1 | // Copyright 2016 The SwiftShader Authors. All Rights Reserved. |
2 | // |
3 | // Licensed under the Apache License, Version 2.0 (the "License"); |
4 | // you may not use this file except in compliance with the License. |
5 | // You may obtain a copy of the License at |
6 | // |
7 | // http://www.apache.org/licenses/LICENSE-2.0 |
8 | // |
9 | // Unless required by applicable law or agreed to in writing, software |
10 | // distributed under the License is distributed on an "AS IS" BASIS, |
11 | // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
12 | // See the License for the specific language governing permissions and |
13 | // limitations under the License. |
14 | |
15 | #include "PixelPipeline.hpp" |
16 | #include "SamplerCore.hpp" |
17 | #include "Renderer/Renderer.hpp" |
18 | |
19 | namespace sw |
20 | { |
21 | extern bool postBlendSRGB; |
22 | |
23 | void PixelPipeline::setBuiltins(Int &x, Int &y, Float4(&z)[4], Float4 &w) |
24 | { |
25 | if(state.color[0].component & 0x1) diffuse.x = convertFixed12(v[0].x); else diffuse.x = Short4(0x1000); |
26 | if(state.color[0].component & 0x2) diffuse.y = convertFixed12(v[0].y); else diffuse.y = Short4(0x1000); |
27 | if(state.color[0].component & 0x4) diffuse.z = convertFixed12(v[0].z); else diffuse.z = Short4(0x1000); |
28 | if(state.color[0].component & 0x8) diffuse.w = convertFixed12(v[0].w); else diffuse.w = Short4(0x1000); |
29 | |
30 | if(state.color[1].component & 0x1) specular.x = convertFixed12(v[1].x); else specular.x = Short4(0x0000); |
31 | if(state.color[1].component & 0x2) specular.y = convertFixed12(v[1].y); else specular.y = Short4(0x0000); |
32 | if(state.color[1].component & 0x4) specular.z = convertFixed12(v[1].z); else specular.z = Short4(0x0000); |
33 | if(state.color[1].component & 0x8) specular.w = convertFixed12(v[1].w); else specular.w = Short4(0x0000); |
34 | } |
35 | |
36 | void PixelPipeline::fixedFunction() |
37 | { |
38 | current = diffuse; |
39 | Vector4s temp(0x0000, 0x0000, 0x0000, 0x0000); |
40 | |
41 | for(int stage = 0; stage < 8; stage++) |
42 | { |
43 | if(state.textureStage[stage].stageOperation == TextureStage::STAGE_DISABLE) |
44 | { |
45 | break; |
46 | } |
47 | |
48 | Vector4s texture; |
49 | |
50 | if(state.textureStage[stage].usesTexture) |
51 | { |
52 | texture = sampleTexture(stage, stage); |
53 | } |
54 | |
55 | blendTexture(temp, texture, stage); |
56 | } |
57 | |
58 | specularPixel(current, specular); |
59 | } |
60 | |
61 | void PixelPipeline::applyShader(Int cMask[4]) |
62 | { |
63 | if(!shader) |
64 | { |
65 | fixedFunction(); |
66 | return; |
67 | } |
68 | |
69 | int pad = 0; // Count number of texm3x3pad instructions |
70 | Vector4s dPairing; // Destination for first pairing instruction |
71 | |
72 | for(size_t i = 0; i < shader->getLength(); i++) |
73 | { |
74 | const Shader::Instruction *instruction = shader->getInstruction(i); |
75 | Shader::Opcode opcode = instruction->opcode; |
76 | |
77 | // #ifndef NDEBUG // FIXME: Centralize debug output control |
78 | // shader->printInstruction(i, "debug.txt"); |
79 | // #endif |
80 | |
81 | if(opcode == Shader::OPCODE_DCL || opcode == Shader::OPCODE_DEF || opcode == Shader::OPCODE_DEFI || opcode == Shader::OPCODE_DEFB) |
82 | { |
83 | continue; |
84 | } |
85 | |
86 | const Dst &dst = instruction->dst; |
87 | const Src &src0 = instruction->src[0]; |
88 | const Src &src1 = instruction->src[1]; |
89 | const Src &src2 = instruction->src[2]; |
90 | |
91 | unsigned short shaderModel = shader->getShaderModel(); |
92 | bool pairing = i + 1 < shader->getLength() && shader->getInstruction(i + 1)->coissue; // First instruction of pair |
93 | bool coissue = instruction->coissue; // Second instruction of pair |
94 | |
95 | Vector4s d; |
96 | Vector4s s0; |
97 | Vector4s s1; |
98 | Vector4s s2; |
99 | |
100 | if(src0.type != Shader::PARAMETER_VOID) s0 = fetchRegister(src0); |
101 | if(src1.type != Shader::PARAMETER_VOID) s1 = fetchRegister(src1); |
102 | if(src2.type != Shader::PARAMETER_VOID) s2 = fetchRegister(src2); |
103 | |
104 | Float4 x = shaderModel < 0x0104 ? v[2 + dst.index].x : v[2 + src0.index].x; |
105 | Float4 y = shaderModel < 0x0104 ? v[2 + dst.index].y : v[2 + src0.index].y; |
106 | Float4 z = shaderModel < 0x0104 ? v[2 + dst.index].z : v[2 + src0.index].z; |
107 | Float4 w = shaderModel < 0x0104 ? v[2 + dst.index].w : v[2 + src0.index].w; |
108 | |
109 | switch(opcode) |
110 | { |
111 | case Shader::OPCODE_PS_1_0: break; |
112 | case Shader::OPCODE_PS_1_1: break; |
113 | case Shader::OPCODE_PS_1_2: break; |
114 | case Shader::OPCODE_PS_1_3: break; |
115 | case Shader::OPCODE_PS_1_4: break; |
116 | |
117 | case Shader::OPCODE_DEF: break; |
118 | |
119 | case Shader::OPCODE_NOP: break; |
120 | case Shader::OPCODE_MOV: MOV(d, s0); break; |
121 | case Shader::OPCODE_ADD: ADD(d, s0, s1); break; |
122 | case Shader::OPCODE_SUB: SUB(d, s0, s1); break; |
123 | case Shader::OPCODE_MAD: MAD(d, s0, s1, s2); break; |
124 | case Shader::OPCODE_MUL: MUL(d, s0, s1); break; |
125 | case Shader::OPCODE_DP3: DP3(d, s0, s1); break; |
126 | case Shader::OPCODE_DP4: DP4(d, s0, s1); break; |
127 | case Shader::OPCODE_LRP: LRP(d, s0, s1, s2); break; |
128 | case Shader::OPCODE_TEXCOORD: |
129 | if(shaderModel < 0x0104) |
130 | { |
131 | TEXCOORD(d, x, y, z, dst.index); |
132 | } |
133 | else |
134 | { |
135 | if((src0.swizzle & 0x30) == 0x20) // .xyz |
136 | { |
137 | TEXCRD(d, x, y, z, src0.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW); |
138 | } |
139 | else // .xwy |
140 | { |
141 | TEXCRD(d, x, y, w, src0.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW); |
142 | } |
143 | } |
144 | break; |
145 | case Shader::OPCODE_TEXKILL: |
146 | if(shaderModel < 0x0104) |
147 | { |
148 | TEXKILL(cMask, x, y, z); |
149 | } |
150 | else if(shaderModel == 0x0104) |
151 | { |
152 | if(dst.type == Shader::PARAMETER_TEXTURE) |
153 | { |
154 | TEXKILL(cMask, x, y, z); |
155 | } |
156 | else |
157 | { |
158 | TEXKILL(cMask, rs[dst.index]); |
159 | } |
160 | } |
161 | else ASSERT(false); |
162 | break; |
163 | case Shader::OPCODE_TEX: |
164 | if(shaderModel < 0x0104) |
165 | { |
166 | TEX(d, x, y, z, dst.index, false); |
167 | } |
168 | else if(shaderModel == 0x0104) |
169 | { |
170 | if(src0.type == Shader::PARAMETER_TEXTURE) |
171 | { |
172 | if((src0.swizzle & 0x30) == 0x20) // .xyz |
173 | { |
174 | TEX(d, x, y, z, dst.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW); |
175 | } |
176 | else // .xyw |
177 | { |
178 | TEX(d, x, y, w, dst.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW); |
179 | } |
180 | } |
181 | else |
182 | { |
183 | TEXLD(d, s0, dst.index, src0.modifier == Shader::MODIFIER_DZ || src0.modifier == Shader::MODIFIER_DW); |
184 | } |
185 | } |
186 | else ASSERT(false); |
187 | break; |
188 | case Shader::OPCODE_TEXBEM: TEXBEM(d, s0, x, y, z, dst.index); break; |
189 | case Shader::OPCODE_TEXBEML: TEXBEML(d, s0, x, y, z, dst.index); break; |
190 | case Shader::OPCODE_TEXREG2AR: TEXREG2AR(d, s0, dst.index); break; |
191 | case Shader::OPCODE_TEXREG2GB: TEXREG2GB(d, s0, dst.index); break; |
192 | case Shader::OPCODE_TEXM3X2PAD: TEXM3X2PAD(x, y, z, s0, 0, src0.modifier == Shader::MODIFIER_SIGN); break; |
193 | case Shader::OPCODE_TEXM3X2TEX: TEXM3X2TEX(d, x, y, z, dst.index, s0, src0.modifier == Shader::MODIFIER_SIGN); break; |
194 | case Shader::OPCODE_TEXM3X3PAD: TEXM3X3PAD(x, y, z, s0, pad++ % 2, src0.modifier == Shader::MODIFIER_SIGN); break; |
195 | case Shader::OPCODE_TEXM3X3TEX: TEXM3X3TEX(d, x, y, z, dst.index, s0, src0.modifier == Shader::MODIFIER_SIGN); break; |
196 | case Shader::OPCODE_TEXM3X3SPEC: TEXM3X3SPEC(d, x, y, z, dst.index, s0, s1); break; |
197 | case Shader::OPCODE_TEXM3X3VSPEC: TEXM3X3VSPEC(d, x, y, z, dst.index, s0); break; |
198 | case Shader::OPCODE_CND: CND(d, s0, s1, s2); break; |
199 | case Shader::OPCODE_TEXREG2RGB: TEXREG2RGB(d, s0, dst.index); break; |
200 | case Shader::OPCODE_TEXDP3TEX: TEXDP3TEX(d, x, y, z, dst.index, s0); break; |
201 | case Shader::OPCODE_TEXM3X2DEPTH: TEXM3X2DEPTH(d, x, y, z, s0, src0.modifier == Shader::MODIFIER_SIGN); break; |
202 | case Shader::OPCODE_TEXDP3: TEXDP3(d, x, y, z, s0); break; |
203 | case Shader::OPCODE_TEXM3X3: TEXM3X3(d, x, y, z, s0, src0.modifier == Shader::MODIFIER_SIGN); break; |
204 | case Shader::OPCODE_TEXDEPTH: TEXDEPTH(); break; |
205 | case Shader::OPCODE_CMP0: CMP(d, s0, s1, s2); break; |
206 | case Shader::OPCODE_BEM: BEM(d, s0, s1, dst.index); break; |
207 | case Shader::OPCODE_PHASE: break; |
208 | case Shader::OPCODE_END: break; |
209 | default: |
210 | ASSERT(false); |
211 | } |
212 | |
213 | if(dst.type != Shader::PARAMETER_VOID && opcode != Shader::OPCODE_TEXKILL) |
214 | { |
215 | if(dst.shift > 0) |
216 | { |
217 | if(dst.mask & 0x1) { d.x = AddSat(d.x, d.x); if(dst.shift > 1) d.x = AddSat(d.x, d.x); if(dst.shift > 2) d.x = AddSat(d.x, d.x); } |
218 | if(dst.mask & 0x2) { d.y = AddSat(d.y, d.y); if(dst.shift > 1) d.y = AddSat(d.y, d.y); if(dst.shift > 2) d.y = AddSat(d.y, d.y); } |
219 | if(dst.mask & 0x4) { d.z = AddSat(d.z, d.z); if(dst.shift > 1) d.z = AddSat(d.z, d.z); if(dst.shift > 2) d.z = AddSat(d.z, d.z); } |
220 | if(dst.mask & 0x8) { d.w = AddSat(d.w, d.w); if(dst.shift > 1) d.w = AddSat(d.w, d.w); if(dst.shift > 2) d.w = AddSat(d.w, d.w); } |
221 | } |
222 | else if(dst.shift < 0) |
223 | { |
224 | if(dst.mask & 0x1) d.x = d.x >> -dst.shift; |
