| 1 | /* |
| 2 | * Copyright (C) 2011 Gilead Kutnick "Exophase" <exophase@gmail.com> |
| 3 | * |
| 4 | * This program is free software; you can redistribute it and/or |
| 5 | * modify it under the terms of the GNU General Public License as |
| 6 | * published by the Free Software Foundation; either version 2 of |
| 7 | * the License, or (at your option) any later version. |
| 8 | * |
| 9 | * This program is distributed in the hope that it will be useful, |
| 10 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 12 | * General Public License for more details. |
| 13 | */ |
| 14 | |
| 15 | #include <stdio.h> |
| 16 | #include <stdlib.h> |
| 17 | #include <stdint.h> |
| 18 | #include <string.h> |
| 19 | #include <assert.h> |
| 20 | |
| 21 | #include "common.h" |
| 22 | #ifndef NEON_BUILD |
| 23 | #include "vector_ops.h" |
| 24 | #endif |
| 25 | #include "psx_gpu_simd.h" |
| 26 | |
| 27 | #if 0 |
| 28 | void dump_r_d(const char *name, void *dump); |
| 29 | void dump_r_q(const char *name, void *dump); |
| 30 | #define dumprd(n) dump_r_d(#n, n.e) |
| 31 | #define dumprq(n) dump_r_q(#n, n.e) |
| 32 | #endif |
| 33 | |
| 34 | u32 span_pixels = 0; |
| 35 | u32 span_pixel_blocks = 0; |
| 36 | u32 spans = 0; |
| 37 | u32 triangles = 0; |
| 38 | u32 sprites = 0; |
| 39 | u32 sprites_4bpp = 0; |
| 40 | u32 sprites_8bpp = 0; |
| 41 | u32 sprites_16bpp = 0; |
| 42 | u32 sprite_blocks = 0; |
| 43 | u32 sprites_untextured = 0; |
| 44 | u32 lines = 0; |
| 45 | u32 trivial_rejects = 0; |
| 46 | u32 texels_4bpp = 0; |
| 47 | u32 texels_8bpp = 0; |
| 48 | u32 texels_16bpp = 0; |
| 49 | u32 texel_blocks_4bpp = 0; |
| 50 | u32 texel_blocks_8bpp = 0; |
| 51 | u32 texel_blocks_16bpp = 0; |
| 52 | u32 texel_blocks_untextured = 0; |
| 53 | u32 blend_blocks = 0; |
| 54 | u32 render_buffer_flushes = 0; |
| 55 | u32 state_changes = 0; |
| 56 | u32 left_split_triangles = 0; |
| 57 | u32 flat_triangles = 0; |
| 58 | u32 clipped_triangles = 0; |
| 59 | u32 zero_block_spans = 0; |
| 60 | u32 texture_cache_loads = 0; |
| 61 | u32 false_modulated_blocks = 0; |
| 62 | |
| 63 | #define stats_add(stat, count) // stat += count |
| 64 | |
| 65 | /* double size for enhancement */ |
| 66 | u32 reciprocal_table[512 * 2]; |
| 67 | |
| 68 | |
| 69 | typedef s32 fixed_type; |
| 70 | |
| 71 | #define EDGE_STEP_BITS 32 |
| 72 | #define FIXED_BITS 12 |
| 73 | |
| 74 | #define fixed_center(value) \ |
| 75 | ((((fixed_type)(value)) << FIXED_BITS) + (1 << (FIXED_BITS - 1))) \ |
| 76 | |
| 77 | #define int_to_fixed(value) \ |
| 78 | (((fixed_type)(value)) << FIXED_BITS) \ |
| 79 | |
| 80 | #define fixed_to_int(value) \ |
| 81 | ((value) >> FIXED_BITS) \ |
| 82 | |
| 83 | #define fixed_to_double(value) \ |
| 84 | ((value) / (double)(1 << FIXED_BITS)) \ |
| 85 | |
| 86 | #define double_to_fixed(value) \ |
| 87 | (fixed_type)(((value) * (double)(1 << FIXED_BITS))) \ |
| 88 | |
| 89 | typedef void (setup_blocks_function_type)(psx_gpu_struct *psx_gpu); |
| 90 | typedef void (texture_blocks_function_type)(psx_gpu_struct *psx_gpu); |
| 91 | typedef void (shade_blocks_function_type)(psx_gpu_struct *psx_gpu); |
| 92 | typedef void (blend_blocks_function_type)(psx_gpu_struct *psx_gpu); |
| 93 | |
| 94 | typedef void (setup_sprite_function_type)(psx_gpu_struct *psx_gpu, s32 x, |
| 95 | s32 y, s32 u, s32 v, s32 width, s32 height, u32 color); |
| 96 | |
| 97 | struct render_block_handler_struct |
| 98 | { |
| 99 | void *setup_blocks; |
| 100 | texture_blocks_function_type *texture_blocks; |
| 101 | shade_blocks_function_type *shade_blocks; |
| 102 | blend_blocks_function_type *blend_blocks; |
| 103 | }; |
| 104 | |
| 105 | #ifndef NEON_BUILD |
| 106 | |
| 107 | u32 fixed_reciprocal(u32 denominator, u32 *_shift) |
| 108 | { |
| 109 | u32 shift = __builtin_clz(denominator); |
| 110 | u32 denominator_normalized = denominator << shift; |
| 111 | |
| 112 | double numerator = (1ULL << 62) + denominator_normalized; |
| 113 | double numerator_b; |
| 114 | |
| 115 | double denominator_normalized_dp_b; |
| 116 | u64 denominator_normalized_dp_u64; |
| 117 | |
| 118 | u32 reciprocal; |
| 119 | double reciprocal_dp; |
| 120 | |
| 121 | u64 numerator_u64 = (denominator_normalized >> 10) | |
| 122 | ((u64)(62 + 1023) << 52); |
| 123 | *((u64 *)(&numerator_b)) = numerator_u64; |
| 124 | |
| 125 | denominator_normalized_dp_u64 = |
| 126 | (u64)(denominator_normalized << 21) | |
| 127 | ((u64)((denominator_normalized >> 11) + ((1022 + 31) << 20)) << 32); |
| 128 | *((u64 *)(&denominator_normalized_dp_b)) = denominator_normalized_dp_u64; |
| 129 | |
| 130 | // Implement with a DP divide |
| 131 | reciprocal_dp = numerator / denominator_normalized_dp_b; |
| 132 | reciprocal = reciprocal_dp; |
| 133 | |
| 134 | if(reciprocal == 0x80000001) |
| 135 | reciprocal = 0x80000000; |
| 136 | |
| 137 | *_shift = 62 - shift; |
| 138 | return reciprocal; |
| 139 | } |
| 140 | |
| 141 | double reciprocal_estimate(double a) |
| 142 | { |
| 143 | int q, s; |
| 144 | double r; |
| 145 | |
| 146 | q = (int)(a * 512.0); |
| 147 | /* a in units of 1/512 rounded down */ |
| 148 | r = 1.0 / (((double)q + 0.5) / 512.0); /* reciprocal r */ |
| 149 | s = (int)(256.0 * r + 0.5); |
| 150 | |
| 151 | /* r in units of 1/256 rounded to nearest */ |
| 152 | |
| 153 | return (double)s / 256.0; |
| 154 | } |
| 155 | |
| 156 | u32 reciprocal_estimate_u32(u32 value) |
| 157 | { |
| 158 | u64 dp_value_u64; |
| 159 | volatile double dp_value; |
| 160 | volatile u64 *dp_value_ptr = (volatile u64 *)&dp_value; |
| 161 | |
| 162 | if((value >> 31) == 0) |
| 163 | return 0xFFFFFFFF; |
| 164 | |
| 165 | dp_value_u64 = (0x3FEULL << (31 + 21)) | ((u64)(value & 0x7FFFFFFF) << 21); |
| 166 | |
| 167 | *dp_value_ptr = dp_value_u64; |
| 168 | |
| 169 | dp_value = reciprocal_estimate(dp_value); |
| 170 | dp_value_u64 = *dp_value_ptr; |
| 171 | |
| 172 | return (0x80000000 | ((dp_value_u64 >> 21) & 0x7FFFFFFF)); |
| 173 | } |
| 174 | |
| 175 | u32 fixed_reciprocal_nr(u32 value, u32 *_shift) |
| 176 | { |
| 177 | u32 shift = __builtin_clz(value); |
| 178 | u32 value_normalized = value << shift; |
| 179 | |
| 180 | *_shift = 62 - shift; |
| 181 | |
| 182 | value_normalized -= 2; |
| 183 | |
| 184 | u32 reciprocal_normalized = reciprocal_estimate_u32(value_normalized) >> 1; |
| 185 | |
| 186 | u32 temp = -(((u64)value_normalized * (u32)reciprocal_normalized) >> 31); |
| 187 | reciprocal_normalized = (((u64)reciprocal_normalized * temp) >> 31); |
| 188 | temp = -(((u64)value_normalized * (u32)reciprocal_normalized) >> 31); |
| 189 | reciprocal_normalized = (((u64)reciprocal_normalized * temp) >> 31); |
| 190 | temp = -(((u64)value_normalized * (u32)reciprocal_normalized) >> 31); |
| 191 | reciprocal_normalized = (((u64)reciprocal_normalized * temp) >> 31); |
| 192 | |
| 193 | return reciprocal_normalized; |
| 194 | } |
| 195 | |
| 196 | #endif |
| 197 | |
| 198 | |
| 199 | s32 triangle_signed_area_x2(s32 x0, s32 y0, s32 x1, s32 y1, s32 x2, s32 y2) |
| 200 | { |
| 201 | return ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)); |
| 202 | } |
| 203 | |
| 204 | u32 texture_region_mask(s32 x1, s32 y1, s32 x2, s32 y2) |
| 205 | { |
| 206 | s32 coverage_x, coverage_y; |
| 207 | |
| 208 | u32 mask_up_left; |
| 209 | u32 mask_down_right; |
| 210 | |
| 211 | coverage_x = x2 >> 6; |
| 212 | coverage_y = y2 >> 8; |
| 213 | |
| 214 | if(coverage_x < 0) |
| 215 | coverage_x = 0; |
| 216 | |
| 217 | if(coverage_x > 31) |
| 218 | coverage_x = 31; |
| 219 | |
| 220 | mask_down_right = ~(0xFFFFFFFF << (coverage_x + 1)) & 0xFFFF; |
| 221 | |
| 222 | if(coverage_y >= 1) |
| 223 | mask_down_right |= mask_down_right << 16; |
| 224 | |
| 225 | coverage_x = x1 >> 6; |
| 226 | |
| 227 | mask_up_left = 0xFFFF0000 << coverage_x; |
| 228 | if(coverage_x < 0) |
| 229 | mask_up_left = 0xFFFF0000; |
| 230 | |
| 231 | coverage_y = y1 >> 8; |
| 232 | if(coverage_y <= 0) |
| 233 | mask_up_left |= mask_up_left >> 16; |
| 234 | |
| 235 | return mask_up_left & mask_down_right; |
| 236 | } |
| 237 | |
| 238 | u32 invalidate_texture_cache_region(psx_gpu_struct *psx_gpu, u32 x1, u32 y1, |
| 239 | u32 x2, u32 y2) |
| 240 | { |
| 241 | u32 mask = texture_region_mask(x1, y1, x2, y2); |
| 242 | |
| 243 | psx_gpu->dirty_textures_4bpp_mask |= mask; |
| 244 | psx_gpu->dirty_textures_8bpp_mask |= mask; |
| 245 | psx_gpu->dirty_textures_8bpp_alternate_mask |= mask; |
| 246 | |
| 247 | return mask; |
| 248 | } |
| 249 | |
| 250 | u32 invalidate_texture_cache_region_viewport(psx_gpu_struct *psx_gpu, u32 x1, |
| 251 | u32 y1, u32 x2, u32 y2) |
| 252 | { |
| 253 | u32 mask = texture_region_mask(x1, y1, x2, y2) & |
| 254 | psx_gpu->viewport_mask; |
| 255 | |
| 256 | psx_gpu->dirty_textures_4bpp_mask |= mask; |
| 257 | psx_gpu->dirty_textures_8bpp_mask |= mask; |
| 258 | psx_gpu->dirty_textures_8bpp_alternate_mask |= mask; |
| 259 | |
| 260 | return mask; |
| 261 | } |
| 262 | |
| 263 | static void update_texture_cache_region_(psx_gpu_struct *psx_gpu, |
| 264 | u32 x1, u32 y1, u32 x2, u32 y2) |
| 265 | { |
| 266 | u32 mask = texture_region_mask(x1, y1, x2, y2); |
| 267 | u32 texture_page; |
| 268 | u8 *texture_page_ptr; |
| 269 | u16 *vram_ptr; |
| 270 | u32 texel_block; |
| 271 | u32 sub_x, sub_y; |
| 272 | |
| 273 | psx_gpu->dirty_textures_8bpp_mask |= mask; |
| 274 | psx_gpu->dirty_textures_8bpp_alternate_mask |= mask; |
| 275 | |
| 276 | if ((psx_gpu->dirty_textures_4bpp_mask & mask) == 0 && |
| 277 | (x1 & 3) == 0 && (y1 & 15) == 0 && x2 - x1 < 4 && y2 - y1 < 16) |
| 278 | { |
| 279 | texture_page = ((x1 / 64) & 15) + (y1 / 256) * 16; |
| 280 | texture_page_ptr = psx_gpu->texture_4bpp_cache[texture_page]; |
| 281 | texture_page_ptr += (x1 / 4 & 15) * 16*16 + (y1 / 16 & 15) * 16*16*16; |
| 282 | vram_ptr = psx_gpu->vram_ptr + x1 + y1 * 1024; |
| 283 | sub_x = 4; |
| 284 | sub_y = 16; |
| 285 | |
| 286 | while(sub_y) |
| 287 | { |
| 288 | while(sub_x) |
| 289 | { |
| 290 | texel_block = *vram_ptr; |
| 291 | |
| 292 | texture_page_ptr[0] = texel_block & 0xF; |
| 293 | texture_page_ptr[1] = (texel_block >> 4) & 0xF; |
| 294 | texture_page_ptr[2] = (texel_block >> 8) & 0xF; |
| 295 | texture_page_ptr[3] = texel_block >> 12; |
| 296 | |
| 297 | vram_ptr++; |
| 298 | texture_page_ptr += 4; |
| 299 | |
| 300 | sub_x--; |
| 301 | } |
| 302 | |
| 303 | vram_ptr -= 4; |
| 304 | sub_x = 4; |
| 305 | |
| 306 | sub_y--; |
| 307 | vram_ptr += 1024; |
| 308 | } |
| 309 | } |
| 310 | else |
| 311 | { |
| 312 | psx_gpu->dirty_textures_4bpp_mask |= mask; |
| 313 | } |
| 314 | } |
| 315 | |
| 316 | void update_texture_cache_region(psx_gpu_struct *psx_gpu, u32 x1, u32 y1, |
| 317 | u32 x2, u32 y2) |
| 318 | { |
| 319 | s32 w = x2 - x1; |
| 320 | do |
| 321 | { |
| 322 | x2 = x1 + w; |
| 323 | if (x2 > 1023) |
| 324 | x2 = 1023; |
| 325 | update_texture_cache_region_(psx_gpu, x1, y1, x2, y2); |
| 326 | w -= x2 - x1; |
| 327 | x1 = 0; |
| 328 | } |
| 329 | while (unlikely(w > 0)); |
| 330 | } |
| 331 | |
| 332 | #ifndef NEON_BUILD |
| 333 | |
| 334 | void update_texture_4bpp_cache(psx_gpu_struct *psx_gpu) |
| 335 | { |
