w;
1827 g_vtxTransformed[vertex].y = y*g_vtxTransformed[vertex].w;
1828 g_vtxTransformed[vertex].z = z*g_vtxTransformed[vertex].w;
1829}
1830
1831void ModifyVertexInfo(uint32 where, uint32 vertex, uint32 val)
1832{
1833 switch (where)
1834 {
1835 case RSP_MV_WORD_OFFSET_POINT_RGBA: // Modify RGBA
1836 {
1837 uint32 r = (val>>24)&0xFF;
1838 uint32 g = (val>>16)&0xFF;
1839 uint32 b = (val>>8)&0xFF;
1840 uint32 a = val&0xFF;
1841 g_dwVtxDifColor[vertex] = COLOR_RGBA(r, g, b, a);
1842 LOG_UCODE("Modify vert %d color, 0x%08x", vertex, g_dwVtxDifColor[vertex]);
1843 }
1844 break;
1845 case RSP_MV_WORD_OFFSET_POINT_XYSCREEN: // Modify X,Y
1846 {
1847 uint16 nX = (uint16)(val>>16);
1848 short x = *((short*)&nX);
1849 x /= 4;
1850
1851 uint16 nY = (uint16)(val&0xFFFF);
1852 short y = *((short*)&nY);
1853 y /= 4;
1854
1855 // Should do viewport transform.
1856
1857
1858 x -= windowSetting.uViWidth/2;
1859 y = windowSetting.uViHeight/2-y;
1860
1861 if( options.bEnableHacks && ((*g_GraphicsInfo.VI_X_SCALE_REG)&0xF) != 0 )
1862 {
1863 // Tarzan
1864 // I don't know why Tarzan is different
1865 SetVertexXYZ(vertex, x/windowSetting.fViWidth, y/windowSetting.fViHeight, g_vecProjected[vertex].z);
1866 }
1867 else
1868 {
1869 // Toy Story 2 and other games
1870 SetVertexXYZ(vertex, x*2/windowSetting.fViWidth, y*2/windowSetting.fViHeight, g_vecProjected[vertex].z);
1871 }
1872
1873 LOG_UCODE("Modify vert %d: x=%d, y=%d", vertex, x, y);
1874 VTX_DUMP(TRACE3("Modify vert %d: (%d,%d)", vertex, x, y));
1875 }
1876 break;
1877 case RSP_MV_WORD_OFFSET_POINT_ZSCREEN: // Modify C
1878 {
1879 int z = val>>16;
1880
1881 SetVertexXYZ(vertex, g_vecProjected[vertex].x, g_vecProjected[vertex].y, (((float)z/0x03FF)+0.5f)/2.0f );
1882 LOG_UCODE("Modify vert %d: z=%d", vertex, z);
1883 VTX_DUMP(TRACE2("Modify vert %d: z=%d", vertex, z));
1884 }
1885 break;
1886 case RSP_MV_WORD_OFFSET_POINT_ST: // Texture
1887 {
1888 short tu = short(val>>16);
1889 short tv = short(val & 0xFFFF);
1890 float ftu = tu / 32.0f;
1891 float ftv = tv / 32.0f;
1892 LOG_UCODE(" Setting vertex %d tu/tv to %f, %f", vertex, (float)tu, (float)tv);
1893 CRender::g_pRender->SetVtxTextureCoord(vertex, ftu/gRSP.fTexScaleX, ftv/gRSP.fTexScaleY);
1894 }
1895 break;
1896 }
1897 DEBUGGER_PAUSE_AND_DUMP(NEXT_VERTEX_CMD,{TRACE0("Paused at ModVertex Cmd");});
1898}
1899
1900void ProcessVertexDataDKR(uint32 dwAddr, uint32 dwV0, uint32 dwNum)
1901{
1902 UpdateCombinedMatrix();
1903
1904 long long pVtxBase = (long long) (g_pRDRAMu8 + dwAddr);
1905 g_pVtxBase = (FiddledVtx*)pVtxBase;
1906
1907 Matrix &matWorldProject = gRSP.DKRMatrixes[gRSP.DKRCMatrixIndex];
1908
1909 int nOff;
1910
1911 bool addbase=false;
1912 if ((!gRSP.DKRBillBoard) || (gRSP.DKRCMatrixIndex != 2) )