225 | if(dst.mask & 0x2) d.y = d.y >> -dst.shift; |
226 | if(dst.mask & 0x4) d.z = d.z >> -dst.shift; |
227 | if(dst.mask & 0x8) d.w = d.w >> -dst.shift; |
228 | } |
229 | |
230 | if(dst.saturate) |
231 | { |
232 | if(dst.mask & 0x1) { d.x = Min(d.x, Short4(0x1000)); d.x = Max(d.x, Short4(0x0000)); } |
233 | if(dst.mask & 0x2) { d.y = Min(d.y, Short4(0x1000)); d.y = Max(d.y, Short4(0x0000)); } |
234 | if(dst.mask & 0x4) { d.z = Min(d.z, Short4(0x1000)); d.z = Max(d.z, Short4(0x0000)); } |
235 | if(dst.mask & 0x8) { d.w = Min(d.w, Short4(0x1000)); d.w = Max(d.w, Short4(0x0000)); } |
236 | } |
237 | |
238 | if(pairing) |
239 | { |
240 | if(dst.mask & 0x1) dPairing.x = d.x; |
241 | if(dst.mask & 0x2) dPairing.y = d.y; |
242 | if(dst.mask & 0x4) dPairing.z = d.z; |
243 | if(dst.mask & 0x8) dPairing.w = d.w; |
244 | } |
245 | |
246 | if(coissue) |
247 | { |
248 | const Dst &dst = shader->getInstruction(i - 1)->dst; |
249 | |
250 | writeDestination(dPairing, dst); |
251 | } |
252 | |
253 | if(!pairing) |
254 | { |
255 | writeDestination(d, dst); |
256 | } |
257 | } |
258 | } |
259 | |
260 | current.x = Min(current.x, Short4(0x0FFF)); current.x = Max(current.x, Short4(0x0000)); |
261 | current.y = Min(current.y, Short4(0x0FFF)); current.y = Max(current.y, Short4(0x0000)); |
262 | current.z = Min(current.z, Short4(0x0FFF)); current.z = Max(current.z, Short4(0x0000)); |
263 | current.w = Min(current.w, Short4(0x0FFF)); current.w = Max(current.w, Short4(0x0000)); |
264 | } |
265 | |
266 | Bool PixelPipeline::alphaTest(Int cMask[4]) |
267 | { |
268 | if(!state.alphaTestActive()) |
269 | { |
270 | return true; |
271 | } |
272 | |
273 | Int aMask; |
274 | |
275 | if(state.transparencyAntialiasing == TRANSPARENCY_NONE) |
276 | { |
277 | PixelRoutine::alphaTest(aMask, current.w); |
278 | |
279 | for(unsigned int q = 0; q < state.multiSample; q++) |
280 | { |
281 | cMask[q] &= aMask; |
282 | } |
283 | } |
284 | else if(state.transparencyAntialiasing == TRANSPARENCY_ALPHA_TO_COVERAGE) |
285 | { |
286 | Float4 alpha = Float4(current.w) * Float4(1.0f / 0x1000); |
287 | |
288 | alphaToCoverage(cMask, alpha); |
289 | } |
290 | else ASSERT(false); |
291 | |
292 | Int pass = cMask[0]; |
293 | |
294 | for(unsigned int q = 1; q < state.multiSample; q++) |
295 | { |
296 | pass = pass | cMask[q]; |
297 | } |
298 | |
299 | return pass != 0x0; |
300 | } |
301 | |
302 | void PixelPipeline::rasterOperation(Float4 &fog, Pointer<Byte> cBuffer[4], Int &x, Int sMask[4], Int zMask[4], Int cMask[4]) |
303 | { |
304 | if(!state.colorWriteActive(0)) |
305 | { |
306 | return; |
307 | } |
308 | |
309 | Vector4f oC; |
310 | |
311 | switch(state.targetFormat[0]) |
312 | { |
313 | case FORMAT_R5G6B5: |
314 | case FORMAT_X8R8G8B8: |
315 | case FORMAT_X8B8G8R8: |
316 | case FORMAT_A8R8G8B8: |
317 | case FORMAT_A8B8G8R8: |
318 | case FORMAT_A8: |
319 | case FORMAT_G16R16: |
320 | case FORMAT_A16B16G16R16: |
321 | if(!postBlendSRGB && state.writeSRGB) |
322 | { |
323 | linearToSRGB12_16(current); |
324 | } |
325 | else |
326 | { |
327 | current.x <<= 4; |
328 | current.y <<= 4; |
329 | current.z <<= 4; |
330 | current.w <<= 4; |
331 | } |
332 | |
333 | if(state.targetFormat[0] == FORMAT_R5G6B5) |
334 | { |
335 | current.x &= Short4(0xF800u); |
336 | current.y &= Short4(0xFC00u); |
337 | current.z &= Short4(0xF800u); |
338 | } |
339 | |
340 | fogBlend(current, fog); |
341 | |
342 | for(unsigned int q = 0; q < state.multiSample; q++) |
343 | { |
344 | Pointer<Byte> buffer = cBuffer[0] + q * *Pointer<Int>(data + OFFSET(DrawData, colorSliceB[0])); |
345 | Vector4s color = current; |
346 | |
347 | if(state.multiSampleMask & (1 << q)) |
348 | { |
349 | alphaBlend(0, buffer, color, x); |
350 | logicOperation(0, buffer, color, x); |
351 | writeColor(0, buffer, x, color, sMask[q], zMask[q], cMask[q]); |
352 | } |
353 | } |
354 | break; |
355 | case FORMAT_R32F: |
356 | case FORMAT_G32R32F: |
357 | case FORMAT_X32B32G32R32F: |
358 | case FORMAT_A32B32G32R32F: |
359 | // case FORMAT_X32B32G32R32F_UNSIGNED: // Not renderable in any fixed-function API. |
360 | convertSigned12(oC, current); |
361 | PixelRoutine::fogBlend(oC, fog); |
362 | |
363 | for(unsigned int q = 0; q < state.multiSample; q++) |
364 | { |
365 | Pointer<Byte> buffer = cBuffer[0] + q * *Pointer<Int>(data + OFFSET(DrawData, colorSliceB[0])); |
366 | Vector4f color = oC; |
367 | |
368 | if(state.multiSampleMask & (1 << q)) |
369 | { |
370 | alphaBlend(0, buffer, color, x); |
371 | writeColor(0, buffer, x, color, sMask[q], zMask[q], cMask[q]); |
372 | } |
373 | } |
374 | break; |
375 | default: |
376 | ASSERT(false); |
377 | } |
378 | } |
379 | |
380 | void PixelPipeline::blendTexture(Vector4s &temp, Vector4s &texture, int stage) |
381 | { |
382 | Vector4s *arg1 = nullptr; |
383 | Vector4s *arg2 = nullptr; |
384 | Vector4s *arg3 = nullptr; |
385 | Vector4s res; |
386 | |
387 | Vector4s constant; |
388 | Vector4s tfactor; |
389 | |
390 | const TextureStage::State &textureStage = state.textureStage[stage]; |
391 | |
392 | if(textureStage.firstArgument == TextureStage::SOURCE_CONSTANT || |
393 | textureStage.firstArgumentAlpha == TextureStage::SOURCE_CONSTANT || |
394 | textureStage.secondArgument == TextureStage::SOURCE_CONSTANT || |
395 | textureStage.secondArgumentAlpha == TextureStage::SOURCE_CONSTANT || |
396 | textureStage.thirdArgument == TextureStage::SOURCE_CONSTANT || |
397 | textureStage.thirdArgumentAlpha == TextureStage::SOURCE_CONSTANT) |
398 | { |
399 | constant.x = *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].constantColor4[0])); |
400 | constant.y = *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].constantColor4[1])); |
401 | constant.z = *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].constantColor4[2])); |
402 | constant.w = *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].constantColor4[3])); |
403 | } |
404 | |
405 | if(textureStage.firstArgument == TextureStage::SOURCE_TFACTOR || |
406 | textureStage.firstArgumentAlpha == TextureStage::SOURCE_TFACTOR || |
407 | textureStage.secondArgument == TextureStage::SOURCE_TFACTOR || |
408 | textureStage.secondArgumentAlpha == TextureStage::SOURCE_TFACTOR || |
409 | textureStage.thirdArgument == TextureStage::SOURCE_TFACTOR || |
410 | textureStage.thirdArgumentAlpha == TextureStage::SOURCE_TFACTOR) |
411 | { |
412 | tfactor.x = *Pointer<Short4>(data + OFFSET(DrawData, factor.textureFactor4[0])); |
413 | tfactor.y = *Pointer<Short4>(data + OFFSET(DrawData, factor.textureFactor4[1])); |
414 | tfactor.z = *Pointer<Short4>(data + OFFSET(DrawData, factor.textureFactor4[2])); |
415 | tfactor.w = *Pointer<Short4>(data + OFFSET(DrawData, factor.textureFactor4[3])); |
416 | } |
417 | |
418 | // Premodulate |
419 | if(stage > 0 && textureStage.usesTexture) |
420 | { |
421 | if(state.textureStage[stage - 1].stageOperation == TextureStage::STAGE_PREMODULATE) |
422 | { |
423 | current.x = MulHigh(current.x, texture.x) << 4; |
424 | current.y = MulHigh(current.y, texture.y) << 4; |
425 | current.z = MulHigh(current.z, texture.z) << 4; |
426 | } |
427 | |
428 | if(state.textureStage[stage - 1].stageOperationAlpha == TextureStage::STAGE_PREMODULATE) |
429 | { |
430 | current.w = MulHigh(current.w, texture.w) << 4; |
431 | } |
432 | } |
433 | |
434 | if(luminance) |
435 | { |
436 | texture.x = MulHigh(texture.x, L) << 4; |
437 | texture.y = MulHigh(texture.y, L) << 4; |
438 | texture.z = MulHigh(texture.z, L) << 4; |
439 | |
440 | luminance = false; |
441 | } |
442 | |
443 | switch(textureStage.firstArgument) |
444 | { |
445 | case TextureStage::SOURCE_TEXTURE: arg1 = &texture; break; |
446 | case TextureStage::SOURCE_CONSTANT: arg1 = &constant; break; |
447 | case TextureStage::SOURCE_CURRENT: arg1 = ¤t; break; |
448 | case TextureStage::SOURCE_DIFFUSE: arg1 = &diffuse; break; |
449 | case TextureStage::SOURCE_SPECULAR: arg1 = &specular; break; |
450 | case TextureStage::SOURCE_TEMP: arg1 = &temp; break; |
451 | case TextureStage::SOURCE_TFACTOR: arg1 = &tfactor; break; |
452 | default: |
453 | ASSERT(false); |
454 | } |
455 | |
456 | switch(textureStage.secondArgument) |
457 | { |
458 | case TextureStage::SOURCE_TEXTURE: arg2 = &texture; break; |
459 | case TextureStage::SOURCE_CONSTANT: arg2 = &constant; break; |
460 | case TextureStage::SOURCE_CURRENT: arg2 = ¤t; break; |
461 | case TextureStage::SOURCE_DIFFUSE: arg2 = &diffuse; break; |
462 | case TextureStage::SOURCE_SPECULAR: arg2 = &specular; break; |
463 | case TextureStage::SOURCE_TEMP: arg2 = &temp; break; |
464 | case TextureStage::SOURCE_TFACTOR: arg2 = &tfactor; break; |
465 | default: |
466 | ASSERT(false); |
467 | } |
468 | |
469 | switch(textureStage.thirdArgument) |
470 | { |
471 | case TextureStage::SOURCE_TEXTURE: arg3 = &texture; break; |
472 | case TextureStage::SOURCE_CONSTANT: arg3 = &constant; break; |
473 | case TextureStage::SOURCE_CURRENT: arg3 = ¤t; break; |
474 | case TextureStage::SOURCE_DIFFUSE: arg3 = &diffuse; break; |
475 | case TextureStage::SOURCE_SPECULAR: arg3 = &specular; break; |
476 | case TextureStage::SOURCE_TEMP: arg3 = &temp; break; |
477 | case TextureStage::SOURCE_TFACTOR: arg3 = &tfactor; break; |
478 | default: |
479 | ASSERT(false); |
480 | } |
481 | |
482 | Vector4s mod1; |
483 | Vector4s mod2; |
484 | Vector4s mod3; |
485 | |
486 | switch(textureStage.firstModifier) |
487 | { |
488 | case TextureStage::MODIFIER_COLOR: |