| 336 | u32 current_texture_page = psx_gpu->current_texture_page; |
| 337 | u8 *texture_page_ptr = psx_gpu->texture_page_base; |
| 338 | u16 *vram_ptr = psx_gpu->vram_ptr; |
| 339 | |
| 340 | u32 texel_block; |
| 341 | u32 tile_x, tile_y; |
| 342 | u32 sub_x, sub_y; |
| 343 | |
| 344 | vram_ptr += (current_texture_page >> 4) * 256 * 1024; |
| 345 | vram_ptr += (current_texture_page & 0xF) * 64; |
| 346 | |
| 347 | texture_cache_loads++; |
| 348 | |
| 349 | tile_y = 16; |
| 350 | tile_x = 16; |
| 351 | sub_x = 4; |
| 352 | sub_y = 16; |
| 353 | |
| 354 | psx_gpu->dirty_textures_4bpp_mask &= ~(psx_gpu->current_texture_mask); |
| 355 | |
| 356 | while(tile_y) |
| 357 | { |
| 358 | while(tile_x) |
| 359 | { |
| 360 | while(sub_y) |
| 361 | { |
| 362 | while(sub_x) |
| 363 | { |
| 364 | texel_block = *vram_ptr; |
| 365 | |
| 366 | texture_page_ptr[0] = texel_block & 0xF; |
| 367 | texture_page_ptr[1] = (texel_block >> 4) & 0xF; |
| 368 | texture_page_ptr[2] = (texel_block >> 8) & 0xF; |
| 369 | texture_page_ptr[3] = texel_block >> 12; |
| 370 | |
| 371 | vram_ptr++; |
| 372 | texture_page_ptr += 4; |
| 373 | |
| 374 | sub_x--; |
| 375 | } |
| 376 | |
| 377 | vram_ptr -= 4; |
| 378 | sub_x = 4; |
| 379 | |
| 380 | sub_y--; |
| 381 | vram_ptr += 1024; |
| 382 | } |
| 383 | |
| 384 | sub_y = 16; |
| 385 | |
| 386 | vram_ptr -= (1024 * 16) - 4; |
| 387 | tile_x--; |
| 388 | } |
| 389 | |
| 390 | tile_x = 16; |
| 391 | |
| 392 | vram_ptr += (16 * 1024) - (4 * 16); |
| 393 | tile_y--; |
| 394 | } |
| 395 | } |
| 396 | |
| 397 | void update_texture_8bpp_cache_slice(psx_gpu_struct *psx_gpu, |
| 398 | u32 texture_page) |
| 399 | { |
| 400 | u16 *texture_page_ptr = psx_gpu->texture_page_base; |
| 401 | u16 *vram_ptr = psx_gpu->vram_ptr; |
| 402 | |
| 403 | u32 tile_x, tile_y; |
| 404 | u32 sub_y; |
| 405 | |
| 406 | vec_8x16u texels; |
| 407 | |
| 408 | texture_cache_loads++; |
| 409 | |
| 410 | vram_ptr += (texture_page >> 4) * 256 * 1024; |
| 411 | vram_ptr += (texture_page & 0xF) * 64; |
| 412 | |
| 413 | if((texture_page ^ psx_gpu->current_texture_page) & 0x1) |
| 414 | texture_page_ptr += (8 * 16) * 8; |
| 415 | |
| 416 | tile_x = 8; |
| 417 | tile_y = 16; |
| 418 | |
| 419 | sub_y = 16; |
| 420 | |
| 421 | while(tile_y) |
| 422 | { |
| 423 | while(tile_x) |
| 424 | { |
| 425 | while(sub_y) |
| 426 | { |
| 427 | load_128b(texels, vram_ptr); |
| 428 | store_128b(texels, texture_page_ptr); |
| 429 | |
| 430 | texture_page_ptr += 8; |
| 431 | vram_ptr += 1024; |
| 432 | |
| 433 | sub_y--; |
| 434 | } |
| 435 | |
| 436 | sub_y = 16; |
| 437 | |
| 438 | vram_ptr -= (1024 * 16); |
| 439 | vram_ptr += 8; |
| 440 | |
| 441 | tile_x--; |
| 442 | } |
| 443 | |
| 444 | tile_x = 8; |
| 445 | |
| 446 | vram_ptr -= (8 * 8); |
| 447 | vram_ptr += (16 * 1024); |
| 448 | |
| 449 | texture_page_ptr += (8 * 16) * 8; |
| 450 | tile_y--; |
| 451 | } |
| 452 | } |
| 453 | |
| 454 | #endif |
| 455 | |
| 456 | |
| 457 | void update_texture_8bpp_cache(psx_gpu_struct *psx_gpu) |
| 458 | { |
| 459 | u32 current_texture_page = psx_gpu->current_texture_page; |
| 460 | u32 update_textures = |
| 461 | psx_gpu->dirty_textures_8bpp_mask & psx_gpu->current_texture_mask; |
| 462 | |
| 463 | psx_gpu->dirty_textures_8bpp_mask &= ~update_textures; |
| 464 | |
| 465 | if(update_textures & (1 << current_texture_page)) |
| 466 | { |
| 467 | update_texture_8bpp_cache_slice(psx_gpu, current_texture_page); |
| 468 | update_textures &= ~(1 << current_texture_page); |
| 469 | } |
| 470 | |
| 471 | if(update_textures) |
| 472 | { |
| 473 | u32 adjacent_texture_page = ((current_texture_page + 1) & 0xF) | |
| 474 | (current_texture_page & 0x10); |
| 475 | |
| 476 | update_texture_8bpp_cache_slice(psx_gpu, adjacent_texture_page); |
| 477 | } |
| 478 | } |
| 479 | |
| 480 | void flush_render_block_buffer(psx_gpu_struct *psx_gpu) |
| 481 | { |
| 482 | if((psx_gpu->render_mode & RENDER_INTERLACE_ENABLED) && |
| 483 | (psx_gpu->primitive_type == PRIMITIVE_TYPE_SPRITE)) |
| 484 | { |
| 485 | u32 num_blocks_dest = 0; |
| 486 | block_struct *block_src = psx_gpu->blocks; |
| 487 | block_struct *block_dest = psx_gpu->blocks; |
| 488 | |
| 489 | u16 *vram_ptr = psx_gpu->vram_ptr; |
| 490 | u32 i; |
| 491 | |
| 492 | if(psx_gpu->render_mode & RENDER_INTERLACE_ODD) |
| 493 | { |
| 494 | for(i = 0; i < psx_gpu->num_blocks; i++) |
| 495 | { |
| 496 | u32 fb_offset = (u32)((u8 *)block_src->fb_ptr - (u8 *)vram_ptr); |
| 497 | if(fb_offset & (1 << 11)) |
| 498 | { |
| 499 | *block_dest = *block_src; |
| 500 | num_blocks_dest++; |
| 501 | block_dest++; |
| 502 | } |
| 503 | block_src++; |
| 504 | } |
| 505 | } |
| 506 | else |
| 507 | { |
| 508 | for(i = 0; i < psx_gpu->num_blocks; i++) |
| 509 | { |
| 510 | u32 fb_offset = (u32)((u8 *)block_src->fb_ptr - (u8 *)vram_ptr); |
| 511 | if((fb_offset & (1 << 11)) == 0) |
| 512 | { |
| 513 | *block_dest = *block_src; |
| 514 | num_blocks_dest++; |
| 515 | block_dest++; |
| 516 | } |
| 517 | block_src++; |
| 518 | } |
| 519 | } |
| 520 | |
| 521 | psx_gpu->num_blocks = num_blocks_dest; |
| 522 | } |
| 523 | |
| 524 | if(psx_gpu->num_blocks) |
| 525 | { |
| 526 | render_block_handler_struct *render_block_handler = |
| 527 | psx_gpu->render_block_handler; |
| 528 | |
| 529 | render_block_handler->texture_blocks(psx_gpu); |
| 530 | render_block_handler->shade_blocks(psx_gpu); |
| 531 | render_block_handler->blend_blocks(psx_gpu); |
| 532 | |
| 533 | #ifdef PROFILE |
| 534 | span_pixel_blocks += psx_gpu->num_blocks; |
| 535 | render_buffer_flushes++; |
| 536 | #endif |
| 537 | |
| 538 | psx_gpu->num_blocks = 0; |
| 539 | } |
| 540 | } |
| 541 | |
| 542 | |
| 543 | #ifndef NEON_BUILD |
| 544 | |
| 545 | #define setup_gradient_calculation_input(set, vertex) \ |
| 546 | /* First type is: uvrg bxxx xxxx */\ |
| 547 | /* Second type is: yyyy ybyy uvrg */\ |
| 548 | /* Since x_a and y_c are the same the same variable is used for both. */\ |
| 549 | x##set##_a_y##set##_c.e[0] = vertex->u; \ |
| 550 | x##set##_a_y##set##_c.e[1] = vertex->v; \ |
| 551 | x##set##_a_y##set##_c.e[2] = vertex->r; \ |
| 552 | x##set##_a_y##set##_c.e[3] = vertex->g; \ |
| 553 | dup_4x16b(x##set##_b, vertex->x); \ |
| 554 | dup_4x16b(x##set##_c, vertex->x); \ |
| 555 | dup_4x16b(y##set##_a, vertex->y); \ |
| 556 | dup_4x16b(y##set##_b, vertex->y); \ |
| 557 | x##set##_b.e[0] = vertex->b; \ |
| 558 | y##set##_b.e[1] = vertex->b \ |
| 559 | |
| 560 | |
| 561 | void compute_all_gradients(psx_gpu_struct *psx_gpu, vertex_struct *a, |
| 562 | vertex_struct *b, vertex_struct *c) |
| 563 | { |
| 564 | u32 triangle_area = psx_gpu->triangle_area; |
| 565 | u32 winding_mask_scalar; |
| 566 | |
| 567 | u32 triangle_area_shift; |
| 568 | u64 triangle_area_reciprocal = |
| 569 | fixed_reciprocal(triangle_area, &triangle_area_shift); |
| 570 | triangle_area_shift = -(triangle_area_shift - FIXED_BITS); |
| 571 | |
| 572 | // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) = |
| 573 | // ( d0 * d1 ) - ( d2 * d3 ) = |
| 574 | // ( m0 ) - ( m1 ) = gradient |
| 575 | |
| 576 | // This is split to do 12 elements at a time over three sets: a, b, and c. |
| 577 | // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so |
| 578 | // two of the slots are unused. |
| 579 | |
| 580 | // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as |
| 581 | // is g. |
| 582 | |
| 583 | vec_4x16s x0_a_y0_c, x0_b, x0_c; |
| 584 | vec_4x16s y0_a, y0_b; |
| 585 | vec_4x16s x1_a_y1_c, x1_b, x1_c; |
| 586 | vec_4x16s y1_a, y1_b; |
| 587 | vec_4x16s x2_a_y2_c, x2_b, x2_c; |
| 588 | vec_4x16s y2_a, y2_b; |
| 589 | |
| 590 | vec_4x32u uvrg_base; |
| 591 | vec_4x32u b_base; |
| 592 | vec_4x32u uvrgb_phase; |
| 593 | |
| 594 | vec_4x16s d0_a_d3_c, d0_b, d0_c; |
| 595 | vec_4x16s d1_a, d1_b, d1_c_d2_a; |
| 596 | vec_4x16s d2_b, d2_c; |
| 597 | vec_4x16s d3_a, d3_b; |
| 598 | |
| 599 | vec_4x32s m0_a, m0_b, m0_c; |
| 600 | vec_4x32s m1_a, m1_b, m1_c; |
| 601 | |
| 602 | vec_4x32u gradient_area_a, gradient_area_c; |
| 603 | vec_2x32u gradient_area_b; |
| 604 | |
| 605 | vec_4x32u gradient_area_sign_a, gradient_area_sign_c; |
| 606 | vec_2x32u gradient_area_sign_b; |
| 607 | vec_4x32u winding_mask; |
| 608 | |
| 609 | vec_2x64u gradient_wide_a0, gradient_wide_a1; |
| 610 | vec_2x64u gradient_wide_c0, gradient_wide_c1; |
| 611 | vec_2x64u gradient_wide_b; |
| 612 | |
| 613 | vec_4x32u gradient_a, gradient_c; |
| 614 | vec_2x32u gradient_b; |
| 615 | vec_16x8s gradient_shift; |
| 616 | |
| 617 | setup_gradient_calculation_input(0, a); |
| 618 | setup_gradient_calculation_input(1, b); |
| 619 | setup_gradient_calculation_input(2, c); |
| 620 | |
| 621 | dup_4x32b(uvrgb_phase, psx_gpu->uvrgb_phase); |
| 622 | shl_long_4x16b(uvrg_base, x0_a_y0_c, 16); |
| 623 | shl_long_4x16b(b_base, x0_b, 16); |
| 624 | |
| 625 | add_4x32b(uvrg_base, uvrg_base, uvrgb_phase); |
| 626 | add_4x32b(b_base, b_base, uvrgb_phase); |
| 627 | |
| 628 | // Can probably pair these, but it'll require careful register allocation |
| 629 | sub_4x16b(d0_a_d3_c, x1_a_y1_c, x0_a_y0_c); |
| 630 | sub_4x16b(d1_c_d2_a, x2_a_y2_c, x1_a_y1_c); |
| 631 | |
| 632 | sub_4x16b(d0_b, x1_b, x0_b); |
| 633 | sub_4x16b(d0_c, x1_c, x0_c); |
| 634 | |
| 635 | sub_4x16b(d1_a, y2_a, y1_a); |
| 636 | sub_4x16b(d1_b, y2_b, y1_b); |
| 637 | |
| 638 | sub_4x16b(d2_b, x2_b, x1_b); |
| 639 | sub_4x16b(d2_c, x2_c, x1_c); |
| 640 | |
| 641 | sub_4x16b(d3_a, y1_a, y0_a); |
| 642 | sub_4x16b(d3_b, y1_b, y0_b); |
| 643 | |
| 644 | mul_long_4x16b(m0_a, d0_a_d3_c, d1_a); |
| 645 | mul_long_4x16b(m0_b, d0_b, d1_b); |
| 646 | mul_long_4x16b(m0_c, d0_c, d1_c_d2_a); |
| 647 | |
| 648 | mul_long_4x16b(m1_a, d1_c_d2_a, d3_a); |
| 649 | mul_long_4x16b(m1_b, d2_b, d3_b); |
| 650 | mul_long_4x16b(m1_c, d2_c, d0_a_d3_c); |
| 651 | |
| 652 | sub_4x32b(gradient_area_a, m0_a, m1_a); |
| 653 | sub_2x32b(gradient_area_b, m0_b.low, m1_b.low); |
| 654 | sub_4x32b(gradient_area_c, m0_c, m1_c); |
| 655 | |
| 656 | cmpltz_4x32b(gradient_area_sign_a, gradient_area_a); |
| 657 | cmpltz_2x32b(gradient_area_sign_b, gradient_area_b); |
| 658 | cmpltz_4x32b(gradient_area_sign_c, gradient_area_c); |
| 659 | |
| 660 | abs_4x32b(gradient_area_a, gradient_area_a); |
| 661 | abs_2x32b(gradient_area_b, gradient_area_b); |
| 662 | abs_4x32b(gradient_area_c, gradient_area_c); |
| 663 | |
| 664 | winding_mask_scalar = -psx_gpu->triangle_winding; |
| 665 | |
| 666 | dup_4x32b(winding_mask, winding_mask_scalar); |
| 667 | eor_4x32b(gradient_area_sign_a, gradient_area_sign_a, winding_mask); |
| 668 | eor_2x32b(gradient_area_sign_b, gradient_area_sign_b, winding_mask); |
| 669 | eor_4x32b(gradient_area_sign_c, gradient_area_sign_c, winding_mask); |
| 670 | |
| 671 | mul_scalar_long_2x32b(gradient_wide_a0, |
| 672 | vector_cast(vec_2x32s, gradient_area_a.low), |
| 673 | (s64)triangle_area_reciprocal); |
| 674 | mul_scalar_long_2x32b(gradient_wide_a1, |
| 675 | vector_cast(vec_2x32s, gradient_area_a.high), |
| 676 | (s64)triangle_area_reciprocal); |
| 677 | mul_scalar_long_2x32b(gradient_wide_b, |
| 678 | vector_cast(vec_2x32s, gradient_area_b), |