1913 addbase = false;
1914 else
1915 addbase = true;
1916
1917 if( addbase && gRSP.DKRVtxCount == 0 && dwNum > 1 )
1918 {
1919 gRSP.DKRVtxCount++;
1920 }
1921
1922 LOG_UCODE(" ProcessVertexDataDKR, CMatrix = %d, Add base=%s", gRSP.DKRCMatrixIndex, gRSP.DKRBillBoard?"true":"false");
1923 VTX_DUMP(TRACE2("DKR Setting Vertexes\nCMatrix = %d, Add base=%s", gRSP.DKRCMatrixIndex, gRSP.DKRBillBoard?"true":"false"));
1924
1925 nOff = 0;
1926 uint32 end = dwV0 + dwNum;
1927 for (uint32 i = dwV0; i < end; i++)
1928 {
1929 XVECTOR3 w;
1930
1931 g_vtxNonTransformed[i].x = (float)*(short*)((pVtxBase+nOff + 0) ^ 2);
1932 g_vtxNonTransformed[i].y = (float)*(short*)((pVtxBase+nOff + 2) ^ 2);
1933 g_vtxNonTransformed[i].z = (float)*(short*)((pVtxBase+nOff + 4) ^ 2);
1934
1935 //if( status.isSSEEnabled )
1936 // SSEVec3TransformDKR(g_vtxTransformed[i], g_vtxNonTransformed[i]);
1937 //else
1938 Vec3Transform(&g_vtxTransformed[i], (XVECTOR3*)&g_vtxNonTransformed[i], &matWorldProject); // Convert to w=1
1939
1940 if( gRSP.DKRVtxCount == 0 && dwNum==1 )
1941 {
1942 gRSP.DKRBaseVec.x = g_vtxTransformed[i].x;
1943 gRSP.DKRBaseVec.y = g_vtxTransformed[i].y;
1944 gRSP.DKRBaseVec.z = g_vtxTransformed[i].z;
1945 gRSP.DKRBaseVec.w = g_vtxTransformed[i].w;
1946 }
1947 else if( addbase )
1948 {
1949 g_vtxTransformed[i].x += gRSP.DKRBaseVec.x;
1950 g_vtxTransformed[i].y += gRSP.DKRBaseVec.y;
1951 g_vtxTransformed[i].z += gRSP.DKRBaseVec.z;
1952 g_vtxTransformed[i].w = gRSP.DKRBaseVec.w;
1953 }
1954
1955 g_vecProjected[i].w = 1.0f / g_vtxTransformed[i].w;
1956 g_vecProjected[i].x = g_vtxTransformed[i].x * g_vecProjected[i].w;
1957 g_vecProjected[i].y = g_vtxTransformed[i].y * g_vecProjected[i].w;
1958 g_vecProjected[i].z = g_vtxTransformed[i].z * g_vecProjected[i].w;
1959
1960 gRSP.DKRVtxCount++;
1961
1962 VTX_DUMP(TRACE5("vtx %d: %f, %f, %f, %f", i,
1963 g_vtxTransformed[i].x,g_vtxTransformed[i].y,g_vtxTransformed[i].z,g_vtxTransformed[i].w));
1964
1965 if( gRSP.bFogEnabled )
1966 {
1967 g_fFogCoord[i] = g_vecProjected[i].z;
1968 if( g_vecProjected[i].w < 0 || g_vecProjected[i].z < 0 || g_fFogCoord[i] < gRSPfFogMin )
1969 g_fFogCoord[i] = gRSPfFogMin;
1970 }
1971
1972 RSP_Vtx_Clipping(i);
1973
1974 short wA = *(short*)((pVtxBase+nOff + 6) ^ 2);
1975 short wB = *(short*)((pVtxBase+nOff + 8) ^ 2);
1976
1977 s8 r = (s8)(wA >> 8);
1978 s8 g = (s8)(wA);
1979 s8 b = (s8)(wB >> 8);
1980 s8 a = (s8)(wB);
1981
1982 if (gRSP.bLightingEnable)
1983 {
1984 g_normal.x = (char)r; //norma.nx;
1985 g_normal.y = (char)g; //norma.ny;
1986 g_normal.z = (char)b; //norma.nz;
1987
1988 Vec3TransformNormal(g_normal, matWorldProject)
1989#if !defined(NO_ASM)
1990 if( status.isSSEEnabled )
1991 g_dwVtxDifColor[i] = SSELightVert();