489 | break; |
490 | case TextureStage::MODIFIER_INVCOLOR: |
491 | mod1.x = SubSat(Short4(0x1000), arg1->x); |
492 | mod1.y = SubSat(Short4(0x1000), arg1->y); |
493 | mod1.z = SubSat(Short4(0x1000), arg1->z); |
494 | mod1.w = SubSat(Short4(0x1000), arg1->w); |
495 | |
496 | arg1 = &mod1; |
497 | break; |
498 | case TextureStage::MODIFIER_ALPHA: |
499 | mod1.x = arg1->w; |
500 | mod1.y = arg1->w; |
501 | mod1.z = arg1->w; |
502 | mod1.w = arg1->w; |
503 | |
504 | arg1 = &mod1; |
505 | break; |
506 | case TextureStage::MODIFIER_INVALPHA: |
507 | mod1.x = SubSat(Short4(0x1000), arg1->w); |
508 | mod1.y = SubSat(Short4(0x1000), arg1->w); |
509 | mod1.z = SubSat(Short4(0x1000), arg1->w); |
510 | mod1.w = SubSat(Short4(0x1000), arg1->w); |
511 | |
512 | arg1 = &mod1; |
513 | break; |
514 | default: |
515 | ASSERT(false); |
516 | } |
517 | |
518 | switch(textureStage.secondModifier) |
519 | { |
520 | case TextureStage::MODIFIER_COLOR: |
521 | break; |
522 | case TextureStage::MODIFIER_INVCOLOR: |
523 | mod2.x = SubSat(Short4(0x1000), arg2->x); |
524 | mod2.y = SubSat(Short4(0x1000), arg2->y); |
525 | mod2.z = SubSat(Short4(0x1000), arg2->z); |
526 | mod2.w = SubSat(Short4(0x1000), arg2->w); |
527 | |
528 | arg2 = &mod2; |
529 | break; |
530 | case TextureStage::MODIFIER_ALPHA: |
531 | mod2.x = arg2->w; |
532 | mod2.y = arg2->w; |
533 | mod2.z = arg2->w; |
534 | mod2.w = arg2->w; |
535 | |
536 | arg2 = &mod2; |
537 | break; |
538 | case TextureStage::MODIFIER_INVALPHA: |
539 | mod2.x = SubSat(Short4(0x1000), arg2->w); |
540 | mod2.y = SubSat(Short4(0x1000), arg2->w); |
541 | mod2.z = SubSat(Short4(0x1000), arg2->w); |
542 | mod2.w = SubSat(Short4(0x1000), arg2->w); |
543 | |
544 | arg2 = &mod2; |
545 | break; |
546 | default: |
547 | ASSERT(false); |
548 | } |
549 | |
550 | switch(textureStage.thirdModifier) |
551 | { |
552 | case TextureStage::MODIFIER_COLOR: |
553 | break; |
554 | case TextureStage::MODIFIER_INVCOLOR: |
555 | mod3.x = SubSat(Short4(0x1000), arg3->x); |
556 | mod3.y = SubSat(Short4(0x1000), arg3->y); |
557 | mod3.z = SubSat(Short4(0x1000), arg3->z); |
558 | mod3.w = SubSat(Short4(0x1000), arg3->w); |
559 | |
560 | arg3 = &mod3; |
561 | break; |
562 | case TextureStage::MODIFIER_ALPHA: |
563 | mod3.x = arg3->w; |
564 | mod3.y = arg3->w; |
565 | mod3.z = arg3->w; |
566 | mod3.w = arg3->w; |
567 | |
568 | arg3 = &mod3; |
569 | break; |
570 | case TextureStage::MODIFIER_INVALPHA: |
571 | mod3.x = SubSat(Short4(0x1000), arg3->w); |
572 | mod3.y = SubSat(Short4(0x1000), arg3->w); |
573 | mod3.z = SubSat(Short4(0x1000), arg3->w); |
574 | mod3.w = SubSat(Short4(0x1000), arg3->w); |
575 | |
576 | arg3 = &mod3; |
577 | break; |
578 | default: |
579 | ASSERT(false); |
580 | } |
581 | |
582 | switch(textureStage.stageOperation) |
583 | { |
584 | case TextureStage::STAGE_DISABLE: |
585 | break; |
586 | case TextureStage::STAGE_SELECTARG1: // Arg1 |
587 | res.x = arg1->x; |
588 | res.y = arg1->y; |
589 | res.z = arg1->z; |
590 | break; |
591 | case TextureStage::STAGE_SELECTARG2: // Arg2 |
592 | res.x = arg2->x; |
593 | res.y = arg2->y; |
594 | res.z = arg2->z; |
595 | break; |
596 | case TextureStage::STAGE_SELECTARG3: // Arg3 |
597 | res.x = arg3->x; |
598 | res.y = arg3->y; |
599 | res.z = arg3->z; |
600 | break; |
601 | case TextureStage::STAGE_MODULATE: // Arg1 * Arg2 |
602 | res.x = MulHigh(arg1->x, arg2->x) << 4; |
603 | res.y = MulHigh(arg1->y, arg2->y) << 4; |
604 | res.z = MulHigh(arg1->z, arg2->z) << 4; |
605 | break; |
606 | case TextureStage::STAGE_MODULATE2X: // Arg1 * Arg2 * 2 |
607 | res.x = MulHigh(arg1->x, arg2->x) << 5; |
608 | res.y = MulHigh(arg1->y, arg2->y) << 5; |
609 | res.z = MulHigh(arg1->z, arg2->z) << 5; |
610 | break; |
611 | case TextureStage::STAGE_MODULATE4X: // Arg1 * Arg2 * 4 |
612 | res.x = MulHigh(arg1->x, arg2->x) << 6; |
613 | res.y = MulHigh(arg1->y, arg2->y) << 6; |
614 | res.z = MulHigh(arg1->z, arg2->z) << 6; |
615 | break; |
616 | case TextureStage::STAGE_ADD: // Arg1 + Arg2 |
617 | res.x = AddSat(arg1->x, arg2->x); |
618 | res.y = AddSat(arg1->y, arg2->y); |
619 | res.z = AddSat(arg1->z, arg2->z); |
620 | break; |
621 | case TextureStage::STAGE_ADDSIGNED: // Arg1 + Arg2 - 0.5 |
622 | res.x = AddSat(arg1->x, arg2->x); |
623 | res.y = AddSat(arg1->y, arg2->y); |
624 | res.z = AddSat(arg1->z, arg2->z); |
625 | |
626 | res.x = SubSat(res.x, Short4(0x0800)); |
627 | res.y = SubSat(res.y, Short4(0x0800)); |
628 | res.z = SubSat(res.z, Short4(0x0800)); |
629 | break; |
630 | case TextureStage::STAGE_ADDSIGNED2X: // (Arg1 + Arg2 - 0.5) << 1 |
631 | res.x = AddSat(arg1->x, arg2->x); |
632 | res.y = AddSat(arg1->y, arg2->y); |
633 | res.z = AddSat(arg1->z, arg2->z); |
634 | |
635 | res.x = SubSat(res.x, Short4(0x0800)); |
636 | res.y = SubSat(res.y, Short4(0x0800)); |
637 | res.z = SubSat(res.z, Short4(0x0800)); |
638 | |
639 | res.x = AddSat(res.x, res.x); |
640 | res.y = AddSat(res.y, res.y); |
641 | res.z = AddSat(res.z, res.z); |
642 | break; |
643 | case TextureStage::STAGE_SUBTRACT: // Arg1 - Arg2 |
644 | res.x = SubSat(arg1->x, arg2->x); |
645 | res.y = SubSat(arg1->y, arg2->y); |
646 | res.z = SubSat(arg1->z, arg2->z); |
647 | break; |
648 | case TextureStage::STAGE_ADDSMOOTH: // Arg1 + Arg2 - Arg1 * Arg2 |
649 | { |
650 | Short4 tmp; |
651 | |
652 | tmp = MulHigh(arg1->x, arg2->x) << 4; res.x = AddSat(arg1->x, arg2->x); res.x = SubSat(res.x, tmp); |
653 | tmp = MulHigh(arg1->y, arg2->y) << 4; res.y = AddSat(arg1->y, arg2->y); res.y = SubSat(res.y, tmp); |
654 | tmp = MulHigh(arg1->z, arg2->z) << 4; res.z = AddSat(arg1->z, arg2->z); res.z = SubSat(res.z, tmp); |
655 | } |
656 | break; |
657 | case TextureStage::STAGE_MULTIPLYADD: // Arg3 + Arg1 * Arg2 |
658 | res.x = MulHigh(arg1->x, arg2->x) << 4; res.x = AddSat(res.x, arg3->x); |
659 | res.y = MulHigh(arg1->y, arg2->y) << 4; res.y = AddSat(res.y, arg3->y); |
660 | res.z = MulHigh(arg1->z, arg2->z) << 4; res.z = AddSat(res.z, arg3->z); |
661 | break; |
662 | case TextureStage::STAGE_LERP: // Arg3 * (Arg1 - Arg2) + Arg2 |
663 | res.x = SubSat(arg1->x, arg2->x); res.x = MulHigh(res.x, arg3->x) << 4; res.x = AddSat(res.x, arg2->x); |
664 | res.y = SubSat(arg1->y, arg2->y); res.y = MulHigh(res.y, arg3->y) << 4; res.y = AddSat(res.y, arg2->y); |
665 | res.z = SubSat(arg1->z, arg2->z); res.z = MulHigh(res.z, arg3->z) << 4; res.z = AddSat(res.z, arg2->z); |
666 | break; |
667 | case TextureStage::STAGE_DOT3: // 2 * (Arg1.x - 0.5) * 2 * (Arg2.x - 0.5) + 2 * (Arg1.y - 0.5) * 2 * (Arg2.y - 0.5) + 2 * (Arg1.z - 0.5) * 2 * (Arg2.z - 0.5) |
668 | { |
669 | Short4 tmp; |
670 | |
671 | res.x = SubSat(arg1->x, Short4(0x0800)); tmp = SubSat(arg2->x, Short4(0x0800)); res.x = MulHigh(res.x, tmp); |
672 | res.y = SubSat(arg1->y, Short4(0x0800)); tmp = SubSat(arg2->y, Short4(0x0800)); res.y = MulHigh(res.y, tmp); |
673 | res.z = SubSat(arg1->z, Short4(0x0800)); tmp = SubSat(arg2->z, Short4(0x0800)); res.z = MulHigh(res.z, tmp); |
674 | |
675 | res.x = res.x << 6; |
676 | res.y = res.y << 6; |
677 | res.z = res.z << 6; |
678 | |
679 | res.x = AddSat(res.x, res.y); |
680 | res.x = AddSat(res.x, res.z); |
681 | |
682 | // Clamp to [0, 1] |
683 | res.x = Max(res.x, Short4(0x0000)); |
684 | res.x = Min(res.x, Short4(0x1000)); |
685 | |
686 | res.y = res.x; |
687 | res.z = res.x; |
688 | res.w = res.x; |
689 | } |
690 | break; |
691 | case TextureStage::STAGE_BLENDCURRENTALPHA: // Alpha * (Arg1 - Arg2) + Arg2 |
692 | res.x = SubSat(arg1->x, arg2->x); res.x = MulHigh(res.x, current.w) << 4; res.x = AddSat(res.x, arg2->x); |
693 | res.y = SubSat(arg1->y, arg2->y); res.y = MulHigh(res.y, current.w) << 4; res.y = AddSat(res.y, arg2->y); |
694 | res.z = SubSat(arg1->z, arg2->z); res.z = MulHigh(res.z, current.w) << 4; res.z = AddSat(res.z, arg2->z); |
695 | break; |
696 | case TextureStage::STAGE_BLENDDIFFUSEALPHA: // Alpha * (Arg1 - Arg2) + Arg2 |
697 | res.x = SubSat(arg1->x, arg2->x); res.x = MulHigh(res.x, diffuse.w) << 4; res.x = AddSat(res.x, arg2->x); |
698 | res.y = SubSat(arg1->y, arg2->y); res.y = MulHigh(res.y, diffuse.w) << 4; res.y = AddSat(res.y, arg2->y); |
699 | res.z = SubSat(arg1->z, arg2->z); res.z = MulHigh(res.z, diffuse.w) << 4; res.z = AddSat(res.z, arg2->z); |
700 | break; |
701 | case TextureStage::STAGE_BLENDFACTORALPHA: // Alpha * (Arg1 - Arg2) + Arg2 |
702 | res.x = SubSat(arg1->x, arg2->x); res.x = MulHigh(res.x, *Pointer<Short4>(data + OFFSET(DrawData, factor.textureFactor4[3]))) << 4; res.x = AddSat(res.x, arg2->x); |
703 | res.y = SubSat(arg1->y, arg2->y); res.y = MulHigh(res.y, *Pointer<Short4>(data + OFFSET(DrawData, factor.textureFactor4[3]))) << 4; res.y = AddSat(res.y, arg2->y); |
704 | res.z = SubSat(arg1->z, arg2->z); res.z = MulHigh(res.z, *Pointer<Short4>(data + OFFSET(DrawData, factor.textureFactor4[3]))) << 4; res.z = AddSat(res.z, arg2->z); |
705 | break; |
706 | case TextureStage::STAGE_BLENDTEXTUREALPHA: // Alpha * (Arg1 - Arg2) + Arg2 |
707 | res.x = SubSat(arg1->x, arg2->x); res.x = MulHigh(res.x, texture.w) << 4; res.x = AddSat(res.x, arg2->x); |
708 | res.y = SubSat(arg1->y, arg2->y); res.y = MulHigh(res.y, texture.w) << 4; res.y = AddSat(res.y, arg2->y); |
709 | res.z = SubSat(arg1->z, arg2->z); res.z = MulHigh(res.z, texture.w) << 4; res.z = AddSat(res.z, arg2->z); |
710 | break; |
711 | case TextureStage::STAGE_BLENDTEXTUREALPHAPM: // Arg1 + Arg2 * (1 - Alpha) |