| 679 | (s64)triangle_area_reciprocal); |
| 680 | mul_scalar_long_2x32b(gradient_wide_c0, |
| 681 | vector_cast(vec_2x32s, gradient_area_c.low), |
| 682 | (s64)triangle_area_reciprocal); |
| 683 | mul_scalar_long_2x32b(gradient_wide_c1, |
| 684 | vector_cast(vec_2x32s, gradient_area_c.high), |
| 685 | (s64)triangle_area_reciprocal); |
| 686 | |
| 687 | dup_16x8b(gradient_shift, triangle_area_shift); |
| 688 | shl_reg_2x64b(gradient_wide_a0, gradient_wide_a0, |
| 689 | vector_cast(vec_2x64u, gradient_shift)); |
| 690 | shl_reg_2x64b(gradient_wide_a1, gradient_wide_a1, |
| 691 | vector_cast(vec_2x64u, gradient_shift)); |
| 692 | shl_reg_2x64b(gradient_wide_b, gradient_wide_b, |
| 693 | vector_cast(vec_2x64u, gradient_shift)); |
| 694 | shl_reg_2x64b(gradient_wide_c0, gradient_wide_c0, |
| 695 | vector_cast(vec_2x64u, gradient_shift)); |
| 696 | shl_reg_2x64b(gradient_wide_c1, gradient_wide_c1, |
| 697 | vector_cast(vec_2x64u, gradient_shift)); |
| 698 | |
| 699 | mov_narrow_2x64b(gradient_a.low, gradient_wide_a0); |
| 700 | mov_narrow_2x64b(gradient_a.high, gradient_wide_a1); |
| 701 | mov_narrow_2x64b(gradient_b, gradient_wide_b); |
| 702 | mov_narrow_2x64b(gradient_c.low, gradient_wide_c0); |
| 703 | mov_narrow_2x64b(gradient_c.high, gradient_wide_c1); |
| 704 | |
| 705 | shl_4x32b(gradient_a, gradient_a, 4); |
| 706 | shl_2x32b(gradient_b, gradient_b, 4); |
| 707 | shl_4x32b(gradient_c, gradient_c, 4); |
| 708 | |
| 709 | eor_4x32b(gradient_a, gradient_a, gradient_area_sign_a); |
| 710 | eor_2x32b(gradient_b, gradient_b, gradient_area_sign_b); |
| 711 | eor_4x32b(gradient_c, gradient_c, gradient_area_sign_c); |
| 712 | |
| 713 | sub_4x32b(gradient_a, gradient_a, gradient_area_sign_a); |
| 714 | sub_2x32b(gradient_b, gradient_b, gradient_area_sign_b); |
| 715 | sub_4x32b(gradient_c, gradient_c, gradient_area_sign_c); |
| 716 | |
| 717 | u32 left_adjust = a->x; |
| 718 | mls_scalar_4x32b(uvrg_base, gradient_a, left_adjust); |
| 719 | mls_scalar_2x32b(b_base.low, gradient_b, left_adjust); |
| 720 | |
| 721 | vec_4x32u uvrg_dx2; |
| 722 | vec_2x32u b_dx2; |
| 723 | |
| 724 | vec_4x32u uvrg_dx3; |
| 725 | vec_2x32u b_dx3; |
| 726 | |
| 727 | vec_4x32u zero; |
| 728 | |
| 729 | eor_4x32b(zero, zero, zero); |
| 730 | add_4x32b(uvrg_dx2, gradient_a, gradient_a); |
| 731 | add_2x32b(b_dx2, gradient_b, gradient_b); |
| 732 | add_4x32b(uvrg_dx3, gradient_a, uvrg_dx2); |
| 733 | add_2x32b(b_dx3, gradient_b, b_dx2); |
| 734 | |
| 735 | // Can be done with vst4, assuming that the zero, dx, dx2, and dx3 are |
| 736 | // lined up properly |
| 737 | psx_gpu->u_block_span.e[0] = zero.e[0]; |
| 738 | psx_gpu->u_block_span.e[1] = gradient_a.e[0]; |
| 739 | psx_gpu->u_block_span.e[2] = uvrg_dx2.e[0]; |
| 740 | psx_gpu->u_block_span.e[3] = uvrg_dx3.e[0]; |
| 741 | |
| 742 | psx_gpu->v_block_span.e[0] = zero.e[1]; |
| 743 | psx_gpu->v_block_span.e[1] = gradient_a.e[1]; |
| 744 | psx_gpu->v_block_span.e[2] = uvrg_dx2.e[1]; |
| 745 | psx_gpu->v_block_span.e[3] = uvrg_dx3.e[1]; |
| 746 | |
| 747 | psx_gpu->r_block_span.e[0] = zero.e[2]; |
| 748 | psx_gpu->r_block_span.e[1] = gradient_a.e[2]; |
| 749 | psx_gpu->r_block_span.e[2] = uvrg_dx2.e[2]; |
| 750 | psx_gpu->r_block_span.e[3] = uvrg_dx3.e[2]; |
| 751 | |
| 752 | psx_gpu->g_block_span.e[0] = zero.e[3]; |
| 753 | psx_gpu->g_block_span.e[1] = gradient_a.e[3]; |
| 754 | psx_gpu->g_block_span.e[2] = uvrg_dx2.e[3]; |
| 755 | psx_gpu->g_block_span.e[3] = uvrg_dx3.e[3]; |
| 756 | |
| 757 | psx_gpu->b_block_span.e[0] = zero.e[0]; |
| 758 | psx_gpu->b_block_span.e[1] = gradient_b.e[0]; |
| 759 | psx_gpu->b_block_span.e[2] = b_dx2.e[0]; |
| 760 | psx_gpu->b_block_span.e[3] = b_dx3.e[0]; |
| 761 | |
| 762 | psx_gpu->uvrg = uvrg_base; |
| 763 | psx_gpu->b = b_base.e[0]; |
| 764 | |
| 765 | psx_gpu->uvrg_dx = gradient_a; |
| 766 | psx_gpu->uvrg_dy = gradient_c; |
| 767 | psx_gpu->b_dy = gradient_b.e[1]; |
| 768 | } |
| 769 | #endif |
| 770 | |
| 771 | #define vector_check(_a, _b) \ |
| 772 | if(memcmp(&_a, &_b, sizeof(_b))) \ |
| 773 | { \ |
| 774 | if(sizeof(_b) == 8) \ |
| 775 | { \ |
| 776 | printf("mismatch on %s vs %s: (%x %x) vs (%x %x)\n", \ |
| 777 | #_a, #_b, _a.e[0], _a.e[1], _b.e[0], _b.e[1]); \ |
| 778 | } \ |
| 779 | else \ |
| 780 | { \ |
| 781 | printf("mismatch on %s vs %s: (%x %x %x %x) vs (%x %x %x %x)\n", \ |
| 782 | #_a, #_b, _a.e[0], _a.e[1], _a.e[2], _a.e[3], _b.e[0], _b.e[1], \ |
| 783 | _b.e[2], _b.e[3]); \ |
| 784 | } \ |
| 785 | } \ |
| 786 | |
| 787 | #define scalar_check(_a, _b) \ |
| 788 | if(_a != _b) \ |
| 789 | printf("mismatch on %s %s: %x vs %x\n", #_a, #_b, _a, _b) \ |
| 790 | |
| 791 | |
| 792 | #if !defined(NEON_BUILD) && !defined(NDEBUG) |
| 793 | static void setup_spans_debug_check(psx_gpu_struct *psx_gpu, |
| 794 | edge_data_struct *span_edge_data_element) |
| 795 | { |
| 796 | u32 _num_spans = span_edge_data_element - psx_gpu->span_edge_data; |
| 797 | if (_num_spans > MAX_SPANS) |
| 798 | *(volatile int *)0 = 1; |
| 799 | if (_num_spans < psx_gpu->num_spans) |
| 800 | { |
| 801 | if(span_edge_data_element->num_blocks > MAX_BLOCKS_PER_ROW) |
| 802 | *(volatile int *)0 = 2; |
| 803 | if(span_edge_data_element->y >= 2048) |
| 804 | *(volatile int *)0 = 3; |
| 805 | } |
| 806 | } |
| 807 | #else |
| 808 | #define setup_spans_debug_check(psx_gpu, span_edge_data_element) |
| 809 | #endif |
| 810 | |
| 811 | #define setup_spans_prologue_alternate_yes() \ |
| 812 | vec_2x64s alternate_x; \ |
| 813 | vec_2x64s alternate_dx_dy; \ |
| 814 | vec_4x32s alternate_x_32; \ |
| 815 | vec_4x16u alternate_x_16; \ |
| 816 | \ |
| 817 | vec_4x16u alternate_select; \ |
| 818 | vec_4x16s y_mid_point; \ |
| 819 | \ |
| 820 | s32 y_b = v_b->y; \ |
| 821 | s64 edge_alt; \ |
| 822 | s32 edge_dx_dy_alt; \ |
| 823 | u32 edge_shift_alt \ |
| 824 | |
| 825 | #define setup_spans_prologue_alternate_no() \ |
| 826 | |
| 827 | #define setup_spans_prologue(alternate_active) \ |
| 828 | edge_data_struct *span_edge_data; \ |
| 829 | vec_4x32u *span_uvrg_offset; \ |
| 830 | u32 *span_b_offset; \ |
| 831 | \ |
| 832 | s32 clip; \ |
| 833 | \ |
| 834 | vec_2x64s edges_xy; \ |
| 835 | vec_2x32s edges_dx_dy; \ |
| 836 | vec_2x32u edge_shifts; \ |
| 837 | \ |
| 838 | vec_2x64s left_x, right_x; \ |
| 839 | vec_2x64s left_dx_dy, right_dx_dy; \ |
| 840 | vec_4x32s left_x_32, right_x_32; \ |
| 841 | vec_8x16s left_right_x_16; \ |
| 842 | vec_4x16s y_x4; \ |
| 843 | vec_8x16s left_edge; \ |
| 844 | vec_8x16s right_edge; \ |
| 845 | vec_4x16u span_shift; \ |
| 846 | \ |
| 847 | vec_2x32u c_0x01; \ |
| 848 | vec_4x16u c_0x04; \ |
| 849 | vec_4x16u c_0xFFFE; \ |
| 850 | vec_4x16u c_0x07; \ |
| 851 | \ |
| 852 | vec_2x32s x_starts; \ |
| 853 | vec_2x32s x_ends; \ |
| 854 | \ |
| 855 | s32 x_a = v_a->x; \ |
| 856 | s32 x_b = v_b->x; \ |
| 857 | s32 x_c = v_c->x; \ |
| 858 | s32 y_a = v_a->y; \ |
| 859 | s32 y_c = v_c->y; \ |
| 860 | \ |
| 861 | vec_4x32u uvrg = psx_gpu->uvrg; \ |
| 862 | vec_4x32u uvrg_dy = psx_gpu->uvrg_dy; \ |
| 863 | u32 b = psx_gpu->b; \ |
| 864 | u32 b_dy = psx_gpu->b_dy; \ |
| 865 | \ |
| 866 | dup_2x32b(c_0x01, 0x01); \ |
| 867 | setup_spans_prologue_alternate_##alternate_active() \ |
| 868 | |
| 869 | #define setup_spans_prologue_b() \ |
| 870 | span_edge_data = psx_gpu->span_edge_data; \ |
| 871 | span_uvrg_offset = psx_gpu->span_uvrg_offset; \ |
| 872 | span_b_offset = psx_gpu->span_b_offset; \ |
| 873 | \ |
| 874 | vec_8x16u c_0x0001; \ |
| 875 | vec_4x16u c_max_blocks_per_row; \ |
| 876 | \ |
| 877 | dup_8x16b(c_0x0001, 0x0001); \ |
| 878 | dup_8x16b(left_edge, psx_gpu->viewport_start_x); \ |
| 879 | dup_8x16b(right_edge, psx_gpu->viewport_end_x); \ |
| 880 | add_8x16b(right_edge, right_edge, c_0x0001); \ |
| 881 | dup_4x16b(c_0x04, 0x04); \ |
| 882 | dup_4x16b(c_0x07, 0x07); \ |
| 883 | dup_4x16b(c_0xFFFE, 0xFFFE); \ |
| 884 | dup_4x16b(c_max_blocks_per_row, MAX_BLOCKS_PER_ROW); \ |
| 885 | |
| 886 | |
| 887 | #define compute_edge_delta_x2() \ |
| 888 | { \ |
| 889 | vec_2x32s heights; \ |
| 890 | vec_2x32s height_reciprocals; \ |
| 891 | vec_2x32s heights_b; \ |
| 892 | vec_4x32u widths; \ |
| 893 | \ |
| 894 | u32 edge_shift = reciprocal_table[height]; \ |
| 895 | \ |
| 896 | dup_2x32b(heights, height); \ |
| 897 | sub_2x32b(widths, x_ends, x_starts); \ |
| 898 | \ |
| 899 | dup_2x32b(edge_shifts, edge_shift); \ |
| 900 | sub_2x32b(heights_b, heights, c_0x01); \ |
| 901 | shr_2x32b(height_reciprocals, edge_shifts, 10); \ |
| 902 | \ |
| 903 | mla_2x32b(heights_b, x_starts, heights); \ |
| 904 | bic_immediate_4x16b(vector_cast(vec_4x16u, edge_shifts), 0xE0); \ |
| 905 | mul_2x32b(edges_dx_dy, widths, height_reciprocals); \ |
| 906 | mul_long_2x32b(edges_xy, heights_b, height_reciprocals); \ |
| 907 | } \ |
| 908 | |
| 909 | #define compute_edge_delta_x3(start_c, height_a, height_b) \ |
| 910 | { \ |
| 911 | vec_2x32s heights; \ |
| 912 | vec_2x32s height_reciprocals; \ |
| 913 | vec_2x32s heights_b; \ |
| 914 | vec_2x32u widths; \ |
| 915 | \ |
| 916 | u32 width_alt; \ |
| 917 | s32 height_b_alt; \ |
| 918 | u32 height_reciprocal_alt; \ |
| 919 | \ |
| 920 | heights.e[0] = height_a; \ |
| 921 | heights.e[1] = height_b; \ |
| 922 | \ |
| 923 | edge_shifts.e[0] = reciprocal_table[height_a]; \ |
| 924 | edge_shifts.e[1] = reciprocal_table[height_b]; \ |
| 925 | edge_shift_alt = reciprocal_table[height_minor_b]; \ |
| 926 | \ |
| 927 | sub_2x32b(widths, x_ends, x_starts); \ |
| 928 | width_alt = x_c - start_c; \ |
| 929 | \ |
| 930 | shr_2x32b(height_reciprocals, edge_shifts, 10); \ |
| 931 | height_reciprocal_alt = edge_shift_alt >> 10; \ |
| 932 | \ |
| 933 | bic_immediate_4x16b(vector_cast(vec_4x16u, edge_shifts), 0xE0); \ |
| 934 | edge_shift_alt &= 0x1F; \ |
| 935 | \ |
| 936 | sub_2x32b(heights_b, heights, c_0x01); \ |
| 937 | height_b_alt = height_minor_b - 1; \ |
| 938 | \ |
| 939 | mla_2x32b(heights_b, x_starts, heights); \ |
| 940 | height_b_alt += height_minor_b * start_c; \ |
| 941 | \ |
| 942 | mul_long_2x32b(edges_xy, heights_b, height_reciprocals); \ |
| 943 | edge_alt = (s64)height_b_alt * height_reciprocal_alt; \ |
| 944 | \ |
| 945 | mul_2x32b(edges_dx_dy, widths, height_reciprocals); \ |
| 946 | edge_dx_dy_alt = width_alt * height_reciprocal_alt; \ |
| 947 | } \ |
| 948 | |
| 949 | |
| 950 | #define setup_spans_adjust_y_up() \ |
| 951 | sub_4x32b(y_x4, y_x4, c_0x04) \ |
| 952 | |
| 953 | #define setup_spans_adjust_y_down() \ |
| 954 | add_4x32b(y_x4, y_x4, c_0x04) \ |
| 955 | |
| 956 | #define setup_spans_adjust_interpolants_up() \ |
| 957 | sub_4x32b(uvrg, uvrg, uvrg_dy); \ |
| 958 | b -= b_dy \ |
| 959 | |
| 960 | #define setup_spans_adjust_interpolants_down() \ |
| 961 | add_4x32b(uvrg, uvrg, uvrg_dy); \ |
| 962 | b += b_dy \ |
| 963 | |
| 964 | |
| 965 | #define setup_spans_clip_interpolants_increment() \ |
| 966 | mla_scalar_4x32b(uvrg, uvrg_dy, clip); \ |
| 967 | b += b_dy * clip \ |
| 968 | |
| 969 | #define setup_spans_clip_interpolants_decrement() \ |
| 970 | mls_scalar_4x32b(uvrg, uvrg_dy, clip); \ |
| 971 | b -= b_dy * clip \ |
| 972 | |
| 973 | #define setup_spans_clip_alternate_yes() \ |