1992 else
1993#endif
1994 g_dwVtxDifColor[i] = LightVert(g_normal, i);
1995 }
1996 else
1997 {
1998 int nR, nG, nB, nA;
1999
2000 nR = r;
2001 nG = g;
2002 nB = b;
2003 nA = a;
2004 // Assign true vert colour after lighting/fogging
2005 g_dwVtxDifColor[i] = COLOR_RGBA(nR, nG, nB, nA);
2006 }
2007
2008 ReplaceAlphaWithFogFactor(i);
2009
2010 g_fVtxTxtCoords[i].x = g_fVtxTxtCoords[i].y = 1;
2011
2012 nOff += 10;
2013 }
2014
2015
2016 DEBUGGER_PAUSE_AND_DUMP(NEXT_VERTEX_CMD,{DebuggerAppendMsg("Paused at DKR Vertex Cmd, v0=%d, vn=%d, addr=%08X", dwV0, dwNum, dwAddr);});
2017}
2018
2019
2020extern uint32 dwPDCIAddr;
2021void ProcessVertexDataPD(uint32 dwAddr, uint32 dwV0, uint32 dwNum)
2022{
2023 UpdateCombinedMatrix();
2024
2025 N64VtxPD * pVtxBase = (N64VtxPD*)(g_pRDRAMu8 + dwAddr);
2026 g_pVtxBase = (FiddledVtx*)pVtxBase; // Fix me
2027
2028 for (uint32 i = dwV0; i < dwV0 + dwNum; i++)
2029 {
2030 N64VtxPD &vert = pVtxBase[i - dwV0];
2031
2032 g_vtxNonTransformed[i].x = (float)vert.x;
2033 g_vtxNonTransformed[i].y = (float)vert.y;
2034 g_vtxNonTransformed[i].z = (float)vert.z;
2035
2036#if !defined(NO_ASM)
2037 if( status.isSSEEnabled )
2038 SSEVec3Transform(i);
2039 else
2040#endif
2041 {
2042 Vec3Transform(&g_vtxTransformed[i], (XVECTOR3*)&g_vtxNonTransformed[i], &gRSPworldProject); // Convert to w=1
2043 g_vecProjected[i].w = 1.0f / g_vtxTransformed[i].w;
2044 g_vecProjected[i].x = g_vtxTransformed[i].x * g_vecProjected[i].w;
2045 g_vecProjected[i].y = g_vtxTransformed[i].y * g_vecProjected[i].w;
2046 g_vecProjected[i].z = g_vtxTransformed[i].z * g_vecProjected[i].w;
2047 }
2048
2049 g_fFogCoord[i] = g_vecProjected[i].z;
2050 if( g_vecProjected[i].w < 0 || g_vecProjected[i].z < 0 || g_fFogCoord[i] < gRSPfFogMin )
2051 g_fFogCoord[i] = gRSPfFogMin;
2052
2053 RSP_Vtx_Clipping(i);
2054
2055 uint8 *addr = g_pRDRAMu8+dwPDCIAddr+ (vert.cidx&0xFF);
2056 uint32 a = addr[0];
2057 uint32 r = addr[3];
2058 uint32 g = addr[2];
2059 uint32 b = addr[1];
2060
2061 if( gRSP.bLightingEnable )
2062 {
2063 g_normal.x = (char)r;
2064 g_normal.y = (char)g;
2065 g_normal.z = (char)b;
2066#if !defined(NO_ASM)
2067 if( status.isSSEEnabled )
2068 {
2069 SSEVec3TransformNormal();
2070 g_dwVtxDifColor[i] = SSELightVert();
2071 }
2072 else
2073#endif
2074 {
2075 Vec3TransformNormal(g_normal, gRSPmodelViewTop);
2076 g_dwVtxDifColor[i] = LightVert(g_normal, i);
2077 }
2078 *(((uint8*)&(g_dwVtxDifColor[i]))+3) = (uint8)a; // still use alpha from the vertex
2079 }
2080 else
2081 {
2082 if( (gRDP.geometryMode & G_SHADE) == 0 && gRSP.ucode < 5 ) //Shade is disabled
2083 {
2084 g_dwVtxDifColor[i] = gRDP.primitiveColor;
2085 }
2086 else //FLAT shade
2087 {
2088 g_dwVtxDifColor[i] = COLOR_RGBA(r, g, b, a);
2089 }
2090 }
2091