712 | res.x = SubSat(Short4(0x1000), texture.w); res.x = MulHigh(res.x, arg2->x) << 4; res.x = AddSat(res.x, arg1->x); |
713 | res.y = SubSat(Short4(0x1000), texture.w); res.y = MulHigh(res.y, arg2->y) << 4; res.y = AddSat(res.y, arg1->y); |
714 | res.z = SubSat(Short4(0x1000), texture.w); res.z = MulHigh(res.z, arg2->z) << 4; res.z = AddSat(res.z, arg1->z); |
715 | break; |
716 | case TextureStage::STAGE_PREMODULATE: |
717 | res.x = arg1->x; |
718 | res.y = arg1->y; |
719 | res.z = arg1->z; |
720 | break; |
721 | case TextureStage::STAGE_MODULATEALPHA_ADDCOLOR: // Arg1 + Arg1.w * Arg2 |
722 | res.x = MulHigh(arg1->w, arg2->x) << 4; res.x = AddSat(res.x, arg1->x); |
723 | res.y = MulHigh(arg1->w, arg2->y) << 4; res.y = AddSat(res.y, arg1->y); |
724 | res.z = MulHigh(arg1->w, arg2->z) << 4; res.z = AddSat(res.z, arg1->z); |
725 | break; |
726 | case TextureStage::STAGE_MODULATECOLOR_ADDALPHA: // Arg1 * Arg2 + Arg1.w |
727 | res.x = MulHigh(arg1->x, arg2->x) << 4; res.x = AddSat(res.x, arg1->w); |
728 | res.y = MulHigh(arg1->y, arg2->y) << 4; res.y = AddSat(res.y, arg1->w); |
729 | res.z = MulHigh(arg1->z, arg2->z) << 4; res.z = AddSat(res.z, arg1->w); |
730 | break; |
731 | case TextureStage::STAGE_MODULATEINVALPHA_ADDCOLOR: // (1 - Arg1.w) * Arg2 + Arg1 |
732 | { |
733 | Short4 tmp; |
734 | |
735 | res.x = AddSat(arg1->x, arg2->x); tmp = MulHigh(arg1->w, arg2->x) << 4; res.x = SubSat(res.x, tmp); |
736 | res.y = AddSat(arg1->y, arg2->y); tmp = MulHigh(arg1->w, arg2->y) << 4; res.y = SubSat(res.y, tmp); |
737 | res.z = AddSat(arg1->z, arg2->z); tmp = MulHigh(arg1->w, arg2->z) << 4; res.z = SubSat(res.z, tmp); |
738 | } |
739 | break; |
740 | case TextureStage::STAGE_MODULATEINVCOLOR_ADDALPHA: // (1 - Arg1) * Arg2 + Arg1.w |
741 | { |
742 | Short4 tmp; |
743 | |
744 | res.x = AddSat(arg1->w, arg2->x); tmp = MulHigh(arg1->x, arg2->x) << 4; res.x = SubSat(res.x, tmp); |
745 | res.y = AddSat(arg1->w, arg2->y); tmp = MulHigh(arg1->y, arg2->y) << 4; res.y = SubSat(res.y, tmp); |
746 | res.z = AddSat(arg1->w, arg2->z); tmp = MulHigh(arg1->z, arg2->z) << 4; res.z = SubSat(res.z, tmp); |
747 | } |
748 | break; |
749 | case TextureStage::STAGE_BUMPENVMAP: |
750 | { |
751 | du = Float4(texture.x) * Float4(1.0f / 0x0FE0); |
752 | dv = Float4(texture.y) * Float4(1.0f / 0x0FE0); |
753 | |
754 | Float4 du2; |
755 | Float4 dv2; |
756 | |
757 | du2 = du; |
758 | dv2 = dv; |
759 | du *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][0])); |
760 | dv2 *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][0])); |
761 | du += dv2; |
762 | dv *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][1])); |
763 | du2 *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][1])); |
764 | dv += du2; |
765 | |
766 | perturbate = true; |
767 | |
768 | res.x = current.x; |
769 | res.y = current.y; |
770 | res.z = current.z; |
771 | res.w = current.w; |
772 | } |
773 | break; |
774 | case TextureStage::STAGE_BUMPENVMAPLUMINANCE: |
775 | { |
776 | du = Float4(texture.x) * Float4(1.0f / 0x0FE0); |
777 | dv = Float4(texture.y) * Float4(1.0f / 0x0FE0); |
778 | |
779 | Float4 du2; |
780 | Float4 dv2; |
781 | |
782 | du2 = du; |
783 | dv2 = dv; |
784 | |
785 | du *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][0])); |
786 | dv2 *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][0])); |
787 | du += dv2; |
788 | dv *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][1])); |
789 | du2 *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][1])); |
790 | dv += du2; |
791 | |
792 | perturbate = true; |
793 | |
794 | L = texture.z; |
795 | L = MulHigh(L, *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].luminanceScale4))); |
796 | L = L << 4; |
797 | L = AddSat(L, *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].luminanceOffset4))); |
798 | L = Max(L, Short4(0x0000)); |
799 | L = Min(L, Short4(0x1000)); |
800 | |
801 | luminance = true; |
802 | |
803 | res.x = current.x; |
804 | res.y = current.y; |
805 | res.z = current.z; |
806 | res.w = current.w; |
807 | } |
808 | break; |
809 | default: |
810 | ASSERT(false); |
811 | } |
812 | |
813 | if(textureStage.stageOperation != TextureStage::STAGE_DOT3) |
814 | { |
815 | switch(textureStage.firstArgumentAlpha) |
816 | { |
817 | case TextureStage::SOURCE_TEXTURE: arg1 = &texture; break; |
818 | case TextureStage::SOURCE_CONSTANT: arg1 = &constant; break; |
819 | case TextureStage::SOURCE_CURRENT: arg1 = ¤t; break; |
820 | case TextureStage::SOURCE_DIFFUSE: arg1 = &diffuse; break; |
821 | case TextureStage::SOURCE_SPECULAR: arg1 = &specular; break; |
822 | case TextureStage::SOURCE_TEMP: arg1 = &temp; break; |
823 | case TextureStage::SOURCE_TFACTOR: arg1 = &tfactor; break; |
824 | default: |
825 | ASSERT(false); |
826 | } |
827 | |
828 | switch(textureStage.secondArgumentAlpha) |
829 | { |
830 | case TextureStage::SOURCE_TEXTURE: arg2 = &texture; break; |
831 | case TextureStage::SOURCE_CONSTANT: arg2 = &constant; break; |
832 | case TextureStage::SOURCE_CURRENT: arg2 = ¤t; break; |
833 | case TextureStage::SOURCE_DIFFUSE: arg2 = &diffuse; break; |
834 | case TextureStage::SOURCE_SPECULAR: arg2 = &specular; break; |
835 | case TextureStage::SOURCE_TEMP: arg2 = &temp; break; |
836 | case TextureStage::SOURCE_TFACTOR: arg2 = &tfactor; break; |
837 | default: |
838 | ASSERT(false); |
839 | } |
840 | |
841 | switch(textureStage.thirdArgumentAlpha) |
842 | { |
843 | case TextureStage::SOURCE_TEXTURE: arg3 = &texture; break; |
844 | case TextureStage::SOURCE_CONSTANT: arg3 = &constant; break; |
845 | case TextureStage::SOURCE_CURRENT: arg3 = ¤t; break; |
846 | case TextureStage::SOURCE_DIFFUSE: arg3 = &diffuse; break; |
847 | case TextureStage::SOURCE_SPECULAR: arg3 = &specular; break; |
848 | case TextureStage::SOURCE_TEMP: arg3 = &temp; break; |
849 | case TextureStage::SOURCE_TFACTOR: arg3 = &tfactor; break; |
850 | default: |
851 | ASSERT(false); |
852 | } |
853 | |
854 | switch(textureStage.firstModifierAlpha) // FIXME: Check if actually used |
855 | { |
856 | case TextureStage::MODIFIER_COLOR: |
857 | break; |
858 | case TextureStage::MODIFIER_INVCOLOR: |
859 | mod1.w = SubSat(Short4(0x1000), arg1->w); |
860 | |
861 | arg1 = &mod1; |
862 | break; |
863 | case TextureStage::MODIFIER_ALPHA: |
864 | // Redudant |
865 | break; |
866 | case TextureStage::MODIFIER_INVALPHA: |
867 | mod1.w = SubSat(Short4(0x1000), arg1->w); |
868 | |
869 | arg1 = &mod1; |
870 | break; |
871 | default: |
872 | ASSERT(false); |
873 | } |
874 | |
875 | switch(textureStage.secondModifierAlpha) // FIXME: Check if actually used |
876 | { |
877 | case TextureStage::MODIFIER_COLOR: |
878 | break; |
879 | case TextureStage::MODIFIER_INVCOLOR: |
880 | mod2.w = SubSat(Short4(0x1000), arg2->w); |
881 | |
882 | arg2 = &mod2; |
883 | break; |
884 | case TextureStage::MODIFIER_ALPHA: |
885 | // Redudant |
886 | break; |
887 | case TextureStage::MODIFIER_INVALPHA: |
888 | mod2.w = SubSat(Short4(0x1000), arg2->w); |
889 | |
890 | arg2 = &mod2; |
891 | break; |
892 | default: |
893 | ASSERT(false); |
894 | } |
895 | |
896 | switch(textureStage.thirdModifierAlpha) // FIXME: Check if actually used |
897 | { |
898 | case TextureStage::MODIFIER_COLOR: |
899 | break; |
900 | case TextureStage::MODIFIER_INVCOLOR: |
901 | mod3.w = SubSat(Short4(0x1000), arg3->w); |
902 | |
903 | arg3 = &mod3; |
904 | break; |
905 | case TextureStage::MODIFIER_ALPHA: |
906 | // Redudant |
907 | break; |
908 | case TextureStage::MODIFIER_INVALPHA: |
909 | mod3.w = SubSat(Short4(0x1000), arg3->w); |
910 | |
911 | arg3 = &mod3; |
912 | break; |
913 | default: |
914 | ASSERT(false); |
915 | } |
916 | |
917 | switch(textureStage.stageOperationAlpha) |
918 | { |
919 | case TextureStage::STAGE_DISABLE: |
920 | break; |
921 | case TextureStage::STAGE_SELECTARG1: // Arg1 |
922 | res.w = arg1->w; |
923 | break; |
924 | case TextureStage::STAGE_SELECTARG2: // Arg2 |
925 | res.w = arg2->w; |
926 | break; |
927 | case TextureStage::STAGE_SELECTARG3: // Arg3 |
928 | res.w = arg3->w; |
929 | break; |
930 | case TextureStage::STAGE_MODULATE: // Arg1 * Arg2 |
931 | res.w = MulHigh(arg1->w, arg2->w) << 4; |
932 | break; |
933 | case TextureStage::STAGE_MODULATE2X: // Arg1 * Arg2 * 2 |
934 | res.w = MulHigh(arg1->w, arg2->w) << 5; |
935 | break; |
936 | case TextureStage::STAGE_MODULATE4X: // Arg1 * Arg2 * 4 |
937 | res.w = MulHigh(arg1->w, arg2->w) << 6; |
938 | break; |
939 | case TextureStage::STAGE_ADD: // Arg1 + Arg2 |
940 | res.w = AddSat(arg1->w, arg2->w); |
941 | break; |
942 | case TextureStage::STAGE_ADDSIGNED: // Arg1 + Arg2 - 0.5 |
943 | res.w = AddSat(arg1->w, arg2->w); |
944 | res.w = SubSat(res.w, Short4(0x0800)); |
945 | break; |
946 | case TextureStage::STAGE_ADDSIGNED2X: // (Arg1 + Arg2 - 0.5) << 1 |
947 | res.w = AddSat(arg1->w, arg2->w); |
948 | res.w = SubSat(res.w, Short4(0x0800)); |
949 | res.w = AddSat(res.w, res.w); |
950 | break; |
951 | case TextureStage::STAGE_SUBTRACT: // Arg1 - Arg2 |
952 | res.w = SubSat(arg1->w, arg2->w); |
953 | break; |
954 | case TextureStage::STAGE_ADDSMOOTH: // Arg1 + Arg2 - Arg1 * Arg2 |
955 | { |
956 | Short4 tmp; |
957 | |
958 | tmp = MulHigh(arg1->w, arg2->w) << 4; res.w = AddSat(arg1->w, arg2->w); res.w = SubSat(res.w, tmp); |
959 | } |
960 | break; |
961 | case TextureStage::STAGE_MULTIPLYADD: // Arg3 + Arg1 * Arg2 |
962 | res.w = MulHigh(arg1->w, arg2->w) << 4; res.w = AddSat(res.w, arg3->w); |