| 974 | edge_alt += edge_dx_dy_alt * (s64)(clip) \ |
| 975 | |
| 976 | #define setup_spans_clip_alternate_no() \ |
| 977 | |
| 978 | #define setup_spans_clip(direction, alternate_active) \ |
| 979 | { \ |
| 980 | clipped_triangles++; \ |
| 981 | mla_scalar_long_2x32b(edges_xy, edges_dx_dy, (s64)clip); \ |
| 982 | setup_spans_clip_alternate_##alternate_active(); \ |
| 983 | setup_spans_clip_interpolants_##direction(); \ |
| 984 | } \ |
| 985 | |
| 986 | |
| 987 | #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \ |
| 988 | { \ |
| 989 | vec_2x64u edge_shifts_64; \ |
| 990 | vec_2x64s edges_dx_dy_64; \ |
| 991 | \ |
| 992 | mov_wide_2x32b(edge_shifts_64, edge_shifts); \ |
| 993 | shl_variable_2x64b(edges_xy, edges_xy, edge_shifts_64); \ |
| 994 | \ |
| 995 | mov_wide_2x32b(edges_dx_dy_64, edges_dx_dy); \ |
| 996 | shl_variable_2x64b(edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64); \ |
| 997 | \ |
| 998 | left_x.e[0] = edges_xy.e[left_index]; \ |
| 999 | right_x.e[0] = edges_xy.e[right_index]; \ |
| 1000 | \ |
| 1001 | left_dx_dy.e[0] = edges_dx_dy_64.e[left_index]; \ |
| 1002 | left_dx_dy.e[1] = edges_dx_dy_64.e[left_index]; \ |
| 1003 | right_dx_dy.e[0] = edges_dx_dy_64.e[right_index]; \ |
| 1004 | right_dx_dy.e[1] = edges_dx_dy_64.e[right_index]; \ |
| 1005 | \ |
| 1006 | add_1x64b(left_x.high, left_x.low, left_dx_dy.low); \ |
| 1007 | add_1x64b(right_x.high, right_x.low, right_dx_dy.low); \ |
| 1008 | \ |
| 1009 | add_2x64b(left_dx_dy, left_dx_dy, left_dx_dy); \ |
| 1010 | add_2x64b(right_dx_dy, right_dx_dy, right_dx_dy); \ |
| 1011 | } \ |
| 1012 | |
| 1013 | #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \ |
| 1014 | { \ |
| 1015 | setup_spans_adjust_edges_alternate_no(left_index, right_index); \ |
| 1016 | s64 edge_dx_dy_alt_64; \ |
| 1017 | \ |
| 1018 | dup_4x16b(y_mid_point, y_b); \ |
| 1019 | \ |
| 1020 | edge_alt <<= edge_shift_alt; \ |
| 1021 | edge_dx_dy_alt_64 = (s64)edge_dx_dy_alt << edge_shift_alt; \ |
| 1022 | \ |
| 1023 | alternate_x.e[0] = edge_alt; \ |
| 1024 | alternate_dx_dy.e[0] = edge_dx_dy_alt_64; \ |
| 1025 | alternate_dx_dy.e[1] = edge_dx_dy_alt_64; \ |
| 1026 | \ |
| 1027 | add_1x64b(alternate_x.high, alternate_x.low, alternate_dx_dy.low); \ |
| 1028 | add_2x64b(alternate_dx_dy, alternate_dx_dy, alternate_dx_dy); \ |
| 1029 | } \ |
| 1030 | |
| 1031 | |
| 1032 | #define setup_spans_y_select_up() \ |
| 1033 | cmplt_4x16b(alternate_select, y_x4, y_mid_point) \ |
| 1034 | |
| 1035 | #define setup_spans_y_select_down() \ |
| 1036 | cmpgt_4x16b(alternate_select, y_x4, y_mid_point) \ |
| 1037 | |
| 1038 | #define setup_spans_y_select_alternate_yes(direction) \ |
| 1039 | setup_spans_y_select_##direction() \ |
| 1040 | |
| 1041 | #define setup_spans_y_select_alternate_no(direction) \ |
| 1042 | |
| 1043 | #define setup_spans_alternate_select_left() \ |
| 1044 | bit_4x16b(left_right_x_16.low, alternate_x_16, alternate_select) \ |
| 1045 | |
| 1046 | #define setup_spans_alternate_select_right() \ |
| 1047 | bit_4x16b(left_right_x_16.high, alternate_x_16, alternate_select) \ |
| 1048 | |
| 1049 | #define setup_spans_alternate_select_none() \ |
| 1050 | |
| 1051 | #define setup_spans_increment_alternate_yes() \ |
| 1052 | shr_narrow_2x64b(alternate_x_32.low, alternate_x, 32); \ |
| 1053 | add_2x64b(alternate_x, alternate_x, alternate_dx_dy); \ |
| 1054 | shr_narrow_2x64b(alternate_x_32.high, alternate_x, 32); \ |
| 1055 | add_2x64b(alternate_x, alternate_x, alternate_dx_dy); \ |
| 1056 | mov_narrow_4x32b(alternate_x_16, alternate_x_32) \ |
| 1057 | |
| 1058 | #define setup_spans_increment_alternate_no() \ |
| 1059 | |
| 1060 | #define setup_spans_set_x4(alternate, direction, alternate_active) \ |
| 1061 | { \ |
| 1062 | span_uvrg_offset[0] = uvrg; \ |
| 1063 | span_b_offset[0] = b; \ |
| 1064 | setup_spans_adjust_interpolants_##direction(); \ |
| 1065 | \ |
| 1066 | span_uvrg_offset[1] = uvrg; \ |
| 1067 | span_b_offset[1] = b; \ |
| 1068 | setup_spans_adjust_interpolants_##direction(); \ |
| 1069 | \ |
| 1070 | span_uvrg_offset[2] = uvrg; \ |
| 1071 | span_b_offset[2] = b; \ |
| 1072 | setup_spans_adjust_interpolants_##direction(); \ |
| 1073 | \ |
| 1074 | span_uvrg_offset[3] = uvrg; \ |
| 1075 | span_b_offset[3] = b; \ |
| 1076 | setup_spans_adjust_interpolants_##direction(); \ |
| 1077 | \ |
| 1078 | span_uvrg_offset += 4; \ |
| 1079 | span_b_offset += 4; \ |
| 1080 | \ |
| 1081 | shr_narrow_2x64b(left_x_32.low, left_x, 32); \ |
| 1082 | shr_narrow_2x64b(right_x_32.low, right_x, 32); \ |
| 1083 | \ |
| 1084 | add_2x64b(left_x, left_x, left_dx_dy); \ |
| 1085 | add_2x64b(right_x, right_x, right_dx_dy); \ |
| 1086 | \ |
| 1087 | shr_narrow_2x64b(left_x_32.high, left_x, 32); \ |
| 1088 | shr_narrow_2x64b(right_x_32.high, right_x, 32); \ |
| 1089 | \ |
| 1090 | add_2x64b(left_x, left_x, left_dx_dy); \ |
| 1091 | add_2x64b(right_x, right_x, right_dx_dy); \ |
| 1092 | \ |
| 1093 | mov_narrow_4x32b(left_right_x_16.low, left_x_32); \ |
| 1094 | mov_narrow_4x32b(left_right_x_16.high, right_x_32); \ |
| 1095 | \ |
| 1096 | setup_spans_increment_alternate_##alternate_active(); \ |
| 1097 | setup_spans_y_select_alternate_##alternate_active(direction); \ |
| 1098 | setup_spans_alternate_select_##alternate(); \ |
| 1099 | \ |
| 1100 | max_8x16b(left_right_x_16, left_right_x_16, left_edge); \ |
| 1101 | min_8x16b(left_right_x_16, left_right_x_16, right_edge); \ |
| 1102 | \ |
| 1103 | sub_4x16b(left_right_x_16.high, left_right_x_16.high, left_right_x_16.low); \ |
| 1104 | add_4x16b(left_right_x_16.high, left_right_x_16.high, c_0x07); \ |
| 1105 | and_4x16b(span_shift, left_right_x_16.high, c_0x07); \ |
| 1106 | shl_variable_4x16b(span_shift, c_0xFFFE, span_shift); \ |
| 1107 | shr_4x16b(left_right_x_16.high, left_right_x_16.high, 3); \ |
| 1108 | min_4x16b(left_right_x_16.high, left_right_x_16.high, c_max_blocks_per_row); \ |
| 1109 | \ |
| 1110 | u32 i; \ |
| 1111 | for(i = 0; i < 4; i++) \ |
| 1112 | { \ |
| 1113 | span_edge_data[i].left_x = left_right_x_16.low.e[i]; \ |
| 1114 | span_edge_data[i].num_blocks = left_right_x_16.high.e[i]; \ |
| 1115 | span_edge_data[i].right_mask = span_shift.e[i]; \ |
| 1116 | span_edge_data[i].y = y_x4.e[i]; \ |
| 1117 | setup_spans_debug_check(psx_gpu, &span_edge_data[i]); \ |
| 1118 | } \ |
| 1119 | \ |
| 1120 | span_edge_data += 4; \ |
| 1121 | \ |
| 1122 | setup_spans_adjust_y_##direction(); \ |
| 1123 | } \ |
| 1124 | |
| 1125 | |
| 1126 | #define setup_spans_alternate_adjust_yes() \ |
| 1127 | edge_alt -= edge_dx_dy_alt * (s64)height_minor_a \ |
| 1128 | |
| 1129 | #define setup_spans_alternate_adjust_no() \ |
| 1130 | |
| 1131 | |
| 1132 | #define setup_spans_down(left_index, right_index, alternate, alternate_active) \ |
| 1133 | setup_spans_alternate_adjust_##alternate_active(); \ |
| 1134 | if(y_c > psx_gpu->viewport_end_y) \ |
| 1135 | height -= y_c - psx_gpu->viewport_end_y - 1; \ |
| 1136 | \ |
| 1137 | clip = psx_gpu->viewport_start_y - y_a; \ |
| 1138 | if(clip > 0) \ |
| 1139 | { \ |
| 1140 | height -= clip; \ |
| 1141 | y_a += clip; \ |
| 1142 | setup_spans_clip(increment, alternate_active); \ |
| 1143 | } \ |
| 1144 | \ |
| 1145 | setup_spans_prologue_b(); \ |
| 1146 | \ |
| 1147 | if (height > 512) \ |
| 1148 | height = 512; \ |
| 1149 | if (height > 0) \ |
| 1150 | { \ |
| 1151 | y_x4.e[0] = y_a; \ |
| 1152 | y_x4.e[1] = y_a + 1; \ |
| 1153 | y_x4.e[2] = y_a + 2; \ |
| 1154 | y_x4.e[3] = y_a + 3; \ |
| 1155 | setup_spans_adjust_edges_alternate_##alternate_active(left_index, \ |
| 1156 | right_index); \ |
| 1157 | \ |
| 1158 | psx_gpu->num_spans = height; \ |
| 1159 | do \ |
| 1160 | { \ |
| 1161 | setup_spans_set_x4(alternate, down, alternate_active); \ |
| 1162 | height -= 4; \ |
| 1163 | } while(height > 0); \ |
| 1164 | } \ |
| 1165 | |
| 1166 | |
| 1167 | #define setup_spans_alternate_pre_increment_yes() \ |
| 1168 | edge_alt += edge_dx_dy_alt \ |
| 1169 | |
| 1170 | #define setup_spans_alternate_pre_increment_no() \ |
| 1171 | |
| 1172 | #define setup_spans_up_decrement_height_yes() \ |
| 1173 | height-- \ |
| 1174 | |
| 1175 | #define setup_spans_up_decrement_height_no() \ |
| 1176 | {} \ |
| 1177 | |
| 1178 | #define setup_spans_up(left_index, right_index, alternate, alternate_active) \ |
| 1179 | setup_spans_alternate_adjust_##alternate_active(); \ |
| 1180 | y_a--; \ |
| 1181 | \ |
| 1182 | if(y_c < psx_gpu->viewport_start_y) \ |
| 1183 | height -= psx_gpu->viewport_start_y - y_c; \ |
| 1184 | else \ |
| 1185 | setup_spans_up_decrement_height_##alternate_active(); \ |
| 1186 | \ |
| 1187 | clip = y_a - psx_gpu->viewport_end_y; \ |
| 1188 | if(clip > 0) \ |
| 1189 | { \ |
| 1190 | height -= clip; \ |
| 1191 | y_a -= clip; \ |
| 1192 | setup_spans_clip(decrement, alternate_active); \ |
| 1193 | } \ |
| 1194 | \ |
| 1195 | setup_spans_prologue_b(); \ |
| 1196 | \ |
| 1197 | if (height > 512) \ |
| 1198 | height = 512; \ |
| 1199 | if (height > 0) \ |
| 1200 | { \ |
| 1201 | y_x4.e[0] = y_a; \ |
| 1202 | y_x4.e[1] = y_a - 1; \ |
| 1203 | y_x4.e[2] = y_a - 2; \ |
| 1204 | y_x4.e[3] = y_a - 3; \ |
| 1205 | add_wide_2x32b(edges_xy, edges_xy, edges_dx_dy); \ |
| 1206 | setup_spans_alternate_pre_increment_##alternate_active(); \ |
| 1207 | setup_spans_adjust_edges_alternate_##alternate_active(left_index, \ |
| 1208 | right_index); \ |
| 1209 | setup_spans_adjust_interpolants_up(); \ |
| 1210 | \ |
| 1211 | psx_gpu->num_spans = height; \ |
| 1212 | while(height > 0) \ |
| 1213 | { \ |
| 1214 | setup_spans_set_x4(alternate, up, alternate_active); \ |
| 1215 | height -= 4; \ |
| 1216 | } \ |
| 1217 | } \ |
| 1218 | |
| 1219 | #define index_left 0 |
| 1220 | #define index_right 1 |
| 1221 | |
| 1222 | #define setup_spans_up_up(minor, major) \ |
| 1223 | setup_spans_prologue(yes); \ |
| 1224 | s32 height_minor_a = y_a - y_b; \ |
| 1225 | s32 height_minor_b = y_b - y_c; \ |
| 1226 | s32 height = y_a - y_c; \ |
| 1227 | \ |
| 1228 | dup_2x32b(x_starts, x_a); \ |
| 1229 | x_ends.e[0] = x_c; \ |
| 1230 | x_ends.e[1] = x_b; \ |
| 1231 | \ |
| 1232 | compute_edge_delta_x3(x_b, height, height_minor_a); \ |
| 1233 | setup_spans_up(index_##major, index_##minor, minor, yes) \ |
| 1234 | |
| 1235 | |
| 1236 | #ifndef NEON_BUILD |
| 1237 | |
| 1238 | void setup_spans_up_left(psx_gpu_struct *psx_gpu, vertex_struct *v_a, |
| 1239 | vertex_struct *v_b, vertex_struct *v_c) |
| 1240 | { |
| 1241 | setup_spans_up_up(left, right); |
| 1242 | } |
| 1243 | |
| 1244 | void setup_spans_up_right(psx_gpu_struct *psx_gpu, vertex_struct *v_a, |
| 1245 | vertex_struct *v_b, vertex_struct *v_c) |
| 1246 | { |
| 1247 | setup_spans_up_up(right, left); |
| 1248 | } |
| 1249 | |
| 1250 | #define setup_spans_down_down(minor, major) \ |
| 1251 | setup_spans_prologue(yes); \ |
| 1252 | s32 height_minor_a = y_b - y_a; \ |
| 1253 | s32 height_minor_b = y_c - y_b; \ |
| 1254 | s32 height = y_c - y_a; \ |
| 1255 | \ |
| 1256 | dup_2x32b(x_starts, x_a); \ |
| 1257 | x_ends.e[0] = x_c; \ |
| 1258 | x_ends.e[1] = x_b; \ |
| 1259 | \ |
| 1260 | compute_edge_delta_x3(x_b, height, height_minor_a); \ |