2092 if( options.bWinFrameMode )
2093 {
2094 g_dwVtxDifColor[i] = COLOR_RGBA(r, g, b, a);
2095 }
2096
2097 ReplaceAlphaWithFogFactor(i);
2098
2099 VECTOR2 & t = g_fVtxTxtCoords[i];
2100 if (gRSP.bTextureGen && gRSP.bLightingEnable )
2101 {
2102 // Not sure if we should transform the normal here
2103 //Matrix & matWV = gRSP.projectionMtxs[gRSP.projectionMtxTop];
2104 //Vec3TransformNormal(g_normal, matWV);
2105
2106 TexGen(g_fVtxTxtCoords[i].x, g_fVtxTxtCoords[i].y);
2107 }
2108 else
2109 {
2110 t.x = vert.s;
2111 t.y = vert.t;
2112 }
2113
2114
2115 VTX_DUMP(
2116 {
2117 DebuggerAppendMsg("vtx %d: %d %d %d", i, vert.x,vert.y,vert.z);
2118 DebuggerAppendMsg(" : %f, %f, %f, %f",
2119 g_vtxTransformed[i].x,g_vtxTransformed[i].y,g_vtxTransformed[i].z,g_vtxTransformed[i].w);
2120 DebuggerAppendMsg(" : %X, %X, %X, %X", r,g,b,a);
2121 DebuggerAppendMsg(" : u=%f, v=%f", t.x, t.y);
2122 });
2123 }
2124
2125 VTX_DUMP(TRACE2("Setting Vertexes: %d - %d\n", dwV0, dwV0+dwNum-1));
2126 DEBUGGER_PAUSE_AND_DUMP(NEXT_VERTEX_CMD,{TRACE0("Paused at Vertex Cmd");});
2127}
2128
2129extern uint32 dwConkerVtxZAddr;
2130void ProcessVertexDataConker(uint32 dwAddr, uint32 dwV0, uint32 dwNum)
2131{
2132 UpdateCombinedMatrix();
2133
2134 FiddledVtx * pVtxBase = (FiddledVtx*)(g_pRDRAMu8 + dwAddr);
2135 g_pVtxBase = pVtxBase;
2136 //short *vertexColoraddr = (short*)(g_pRDRAMu8+dwConkerVtxZAddr);
2137
2138 for (uint32 i = dwV0; i < dwV0 + dwNum; i++)
2139 {
2140 SP_Timing(RSP_GBI0_Vtx);
2141
2142 FiddledVtx & vert = pVtxBase[i - dwV0];
2143
2144 g_vtxNonTransformed[i].x = (float)vert.x;
2145 g_vtxNonTransformed[i].y = (float)vert.y;
2146 g_vtxNonTransformed[i].z = (float)vert.z;
2147
2148#if !defined(NO_ASM)
2149 if( status.isSSEEnabled )
2150 SSEVec3Transform(i);
2151 else
2152#endif
2153 {
2154 Vec3Transform(&g_vtxTransformed[i], (XVECTOR3*)&g_vtxNonTransformed[i], &gRSPworldProject); // Convert to w=1
2155 g_vecProjected[i].w = 1.0f / g_vtxTransformed[i].w;
2156 g_vecProjected[i].x = g_vtxTransformed[i].x * g_vecProjected[i].w;
2157 g_vecProjected[i].y = g_vtxTransformed[i].y * g_vecProjected[i].w;
2158 g_vecProjected[i].z = g_vtxTransformed[i].z * g_vecProjected[i].w;
2159 }
2160
2161 g_fFogCoord[i] = g_vecProjected[i].z;
2162 if( g_vecProjected[i].w < 0 || g_vecProjected[i].z < 0 || g_fFogCoord[i] < gRSPfFogMin )
2163 g_fFogCoord[i] = gRSPfFogMin;
2164
2165 VTX_DUMP(
2166 {
2167 uint32 *dat = (uint32*)(&vert);
2168 DebuggerAppendMsg("vtx %d: %08X %08X %08X %08X", i, dat[0],dat[1],dat[2],dat[3]);
2169 DebuggerAppendMsg(" : %f, %f, %f, %f",
2170 g_vtxTransformed[i].x,g_vtxTransformed[i].y,g_vtxTransformed[i].z,g_vtxTransformed[i].w);
2171 DebuggerAppendMsg(" : %f, %f, %f, %f",