963 | break; |
964 | case TextureStage::STAGE_LERP: // Arg3 * (Arg1 - Arg2) + Arg2 |
965 | res.w = SubSat(arg1->w, arg2->w); res.w = MulHigh(res.w, arg3->w) << 4; res.w = AddSat(res.w, arg2->w); |
966 | break; |
967 | case TextureStage::STAGE_DOT3: |
968 | break; // Already computed in color channel |
969 | case TextureStage::STAGE_BLENDCURRENTALPHA: // Alpha * (Arg1 - Arg2) + Arg2 |
970 | res.w = SubSat(arg1->w, arg2->w); res.w = MulHigh(res.w, current.w) << 4; res.w = AddSat(res.w, arg2->w); |
971 | break; |
972 | case TextureStage::STAGE_BLENDDIFFUSEALPHA: // Arg1 * (Alpha) + Arg2 * (1 - Alpha) |
973 | res.w = SubSat(arg1->w, arg2->w); res.w = MulHigh(res.w, diffuse.w) << 4; res.w = AddSat(res.w, arg2->w); |
974 | break; |
975 | case TextureStage::STAGE_BLENDFACTORALPHA: |
976 | res.w = SubSat(arg1->w, arg2->w); res.w = MulHigh(res.w, *Pointer<Short4>(data + OFFSET(DrawData, factor.textureFactor4[3]))) << 4; res.w = AddSat(res.w, arg2->w); |
977 | break; |
978 | case TextureStage::STAGE_BLENDTEXTUREALPHA: // Arg1 * (Alpha) + Arg2 * (1 - Alpha) |
979 | res.w = SubSat(arg1->w, arg2->w); res.w = MulHigh(res.w, texture.w) << 4; res.w = AddSat(res.w, arg2->w); |
980 | break; |
981 | case TextureStage::STAGE_BLENDTEXTUREALPHAPM: // Arg1 + Arg2 * (1 - Alpha) |
982 | res.w = SubSat(Short4(0x1000), texture.w); res.w = MulHigh(res.w, arg2->w) << 4; res.w = AddSat(res.w, arg1->w); |
983 | break; |
984 | case TextureStage::STAGE_PREMODULATE: |
985 | res.w = arg1->w; |
986 | break; |
987 | case TextureStage::STAGE_MODULATEALPHA_ADDCOLOR: |
988 | case TextureStage::STAGE_MODULATECOLOR_ADDALPHA: |
989 | case TextureStage::STAGE_MODULATEINVALPHA_ADDCOLOR: |
990 | case TextureStage::STAGE_MODULATEINVCOLOR_ADDALPHA: |
991 | case TextureStage::STAGE_BUMPENVMAP: |
992 | case TextureStage::STAGE_BUMPENVMAPLUMINANCE: |
993 | break; // Invalid alpha operations |
994 | default: |
995 | ASSERT(false); |
996 | } |
997 | } |
998 | |
999 | // Clamp result to [0, 1] |
1000 | |
1001 | switch(textureStage.stageOperation) |
1002 | { |
1003 | case TextureStage::STAGE_DISABLE: |
1004 | case TextureStage::STAGE_SELECTARG1: |
1005 | case TextureStage::STAGE_SELECTARG2: |
1006 | case TextureStage::STAGE_SELECTARG3: |
1007 | case TextureStage::STAGE_MODULATE: |
1008 | case TextureStage::STAGE_MODULATE2X: |
1009 | case TextureStage::STAGE_MODULATE4X: |
1010 | case TextureStage::STAGE_ADD: |
1011 | case TextureStage::STAGE_MULTIPLYADD: |
1012 | case TextureStage::STAGE_LERP: |
1013 | case TextureStage::STAGE_BLENDCURRENTALPHA: |
1014 | case TextureStage::STAGE_BLENDDIFFUSEALPHA: |
1015 | case TextureStage::STAGE_BLENDFACTORALPHA: |
1016 | case TextureStage::STAGE_BLENDTEXTUREALPHA: |
1017 | case TextureStage::STAGE_BLENDTEXTUREALPHAPM: |
1018 | case TextureStage::STAGE_DOT3: // Already clamped |
1019 | case TextureStage::STAGE_PREMODULATE: |
1020 | case TextureStage::STAGE_MODULATEALPHA_ADDCOLOR: |
1021 | case TextureStage::STAGE_MODULATECOLOR_ADDALPHA: |
1022 | case TextureStage::STAGE_MODULATEINVALPHA_ADDCOLOR: |
1023 | case TextureStage::STAGE_MODULATEINVCOLOR_ADDALPHA: |
1024 | case TextureStage::STAGE_BUMPENVMAP: |
1025 | case TextureStage::STAGE_BUMPENVMAPLUMINANCE: |
1026 | if(state.textureStage[stage].cantUnderflow) |
1027 | { |
1028 | break; // Can't go below zero |
1029 | } |
1030 | case TextureStage::STAGE_ADDSIGNED: |
1031 | case TextureStage::STAGE_ADDSIGNED2X: |
1032 | case TextureStage::STAGE_SUBTRACT: |
1033 | case TextureStage::STAGE_ADDSMOOTH: |
1034 | res.x = Max(res.x, Short4(0x0000)); |
1035 | res.y = Max(res.y, Short4(0x0000)); |
1036 | res.z = Max(res.z, Short4(0x0000)); |
1037 | break; |
1038 | default: |
1039 | ASSERT(false); |
1040 | } |
1041 | |
1042 | switch(textureStage.stageOperationAlpha) |
1043 | { |
1044 | case TextureStage::STAGE_DISABLE: |
1045 | case TextureStage::STAGE_SELECTARG1: |
1046 | case TextureStage::STAGE_SELECTARG2: |
1047 | case TextureStage::STAGE_SELECTARG3: |
1048 | case TextureStage::STAGE_MODULATE: |
1049 | case TextureStage::STAGE_MODULATE2X: |
1050 | case TextureStage::STAGE_MODULATE4X: |
1051 | case TextureStage::STAGE_ADD: |
1052 | case TextureStage::STAGE_MULTIPLYADD: |
1053 | case TextureStage::STAGE_LERP: |
1054 | case TextureStage::STAGE_BLENDCURRENTALPHA: |
1055 | case TextureStage::STAGE_BLENDDIFFUSEALPHA: |
1056 | case TextureStage::STAGE_BLENDFACTORALPHA: |
1057 | case TextureStage::STAGE_BLENDTEXTUREALPHA: |
1058 | case TextureStage::STAGE_BLENDTEXTUREALPHAPM: |
1059 | case TextureStage::STAGE_DOT3: // Already clamped |
1060 | case TextureStage::STAGE_PREMODULATE: |
1061 | case TextureStage::STAGE_MODULATEALPHA_ADDCOLOR: |
1062 | case TextureStage::STAGE_MODULATECOLOR_ADDALPHA: |
1063 | case TextureStage::STAGE_MODULATEINVALPHA_ADDCOLOR: |
1064 | case TextureStage::STAGE_MODULATEINVCOLOR_ADDALPHA: |
1065 | case TextureStage::STAGE_BUMPENVMAP: |
1066 | case TextureStage::STAGE_BUMPENVMAPLUMINANCE: |
1067 | if(state.textureStage[stage].cantUnderflow) |
1068 | { |
1069 | break; // Can't go below zero |
1070 | } |
1071 | case TextureStage::STAGE_ADDSIGNED: |
1072 | case TextureStage::STAGE_ADDSIGNED2X: |
1073 | case TextureStage::STAGE_SUBTRACT: |
1074 | case TextureStage::STAGE_ADDSMOOTH: |
1075 | res.w = Max(res.w, Short4(0x0000)); |
1076 | break; |
1077 | default: |
1078 | ASSERT(false); |
1079 | } |
1080 | |
1081 | switch(textureStage.stageOperation) |
1082 | { |
1083 | case TextureStage::STAGE_DISABLE: |
1084 | case TextureStage::STAGE_SELECTARG1: |
1085 | case TextureStage::STAGE_SELECTARG2: |
1086 | case TextureStage::STAGE_SELECTARG3: |
1087 | case TextureStage::STAGE_MODULATE: |
1088 | case TextureStage::STAGE_SUBTRACT: |
1089 | case TextureStage::STAGE_ADDSMOOTH: |
1090 | case TextureStage::STAGE_LERP: |
1091 | case TextureStage::STAGE_BLENDCURRENTALPHA: |
1092 | case TextureStage::STAGE_BLENDDIFFUSEALPHA: |
1093 | case TextureStage::STAGE_BLENDFACTORALPHA: |
1094 | case TextureStage::STAGE_BLENDTEXTUREALPHA: |
1095 | case TextureStage::STAGE_DOT3: // Already clamped |
1096 | case TextureStage::STAGE_PREMODULATE: |
1097 | case TextureStage::STAGE_MODULATEINVALPHA_ADDCOLOR: |
1098 | case TextureStage::STAGE_MODULATEINVCOLOR_ADDALPHA: |
1099 | case TextureStage::STAGE_BUMPENVMAP: |
1100 | case TextureStage::STAGE_BUMPENVMAPLUMINANCE: |
1101 | break; // Can't go above one |
1102 | case TextureStage::STAGE_MODULATE2X: |
1103 | case TextureStage::STAGE_MODULATE4X: |
1104 | case TextureStage::STAGE_ADD: |
1105 | case TextureStage::STAGE_ADDSIGNED: |
1106 | case TextureStage::STAGE_ADDSIGNED2X: |
1107 | case TextureStage::STAGE_MULTIPLYADD: |
1108 | case TextureStage::STAGE_BLENDTEXTUREALPHAPM: |
1109 | case TextureStage::STAGE_MODULATEALPHA_ADDCOLOR: |
1110 | case TextureStage::STAGE_MODULATECOLOR_ADDALPHA: |
1111 | res.x = Min(res.x, Short4(0x1000)); |
1112 | res.y = Min(res.y, Short4(0x1000)); |
1113 | res.z = Min(res.z, Short4(0x1000)); |
1114 | break; |
1115 | default: |
1116 | ASSERT(false); |
1117 | } |
1118 | |
1119 | switch(textureStage.stageOperationAlpha) |
1120 | { |
1121 | case TextureStage::STAGE_DISABLE: |
1122 | case TextureStage::STAGE_SELECTARG1: |
1123 | case TextureStage::STAGE_SELECTARG2: |
1124 | case TextureStage::STAGE_SELECTARG3: |
1125 | case TextureStage::STAGE_MODULATE: |
1126 | case TextureStage::STAGE_SUBTRACT: |
1127 | case TextureStage::STAGE_ADDSMOOTH: |
1128 | case TextureStage::STAGE_LERP: |
1129 | case TextureStage::STAGE_BLENDCURRENTALPHA: |
1130 | case TextureStage::STAGE_BLENDDIFFUSEALPHA: |
1131 | case TextureStage::STAGE_BLENDFACTORALPHA: |
1132 | case TextureStage::STAGE_BLENDTEXTUREALPHA: |
1133 | case TextureStage::STAGE_DOT3: // Already clamped |
1134 | case TextureStage::STAGE_PREMODULATE: |
1135 | case TextureStage::STAGE_MODULATEINVALPHA_ADDCOLOR: |
1136 | case TextureStage::STAGE_MODULATEINVCOLOR_ADDALPHA: |
1137 | case TextureStage::STAGE_BUMPENVMAP: |
1138 | case TextureStage::STAGE_BUMPENVMAPLUMINANCE: |
1139 | break; // Can't go above one |
1140 | case TextureStage::STAGE_MODULATE2X: |
1141 | case TextureStage::STAGE_MODULATE4X: |
1142 | case TextureStage::STAGE_ADD: |
1143 | case TextureStage::STAGE_ADDSIGNED: |
1144 | case TextureStage::STAGE_ADDSIGNED2X: |
1145 | case TextureStage::STAGE_MULTIPLYADD: |
1146 | case TextureStage::STAGE_BLENDTEXTUREALPHAPM: |
1147 | case TextureStage::STAGE_MODULATEALPHA_ADDCOLOR: |
1148 | case TextureStage::STAGE_MODULATECOLOR_ADDALPHA: |
1149 | res.w = Min(res.w, Short4(0x1000)); |
1150 | break; |
1151 | default: |
1152 | ASSERT(false); |
1153 | } |
1154 | |
1155 | switch(textureStage.destinationArgument) |
1156 | { |
1157 | case TextureStage::DESTINATION_CURRENT: |
1158 | current.x = res.x; |
1159 | current.y = res.y; |
1160 | current.z = res.z; |
1161 | current.w = res.w; |
1162 | break; |
1163 | case TextureStage::DESTINATION_TEMP: |
1164 | temp.x = res.x; |
1165 | temp.y = res.y; |
1166 | temp.z = res.z; |
1167 | temp.w = res.w; |
1168 | break; |
1169 | default: |
1170 | ASSERT(false); |
1171 | } |
1172 | } |
1173 | |
1174 | void PixelPipeline::fogBlend(Vector4s ¤t, Float4 &f) |
1175 | { |
1176 | if(!state.fogActive) |
1177 | { |
1178 | return; |
1179 | } |
1180 | |
1181 | if(state.pixelFogMode != FOG_NONE) |
1182 | { |
1183 | pixelFog(f); |
1184 | } |
1185 | |
1186 | UShort4 fog = convertFixed16(f, true); |
1187 | |
1188 | current.x = As<Short4>(MulHigh(As<UShort4>(current.x), fog)); |