| 1261 | setup_spans_down(index_##major, index_##minor, minor, yes) \ |
| 1262 | |
| 1263 | void setup_spans_down_left(psx_gpu_struct *psx_gpu, vertex_struct *v_a, |
| 1264 | vertex_struct *v_b, vertex_struct *v_c) |
| 1265 | { |
| 1266 | setup_spans_down_down(left, right); |
| 1267 | } |
| 1268 | |
| 1269 | void setup_spans_down_right(psx_gpu_struct *psx_gpu, vertex_struct *v_a, |
| 1270 | vertex_struct *v_b, vertex_struct *v_c) |
| 1271 | { |
| 1272 | setup_spans_down_down(right, left); |
| 1273 | } |
| 1274 | |
| 1275 | #define setup_spans_up_flat() \ |
| 1276 | s32 height = y_a - y_c; \ |
| 1277 | \ |
| 1278 | flat_triangles++; \ |
| 1279 | compute_edge_delta_x2(); \ |
| 1280 | setup_spans_up(index_left, index_right, none, no) \ |
| 1281 | |
| 1282 | void setup_spans_up_a(psx_gpu_struct *psx_gpu, vertex_struct *v_a, |
| 1283 | vertex_struct *v_b, vertex_struct *v_c) |
| 1284 | { |
| 1285 | setup_spans_prologue(no); |
| 1286 | x_starts.e[0] = x_a; |
| 1287 | x_starts.e[1] = x_b; |
| 1288 | dup_2x32b(x_ends, x_c); |
| 1289 | |
| 1290 | setup_spans_up_flat(); |
| 1291 | } |
| 1292 | |
| 1293 | void setup_spans_up_b(psx_gpu_struct *psx_gpu, vertex_struct *v_a, |
| 1294 | vertex_struct *v_b, vertex_struct *v_c) |
| 1295 | { |
| 1296 | setup_spans_prologue(no); |
| 1297 | dup_2x32b(x_starts, x_a); |
| 1298 | x_ends.e[0] = x_b; |
| 1299 | x_ends.e[1] = x_c; |
| 1300 | |
| 1301 | setup_spans_up_flat(); |
| 1302 | } |
| 1303 | |
| 1304 | #define setup_spans_down_flat() \ |
| 1305 | s32 height = y_c - y_a; \ |
| 1306 | \ |
| 1307 | flat_triangles++; \ |
| 1308 | compute_edge_delta_x2(); \ |
| 1309 | setup_spans_down(index_left, index_right, none, no) \ |
| 1310 | |
| 1311 | void setup_spans_down_a(psx_gpu_struct *psx_gpu, vertex_struct *v_a, |
| 1312 | vertex_struct *v_b, vertex_struct *v_c) |
| 1313 | { |
| 1314 | setup_spans_prologue(no); |
| 1315 | x_starts.e[0] = x_a; |
| 1316 | x_starts.e[1] = x_b; |
| 1317 | dup_2x32b(x_ends, x_c); |
| 1318 | |
| 1319 | setup_spans_down_flat(); |
| 1320 | } |
| 1321 | |
| 1322 | void setup_spans_down_b(psx_gpu_struct *psx_gpu, vertex_struct *v_a, |
| 1323 | vertex_struct *v_b, vertex_struct *v_c) |
| 1324 | { |
| 1325 | setup_spans_prologue(no); |
| 1326 | dup_2x32b(x_starts, x_a); |
| 1327 | x_ends.e[0] = x_b; |
| 1328 | x_ends.e[1] = x_c; |
| 1329 | |
| 1330 | setup_spans_down_flat(); |
| 1331 | } |
| 1332 | |
| 1333 | void setup_spans_up_down(psx_gpu_struct *psx_gpu, vertex_struct *v_a, |
| 1334 | vertex_struct *v_b, vertex_struct *v_c) |
| 1335 | { |
| 1336 | setup_spans_prologue(no); |
| 1337 | |
| 1338 | s32 y_b = v_b->y; |
| 1339 | s64 edge_alt; |
| 1340 | s32 edge_dx_dy_alt; |
| 1341 | u32 edge_shift_alt; |
| 1342 | |
| 1343 | s32 middle_y = y_a; |
| 1344 | s32 height_minor_a = y_a - y_b; |
| 1345 | s32 height_minor_b = y_c - y_a; |
| 1346 | s32 height_major = y_c - y_b; |
| 1347 | |
| 1348 | vec_2x64s edges_xy_b; |
| 1349 | vec_2x32s edges_dx_dy_b; |
| 1350 | vec_2x32u edge_shifts_b; |
| 1351 | |
| 1352 | vec_2x32s height_increment; |
| 1353 | |
| 1354 | x_starts.e[0] = x_a; |
| 1355 | x_starts.e[1] = x_c; |
| 1356 | dup_2x32b(x_ends, x_b); |
| 1357 | |
| 1358 | compute_edge_delta_x3(x_a, height_minor_a, height_major); |
| 1359 | |
| 1360 | height_increment.e[0] = 0; |
| 1361 | height_increment.e[1] = height_minor_b; |
| 1362 | |
| 1363 | mla_long_2x32b(edges_xy, edges_dx_dy, height_increment); |
| 1364 | |
| 1365 | edges_xy_b.e[0] = edge_alt; |
| 1366 | edges_xy_b.e[1] = edges_xy.e[1]; |
| 1367 | |
| 1368 | edge_shifts_b = edge_shifts; |
| 1369 | edge_shifts_b.e[0] = edge_shift_alt; |
| 1370 | |
| 1371 | neg_2x32b(edges_dx_dy_b, edges_dx_dy); |
| 1372 | edges_dx_dy_b.e[0] = edge_dx_dy_alt; |
| 1373 | |
| 1374 | y_a--; |
| 1375 | |
| 1376 | if(y_b < psx_gpu->viewport_start_y) |
| 1377 | height_minor_a -= psx_gpu->viewport_start_y - y_b; |
| 1378 | |
| 1379 | clip = y_a - psx_gpu->viewport_end_y; |
| 1380 | if(clip > 0) |
| 1381 | { |
| 1382 | height_minor_a -= clip; |
| 1383 | y_a -= clip; |
| 1384 | setup_spans_clip(decrement, no); |
| 1385 | } |
| 1386 | |
| 1387 | setup_spans_prologue_b(); |
| 1388 | |
| 1389 | if (height_minor_a > 512) |
| 1390 | height_minor_a = 512; |
| 1391 | if (height_minor_a > 0) |
| 1392 | { |
| 1393 | y_x4.e[0] = y_a; |
| 1394 | y_x4.e[1] = y_a - 1; |
| 1395 | y_x4.e[2] = y_a - 2; |
| 1396 | y_x4.e[3] = y_a - 3; |
| 1397 | add_wide_2x32b(edges_xy, edges_xy, edges_dx_dy); |
| 1398 | setup_spans_adjust_edges_alternate_no(index_left, index_right); |
| 1399 | setup_spans_adjust_interpolants_up(); |
| 1400 | |
| 1401 | psx_gpu->num_spans = height_minor_a; |
| 1402 | while(height_minor_a > 0) |
| 1403 | { |
| 1404 | setup_spans_set_x4(none, up, no); |
| 1405 | height_minor_a -= 4; |
| 1406 | } |
| 1407 | |
| 1408 | span_edge_data += height_minor_a; |
| 1409 | span_uvrg_offset += height_minor_a; |
| 1410 | span_b_offset += height_minor_a; |
| 1411 | } |
| 1412 | |
| 1413 | edges_xy = edges_xy_b; |
| 1414 | edges_dx_dy = edges_dx_dy_b; |
| 1415 | edge_shifts = edge_shifts_b; |
| 1416 | |
| 1417 | uvrg = psx_gpu->uvrg; |
| 1418 | b = psx_gpu->b; |
| 1419 | |
| 1420 | y_a = middle_y; |
| 1421 | |
| 1422 | if(y_c > psx_gpu->viewport_end_y) |
| 1423 | height_minor_b -= y_c - psx_gpu->viewport_end_y - 1; |
| 1424 | |
| 1425 | clip = psx_gpu->viewport_start_y - y_a; |
| 1426 | if(clip > 0) |
| 1427 | { |
| 1428 | height_minor_b -= clip; |
| 1429 | y_a += clip; |
| 1430 | setup_spans_clip(increment, no); |
| 1431 | } |
| 1432 | |
| 1433 | if (height_minor_b > 512) |
| 1434 | height_minor_b = 512; |
| 1435 | if (height_minor_b > 0) |
| 1436 | { |
| 1437 | y_x4.e[0] = y_a; |
| 1438 | y_x4.e[1] = y_a + 1; |
| 1439 | y_x4.e[2] = y_a + 2; |
| 1440 | y_x4.e[3] = y_a + 3; |
| 1441 | setup_spans_adjust_edges_alternate_no(index_left, index_right); |
| 1442 | |
| 1443 | // FIXME: overflow corner case |
| 1444 | if(psx_gpu->num_spans + height_minor_b == MAX_SPANS) |
| 1445 | height_minor_b &= ~3; |
| 1446 | |
| 1447 | psx_gpu->num_spans += height_minor_b; |
| 1448 | while(height_minor_b > 0) |
| 1449 | { |
| 1450 | setup_spans_set_x4(none, down, no); |
| 1451 | height_minor_b -= 4; |
| 1452 | } |
| 1453 | } |
| 1454 | |
| 1455 | left_split_triangles++; |
| 1456 | } |
| 1457 | |
| 1458 | #endif |
| 1459 | |
| 1460 | |
| 1461 | #define dither_table_entry_normal(value) \ |
| 1462 | (value) \ |
| 1463 | |
| 1464 | |
| 1465 | #define setup_blocks_load_msb_mask_indirect() \ |
| 1466 | |
| 1467 | #define setup_blocks_load_msb_mask_direct() \ |
| 1468 | vec_8x16u msb_mask; \ |
| 1469 | dup_8x16b(msb_mask, psx_gpu->mask_msb); \ |
| 1470 | |
| 1471 | |
| 1472 | #define setup_blocks_variables_shaded_textured(target) \ |
| 1473 | vec_4x32u u_block; \ |
| 1474 | vec_4x32u v_block; \ |
| 1475 | vec_4x32u r_block; \ |
| 1476 | vec_4x32u g_block; \ |
| 1477 | vec_4x32u b_block; \ |
| 1478 | vec_4x32u uvrg_dx = psx_gpu->uvrg_dx; \ |
| 1479 | vec_4x32u uvrg_dx4; \ |
| 1480 | vec_4x32u uvrg_dx8; \ |
| 1481 | vec_4x32u uvrg; \ |
| 1482 | u32 b_dx = psx_gpu->b_block_span.e[1]; \ |
| 1483 | u32 b_dx4 = b_dx << 2; \ |
| 1484 | u32 b_dx8 = b_dx << 3; \ |
| 1485 | u32 b; \ |
| 1486 | \ |
| 1487 | vec_16x8u texture_mask; \ |
| 1488 | shl_4x32b(uvrg_dx4, uvrg_dx, 2); \ |
| 1489 | shl_4x32b(uvrg_dx8, uvrg_dx, 3); \ |
| 1490 | dup_8x8b(texture_mask.low, psx_gpu->texture_mask_width); \ |
| 1491 | dup_8x8b(texture_mask.high, psx_gpu->texture_mask_height) \ |
| 1492 | |
| 1493 | #define setup_blocks_variables_shaded_untextured(target) \ |
| 1494 | vec_4x32u r_block; \ |
| 1495 | vec_4x32u g_block; \ |
| 1496 | vec_4x32u b_block; \ |
| 1497 | vec_4x32u rgb_dx; \ |
| 1498 | vec_4x32u rgb_dx4; \ |
| 1499 | vec_4x32u rgb_dx8; \ |
| 1500 | vec_4x32u rgb; \ |
| 1501 | \ |
| 1502 | vec_8x8u d64_0x07; \ |
| 1503 | vec_8x8u d64_1; \ |
| 1504 | vec_8x8u d64_4; \ |
| 1505 | vec_8x8u d64_128; \ |
| 1506 | \ |
| 1507 | dup_8x8b(d64_0x07, 0x07); \ |
| 1508 | dup_8x8b(d64_1, 1); \ |
| 1509 | dup_8x8b(d64_4, 4); \ |
| 1510 | dup_8x8b(d64_128, 128); \ |
| 1511 | \ |
| 1512 | rgb_dx.low = psx_gpu->uvrg_dx.high; \ |
| 1513 | rgb_dx.e[2] = psx_gpu->b_block_span.e[1]; \ |
| 1514 | shl_4x32b(rgb_dx4, rgb_dx, 2); \ |
| 1515 | shl_4x32b(rgb_dx8, rgb_dx, 3) \ |
| 1516 | |
| 1517 | #define setup_blocks_variables_unshaded_textured(target) \ |
| 1518 | vec_4x32u u_block; \ |
| 1519 | vec_4x32u v_block; \ |
| 1520 | vec_2x32u uv_dx = psx_gpu->uvrg_dx.low; \ |
| 1521 | vec_2x32u uv_dx4; \ |
| 1522 | vec_2x32u uv_dx8; \ |
| 1523 | vec_2x32u uv = psx_gpu->uvrg.low; \ |
| 1524 | \ |
| 1525 | vec_16x8u texture_mask; \ |
| 1526 | shl_2x32b(uv_dx4, uv_dx, 2); \ |
| 1527 | shl_2x32b(uv_dx8, uv_dx, 3); \ |
| 1528 | dup_8x8b(texture_mask.low, psx_gpu->texture_mask_width); \ |
| 1529 | dup_8x8b(texture_mask.high, psx_gpu->texture_mask_height) \ |
| 1530 | |
| 1531 | |
| 1532 | #define setup_blocks_variables_unshaded_untextured_direct() \ |
| 1533 | or_8x16b(colors, colors, msb_mask) \ |
| 1534 | |
| 1535 | #define setup_blocks_variables_unshaded_untextured_indirect() \ |
| 1536 | |
| 1537 | #define setup_blocks_variables_unshaded_untextured(target) \ |
| 1538 | u32 color = psx_gpu->triangle_color; \ |
| 1539 | vec_8x16u colors; \ |
| 1540 | \ |
| 1541 | u32 color_r = color & 0xFF; \ |
| 1542 | u32 color_g = (color >> 8) & 0xFF; \ |
| 1543 | u32 color_b = (color >> 16) & 0xFF; \ |
| 1544 | \ |
| 1545 | color = (color_r >> 3) | ((color_g >> 3) << 5) | \ |
| 1546 | ((color_b >> 3) << 10); \ |
| 1547 | dup_8x16b(colors, color); \ |
| 1548 | setup_blocks_variables_unshaded_untextured_##target() \ |
| 1549 | |
| 1550 | #define setup_blocks_span_initialize_dithered_textured() \ |
| 1551 | vec_8x16u dither_offsets; \ |
| 1552 | shl_long_8x8b(dither_offsets, dither_offsets_short, 4) \ |
| 1553 | |
| 1554 | #define setup_blocks_span_initialize_dithered_untextured() \ |
| 1555 | vec_8x8u dither_offsets; \ |
| 1556 | add_8x8b(dither_offsets, dither_offsets_short, d64_4) \ |
| 1557 | |
| 1558 | #define setup_blocks_span_initialize_dithered(texturing) \ |
| 1559 | u32 dither_row = psx_gpu->dither_table[y & 0x3]; \ |
| 1560 | u32 dither_shift = (span_edge_data->left_x & 0x3) * 8; \ |
| 1561 | vec_8x8s dither_offsets_short; \ |
| 1562 | \ |
| 1563 | dither_row = \ |
| 1564 | (dither_row >> dither_shift) | (dither_row << (32 - dither_shift)); \ |
| 1565 | dup_2x32b(vector_cast(vec_2x32u, dither_offsets_short), dither_row); \ |
| 1566 | setup_blocks_span_initialize_dithered_##texturing() \ |
| 1567 | |
| 1568 | #define setup_blocks_span_initialize_undithered(texturing) \ |
| 1569 | |
| 1570 | |
| 1571 | #define setup_blocks_span_initialize_shaded_textured() \ |
| 1572 | { \ |
| 1573 | vec_4x32u block_span; \ |
| 1574 | u32 offset = span_edge_data->left_x; \ |
| 1575 | \ |
| 1576 | uvrg = *span_uvrg_offset; \ |
| 1577 | mla_scalar_4x32b(uvrg, uvrg_dx, offset); \ |
| 1578 | b = *span_b_offset; \ |
| 1579 | b += b_dx * offset; \ |
| 1580 | \ |
| 1581 | dup_4x32b(u_block, uvrg.e[0]); \ |
| 1582 | dup_4x32b(v_block, uvrg.e[1]); \ |
| 1583 | dup_4x32b(r_block, uvrg.e[2]); \ |