2172 g_vecProjected[i].x,g_vecProjected[i].y,g_vecProjected[i].z,g_vecProjected[i].w);
2173 });
2174
2175 RSP_Vtx_Clipping(i);
2176
2177 if( gRSP.bLightingEnable )
2178 {
2179 {
2180 uint32 r= ((gRSP.ambientLightColor>>16)&0xFF);
2181 uint32 g= ((gRSP.ambientLightColor>> 8)&0xFF);
2182 uint32 b= ((gRSP.ambientLightColor )&0xFF);
2183 for( uint32 k=1; k<=gRSPnumLights; k++)
2184 {
2185 r += gRSPlights[k].r;
2186 g += gRSPlights[k].g;
2187 b += gRSPlights[k].b;
2188 }
2189 if( r>255 ) r=255;
2190 if( g>255 ) g=255;
2191 if( b>255 ) b=255;
2192 r *= vert.rgba.r ;
2193 g *= vert.rgba.g ;
2194 b *= vert.rgba.b ;
2195 r >>= 8;
2196 g >>= 8;
2197 b >>= 8;
2198 g_dwVtxDifColor[i] = 0xFF000000;
2199 g_dwVtxDifColor[i] |= (r<<16);
2200 g_dwVtxDifColor[i] |= (g<< 8);
2201 g_dwVtxDifColor[i] |= (b );
2202 }
2203
2204 *(((uint8*)&(g_dwVtxDifColor[i]))+3) = vert.rgba.a; // still use alpha from the vertex
2205 }
2206 else
2207 {
2208 if( (gRDP.geometryMode & G_SHADE) == 0 && gRSP.ucode < 5 ) //Shade is disabled
2209 {
2210 g_dwVtxDifColor[i] = gRDP.primitiveColor;
2211 }
2212 else //FLAT shade
2213 {
2214 g_dwVtxDifColor[i] = COLOR_RGBA(vert.rgba.r, vert.rgba.g, vert.rgba.b, vert.rgba.a);
2215 }
2216 }
2217
2218 if( options.bWinFrameMode )
2219 {
2220 //g_vecProjected[i].z = 0;
2221 g_dwVtxDifColor[i] = COLOR_RGBA(vert.rgba.r, vert.rgba.g, vert.rgba.b, vert.rgba.a);
2222 }
2223
2224 ReplaceAlphaWithFogFactor(i);
2225
2226 // Update texture coords n.b. need to divide tu/tv by bogus scale on addition to buffer
2227 //VECTOR2 & t = g_fVtxTxtCoords[i];
2228
2229 // If the vert is already lit, then there is no normal (and hence we
2230 // can't generate tex coord)
2231 if (gRSP.bTextureGen && gRSP.bLightingEnable )
2232 {
2233 g_normal.x = (float)*(char*)(g_pRDRAMu8+ (((i<<1)+0)^3)+dwConkerVtxZAddr);
2234 g_normal.y = (float)*(char*)(g_pRDRAMu8+ (((i<<1)+1)^3)+dwConkerVtxZAddr);
2235 g_normal.z = (float)*(char*)(g_pRDRAMu8+ (((i<<1)+2)^3)+dwConkerVtxZAddr);
2236 Vec3TransformNormal(g_normal, gRSPmodelViewTop);
2237 TexGen(g_fVtxTxtCoords[i].x, g_fVtxTxtCoords[i].y);
2238 }
2239 else
2240 {
2241 g_fVtxTxtCoords[i].x = (float)vert.tu;
2242 g_fVtxTxtCoords[i].y = (float)vert.tv;
2243 }
2244 }
2245
2246 VTX_DUMP(TRACE2("Setting Vertexes: %d - %d\n", dwV0, dwV0+dwNum-1));
2247 DEBUGGER_PAUSE_AND_DUMP(NEXT_VERTEX_CMD,{DebuggerAppendMsg("Paused at Vertex Cmd");});
2248}
2249
2250
2251typedef struct{
2252 short y;
2253 short x;
2254 short flag;
2255 short z;
2256} RS_Vtx_XYZ;
2257
2258typedef union {
2259 struct {
2260 uint8 a;
2261 uint8 b;
2262 uint8 g;
2263 uint8 r;
2264 };
2265 struct {
2266 char na; //a
2267 char nz; //b
2268 char ny; //g
2269 char nx; //r
2270 };
2271} RS_Vtx_Color;
2272
2273