1189 | current.y = As<Short4>(MulHigh(As<UShort4>(current.y), fog)); |
1190 | current.z = As<Short4>(MulHigh(As<UShort4>(current.z), fog)); |
1191 | |
1192 | UShort4 invFog = UShort4(0xFFFFu) - fog; |
1193 | |
1194 | current.x += As<Short4>(MulHigh(invFog, *Pointer<UShort4>(data + OFFSET(DrawData, fog.color4[0])))); |
1195 | current.y += As<Short4>(MulHigh(invFog, *Pointer<UShort4>(data + OFFSET(DrawData, fog.color4[1])))); |
1196 | current.z += As<Short4>(MulHigh(invFog, *Pointer<UShort4>(data + OFFSET(DrawData, fog.color4[2])))); |
1197 | } |
1198 | |
1199 | void PixelPipeline::specularPixel(Vector4s ¤t, Vector4s &specular) |
1200 | { |
1201 | if(!state.specularAdd) |
1202 | { |
1203 | return; |
1204 | } |
1205 | |
1206 | current.x = AddSat(current.x, specular.x); |
1207 | current.y = AddSat(current.y, specular.y); |
1208 | current.z = AddSat(current.z, specular.z); |
1209 | } |
1210 | |
1211 | Vector4s PixelPipeline::sampleTexture(int coordinates, int stage, bool project) |
1212 | { |
1213 | Float4 x = v[2 + coordinates].x; |
1214 | Float4 y = v[2 + coordinates].y; |
1215 | Float4 z = v[2 + coordinates].z; |
1216 | Float4 w = v[2 + coordinates].w; |
1217 | |
1218 | if(perturbate) |
1219 | { |
1220 | x += du; |
1221 | y += dv; |
1222 | |
1223 | perturbate = false; |
1224 | } |
1225 | |
1226 | return sampleTexture(stage, x, y, z, w, project); |
1227 | } |
1228 | |
1229 | Vector4s PixelPipeline::sampleTexture(int stage, Float4 &u, Float4 &v, Float4 &w, Float4 &q, bool project) |
1230 | { |
1231 | Vector4s c; |
1232 | |
1233 | #if PERF_PROFILE |
1234 | Long texTime = Ticks(); |
1235 | #endif |
1236 | |
1237 | Vector4f dsx; |
1238 | Vector4f dsy; |
1239 | |
1240 | Pointer<Byte> texture = data + OFFSET(DrawData, mipmap) + stage * sizeof(Texture); |
1241 | |
1242 | if(!project) |
1243 | { |
1244 | c = SamplerCore(constants, state.sampler[stage]).sampleTexture(texture, u, v, w, q, q, dsx, dsy); |
1245 | } |
1246 | else |
1247 | { |
1248 | Float4 rq = reciprocal(q); |
1249 | |
1250 | Float4 u_q = u * rq; |
1251 | Float4 v_q = v * rq; |
1252 | Float4 w_q = w * rq; |
1253 | |
1254 | c = SamplerCore(constants, state.sampler[stage]).sampleTexture(texture, u_q, v_q, w_q, q, q, dsx, dsy); |
1255 | } |
1256 | |
1257 | #if PERF_PROFILE |
1258 | cycles[PERF_TEX] += Ticks() - texTime; |
1259 | #endif |
1260 | |
1261 | return c; |
1262 | } |
1263 | |
1264 | Short4 PixelPipeline::convertFixed12(RValue<Float4> cf) |
1265 | { |
1266 | return RoundShort4(cf * Float4(0x1000)); |
1267 | } |
1268 | |
1269 | void PixelPipeline::convertFixed12(Vector4s &cs, Vector4f &cf) |
1270 | { |
1271 | cs.x = convertFixed12(cf.x); |
1272 | cs.y = convertFixed12(cf.y); |
1273 | cs.z = convertFixed12(cf.z); |
1274 | cs.w = convertFixed12(cf.w); |
1275 | } |
1276 | |
1277 | Float4 PixelPipeline::convertSigned12(Short4 &cs) |
1278 | { |
1279 | return Float4(cs) * Float4(1.0f / 0x0FFE); |
1280 | } |
1281 | |
1282 | void PixelPipeline::convertSigned12(Vector4f &cf, Vector4s &cs) |
1283 | { |
1284 | cf.x = convertSigned12(cs.x); |
1285 | cf.y = convertSigned12(cs.y); |
1286 | cf.z = convertSigned12(cs.z); |
1287 | cf.w = convertSigned12(cs.w); |
1288 | } |
1289 | |
1290 | void PixelPipeline::writeDestination(Vector4s &d, const Dst &dst) |
1291 | { |
1292 | switch(dst.type) |
1293 | { |
1294 | case Shader::PARAMETER_TEMP: |
1295 | if(dst.mask & 0x1) rs[dst.index].x = d.x; |
1296 | if(dst.mask & 0x2) rs[dst.index].y = d.y; |
1297 | if(dst.mask & 0x4) rs[dst.index].z = d.z; |
1298 | if(dst.mask & 0x8) rs[dst.index].w = d.w; |
1299 | break; |
1300 | case Shader::PARAMETER_INPUT: |
1301 | if(dst.mask & 0x1) vs[dst.index].x = d.x; |
1302 | if(dst.mask & 0x2) vs[dst.index].y = d.y; |
1303 | if(dst.mask & 0x4) vs[dst.index].z = d.z; |
1304 | if(dst.mask & 0x8) vs[dst.index].w = d.w; |
1305 | break; |
1306 | case Shader::PARAMETER_CONST: ASSERT(false); break; |
1307 | case Shader::PARAMETER_TEXTURE: |
1308 | if(dst.mask & 0x1) ts[dst.index].x = d.x; |
1309 | if(dst.mask & 0x2) ts[dst.index].y = d.y; |
1310 | if(dst.mask & 0x4) ts[dst.index].z = d.z; |
1311 | if(dst.mask & 0x8) ts[dst.index].w = d.w; |
1312 | break; |
1313 | case Shader::PARAMETER_COLOROUT: |
1314 | if(dst.mask & 0x1) vs[dst.index].x = d.x; |
1315 | if(dst.mask & 0x2) vs[dst.index].y = d.y; |
1316 | if(dst.mask & 0x4) vs[dst.index].z = d.z; |
1317 | if(dst.mask & 0x8) vs[dst.index].w = d.w; |
1318 | break; |
1319 | default: |
1320 | ASSERT(false); |
1321 | } |
1322 | } |
1323 | |
1324 | Vector4s PixelPipeline::fetchRegister(const Src &src) |
1325 | { |
1326 | Vector4s *reg; |
1327 | int i = src.index; |
1328 | |
1329 | Vector4s c; |
1330 | |
1331 | if(src.type == Shader::PARAMETER_CONST) |
1332 | { |
1333 | c.x = *Pointer<Short4>(data + OFFSET(DrawData, ps.cW[i][0])); |
1334 | c.y = *Pointer<Short4>(data + OFFSET(DrawData, ps.cW[i][1])); |
1335 | c.z = *Pointer<Short4>(data + OFFSET(DrawData, ps.cW[i][2])); |
1336 | c.w = *Pointer<Short4>(data + OFFSET(DrawData, ps.cW[i][3])); |
1337 | } |
1338 | |
1339 | switch(src.type) |
1340 | { |
1341 | case Shader::PARAMETER_TEMP: reg = &rs[i]; break; |
1342 | case Shader::PARAMETER_INPUT: reg = &vs[i]; break; |
1343 | case Shader::PARAMETER_CONST: reg = &c; break; |
1344 | case Shader::PARAMETER_TEXTURE: reg = &ts[i]; break; |
1345 | case Shader::PARAMETER_VOID: return rs[0]; // Dummy |
1346 | case Shader::PARAMETER_FLOAT4LITERAL: return rs[0]; // Dummy |
1347 | default: ASSERT(false); return rs[0]; |
1348 | } |
1349 | |
1350 | const Short4 &x = (*reg)[(src.swizzle >> 0) & 0x3]; |
1351 | const Short4 &y = (*reg)[(src.swizzle >> 2) & 0x3]; |
1352 | const Short4 &z = (*reg)[(src.swizzle >> 4) & 0x3]; |
1353 | const Short4 &w = (*reg)[(src.swizzle >> 6) & 0x3]; |
1354 | |
1355 | Vector4s mod; |
1356 | |
1357 | switch(src.modifier) |
1358 | { |
1359 | case Shader::MODIFIER_NONE: |
1360 | mod.x = x; |
1361 | mod.y = y; |
1362 | mod.z = z; |
1363 | mod.w = w; |
1364 | break; |
1365 | case Shader::MODIFIER_BIAS: |
1366 | mod.x = SubSat(x, Short4(0x0800)); |
1367 | mod.y = SubSat(y, Short4(0x0800)); |
1368 | mod.z = SubSat(z, Short4(0x0800)); |
1369 | mod.w = SubSat(w, Short4(0x0800)); |
1370 | break; |
1371 | case Shader::MODIFIER_BIAS_NEGATE: |
1372 | mod.x = SubSat(Short4(0x0800), x); |
1373 | mod.y = SubSat(Short4(0x0800), y); |
1374 | mod.z = SubSat(Short4(0x0800), z); |
1375 | mod.w = SubSat(Short4(0x0800), w); |
1376 | break; |
1377 | case Shader::MODIFIER_COMPLEMENT: |
1378 | mod.x = SubSat(Short4(0x1000), x); |
1379 | mod.y = SubSat(Short4(0x1000), y); |
1380 | mod.z = SubSat(Short4(0x1000), z); |
1381 | mod.w = SubSat(Short4(0x1000), w); |
1382 | break; |
1383 | case Shader::MODIFIER_NEGATE: |
1384 | mod.x = -x; |
1385 | mod.y = -y; |
1386 | mod.z = -z; |
1387 | mod.w = -w; |
1388 | break; |
1389 | case Shader::MODIFIER_X2: |
1390 | mod.x = AddSat(x, x); |
1391 | mod.y = AddSat(y, y); |
1392 | mod.z = AddSat(z, z); |
1393 | mod.w = AddSat(w, w); |
1394 | break; |
1395 | case Shader::MODIFIER_X2_NEGATE: |
1396 | mod.x = -AddSat(x, x); |
1397 | mod.y = -AddSat(y, y); |
1398 | mod.z = -AddSat(z, z); |
1399 | mod.w = -AddSat(w, w); |
1400 | break; |
1401 | case Shader::MODIFIER_SIGN: |
1402 | mod.x = SubSat(x, Short4(0x0800)); |
1403 | mod.y = SubSat(y, Short4(0x0800)); |
1404 | mod.z = SubSat(z, Short4(0x0800)); |
1405 | mod.w = SubSat(w, Short4(0x0800)); |
1406 | mod.x = AddSat(mod.x, mod.x); |
1407 | mod.y = AddSat(mod.y, mod.y); |
1408 | mod.z = AddSat(mod.z, mod.z); |
1409 | mod.w = AddSat(mod.w, mod.w); |
1410 | break; |
1411 | case Shader::MODIFIER_SIGN_NEGATE: |
1412 | mod.x = SubSat(Short4(0x0800), x); |
1413 | mod.y = SubSat(Short4(0x0800), y); |
1414 | mod.z = SubSat(Short4(0x0800), z); |
1415 | mod.w = SubSat(Short4(0x0800), w); |
1416 | mod.x = AddSat(mod.x, mod.x); |
1417 | mod.y = AddSat(mod.y, mod.y); |
1418 | mod.z = AddSat(mod.z, mod.z); |
1419 | mod.w = AddSat(mod.w, mod.w); |
1420 | break; |
1421 | case Shader::MODIFIER_DZ: |
1422 | mod.x = x; |
1423 | mod.y = y; |
1424 | mod.z = z; |
1425 | mod.w = w; |
1426 | // Projection performed by texture sampler |
1427 | break; |
1428 | case Shader::MODIFIER_DW: |
1429 | mod.x = x; |
1430 | mod.y = y; |
1431 | mod.z = z; |
1432 | mod.w = w; |
1433 | // Projection performed by texture sampler |
1434 | break; |
1435 | default: |
1436 | ASSERT(false); |
1437 | } |
1438 | |
1439 | if(src.type == Shader::PARAMETER_CONST && (src.modifier == Shader::MODIFIER_X2 || src.modifier == Shader::MODIFIER_X2_NEGATE)) |
1440 | { |
1441 | mod.x = Min(mod.x, Short4(0x1000)); mod.x = Max(mod.x, Short4(-0x1000)); |
1442 | mod.y = Min(mod.y, Short4(0x1000)); mod.y = Max(mod.y, Short4(-0x1000)); |
1443 | mod.z = Min(mod.z, Short4(0x1000)); mod.z = Max(mod.z, Short4(-0x1000)); |
1444 | mod.w = Min(mod.w, Short4(0x1000)); mod.w = Max(mod.w, Short4(-0x1000)); |
1445 | } |
1446 | |
1447 | return mod; |
1448 | } |
1449 | |
1450 | void PixelPipeline::MOV(Vector4s &dst, Vector4s &src0) |
1451 | { |
1452 | dst.x = src0.x; |
1453 | dst.y = src0.y; |
1454 | dst.z = src0.z; |
1455 | dst.w = src0.w; |
1456 | } |
1457 | |
1458 | void PixelPipeline::ADD(Vector4s &dst, Vector4s &src0, Vector4s &src1) |
1459 | { |
1460 | dst.x = AddSat(src0.x, src1.x); |
1461 | dst.y = AddSat(src0.y, src1.y); |
1462 | dst.z = AddSat(src0.z, src1.z); |
1463 | dst.w = AddSat(src0.w, src1.w); |
1464 | } |
1465 | |