| 1584 | dup_4x32b(g_block, uvrg.e[3]); \ |
| 1585 | dup_4x32b(b_block, b); \ |
| 1586 | \ |
| 1587 | block_span = psx_gpu->u_block_span; \ |
| 1588 | add_4x32b(u_block, u_block, block_span); \ |
| 1589 | block_span = psx_gpu->v_block_span; \ |
| 1590 | add_4x32b(v_block, v_block, block_span); \ |
| 1591 | block_span = psx_gpu->r_block_span; \ |
| 1592 | add_4x32b(r_block, r_block, block_span); \ |
| 1593 | block_span = psx_gpu->g_block_span; \ |
| 1594 | add_4x32b(g_block, g_block, block_span); \ |
| 1595 | block_span = psx_gpu->b_block_span; \ |
| 1596 | add_4x32b(b_block, b_block, block_span); \ |
| 1597 | } |
| 1598 | |
| 1599 | #define setup_blocks_span_initialize_shaded_untextured() \ |
| 1600 | { \ |
| 1601 | vec_4x32u block_span; \ |
| 1602 | u32 offset = span_edge_data->left_x; \ |
| 1603 | \ |
| 1604 | rgb.low = span_uvrg_offset->high; \ |
| 1605 | rgb.high.e[0] = *span_b_offset; \ |
| 1606 | mla_scalar_4x32b(rgb, rgb_dx, offset); \ |
| 1607 | \ |
| 1608 | dup_4x32b(r_block, rgb.e[0]); \ |
| 1609 | dup_4x32b(g_block, rgb.e[1]); \ |
| 1610 | dup_4x32b(b_block, rgb.e[2]); \ |
| 1611 | \ |
| 1612 | block_span = psx_gpu->r_block_span; \ |
| 1613 | add_4x32b(r_block, r_block, block_span); \ |
| 1614 | block_span = psx_gpu->g_block_span; \ |
| 1615 | add_4x32b(g_block, g_block, block_span); \ |
| 1616 | block_span = psx_gpu->b_block_span; \ |
| 1617 | add_4x32b(b_block, b_block, block_span); \ |
| 1618 | } \ |
| 1619 | |
| 1620 | #define setup_blocks_span_initialize_unshaded_textured() \ |
| 1621 | { \ |
| 1622 | vec_4x32u block_span; \ |
| 1623 | u32 offset = span_edge_data->left_x; \ |
| 1624 | \ |
| 1625 | uv = span_uvrg_offset->low; \ |
| 1626 | mla_scalar_2x32b(uv, uv_dx, offset); \ |
| 1627 | \ |
| 1628 | dup_4x32b(u_block, uv.e[0]); \ |
| 1629 | dup_4x32b(v_block, uv.e[1]); \ |
| 1630 | \ |
| 1631 | block_span = psx_gpu->u_block_span; \ |
| 1632 | add_4x32b(u_block, u_block, block_span); \ |
| 1633 | block_span = psx_gpu->v_block_span; \ |
| 1634 | add_4x32b(v_block, v_block, block_span); \ |
| 1635 | } \ |
| 1636 | |
| 1637 | #define setup_blocks_span_initialize_unshaded_untextured() \ |
| 1638 | |
| 1639 | |
| 1640 | #define setup_blocks_texture_swizzled() \ |
| 1641 | { \ |
| 1642 | vec_8x8u u_saved = u; \ |
| 1643 | sli_8x8b(u, v, 4); \ |
| 1644 | sri_8x8b(v, u_saved, 4); \ |
| 1645 | } \ |
| 1646 | |
| 1647 | #define setup_blocks_texture_unswizzled() \ |
| 1648 | |
| 1649 | #define setup_blocks_store_shaded_textured(swizzling, dithering, target, \ |
| 1650 | edge_type) \ |
| 1651 | { \ |
| 1652 | vec_8x16u u_whole; \ |
| 1653 | vec_8x16u v_whole; \ |
| 1654 | vec_8x16u r_whole; \ |
| 1655 | vec_8x16u g_whole; \ |
| 1656 | vec_8x16u b_whole; \ |
| 1657 | \ |
| 1658 | vec_8x8u u; \ |
| 1659 | vec_8x8u v; \ |
| 1660 | vec_8x8u r; \ |
| 1661 | vec_8x8u g; \ |
| 1662 | vec_8x8u b; \ |
| 1663 | vec_8x16u uv; \ |
| 1664 | \ |
| 1665 | vec_4x32u dx4; \ |
| 1666 | vec_4x32u dx8; \ |
| 1667 | \ |
| 1668 | shr_narrow_4x32b(u_whole.low, u_block, 16); \ |
| 1669 | shr_narrow_4x32b(v_whole.low, v_block, 16); \ |
| 1670 | shr_narrow_4x32b(r_whole.low, r_block, 16); \ |
| 1671 | shr_narrow_4x32b(g_whole.low, g_block, 16); \ |
| 1672 | shr_narrow_4x32b(b_whole.low, b_block, 16); \ |
| 1673 | \ |
| 1674 | dup_4x32b(dx4, uvrg_dx4.e[0]); \ |
| 1675 | add_high_narrow_4x32b(u_whole.high, u_block, dx4); \ |
| 1676 | dup_4x32b(dx4, uvrg_dx4.e[1]); \ |
| 1677 | add_high_narrow_4x32b(v_whole.high, v_block, dx4); \ |
| 1678 | dup_4x32b(dx4, uvrg_dx4.e[2]); \ |
| 1679 | add_high_narrow_4x32b(r_whole.high, r_block, dx4); \ |
| 1680 | dup_4x32b(dx4, uvrg_dx4.e[3]); \ |
| 1681 | add_high_narrow_4x32b(g_whole.high, g_block, dx4); \ |
| 1682 | dup_4x32b(dx4, b_dx4); \ |
| 1683 | add_high_narrow_4x32b(b_whole.high, b_block, dx4); \ |
| 1684 | \ |
| 1685 | mov_narrow_8x16b(u, u_whole); \ |
| 1686 | mov_narrow_8x16b(v, v_whole); \ |
| 1687 | mov_narrow_8x16b(r, r_whole); \ |
| 1688 | mov_narrow_8x16b(g, g_whole); \ |
| 1689 | mov_narrow_8x16b(b, b_whole); \ |
| 1690 | \ |
| 1691 | dup_4x32b(dx8, uvrg_dx8.e[0]); \ |
| 1692 | add_4x32b(u_block, u_block, dx8); \ |
| 1693 | dup_4x32b(dx8, uvrg_dx8.e[1]); \ |
| 1694 | add_4x32b(v_block, v_block, dx8); \ |
| 1695 | dup_4x32b(dx8, uvrg_dx8.e[2]); \ |
| 1696 | add_4x32b(r_block, r_block, dx8); \ |
| 1697 | dup_4x32b(dx8, uvrg_dx8.e[3]); \ |
| 1698 | add_4x32b(g_block, g_block, dx8); \ |
| 1699 | dup_4x32b(dx8, b_dx8); \ |
| 1700 | add_4x32b(b_block, b_block, dx8); \ |
| 1701 | \ |
| 1702 | and_8x8b(u, u, texture_mask.low); \ |
| 1703 | and_8x8b(v, v, texture_mask.high); \ |
| 1704 | setup_blocks_texture_##swizzling(); \ |
| 1705 | \ |
| 1706 | zip_8x16b(uv, u, v); \ |
| 1707 | block->uv = uv; \ |
| 1708 | block->r = r; \ |
| 1709 | block->g = g; \ |
| 1710 | block->b = b; \ |
| 1711 | block->dither_offsets = vector_cast(vec_8x16u, dither_offsets); \ |
| 1712 | block->fb_ptr = fb_ptr; \ |
| 1713 | } \ |
| 1714 | |
| 1715 | #define setup_blocks_store_unshaded_textured(swizzling, dithering, target, \ |
| 1716 | edge_type) \ |
| 1717 | { \ |
| 1718 | vec_8x16u u_whole; \ |
| 1719 | vec_8x16u v_whole; \ |
| 1720 | \ |
| 1721 | vec_8x8u u; \ |
| 1722 | vec_8x8u v; \ |
| 1723 | vec_8x16u uv; \ |
| 1724 | \ |
| 1725 | vec_4x32u dx4; \ |
| 1726 | vec_4x32u dx8; \ |
| 1727 | \ |
| 1728 | shr_narrow_4x32b(u_whole.low, u_block, 16); \ |
| 1729 | shr_narrow_4x32b(v_whole.low, v_block, 16); \ |
| 1730 | \ |
| 1731 | dup_4x32b(dx4, uv_dx4.e[0]); \ |
| 1732 | add_high_narrow_4x32b(u_whole.high, u_block, dx4); \ |
| 1733 | dup_4x32b(dx4, uv_dx4.e[1]); \ |
| 1734 | add_high_narrow_4x32b(v_whole.high, v_block, dx4); \ |
| 1735 | \ |
| 1736 | mov_narrow_8x16b(u, u_whole); \ |
| 1737 | mov_narrow_8x16b(v, v_whole); \ |
| 1738 | \ |
| 1739 | dup_4x32b(dx8, uv_dx8.e[0]); \ |
| 1740 | add_4x32b(u_block, u_block, dx8); \ |
| 1741 | dup_4x32b(dx8, uv_dx8.e[1]); \ |
| 1742 | add_4x32b(v_block, v_block, dx8); \ |
| 1743 | \ |
| 1744 | and_8x8b(u, u, texture_mask.low); \ |
| 1745 | and_8x8b(v, v, texture_mask.high); \ |
| 1746 | setup_blocks_texture_##swizzling(); \ |
| 1747 | \ |
| 1748 | zip_8x16b(uv, u, v); \ |
| 1749 | block->uv = uv; \ |
| 1750 | block->dither_offsets = vector_cast(vec_8x16u, dither_offsets); \ |
| 1751 | block->fb_ptr = fb_ptr; \ |
| 1752 | } \ |
| 1753 | |
| 1754 | #define setup_blocks_store_shaded_untextured_dithered() \ |
| 1755 | addq_8x8b(r, r, dither_offsets); \ |
| 1756 | addq_8x8b(g, g, dither_offsets); \ |
| 1757 | addq_8x8b(b, b, dither_offsets); \ |
| 1758 | \ |
| 1759 | subq_8x8b(r, r, d64_4); \ |
| 1760 | subq_8x8b(g, g, d64_4); \ |
| 1761 | subq_8x8b(b, b, d64_4) \ |
| 1762 | |
| 1763 | #define setup_blocks_store_shaded_untextured_undithered() \ |
| 1764 | |
| 1765 | |
| 1766 | #define setup_blocks_store_untextured_pixels_indirect_full(_pixels) \ |
| 1767 | block->pixels = _pixels; \ |
| 1768 | block->fb_ptr = fb_ptr \ |
| 1769 | |
| 1770 | #define setup_blocks_store_untextured_pixels_indirect_edge(_pixels) \ |
| 1771 | block->pixels = _pixels; \ |
| 1772 | block->fb_ptr = fb_ptr \ |
| 1773 | |
| 1774 | #define setup_blocks_store_shaded_untextured_seed_pixels_indirect() \ |
| 1775 | mul_long_8x8b(pixels, r, d64_1) \ |
| 1776 | |
| 1777 | |
| 1778 | #define setup_blocks_store_untextured_pixels_direct_full(_pixels) \ |
| 1779 | store_8x16b(_pixels, fb_ptr) \ |
| 1780 | |
| 1781 | #define setup_blocks_store_untextured_pixels_direct_edge(_pixels) \ |
| 1782 | { \ |
| 1783 | vec_8x16u fb_pixels; \ |
| 1784 | vec_8x16u draw_mask; \ |
| 1785 | vec_8x16u test_mask = psx_gpu->test_mask; \ |
| 1786 | \ |
| 1787 | load_8x16b(fb_pixels, fb_ptr); \ |
| 1788 | dup_8x16b(draw_mask, span_edge_data->right_mask); \ |
| 1789 | tst_8x16b(draw_mask, draw_mask, test_mask); \ |
| 1790 | bif_8x16b(fb_pixels, _pixels, draw_mask); \ |
| 1791 | store_8x16b(fb_pixels, fb_ptr); \ |
| 1792 | } \ |
| 1793 | |
| 1794 | #define setup_blocks_store_shaded_untextured_seed_pixels_direct() \ |
| 1795 | pixels = msb_mask; \ |
| 1796 | mla_long_8x8b(pixels, r, d64_1) \ |
| 1797 | |
| 1798 | |
| 1799 | #define setup_blocks_store_shaded_untextured(swizzling, dithering, target, \ |
| 1800 | edge_type) \ |
| 1801 | { \ |
| 1802 | vec_8x16u r_whole; \ |
| 1803 | vec_8x16u g_whole; \ |
| 1804 | vec_8x16u b_whole; \ |
| 1805 | \ |
| 1806 | vec_8x8u r; \ |
| 1807 | vec_8x8u g; \ |
| 1808 | vec_8x8u b; \ |
| 1809 | \ |
| 1810 | vec_4x32u dx4; \ |
| 1811 | vec_4x32u dx8; \ |
| 1812 | \ |
| 1813 | vec_8x16u pixels; \ |
| 1814 | \ |
| 1815 | shr_narrow_4x32b(r_whole.low, r_block, 16); \ |
| 1816 | shr_narrow_4x32b(g_whole.low, g_block, 16); \ |
| 1817 | shr_narrow_4x32b(b_whole.low, b_block, 16); \ |
| 1818 | \ |
| 1819 | dup_4x32b(dx4, rgb_dx4.e[0]); \ |
| 1820 | add_high_narrow_4x32b(r_whole.high, r_block, dx4); \ |
| 1821 | dup_4x32b(dx4, rgb_dx4.e[1]); \ |
| 1822 | add_high_narrow_4x32b(g_whole.high, g_block, dx4); \ |
| 1823 | dup_4x32b(dx4, rgb_dx4.e[2]); \ |
| 1824 | add_high_narrow_4x32b(b_whole.high, b_block, dx4); \ |
| 1825 | \ |
| 1826 | mov_narrow_8x16b(r, r_whole); \ |
| 1827 | mov_narrow_8x16b(g, g_whole); \ |
| 1828 | mov_narrow_8x16b(b, b_whole); \ |
| 1829 | \ |
| 1830 | dup_4x32b(dx8, rgb_dx8.e[0]); \ |
| 1831 | add_4x32b(r_block, r_block, dx8); \ |
| 1832 | dup_4x32b(dx8, rgb_dx8.e[1]); \ |
| 1833 | add_4x32b(g_block, g_block, dx8); \ |
| 1834 | dup_4x32b(dx8, rgb_dx8.e[2]); \ |
| 1835 | add_4x32b(b_block, b_block, dx8); \ |
| 1836 | \ |
| 1837 | setup_blocks_store_shaded_untextured_##dithering(); \ |
| 1838 | \ |
| 1839 | shr_8x8b(r, r, 3); \ |
| 1840 | bic_8x8b(g, g, d64_0x07); \ |
| 1841 | bic_8x8b(b, b, d64_0x07); \ |
| 1842 | \ |
| 1843 | setup_blocks_store_shaded_untextured_seed_pixels_##target(); \ |
| 1844 | mla_long_8x8b(pixels, g, d64_4); \ |
| 1845 | mla_long_8x8b(pixels, b, d64_128) \ |
| 1846 | \ |
| 1847 | setup_blocks_store_untextured_pixels_##target##_##edge_type(pixels); \ |
| 1848 | } \ |
| 1849 | |
| 1850 | #define setup_blocks_store_unshaded_untextured(swizzling, dithering, target, \ |
| 1851 | edge_type) \ |
| 1852 | setup_blocks_store_untextured_pixels_##target##_##edge_type(colors) \ |
| 1853 | |
| 1854 | |
| 1855 | #define setup_blocks_store_draw_mask_textured_indirect(_block, bits) \ |
| 1856 | (_block)->draw_mask_bits = bits \ |
| 1857 | |
| 1858 | #define setup_blocks_store_draw_mask_untextured_indirect(_block, bits) \ |
| 1859 | { \ |
| 1860 | vec_8x16u bits_mask; \ |
| 1861 | vec_8x16u test_mask = psx_gpu->test_mask; \ |
| 1862 | dup_8x16b(bits_mask, bits); \ |
| 1863 | tst_8x16b(bits_mask, bits_mask, test_mask); \ |
| 1864 | (_block)->draw_mask = bits_mask; \ |
| 1865 | } \ |
| 1866 | |
| 1867 | #define setup_blocks_store_draw_mask_untextured_direct(_block, bits) \ |
| 1868 | |
| 1869 | |
| 1870 | #define setup_blocks_add_blocks_indirect() \ |
| 1871 | num_blocks += span_num_blocks; \ |
| 1872 | \ |
| 1873 | if(num_blocks > MAX_BLOCKS) \ |
| 1874 | { \ |
| 1875 | psx_gpu->num_blocks = num_blocks - span_num_blocks; \ |
| 1876 | flush_render_block_buffer(psx_gpu); \ |
| 1877 | num_blocks = span_num_blocks; \ |