2274void ProcessVertexData_Rogue_Squadron(uint32 dwXYZAddr, uint32 dwColorAddr, uint32 dwXYZCmd, uint32 dwColorCmd)
2275{
2276 UpdateCombinedMatrix();
2277
2278 uint32 dwV0 = 0;
2279 uint32 dwNum = (dwXYZCmd&0xFF00)>>10;
2280
2281 RS_Vtx_XYZ * pVtxXYZBase = (RS_Vtx_XYZ*)(g_pRDRAMu8 + dwXYZAddr);
2282 RS_Vtx_Color * pVtxColorBase = (RS_Vtx_Color*)(g_pRDRAMu8 + dwColorAddr);
2283
2284 uint32 i;
2285 for (i = dwV0; i < dwV0 + dwNum; i++)
2286 {
2287 RS_Vtx_XYZ & vertxyz = pVtxXYZBase[i - dwV0];
2288 RS_Vtx_Color & vertcolors = pVtxColorBase[i - dwV0];
2289
2290 g_vtxNonTransformed[i].x = (float)vertxyz.x;
2291 g_vtxNonTransformed[i].y = (float)vertxyz.y;
2292 g_vtxNonTransformed[i].z = (float)vertxyz.z;
2293
2294#if !defined(NO_ASM)
2295 if( status.isSSEEnabled )
2296 SSEVec3Transform(i);
2297 else
2298#endif
2299 {
2300 Vec3Transform(&g_vtxTransformed[i], (XVECTOR3*)&g_vtxNonTransformed[i], &gRSPworldProject); // Convert to w=1
2301 g_vecProjected[i].w = 1.0f / g_vtxTransformed[i].w;
2302 g_vecProjected[i].x = g_vtxTransformed[i].x * g_vecProjected[i].w;
2303 g_vecProjected[i].y = g_vtxTransformed[i].y * g_vecProjected[i].w;
2304 g_vecProjected[i].z = g_vtxTransformed[i].z * g_vecProjected[i].w;
2305 }
2306
2307 VTX_DUMP(
2308 {
2309 DebuggerAppendMsg(" : %f, %f, %f, %f",
2310 g_vtxTransformed[i].x,g_vtxTransformed[i].y,g_vtxTransformed[i].z,g_vtxTransformed[i].w);
2311 DebuggerAppendMsg(" : %f, %f, %f, %f",
2312 g_vecProjected[i].x,g_vecProjected[i].y,g_vecProjected[i].z,g_vecProjected[i].w);
2313 });
2314
2315 g_fFogCoord[i] = g_vecProjected[i].z;
2316 if( g_vecProjected[i].w < 0 || g_vecProjected[i].z < 0 || g_fFogCoord[i] < gRSPfFogMin )
2317 g_fFogCoord[i] = gRSPfFogMin;
2318
2319 RSP_Vtx_Clipping(i);
2320
2321 if( gRSP.bLightingEnable )
2322 {
2323 g_normal.x = (float)vertcolors.nx;
2324 g_normal.y = (float)vertcolors.ny;
2325 g_normal.z = (float)vertcolors.nz;
2326
2327#if !defined(NO_ASM)
2328 if( status.isSSEEnabled )
2329 {
2330 SSEVec3TransformNormal();
2331 g_dwVtxDifColor[i] = SSELightVert();
2332 }
2333 else
2334#endif
2335 {
2336 Vec3TransformNormal(g_normal, gRSPmodelViewTop);
2337 g_dwVtxDifColor[i] = LightVert(g_normal, i);
2338 }
2339 *(((uint8*)&(g_dwVtxDifColor[i]))+3) = vertcolors.a; // still use alpha from the vertex
2340 }
2341 else
2342 {
2343 if( (gRDP.geometryMode & G_SHADE) == 0 && gRSP.ucode < 5 ) //Shade is disabled
2344 {
2345 g_dwVtxDifColor[i] = gRDP.primitiveColor;
2346 }
2347 else //FLAT shade
2348 {
2349 g_dwVtxDifColor[i] = COLOR_RGBA(vertcolors.r, vertcolors.g, vertcolors.b, vertcolors.a);
2350 }
2351 }
2352
2353 if( options.bWinFrameMode )
2354 {
2355 g_dwVtxDifColor[i] = COLOR_RGBA(vertcolors.r, vertcolors.g, vertcolors.b, vertcolors.a);
2356 }
2357
2358 ReplaceAlphaWithFogFactor(i);
2359
2360 /*