1466 | void PixelPipeline::SUB(Vector4s &dst, Vector4s &src0, Vector4s &src1) |
1467 | { |
1468 | dst.x = SubSat(src0.x, src1.x); |
1469 | dst.y = SubSat(src0.y, src1.y); |
1470 | dst.z = SubSat(src0.z, src1.z); |
1471 | dst.w = SubSat(src0.w, src1.w); |
1472 | } |
1473 | |
1474 | void PixelPipeline::MAD(Vector4s &dst, Vector4s &src0, Vector4s &src1, Vector4s &src2) |
1475 | { |
1476 | // FIXME: Long fixed-point multiply fixup |
1477 | { dst.x = MulHigh(src0.x, src1.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, src2.x); } |
1478 | { dst.y = MulHigh(src0.y, src1.y); dst.y = AddSat(dst.y, dst.y); dst.y = AddSat(dst.y, dst.y); dst.y = AddSat(dst.y, dst.y); dst.y = AddSat(dst.y, dst.y); dst.y = AddSat(dst.y, src2.y); } |
1479 | { dst.z = MulHigh(src0.z, src1.z); dst.z = AddSat(dst.z, dst.z); dst.z = AddSat(dst.z, dst.z); dst.z = AddSat(dst.z, dst.z); dst.z = AddSat(dst.z, dst.z); dst.z = AddSat(dst.z, src2.z); } |
1480 | { dst.w = MulHigh(src0.w, src1.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, src2.w); } |
1481 | } |
1482 | |
1483 | void PixelPipeline::MUL(Vector4s &dst, Vector4s &src0, Vector4s &src1) |
1484 | { |
1485 | // FIXME: Long fixed-point multiply fixup |
1486 | { dst.x = MulHigh(src0.x, src1.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); } |
1487 | { dst.y = MulHigh(src0.y, src1.y); dst.y = AddSat(dst.y, dst.y); dst.y = AddSat(dst.y, dst.y); dst.y = AddSat(dst.y, dst.y); dst.y = AddSat(dst.y, dst.y); } |
1488 | { dst.z = MulHigh(src0.z, src1.z); dst.z = AddSat(dst.z, dst.z); dst.z = AddSat(dst.z, dst.z); dst.z = AddSat(dst.z, dst.z); dst.z = AddSat(dst.z, dst.z); } |
1489 | { dst.w = MulHigh(src0.w, src1.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); } |
1490 | } |
1491 | |
1492 | void PixelPipeline::DP3(Vector4s &dst, Vector4s &src0, Vector4s &src1) |
1493 | { |
1494 | Short4 t0; |
1495 | Short4 t1; |
1496 | |
1497 | // FIXME: Long fixed-point multiply fixup |
1498 | t0 = MulHigh(src0.x, src1.x); t0 = AddSat(t0, t0); t0 = AddSat(t0, t0); t0 = AddSat(t0, t0); t0 = AddSat(t0, t0); |
1499 | t1 = MulHigh(src0.y, src1.y); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); |
1500 | t0 = AddSat(t0, t1); |
1501 | t1 = MulHigh(src0.z, src1.z); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); |
1502 | t0 = AddSat(t0, t1); |
1503 | |
1504 | dst.x = t0; |
1505 | dst.y = t0; |
1506 | dst.z = t0; |
1507 | dst.w = t0; |
1508 | } |
1509 | |
1510 | void PixelPipeline::DP4(Vector4s &dst, Vector4s &src0, Vector4s &src1) |
1511 | { |
1512 | Short4 t0; |
1513 | Short4 t1; |
1514 | |
1515 | // FIXME: Long fixed-point multiply fixup |
1516 | t0 = MulHigh(src0.x, src1.x); t0 = AddSat(t0, t0); t0 = AddSat(t0, t0); t0 = AddSat(t0, t0); t0 = AddSat(t0, t0); |
1517 | t1 = MulHigh(src0.y, src1.y); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); |
1518 | t0 = AddSat(t0, t1); |
1519 | t1 = MulHigh(src0.z, src1.z); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); |
1520 | t0 = AddSat(t0, t1); |
1521 | t1 = MulHigh(src0.w, src1.w); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); t1 = AddSat(t1, t1); |
1522 | t0 = AddSat(t0, t1); |
1523 | |
1524 | dst.x = t0; |
1525 | dst.y = t0; |
1526 | dst.z = t0; |
1527 | dst.w = t0; |
1528 | } |
1529 | |
1530 | void PixelPipeline::LRP(Vector4s &dst, Vector4s &src0, Vector4s &src1, Vector4s &src2) |
1531 | { |
1532 | // FIXME: Long fixed-point multiply fixup |
1533 | { dst.x = SubSat(src1.x, src2.x); dst.x = MulHigh(dst.x, src0.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, dst.x); dst.x = AddSat(dst.x, src2.x); } |
1534 | { |
1535 | dst.y = SubSat(src1.y, src2.y); dst.y = MulHigh(dst.y, src0.y); dst.y = AddSat(dst.y, dst.y); dst.y = AddSat(dst.y, dst.y); dst.y = AddSat(dst.y, dst.y); dst.y = AddSat(dst.y, dst.y); dst.y = AddSat(dst.y, src2.y); |
1536 | } |
1537 | {dst.z = SubSat(src1.z, src2.z); dst.z = MulHigh(dst.z, src0.z); dst.z = AddSat(dst.z, dst.z); dst.z = AddSat(dst.z, dst.z); dst.z = AddSat(dst.z, dst.z); dst.z = AddSat(dst.z, dst.z); dst.z = AddSat(dst.z, src2.z); } |
1538 | {dst.w = SubSat(src1.w, src2.w); dst.w = MulHigh(dst.w, src0.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, dst.w); dst.w = AddSat(dst.w, src2.w); } |
1539 | } |
1540 | |
1541 | void PixelPipeline::TEXCOORD(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int coordinate) |
1542 | { |
1543 | Float4 uw; |
1544 | Float4 vw; |
1545 | Float4 sw; |
1546 | |
1547 | if(state.interpolant[2 + coordinate].component & 0x01) |
1548 | { |
1549 | uw = Max(u, Float4(0.0f)); |
1550 | uw = Min(uw, Float4(1.0f)); |
1551 | dst.x = convertFixed12(uw); |
1552 | } |
1553 | else |
1554 | { |
1555 | dst.x = Short4(0x0000); |
1556 | } |
1557 | |
1558 | if(state.interpolant[2 + coordinate].component & 0x02) |
1559 | { |
1560 | vw = Max(v, Float4(0.0f)); |
1561 | vw = Min(vw, Float4(1.0f)); |
1562 | dst.y = convertFixed12(vw); |
1563 | } |
1564 | else |
1565 | { |
1566 | dst.y = Short4(0x0000); |
1567 | } |
1568 | |
1569 | if(state.interpolant[2 + coordinate].component & 0x04) |
1570 | { |
1571 | sw = Max(s, Float4(0.0f)); |
1572 | sw = Min(sw, Float4(1.0f)); |
1573 | dst.z = convertFixed12(sw); |
1574 | } |
1575 | else |
1576 | { |
1577 | dst.z = Short4(0x0000); |
1578 | } |
1579 | |
1580 | dst.w = Short4(0x1000); |
1581 | } |
1582 | |
1583 | void PixelPipeline::TEXCRD(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int coordinate, bool project) |
1584 | { |
1585 | Float4 uw = u; |
1586 | Float4 vw = v; |
1587 | Float4 sw = s; |
1588 | |
1589 | if(project) |
1590 | { |
1591 | uw *= Rcp_pp(s); |
1592 | vw *= Rcp_pp(s); |
1593 | } |
1594 | |
1595 | if(state.interpolant[2 + coordinate].component & 0x01) |
1596 | { |
1597 | uw *= Float4(0x1000); |
1598 | uw = Max(uw, Float4(-0x8000)); |
1599 | uw = Min(uw, Float4(0x7FFF)); |
1600 | dst.x = RoundShort4(uw); |
1601 | } |
1602 | else |
1603 | { |
1604 | dst.x = Short4(0x0000); |
1605 | } |
1606 | |
1607 | if(state.interpolant[2 + coordinate].component & 0x02) |
1608 | { |
1609 | vw *= Float4(0x1000); |
1610 | vw = Max(vw, Float4(-0x8000)); |
1611 | vw = Min(vw, Float4(0x7FFF)); |
1612 | dst.y = RoundShort4(vw); |
1613 | } |
1614 | else |
1615 | { |
1616 | dst.y = Short4(0x0000); |
1617 | } |
1618 | |
1619 | if(state.interpolant[2 + coordinate].component & 0x04) |
1620 | { |
1621 | sw *= Float4(0x1000); |
1622 | sw = Max(sw, Float4(-0x8000)); |
1623 | sw = Min(sw, Float4(0x7FFF)); |
1624 | dst.z = RoundShort4(sw); |
1625 | } |
1626 | else |
1627 | { |
1628 | dst.z = Short4(0x0000); |
1629 | } |
1630 | } |
1631 | |
1632 | void PixelPipeline::TEXDP3(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, Vector4s &src) |
1633 | { |
1634 | TEXM3X3PAD(u, v, s, src, 0, false); |
1635 | |
1636 | Short4 t0 = RoundShort4(u_ * Float4(0x1000)); |
1637 | |
1638 | dst.x = t0; |
1639 | dst.y = t0; |
1640 | dst.z = t0; |
1641 | dst.w = t0; |
1642 | } |
1643 | |
1644 | void PixelPipeline::TEXDP3TEX(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0) |
1645 | { |
1646 | TEXM3X3PAD(u, v, s, src0, 0, false); |
1647 | |
1648 | v_ = Float4(0.0f); |
1649 | w_ = Float4(0.0f); |
1650 | |
1651 | dst = sampleTexture(stage, u_, v_, w_, w_); |
1652 | } |
1653 | |
1654 | void PixelPipeline::TEXKILL(Int cMask[4], Float4 &u, Float4 &v, Float4 &s) |
1655 | { |
1656 | Int kill = SignMask(CmpNLT(u, Float4(0.0f))) & |
1657 | SignMask(CmpNLT(v, Float4(0.0f))) & |
1658 | SignMask(CmpNLT(s, Float4(0.0f))); |
1659 | |
1660 | for(unsigned int q = 0; q < state.multiSample; q++) |
1661 | { |
1662 | cMask[q] &= kill; |
1663 | } |
1664 | } |
1665 | |
1666 | void PixelPipeline::TEXKILL(Int cMask[4], Vector4s &src) |
1667 | { |
1668 | Short4 test = src.x | src.y | src.z; |
1669 | Int kill = SignMask(PackSigned(test, test)) ^ 0x0000000F; |
1670 | |
1671 | for(unsigned int q = 0; q < state.multiSample; q++) |
1672 | { |
1673 | cMask[q] &= kill; |
1674 | } |
1675 | } |
1676 | |
1677 | void PixelPipeline::TEX(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int sampler, bool project) |
1678 | { |
1679 | dst = sampleTexture(sampler, u, v, s, s, project); |
1680 | } |
1681 | |
1682 | void PixelPipeline::TEXLD(Vector4s &dst, Vector4s &src, int sampler, bool project) |
1683 | { |
1684 | Float4 u = Float4(src.x) * Float4(1.0f / 0x0FFE); |
1685 | Float4 v = Float4(src.y) * Float4(1.0f / 0x0FFE); |
1686 | Float4 s = Float4(src.z) * Float4(1.0f / 0x0FFE); |
1687 | |
1688 | dst = sampleTexture(sampler, u, v, s, s, project); |
1689 | } |
1690 | |
1691 | void PixelPipeline::TEXBEM(Vector4s &dst, Vector4s &src, Float4 &u, Float4 &v, Float4 &s, int stage) |
1692 | { |
1693 | Float4 du = Float4(src.x) * Float4(1.0f / 0x0FFE); |
1694 | Float4 dv = Float4(src.y) * Float4(1.0f / 0x0FFE); |
1695 | |
1696 | Float4 du2 = du; |
1697 | Float4 dv2 = dv; |
1698 | |
1699 | du *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][0])); |
1700 | dv2 *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][0])); |
1701 | du += dv2; |
1702 | dv *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][1])); |
1703 | du2 *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][1])); |
1704 | dv += du2; |
1705 | |
1706 | Float4 u_ = u + du; |
1707 | Float4 v_ = v + dv; |
1708 | |
1709 | dst = sampleTexture(stage, u_, v_, s, s); |
1710 | } |
1711 | |
1712 | void PixelPipeline::TEXBEML(Vector4s &dst, Vector4s &src, Float4 &u, Float4 &v, Float4 &s, int stage) |
1713 | { |