| 1878 | block = psx_gpu->blocks; \ |
| 1879 | } \ |
| 1880 | |
| 1881 | #define setup_blocks_add_blocks_direct() \ |
| 1882 | stats_add(texel_blocks_untextured, span_num_blocks); \ |
| 1883 | span_pixel_blocks += span_num_blocks \ |
| 1884 | |
| 1885 | |
| 1886 | #define setup_blocks_builder(shading, texturing, dithering, sw, target) \ |
| 1887 | void setup_blocks_##shading##_##texturing##_##dithering##_##sw##_##target( \ |
| 1888 | psx_gpu_struct *psx_gpu) \ |
| 1889 | { \ |
| 1890 | setup_blocks_load_msb_mask_##target(); \ |
| 1891 | setup_blocks_variables_##shading##_##texturing(target); \ |
| 1892 | \ |
| 1893 | edge_data_struct *span_edge_data = psx_gpu->span_edge_data; \ |
| 1894 | vec_4x32u *span_uvrg_offset = psx_gpu->span_uvrg_offset; \ |
| 1895 | u32 *span_b_offset = psx_gpu->span_b_offset; \ |
| 1896 | \ |
| 1897 | block_struct *block = psx_gpu->blocks + psx_gpu->num_blocks; \ |
| 1898 | \ |
| 1899 | u32 num_spans = psx_gpu->num_spans; \ |
| 1900 | \ |
| 1901 | u16 *fb_ptr; \ |
| 1902 | u32 y; \ |
| 1903 | \ |
| 1904 | u32 num_blocks = psx_gpu->num_blocks; \ |
| 1905 | u32 span_num_blocks; \ |
| 1906 | \ |
| 1907 | while(num_spans) \ |
| 1908 | { \ |
| 1909 | span_num_blocks = span_edge_data->num_blocks; \ |
| 1910 | if(span_num_blocks) \ |
| 1911 | { \ |
| 1912 | y = span_edge_data->y; \ |
| 1913 | fb_ptr = psx_gpu->vram_out_ptr + span_edge_data->left_x + (y * 1024); \ |
| 1914 | \ |
| 1915 | setup_blocks_span_initialize_##shading##_##texturing(); \ |
| 1916 | setup_blocks_span_initialize_##dithering(texturing); \ |
| 1917 | \ |
| 1918 | setup_blocks_add_blocks_##target(); \ |
| 1919 | \ |
| 1920 | s32 pixel_span = span_num_blocks * 8; \ |
| 1921 | pixel_span -= __builtin_popcount(span_edge_data->right_mask & 0xFF); \ |
| 1922 | span_pixels += pixel_span; \ |
| 1923 | \ |
| 1924 | span_num_blocks--; \ |
| 1925 | while(span_num_blocks) \ |
| 1926 | { \ |
| 1927 | setup_blocks_store_##shading##_##texturing(sw, dithering, target, \ |
| 1928 | full); \ |
| 1929 | setup_blocks_store_draw_mask_##texturing##_##target(block, 0x00); \ |
| 1930 | \ |
| 1931 | fb_ptr += 8; \ |
| 1932 | block++; \ |
| 1933 | span_num_blocks--; \ |
| 1934 | } \ |
| 1935 | \ |
| 1936 | setup_blocks_store_##shading##_##texturing(sw, dithering, target, edge); \ |
| 1937 | setup_blocks_store_draw_mask_##texturing##_##target(block, \ |
| 1938 | span_edge_data->right_mask); \ |
| 1939 | \ |
| 1940 | block++; \ |
| 1941 | } \ |
| 1942 | else \ |
| 1943 | { \ |
| 1944 | zero_block_spans++; \ |
| 1945 | } \ |
| 1946 | \ |
| 1947 | num_spans--; \ |
| 1948 | span_edge_data++; \ |
| 1949 | span_uvrg_offset++; \ |
| 1950 | span_b_offset++; \ |
| 1951 | } \ |
| 1952 | \ |
| 1953 | psx_gpu->num_blocks = num_blocks; \ |
| 1954 | } \ |
| 1955 | |
| 1956 | |
| 1957 | //setup_blocks_builder(unshaded, untextured, undithered, unswizzled, direct); |
| 1958 | |
| 1959 | #ifndef NEON_BUILD |
| 1960 | |
| 1961 | setup_blocks_builder(shaded, textured, dithered, swizzled, indirect); |
| 1962 | setup_blocks_builder(shaded, textured, dithered, unswizzled, indirect); |
| 1963 | |
| 1964 | setup_blocks_builder(unshaded, textured, dithered, unswizzled, indirect); |
| 1965 | setup_blocks_builder(unshaded, textured, dithered, swizzled, indirect); |
| 1966 | |
| 1967 | setup_blocks_builder(shaded, untextured, undithered, unswizzled, indirect); |
| 1968 | setup_blocks_builder(shaded, untextured, dithered, unswizzled, indirect); |
| 1969 | setup_blocks_builder(shaded, untextured, undithered, unswizzled, direct); |
| 1970 | setup_blocks_builder(shaded, untextured, dithered, unswizzled, direct); |
| 1971 | |
| 1972 | setup_blocks_builder(unshaded, untextured, undithered, unswizzled, indirect); |
| 1973 | setup_blocks_builder(unshaded, untextured, undithered, unswizzled, direct); |
| 1974 | |
| 1975 | void texture_blocks_untextured(psx_gpu_struct *psx_gpu) |
| 1976 | { |
| 1977 | if(psx_gpu->primitive_type != PRIMITIVE_TYPE_SPRITE) |
| 1978 | stats_add(texel_blocks_untextured, psx_gpu->num_blocks); |
| 1979 | } |
| 1980 | |
| 1981 | void texture_blocks_4bpp(psx_gpu_struct *psx_gpu) |
| 1982 | { |
| 1983 | block_struct *block = psx_gpu->blocks; |
| 1984 | u32 num_blocks = psx_gpu->num_blocks; |
| 1985 | stats_add(texel_blocks_4bpp, num_blocks); |
| 1986 | |
| 1987 | vec_8x8u texels_low; |
| 1988 | vec_8x8u texels_high; |
| 1989 | vec_8x8u texels; |
| 1990 | vec_8x16u pixels; |
| 1991 | |
| 1992 | vec_8x16u clut_a; |
| 1993 | vec_8x16u clut_b; |
| 1994 | vec_16x8u clut_low; |
| 1995 | vec_16x8u clut_high; |
| 1996 | |
| 1997 | u8 *texture_ptr_8bpp = psx_gpu->texture_page_ptr; |
| 1998 | u16 *clut_ptr = psx_gpu->clut_ptr; |
| 1999 | |
| 2000 | // Can be done with one deinterleaving load on NEON |
| 2001 | load_8x16b(clut_a, clut_ptr); |
| 2002 | load_8x16b(clut_b, clut_ptr + 8); |
| 2003 | unzip_16x8b(clut_low, clut_high, clut_a, clut_b); |
| 2004 | |
| 2005 | if(psx_gpu->current_texture_mask & psx_gpu->dirty_textures_4bpp_mask) |
| 2006 | update_texture_4bpp_cache(psx_gpu); |
| 2007 | |
| 2008 | while(num_blocks) |
| 2009 | { |
| 2010 | texels.e[0] = texture_ptr_8bpp[block->uv.e[0]]; |
| 2011 | texels.e[1] = texture_ptr_8bpp[block->uv.e[1]]; |
| 2012 | texels.e[2] = texture_ptr_8bpp[block->uv.e[2]]; |
| 2013 | texels.e[3] = texture_ptr_8bpp[block->uv.e[3]]; |
| 2014 | texels.e[4] = texture_ptr_8bpp[block->uv.e[4]]; |
| 2015 | texels.e[5] = texture_ptr_8bpp[block->uv.e[5]]; |
| 2016 | texels.e[6] = texture_ptr_8bpp[block->uv.e[6]]; |
| 2017 | texels.e[7] = texture_ptr_8bpp[block->uv.e[7]]; |
| 2018 | |
| 2019 | tbl_16(texels_low, texels, clut_low); |
| 2020 | tbl_16(texels_high, texels, clut_high); |
| 2021 | |
| 2022 | // Can be done with an interleaving store on NEON |
| 2023 | zip_8x16b(pixels, texels_low, texels_high); |
| 2024 | |
| 2025 | block->texels = pixels; |
| 2026 | |
| 2027 | num_blocks--; |
| 2028 | block++; |
| 2029 | } |
| 2030 | } |
| 2031 | |
| 2032 | void texture_blocks_8bpp(psx_gpu_struct *psx_gpu) |
| 2033 | { |
| 2034 | block_struct *block = psx_gpu->blocks; |
| 2035 | u32 num_blocks = psx_gpu->num_blocks; |
| 2036 | |
| 2037 | stats_add(texel_blocks_8bpp, num_blocks); |
| 2038 | |
| 2039 | if(psx_gpu->current_texture_mask & psx_gpu->dirty_textures_8bpp_mask) |
| 2040 | update_texture_8bpp_cache(psx_gpu); |
| 2041 | |
| 2042 | vec_8x16u texels; |
| 2043 | u8 *texture_ptr_8bpp = psx_gpu->texture_page_ptr; |
| 2044 | |
| 2045 | u32 texel; |
| 2046 | u32 offset; |
| 2047 | u32 i; |
| 2048 | |
| 2049 | while(num_blocks) |
| 2050 | { |
| 2051 | for(i = 0; i < 8; i++) |
| 2052 | { |
| 2053 | offset = block->uv.e[i]; |
| 2054 | |
| 2055 | texel = texture_ptr_8bpp[offset]; |
| 2056 | texels.e[i] = psx_gpu->clut_ptr[texel]; |
| 2057 | } |
| 2058 | |
| 2059 | block->texels = texels; |
| 2060 | |
| 2061 | num_blocks--; |
| 2062 | block++; |
| 2063 | } |
| 2064 | } |
| 2065 | |
| 2066 | void texture_blocks_16bpp(psx_gpu_struct *psx_gpu) |
| 2067 | { |
| 2068 | block_struct *block = psx_gpu->blocks; |
| 2069 | u32 num_blocks = psx_gpu->num_blocks; |
| 2070 | |
| 2071 | stats_add(texel_blocks_16bpp, num_blocks); |
| 2072 | |
| 2073 | vec_8x16u texels; |
| 2074 | |
| 2075 | u16 *texture_ptr_16bpp = psx_gpu->texture_page_ptr; |
| 2076 | u32 offset; |
| 2077 | u32 i; |
| 2078 | |
| 2079 | while(num_blocks) |
| 2080 | { |
| 2081 | for(i = 0; i < 8; i++) |
| 2082 | { |
| 2083 | offset = block->uv.e[i]; |
| 2084 | offset += ((offset & 0xFF00) * 3); |
| 2085 | |
| 2086 | texels.e[i] = texture_ptr_16bpp[offset]; |
| 2087 | } |
| 2088 | |
| 2089 | block->texels = texels; |
| 2090 | |
| 2091 | num_blocks--; |
| 2092 | block++; |
| 2093 | } |
| 2094 | } |
| 2095 | |
| 2096 | #endif |
| 2097 | |
| 2098 | |
| 2099 | #define shade_blocks_load_msb_mask_indirect() \ |
| 2100 | |
| 2101 | #define shade_blocks_load_msb_mask_direct() \ |
| 2102 | vec_8x16u msb_mask; \ |
| 2103 | dup_8x16b(msb_mask, psx_gpu->mask_msb); \ |
| 2104 | |
| 2105 | #define shade_blocks_store_indirect(_draw_mask, _pixels) \ |
| 2106 | block->draw_mask = _draw_mask; \ |
| 2107 | block->pixels = _pixels \ |
| 2108 | |
| 2109 | #define shade_blocks_store_direct(_draw_mask, _pixels) \ |
| 2110 | { \ |
| 2111 | vec_8x16u fb_pixels; \ |
| 2112 | or_8x16b(_pixels, _pixels, msb_mask); \ |
| 2113 | load_8x16b(fb_pixels, block->fb_ptr); \ |
| 2114 | bif_8x16b(fb_pixels, _pixels, _draw_mask); \ |
| 2115 | store_8x16b(fb_pixels, block->fb_ptr); \ |
| 2116 | } \ |
| 2117 | |
| 2118 | |
| 2119 | #define shade_blocks_textured_false_modulated_check_dithered(target) \ |
| 2120 | if(psx_gpu->triangle_color == 0x808080) \ |
| 2121 | { \ |
| 2122 | false_modulated_blocks += num_blocks; \ |
| 2123 | } \ |
| 2124 | |
| 2125 | #define shade_blocks_textured_false_modulated_check_undithered(target) \ |
| 2126 | if(psx_gpu->triangle_color == 0x808080) \ |
| 2127 | { \ |
| 2128 | \ |
| 2129 | shade_blocks_textured_unmodulated_##target(psx_gpu); \ |
| 2130 | false_modulated_blocks += num_blocks; \ |
| 2131 | return; \ |
| 2132 | } \ |
| 2133 | |
| 2134 | |
| 2135 | #define shade_blocks_textured_modulated_shaded_primitive_load(dithering, \ |
| 2136 | target) \ |
| 2137 | |
| 2138 | #define shade_blocks_textured_modulated_unshaded_primitive_load(dithering, \ |
| 2139 | target) \ |
| 2140 | { \ |
| 2141 | u32 color = psx_gpu->triangle_color; \ |
| 2142 | dup_8x8b(colors_r, color); \ |
| 2143 | dup_8x8b(colors_g, color >> 8); \ |
| 2144 | dup_8x8b(colors_b, color >> 16); \ |
| 2145 | shade_blocks_textured_false_modulated_check_##dithering(target); \ |
| 2146 | } \ |
| 2147 | |
| 2148 | #define shade_blocks_textured_modulated_shaded_block_load() \ |
| 2149 | colors_r = block->r; \ |
| 2150 | colors_g = block->g; \ |
| 2151 | colors_b = block->b \ |
| 2152 | |
| 2153 | #define shade_blocks_textured_modulated_unshaded_block_load() \ |
| 2154 | |
| 2155 | #define shade_blocks_textured_modulate_dithered(component) \ |
| 2156 | pixels_##component = block->dither_offsets; \ |
| 2157 | mla_long_8x8b(pixels_##component, texels_##component, colors_##component) \ |
| 2158 | |
| 2159 | #define shade_blocks_textured_modulate_undithered(component) \ |
| 2160 | mul_long_8x8b(pixels_##component, texels_##component, colors_##component) \ |
| 2161 | |
| 2162 | #define shade_blocks_textured_modulated_builder(shading, dithering, target) \ |
| 2163 | void shade_blocks_##shading##_textured_modulated_##dithering##_##target( \ |
| 2164 | psx_gpu_struct *psx_gpu) \ |
| 2165 | { \ |
| 2166 | block_struct *block = psx_gpu->blocks; \ |
| 2167 | u32 num_blocks = psx_gpu->num_blocks; \ |
| 2168 | vec_8x16u texels; \ |
| 2169 | \ |
| 2170 | vec_8x8u texels_r; \ |
| 2171 | vec_8x8u texels_g; \ |
| 2172 | vec_8x8u texels_b; \ |
| 2173 | \ |
| 2174 | vec_8x8u colors_r; \ |
| 2175 | vec_8x8u colors_g; \ |
| 2176 | vec_8x8u colors_b; \ |
| 2177 | \ |
| 2178 | vec_8x8u pixels_r_low; \ |
| 2179 | vec_8x8u pixels_g_low; \ |
| 2180 | vec_8x8u pixels_b_low; \ |
| 2181 | vec_8x16u pixels; \ |
| 2182 | \ |
| 2183 | vec_8x16u pixels_r; \ |
| 2184 | vec_8x16u pixels_g; \ |
| 2185 | vec_8x16u pixels_b; \ |