2361 // Update texture coords n.b. need to divide tu/tv by bogus scale on addition to buffer
2362 VECTOR2 & t = g_fVtxTxtCoords[i];
2363
2364 // If the vert is already lit, then there is no normal (and hence we
2365 // can't generate tex coord)
2366 if (gRSP.bTextureGen && gRSP.bLightingEnable && g_textures[gRSP.curTile].m_bTextureEnable )
2367 {
2368 TexGen(g_fVtxTxtCoords[i].x, g_fVtxTxtCoords[i].y);
2369 }
2370 else
2371 {
2372 t.x = (float)vert.tu;
2373 t.y = (float)vert.tv;
2374 }
2375 */
2376 }
2377
2378 VTX_DUMP(TRACE2("Setting Vertexes: %d - %d\n", dwV0, dwV0+dwNum-1));
2379 DEBUGGER_PAUSE_AND_DUMP(NEXT_VERTEX_CMD,{TRACE0("Paused at Vertex Cmd");});
2380}
2381
2382void SetLightCol(uint32 dwLight, uint32 dwCol)
2383{
2384 gRSPlights[dwLight].r = (uint8)((dwCol >> 24)&0xFF);
2385 gRSPlights[dwLight].g = (uint8)((dwCol >> 16)&0xFF);
2386 gRSPlights[dwLight].b = (uint8)((dwCol >> 8)&0xFF);
2387 gRSPlights[dwLight].a = 255; // Ignore light alpha
2388 gRSPlights[dwLight].fr = (float)gRSPlights[dwLight].r;
2389 gRSPlights[dwLight].fg = (float)gRSPlights[dwLight].g;
2390 gRSPlights[dwLight].fb = (float)gRSPlights[dwLight].b;
2391 gRSPlights[dwLight].fa = 255; // Ignore light alpha
2392
2393 //TRACE1("Set light %d color", dwLight);
2394 LIGHT_DUMP(TRACE2("Set Light %d color: %08X", dwLight, dwCol));
2395}
2396
2397void SetLightDirection(uint32 dwLight, float x, float y, float z, float range)
2398{
2399 //gRSP.bLightIsUpdated = true;
2400
2401 //gRSPlights[dwLight].ox = x;
2402 //gRSPlights[dwLight].oy = y;
2403 //gRSPlights[dwLight].oz = z;
2404
2405 register float w = range == 0 ? (float)sqrt(x*x+y*y+z*z) : 1;
2406
2407 gRSPlights[dwLight].x = x/w;
2408 gRSPlights[dwLight].y = y/w;
2409 gRSPlights[dwLight].z = z/w;
2410 gRSPlights[dwLight].range = range;
2411 DEBUGGER_PAUSE_AND_DUMP(NEXT_SET_LIGHT,TRACE5("Set Light %d dir: %.4f, %.4f, %.4f, %.4f", dwLight, x, y, z, range));
2412}
2413
2414static float maxS0, maxT0;
2415static float maxS1, maxT1;
2416static bool validS0, validT0;
2417static bool validS1, validT1;
2418
2419void LogTextureCoords(float fTex0S, float fTex0T, float fTex1S, float fTex1T)
2420{
2421 if( validS0 )
2422 {
2423 if( fTex0S<0 || fTex0S>maxS0 ) validS0 = false;
2424 }
2425 if( validT0 )
2426 {
2427 if( fTex0T<0 || fTex0T>maxT0 ) validT0 = false;
2428 }
2429 if( validS1 )
2430 {
2431 if( fTex1S<0 || fTex1S>maxS1 ) validS1 = false;
2432 }
2433 if( validT1 )
2434 {
2435 if( fTex1T<0 || fTex1T>maxT1 ) validT1 = false;
2436 }
2437}
2438
2439bool CheckTextureCoords(int tex)
2440{
2441 if( tex==0 )
2442 {
2443 return validS0&&validT0;
2444 }
2445 else
2446 {
2447 return validS1&&validT1;
2448 }
2449}
2450
2451void ResetTextureCoordsLog(float maxs0, float maxt0, float maxs1, float maxt1)