1714 | Float4 du = Float4(src.x) * Float4(1.0f / 0x0FFE); |
1715 | Float4 dv = Float4(src.y) * Float4(1.0f / 0x0FFE); |
1716 | |
1717 | Float4 du2 = du; |
1718 | Float4 dv2 = dv; |
1719 | |
1720 | du *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][0])); |
1721 | dv2 *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][0])); |
1722 | du += dv2; |
1723 | dv *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[1][1])); |
1724 | du2 *= *Pointer<Float4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4F[0][1])); |
1725 | dv += du2; |
1726 | |
1727 | Float4 u_ = u + du; |
1728 | Float4 v_ = v + dv; |
1729 | |
1730 | dst = sampleTexture(stage, u_, v_, s, s); |
1731 | |
1732 | Short4 L; |
1733 | |
1734 | L = src.z; |
1735 | L = MulHigh(L, *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].luminanceScale4))); |
1736 | L = L << 4; |
1737 | L = AddSat(L, *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].luminanceOffset4))); |
1738 | L = Max(L, Short4(0x0000)); |
1739 | L = Min(L, Short4(0x1000)); |
1740 | |
1741 | dst.x = MulHigh(dst.x, L); dst.x = dst.x << 4; |
1742 | dst.y = MulHigh(dst.y, L); dst.y = dst.y << 4; |
1743 | dst.z = MulHigh(dst.z, L); dst.z = dst.z << 4; |
1744 | } |
1745 | |
1746 | void PixelPipeline::TEXREG2AR(Vector4s &dst, Vector4s &src0, int stage) |
1747 | { |
1748 | Float4 u = Float4(src0.w) * Float4(1.0f / 0x0FFE); |
1749 | Float4 v = Float4(src0.x) * Float4(1.0f / 0x0FFE); |
1750 | Float4 s = Float4(src0.z) * Float4(1.0f / 0x0FFE); |
1751 | |
1752 | dst = sampleTexture(stage, u, v, s, s); |
1753 | } |
1754 | |
1755 | void PixelPipeline::TEXREG2GB(Vector4s &dst, Vector4s &src0, int stage) |
1756 | { |
1757 | Float4 u = Float4(src0.y) * Float4(1.0f / 0x0FFE); |
1758 | Float4 v = Float4(src0.z) * Float4(1.0f / 0x0FFE); |
1759 | Float4 s = v; |
1760 | |
1761 | dst = sampleTexture(stage, u, v, s, s); |
1762 | } |
1763 | |
1764 | void PixelPipeline::TEXREG2RGB(Vector4s &dst, Vector4s &src0, int stage) |
1765 | { |
1766 | Float4 u = Float4(src0.x) * Float4(1.0f / 0x0FFE); |
1767 | Float4 v = Float4(src0.y) * Float4(1.0f / 0x0FFE); |
1768 | Float4 s = Float4(src0.z) * Float4(1.0f / 0x0FFE); |
1769 | |
1770 | dst = sampleTexture(stage, u, v, s, s); |
1771 | } |
1772 | |
1773 | void PixelPipeline::TEXM3X2DEPTH(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, Vector4s &src, bool signedScaling) |
1774 | { |
1775 | TEXM3X2PAD(u, v, s, src, 1, signedScaling); |
1776 | |
1777 | // z / w |
1778 | u_ *= Rcp_pp(v_); // FIXME: Set result to 1.0 when division by zero |
1779 | |
1780 | oDepth = u_; |
1781 | } |
1782 | |
1783 | void PixelPipeline::TEXM3X2PAD(Float4 &u, Float4 &v, Float4 &s, Vector4s &src0, int component, bool signedScaling) |
1784 | { |
1785 | TEXM3X3PAD(u, v, s, src0, component, signedScaling); |
1786 | } |
1787 | |
1788 | void PixelPipeline::TEXM3X2TEX(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0, bool signedScaling) |
1789 | { |
1790 | TEXM3X2PAD(u, v, s, src0, 1, signedScaling); |
1791 | |
1792 | w_ = Float4(0.0f); |
1793 | |
1794 | dst = sampleTexture(stage, u_, v_, w_, w_); |
1795 | } |
1796 | |
1797 | void PixelPipeline::TEXM3X3(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, Vector4s &src0, bool signedScaling) |
1798 | { |
1799 | TEXM3X3PAD(u, v, s, src0, 2, signedScaling); |
1800 | |
1801 | dst.x = RoundShort4(u_ * Float4(0x1000)); |
1802 | dst.y = RoundShort4(v_ * Float4(0x1000)); |
1803 | dst.z = RoundShort4(w_ * Float4(0x1000)); |
1804 | dst.w = Short4(0x1000); |
1805 | } |
1806 | |
1807 | void PixelPipeline::TEXM3X3PAD(Float4 &u, Float4 &v, Float4 &s, Vector4s &src0, int component, bool signedScaling) |
1808 | { |
1809 | if(component == 0 || previousScaling != signedScaling) // FIXME: Other source modifiers? |
1810 | { |
1811 | U = Float4(src0.x); |
1812 | V = Float4(src0.y); |
1813 | W = Float4(src0.z); |
1814 | |
1815 | previousScaling = signedScaling; |
1816 | } |
1817 | |
1818 | Float4 x = U * u + V * v + W * s; |
1819 | |
1820 | x *= Float4(1.0f / 0x1000); |
1821 | |
1822 | switch(component) |
1823 | { |
1824 | case 0: u_ = x; break; |
1825 | case 1: v_ = x; break; |
1826 | case 2: w_ = x; break; |
1827 | default: ASSERT(false); |
1828 | } |
1829 | } |
1830 | |
1831 | void PixelPipeline::TEXM3X3SPEC(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0, Vector4s &src1) |
1832 | { |
1833 | TEXM3X3PAD(u, v, s, src0, 2, false); |
1834 | |
1835 | Float4 E[3]; // Eye vector |
1836 | |
1837 | E[0] = Float4(src1.x) * Float4(1.0f / 0x0FFE); |
1838 | E[1] = Float4(src1.y) * Float4(1.0f / 0x0FFE); |
1839 | E[2] = Float4(src1.z) * Float4(1.0f / 0x0FFE); |
1840 | |
1841 | // Reflection |
1842 | Float4 u__; |
1843 | Float4 v__; |
1844 | Float4 w__; |
1845 | |
1846 | // (u'', v'', w'') = 2 * (N . E) * N - E * (N . N) |
1847 | u__ = u_ * E[0]; |
1848 | v__ = v_ * E[1]; |
1849 | w__ = w_ * E[2]; |
1850 | u__ += v__ + w__; |
1851 | u__ += u__; |
1852 | v__ = u__; |
1853 | w__ = u__; |
1854 | u__ *= u_; |
1855 | v__ *= v_; |
1856 | w__ *= w_; |
1857 | u_ *= u_; |
1858 | v_ *= v_; |
1859 | w_ *= w_; |
1860 | u_ += v_ + w_; |
1861 | u__ -= E[0] * u_; |
1862 | v__ -= E[1] * u_; |
1863 | w__ -= E[2] * u_; |
1864 | |
1865 | dst = sampleTexture(stage, u__, v__, w__, w__); |
1866 | } |
1867 | |
1868 | void PixelPipeline::TEXM3X3TEX(Vector4s &dst, Float4 &u, Float4 &v, Float4 &s, int stage, Vector4s &src0, bool signedScaling) |
1869 | { |
1870 | TEXM3X3PAD(u, v, s, src0, 2, signedScaling); |
1871 | |
1872 | dst = sampleTexture(stage, u_, v_, w_, w_); |
1873 | } |
1874 | |
1875 | void PixelPipeline::TEXM3X3VSPEC(Vector4s &dst, Float4 &x, Float4 &y, Float4 &z, int stage, Vector4s &src0) |
1876 | { |
1877 | TEXM3X3PAD(x, y, z, src0, 2, false); |
1878 | |
1879 | Float4 E[3]; // Eye vector |
1880 | |
1881 | E[0] = v[2 + stage - 2].w; |
1882 | E[1] = v[2 + stage - 1].w; |
1883 | E[2] = v[2 + stage - 0].w; |
1884 | |
1885 | // Reflection |
1886 | Float4 u__; |
1887 | Float4 v__; |
1888 | Float4 w__; |
1889 | |
1890 | // (u'', v'', w'') = 2 * (N . E) * N - E * (N . N) |
1891 | u__ = u_ * E[0]; |
1892 | v__ = v_ * E[1]; |
1893 | w__ = w_ * E[2]; |
1894 | u__ += v__ + w__; |
1895 | u__ += u__; |
1896 | v__ = u__; |
1897 | w__ = u__; |
1898 | u__ *= u_; |
1899 | v__ *= v_; |
1900 | w__ *= w_; |
1901 | u_ *= u_; |
1902 | v_ *= v_; |
1903 | w_ *= w_; |
1904 | u_ += v_ + w_; |
1905 | u__ -= E[0] * u_; |
1906 | v__ -= E[1] * u_; |
1907 | w__ -= E[2] * u_; |
1908 | |
1909 | dst = sampleTexture(stage, u__, v__, w__, w__); |
1910 | } |
1911 | |
1912 | void PixelPipeline::TEXDEPTH() |
1913 | { |
1914 | u_ = Float4(rs[5].x); |
1915 | v_ = Float4(rs[5].y); |
1916 | |
1917 | // z / w |
1918 | u_ *= Rcp_pp(v_); // FIXME: Set result to 1.0 when division by zero |
1919 | |
1920 | oDepth = u_; |
1921 | } |
1922 | |
1923 | void PixelPipeline::CND(Vector4s &dst, Vector4s &src0, Vector4s &src1, Vector4s &src2) |
1924 | { |
1925 | {Short4 t0; t0 = src0.x; t0 = CmpGT(t0, Short4(0x0800)); Short4 t1; t1 = src1.x; t1 = t1 & t0; t0 = ~t0 & src2.x; t0 = t0 | t1; dst.x = t0; }; |
1926 | {Short4 t0; t0 = src0.y; t0 = CmpGT(t0, Short4(0x0800)); Short4 t1; t1 = src1.y; t1 = t1 & t0; t0 = ~t0 & src2.y; t0 = t0 | t1; dst.y = t0; }; |
1927 | {Short4 t0; t0 = src0.z; t0 = CmpGT(t0, Short4(0x0800)); Short4 t1; t1 = src1.z; t1 = t1 & t0; t0 = ~t0 & src2.z; t0 = t0 | t1; dst.z = t0; }; |
1928 | {Short4 t0; t0 = src0.w; t0 = CmpGT(t0, Short4(0x0800)); Short4 t1; t1 = src1.w; t1 = t1 & t0; t0 = ~t0 & src2.w; t0 = t0 | t1; dst.w = t0; }; |
1929 | } |
1930 | |
1931 | void PixelPipeline::CMP(Vector4s &dst, Vector4s &src0, Vector4s &src1, Vector4s &src2) |
1932 | { |
1933 | {Short4 t0 = CmpGT(Short4(0x0000), src0.x); Short4 t1; t1 = src2.x; t1 &= t0; t0 = ~t0 & src1.x; t0 |= t1; dst.x = t0; }; |
1934 | {Short4 t0 = CmpGT(Short4(0x0000), src0.y); Short4 t1; t1 = src2.y; t1 &= t0; t0 = ~t0 & src1.y; t0 |= t1; dst.y = t0; }; |
1935 | {Short4 t0 = CmpGT(Short4(0x0000), src0.z); Short4 t1; t1 = src2.z; t1 &= t0; t0 = ~t0 & src1.z; t0 |= t1; dst.z = t0; }; |
1936 | {Short4 t0 = CmpGT(Short4(0x0000), src0.w); Short4 t1; t1 = src2.w; t1 &= t0; t0 = ~t0 & src1.w; t0 |= t1; dst.w = t0; }; |
1937 | } |
1938 | |
1939 | void PixelPipeline::BEM(Vector4s &dst, Vector4s &src0, Vector4s &src1, int stage) |
1940 | { |
1941 | Short4 t0; |
1942 | Short4 t1; |
1943 | |
1944 | // dst.x = src0.x + BUMPENVMAT00(stage) * src1.x + BUMPENVMAT10(stage) * src1.y |
1945 | t0 = MulHigh(src1.x, *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4W[0][0]))); t0 = t0 << 4; // FIXME: Matrix components range? Overflow hazard. |
1946 | t1 = MulHigh(src1.y, *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4W[1][0]))); t1 = t1 << 4; // FIXME: Matrix components range? Overflow hazard. |
1947 | t0 = AddSat(t0, t1); |
1948 | t0 = AddSat(t0, src0.x); |
1949 | dst.x = t0; |
1950 | |
1951 | // dst.y = src0.y + BUMPENVMAT01(stage) * src1.x + BUMPENVMAT11(stage) * src1.y |
1952 | t0 = MulHigh(src1.x, *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4W[0][1]))); t0 = t0 << 4; // FIXME: Matrix components range? Overflow hazard. |
1953 | t1 = MulHigh(src1.y, *Pointer<Short4>(data + OFFSET(DrawData, textureStage[stage].bumpmapMatrix4W[1][1]))); t1 = t1 << 4; // FIXME: Matrix components range? Overflow hazard. |
1954 | t0 = AddSat(t0, t1); |
1955 | t0 = AddSat(t0, src0.y); |
1956 | dst.y = t0; |
1957 | } |
1958 | } |
1959 | |
1960 | |