2452{
2453 maxS0 = maxs0;
2454 maxT0 = maxt0;
2455 maxS1 = maxs1;
2456 maxT1 = maxt1;
2457 validS0 = validT0 = true;
2458 validS1 = validT1 = true;
2459}
2460
2461void ForceMainTextureIndex(int dwTile)
2462{
2463 if( dwTile == 1 && !(CRender::g_pRender->IsTexel0Enable()) && CRender::g_pRender->IsTexel1Enable() )
2464 {
2465 // Hack
2466 gRSP.curTile = 0;
2467 }
2468 else
2469 {
2470 gRSP.curTile = dwTile;
2471 }
2472}
2473
2474float HackZ2(float z)
2475{
2476 z = (z+9)/10;
2477 return z;
2478}
2479
2480float HackZ(float z)
2481{
2482 return HackZ2(z);
2483
2484 if( z < 0.1 && z >= 0 )
2485 z = (.1f+z)/2;
2486 else if( z < 0 )
2487 //return (10+z)/100;
2488 z = (expf(z)/20);
2489 return z;
2490}
2491
2492void HackZ(std::vector<XVECTOR3>& points)
2493{
2494 int size = points.size();
2495 for( int i=0; i<size; i++)
2496 {
2497 XVECTOR3 &v = points[i];
2498 v.z = (float)HackZ(v.z);
2499 }
2500}
2501
2502void HackZAll()
2503{
2504 if( CDeviceBuilder::m_deviceGeneralType == DIRECTX_DEVICE )
2505 {
2506 for( uint32 i=0; i<gRSP.numVertices; i++)
2507 {
2508 g_vtxBuffer[i].z = HackZ(g_vtxBuffer[i].z);
2509 }
2510 }
2511 else
2512 {
2513 for( uint32 i=0; i<gRSP.numVertices; i++)
2514 {
2515 float w = g_vtxProjected5[i][3];
2516 g_vtxProjected5[i][2] = HackZ(g_vtxProjected5[i][2]/w)*w;
2517 }
2518 }
2519}
2520
2521
2522extern XMATRIX reverseXY;
2523extern XMATRIX reverseY;
2524
2525void UpdateCombinedMatrix()
2526{
2527 if( gRSP.bMatrixIsUpdated )
2528 {
2529 gRSPworldProject = gRSP.modelviewMtxs[gRSP.modelViewMtxTop] * gRSP.projectionMtxs[gRSP.projectionMtxTop];
2530 gRSP.bMatrixIsUpdated = false;
2531 gRSP.bCombinedMatrixIsUpdated = true;
2532 }
2533
2534 if( gRSP.bCombinedMatrixIsUpdated )
2535 {
2536 if( options.enableHackForGames == HACK_REVERSE_XY_COOR )
2537 {
2538 gRSPworldProject = gRSPworldProject * reverseXY;
2539 }
2540 if( options.enableHackForGames == HACK_REVERSE_Y_COOR )
2541 {
2542 gRSPworldProject = gRSPworldProject * reverseY;
2543 }
2544#if !defined(NO_ASM)
2545 if( status.isSSEEnabled )
2546 {
2547 MatrixTranspose(&gRSPworldProjectTransported, &gRSPworldProject);
2548 }
2549#endif
2550 gRSP.bCombinedMatrixIsUpdated = false;
2551 }
2552
2553 //if( gRSP.bWorldMatrixIsUpdated || gRSP.bLightIsUpdated )
2554 //{
2555 // // Update lights with transported world matrix
2556 // for( unsigned int l=0; l<gRSPnumLights; l++)
2557 // {
2558 // Vec3TransformCoord(&gRSPlights[l].td, &gRSPlights[l].od, &gRSPmodelViewTopTranspose);
2559 // Vec3Normalize(&gRSPlights[l].td,&gRSPlights[l].td);
2560 // }
2561
2562 // gRSP.bWorldMatrixIsUpdated = false;
2563 // gRSP.bLightIsUpdated = false;
2564 //}
2565}
2566
Unnamed repository; edit this file 'description' to name the repository.