| 2186 | \ |
| 2187 | vec_8x16u draw_mask; \ |
| 2188 | vec_8x16u zero_mask; \ |
| 2189 | \ |
| 2190 | vec_8x8u d64_0x07; \ |
| 2191 | vec_8x8u d64_0x1F; \ |
| 2192 | vec_8x8u d64_1; \ |
| 2193 | vec_8x8u d64_4; \ |
| 2194 | vec_8x8u d64_128; \ |
| 2195 | \ |
| 2196 | vec_8x16u d128_0x8000; \ |
| 2197 | \ |
| 2198 | vec_8x16u test_mask = psx_gpu->test_mask; \ |
| 2199 | u32 draw_mask_bits; \ |
| 2200 | shade_blocks_load_msb_mask_##target(); \ |
| 2201 | \ |
| 2202 | dup_8x8b(d64_0x07, 0x07); \ |
| 2203 | dup_8x8b(d64_0x1F, 0x1F); \ |
| 2204 | dup_8x8b(d64_1, 1); \ |
| 2205 | dup_8x8b(d64_4, 4); \ |
| 2206 | dup_8x8b(d64_128, 128); \ |
| 2207 | \ |
| 2208 | dup_8x16b(d128_0x8000, 0x8000); \ |
| 2209 | \ |
| 2210 | shade_blocks_textured_modulated_##shading##_primitive_load(dithering, \ |
| 2211 | target); \ |
| 2212 | \ |
| 2213 | while(num_blocks) \ |
| 2214 | { \ |
| 2215 | draw_mask_bits = block->draw_mask_bits; \ |
| 2216 | dup_8x16b(draw_mask, draw_mask_bits); \ |
| 2217 | tst_8x16b(draw_mask, draw_mask, test_mask); \ |
| 2218 | \ |
| 2219 | shade_blocks_textured_modulated_##shading##_block_load(); \ |
| 2220 | \ |
| 2221 | texels = block->texels; \ |
| 2222 | \ |
| 2223 | mov_narrow_8x16b(texels_r, texels); \ |
| 2224 | shr_narrow_8x16b(texels_g, texels, 5); \ |
| 2225 | shr_narrow_8x16b(texels_b, texels, 7); \ |
| 2226 | \ |
| 2227 | and_8x8b(texels_r, texels_r, d64_0x1F); \ |
| 2228 | and_8x8b(texels_g, texels_g, d64_0x1F); \ |
| 2229 | shr_8x8b(texels_b, texels_b, 3); \ |
| 2230 | \ |
| 2231 | shade_blocks_textured_modulate_##dithering(r); \ |
| 2232 | shade_blocks_textured_modulate_##dithering(g); \ |
| 2233 | shade_blocks_textured_modulate_##dithering(b); \ |
| 2234 | \ |
| 2235 | cmpeqz_8x16b(zero_mask, texels); \ |
| 2236 | and_8x16b(pixels, texels, d128_0x8000); \ |
| 2237 | \ |
| 2238 | shrq_narrow_signed_8x16b(pixels_r_low, pixels_r, 4); \ |
| 2239 | shrq_narrow_signed_8x16b(pixels_g_low, pixels_g, 4); \ |
| 2240 | shrq_narrow_signed_8x16b(pixels_b_low, pixels_b, 4); \ |
| 2241 | \ |
| 2242 | or_8x16b(zero_mask, draw_mask, zero_mask); \ |
| 2243 | \ |
| 2244 | shr_8x8b(pixels_r_low, pixels_r_low, 3); \ |
| 2245 | bic_8x8b(pixels_g_low, pixels_g_low, d64_0x07); \ |
| 2246 | bic_8x8b(pixels_b_low, pixels_b_low, d64_0x07); \ |
| 2247 | \ |
| 2248 | mla_long_8x8b(pixels, pixels_r_low, d64_1); \ |
| 2249 | mla_long_8x8b(pixels, pixels_g_low, d64_4); \ |
| 2250 | mla_long_8x8b(pixels, pixels_b_low, d64_128); \ |
| 2251 | \ |
| 2252 | shade_blocks_store_##target(zero_mask, pixels); \ |
| 2253 | \ |
| 2254 | num_blocks--; \ |
| 2255 | block++; \ |
| 2256 | } \ |
| 2257 | } \ |
| 2258 | |
| 2259 | #ifndef NEON_BUILD |
| 2260 | |
| 2261 | shade_blocks_textured_modulated_builder(shaded, dithered, direct); |
| 2262 | shade_blocks_textured_modulated_builder(shaded, undithered, direct); |
| 2263 | shade_blocks_textured_modulated_builder(unshaded, dithered, direct); |
| 2264 | shade_blocks_textured_modulated_builder(unshaded, undithered, direct); |
| 2265 | |
| 2266 | shade_blocks_textured_modulated_builder(shaded, dithered, indirect); |
| 2267 | shade_blocks_textured_modulated_builder(shaded, undithered, indirect); |
| 2268 | shade_blocks_textured_modulated_builder(unshaded, dithered, indirect); |
| 2269 | shade_blocks_textured_modulated_builder(unshaded, undithered, indirect); |
| 2270 | |
| 2271 | #endif |
| 2272 | |
| 2273 | |
| 2274 | #define shade_blocks_textured_unmodulated_builder(target) \ |
| 2275 | void shade_blocks_textured_unmodulated_##target(psx_gpu_struct *psx_gpu) \ |
| 2276 | { \ |
| 2277 | block_struct *block = psx_gpu->blocks; \ |
| 2278 | u32 num_blocks = psx_gpu->num_blocks; \ |
| 2279 | vec_8x16u draw_mask; \ |
| 2280 | vec_8x16u test_mask = psx_gpu->test_mask; \ |
| 2281 | u32 draw_mask_bits; \ |
| 2282 | \ |
| 2283 | vec_8x16u pixels; \ |
| 2284 | shade_blocks_load_msb_mask_##target(); \ |
| 2285 | \ |
| 2286 | while(num_blocks) \ |
| 2287 | { \ |
| 2288 | vec_8x16u zero_mask; \ |
| 2289 | \ |
| 2290 | draw_mask_bits = block->draw_mask_bits; \ |
| 2291 | dup_8x16b(draw_mask, draw_mask_bits); \ |
| 2292 | tst_8x16b(draw_mask, draw_mask, test_mask); \ |
| 2293 | \ |
| 2294 | pixels = block->texels; \ |
| 2295 | \ |
| 2296 | cmpeqz_8x16b(zero_mask, pixels); \ |
| 2297 | or_8x16b(zero_mask, draw_mask, zero_mask); \ |
| 2298 | \ |
| 2299 | shade_blocks_store_##target(zero_mask, pixels); \ |
| 2300 | \ |
| 2301 | num_blocks--; \ |
| 2302 | block++; \ |
| 2303 | } \ |
| 2304 | } \ |
| 2305 | |
| 2306 | #define shade_blocks_textured_unmodulated_dithered_builder(target) \ |
| 2307 | void shade_blocks_textured_unmodulated_dithered_##target(psx_gpu_struct \ |
| 2308 | *psx_gpu) \ |
| 2309 | { \ |
| 2310 | block_struct *block = psx_gpu->blocks; \ |
| 2311 | u32 num_blocks = psx_gpu->num_blocks; \ |
| 2312 | vec_8x16u draw_mask; \ |
| 2313 | vec_8x16u test_mask = psx_gpu->test_mask; \ |
| 2314 | u32 draw_mask_bits; \ |
| 2315 | \ |
| 2316 | vec_8x16u pixels; \ |
| 2317 | shade_blocks_load_msb_mask_##target(); \ |
| 2318 | \ |
| 2319 | while(num_blocks) \ |
| 2320 | { \ |
| 2321 | vec_8x16u zero_mask; \ |
| 2322 | \ |
| 2323 | draw_mask_bits = block->draw_mask_bits; \ |
| 2324 | dup_8x16b(draw_mask, draw_mask_bits); \ |
| 2325 | tst_8x16b(draw_mask, draw_mask, test_mask); \ |
| 2326 | \ |
| 2327 | pixels = block->texels; \ |
| 2328 | \ |
| 2329 | cmpeqz_8x16b(zero_mask, pixels); \ |
| 2330 | or_8x16b(zero_mask, draw_mask, zero_mask); \ |
| 2331 | \ |
| 2332 | shade_blocks_store_##target(zero_mask, pixels); \ |
| 2333 | \ |
| 2334 | num_blocks--; \ |
| 2335 | block++; \ |
| 2336 | } \ |
| 2337 | } \ |
| 2338 | |
| 2339 | #ifndef NEON_BUILD |
| 2340 | |
| 2341 | shade_blocks_textured_unmodulated_builder(indirect) |
| 2342 | shade_blocks_textured_unmodulated_builder(direct) |
| 2343 | |
| 2344 | void shade_blocks_unshaded_untextured_indirect(psx_gpu_struct *psx_gpu) |
| 2345 | { |
| 2346 | } |
| 2347 | |
| 2348 | void shade_blocks_unshaded_untextured_direct(psx_gpu_struct *psx_gpu) |
| 2349 | { |
| 2350 | block_struct *block = psx_gpu->blocks; |
| 2351 | u32 num_blocks = psx_gpu->num_blocks; |
| 2352 | |
| 2353 | vec_8x16u pixels = block->pixels; |
| 2354 | shade_blocks_load_msb_mask_direct(); |
| 2355 | |
| 2356 | while(num_blocks) |
| 2357 | { |
| 2358 | shade_blocks_store_direct(block->draw_mask, pixels); |
| 2359 | |
| 2360 | num_blocks--; |
| 2361 | block++; |
| 2362 | } |
| 2363 | } |
| 2364 | |
| 2365 | #endif |
| 2366 | |
| 2367 | void shade_blocks_shaded_untextured(psx_gpu_struct *psx_gpu) |
| 2368 | { |
| 2369 | } |
| 2370 | |
| 2371 | |
| 2372 | #define blend_blocks_mask_evaluate_on() \ |
| 2373 | vec_8x16u mask_pixels; \ |
| 2374 | cmpltz_8x16b(mask_pixels, framebuffer_pixels); \ |
| 2375 | or_8x16b(draw_mask, draw_mask, mask_pixels) \ |
| 2376 | |
| 2377 | #define blend_blocks_mask_evaluate_off() \ |
| 2378 | |
| 2379 | #define blend_blocks_average() \ |
| 2380 | { \ |
| 2381 | vec_8x16u pixels_no_msb; \ |
| 2382 | vec_8x16u fb_pixels_no_msb; \ |
| 2383 | \ |
| 2384 | vec_8x16u d128_0x0421; \ |
| 2385 | vec_8x16u d128_0x8000; \ |
| 2386 | \ |
| 2387 | dup_8x16b(d128_0x0421, 0x0421); \ |
| 2388 | dup_8x16b(d128_0x8000, 0x8000); \ |
| 2389 | \ |
| 2390 | eor_8x16b(blend_pixels, pixels, framebuffer_pixels); \ |
| 2391 | bic_8x16b(pixels_no_msb, pixels, d128_0x8000); \ |
| 2392 | and_8x16b(blend_pixels, blend_pixels, d128_0x0421); \ |
| 2393 | sub_8x16b(blend_pixels, pixels_no_msb, blend_pixels); \ |
| 2394 | bic_8x16b(fb_pixels_no_msb, framebuffer_pixels, d128_0x8000); \ |
| 2395 | average_8x16b(blend_pixels, fb_pixels_no_msb, blend_pixels); \ |
| 2396 | } \ |
| 2397 | |
| 2398 | #define blend_blocks_add() \ |
| 2399 | { \ |
| 2400 | vec_8x16u pixels_rb, pixels_g; \ |
| 2401 | vec_8x16u fb_rb, fb_g; \ |
| 2402 | \ |
| 2403 | vec_8x16u d128_0x7C1F; \ |
| 2404 | vec_8x16u d128_0x03E0; \ |
| 2405 | \ |
| 2406 | dup_8x16b(d128_0x7C1F, 0x7C1F); \ |
| 2407 | dup_8x16b(d128_0x03E0, 0x03E0); \ |
| 2408 | \ |
| 2409 | and_8x16b(pixels_rb, pixels, d128_0x7C1F); \ |
| 2410 | and_8x16b(pixels_g, pixels, d128_0x03E0); \ |
| 2411 | \ |
| 2412 | and_8x16b(fb_rb, framebuffer_pixels, d128_0x7C1F); \ |
| 2413 | and_8x16b(fb_g, framebuffer_pixels, d128_0x03E0); \ |
| 2414 | \ |
| 2415 | add_8x16b(fb_rb, fb_rb, pixels_rb); \ |
| 2416 | add_8x16b(fb_g, fb_g, pixels_g); \ |
| 2417 | \ |
| 2418 | min_16x8b(vector_cast(vec_16x8u, fb_rb), vector_cast(vec_16x8u, fb_rb), \ |
| 2419 | vector_cast(vec_16x8u, d128_0x7C1F)); \ |
| 2420 | min_8x16b(fb_g, fb_g, d128_0x03E0); \ |
| 2421 | \ |
| 2422 | or_8x16b(blend_pixels, fb_rb, fb_g); \ |
| 2423 | } \ |
| 2424 | |
| 2425 | #define blend_blocks_subtract() \ |
| 2426 | { \ |
| 2427 | vec_8x16u pixels_rb, pixels_g; \ |
| 2428 | vec_8x16u fb_rb, fb_g; \ |
| 2429 | \ |
| 2430 | vec_8x16u d128_0x7C1F; \ |
| 2431 | vec_8x16u d128_0x03E0; \ |
| 2432 | \ |
| 2433 | dup_8x16b(d128_0x7C1F, 0x7C1F); \ |
| 2434 | dup_8x16b(d128_0x03E0, 0x03E0); \ |
| 2435 | \ |
| 2436 | and_8x16b(pixels_rb, pixels, d128_0x7C1F); \ |
| 2437 | and_8x16b(pixels_g, pixels, d128_0x03E0); \ |
| 2438 | \ |
| 2439 | and_8x16b(fb_rb, framebuffer_pixels, d128_0x7C1F); \ |
| 2440 | and_8x16b(fb_g, framebuffer_pixels, d128_0x03E0); \ |
| 2441 | \ |
| 2442 | subs_16x8b(vector_cast(vec_16x8u, fb_rb), \ |
| 2443 | vector_cast(vec_16x8u, fb_rb), vector_cast(vec_16x8u, pixels_rb)); \ |
| 2444 | subs_8x16b(fb_g, fb_g, pixels_g); \ |
| 2445 | \ |
| 2446 | or_8x16b(blend_pixels, fb_rb, fb_g); \ |
| 2447 | } \ |
| 2448 | |
| 2449 | #define blend_blocks_add_fourth() \ |
| 2450 | { \ |
| 2451 | vec_8x16u pixels_rb, pixels_g; \ |
| 2452 | vec_8x16u pixels_fourth; \ |
| 2453 | vec_8x16u fb_rb, fb_g; \ |
| 2454 | \ |
| 2455 | vec_8x16u d128_0x7C1F; \ |
| 2456 | vec_8x16u d128_0x1C07; \ |
| 2457 | vec_8x16u d128_0x03E0; \ |
| 2458 | vec_8x16u d128_0x00E0; \ |
| 2459 | \ |
| 2460 | dup_8x16b(d128_0x7C1F, 0x7C1F); \ |
| 2461 | dup_8x16b(d128_0x1C07, 0x1C07); \ |
| 2462 | dup_8x16b(d128_0x03E0, 0x03E0); \ |
| 2463 | dup_8x16b(d128_0x00E0, 0x00E0); \ |
| 2464 | \ |
| 2465 | shr_8x16b(pixels_fourth, vector_cast(vec_8x16s, pixels), 2); \ |
| 2466 | \ |
| 2467 | and_8x16b(fb_rb, framebuffer_pixels, d128_0x7C1F); \ |
| 2468 | and_8x16b(fb_g, framebuffer_pixels, d128_0x03E0); \ |
| 2469 | \ |
| 2470 | and_8x16b(pixels_rb, pixels_fourth, d128_0x1C07); \ |
| 2471 | and_8x16b(pixels_g, pixels_fourth, d128_0x00E0); \ |
| 2472 | \ |
| 2473 | add_8x16b(fb_rb, fb_rb, pixels_rb); \ |
| 2474 | add_8x16b(fb_g, fb_g, pixels_g); \ |
| 2475 | \ |
| 2476 | min_16x8b(vector_cast(vec_16x8u, fb_rb), vector_cast(vec_16x8u, fb_rb), \ |
| 2477 | vector_cast(vec_16x8u, d128_0x7C1F)); \ |
| 2478 | min_8x16b(fb_g, fb_g, d128_0x03E0); \ |
| 2479 | \ |
| 2480 | or_8x16b(blend_pixels, fb_rb, fb_g); \ |
| 2481 | } \ |
| 2482 | |
| 2483 | #define blend_blocks_blended_combine_textured() \ |
| 2484 | { \ |
| 2485 | vec_8x16u blend_mask; \ |
| 2486 | cmpltz_8x16b(blend_mask, pixels); \ |
| 2487 | \ |
| 2488 | or_immediate_8x16b(blend_pixels, blend_pixels, 0x8000); \ |
| 2489 | bif_8x16b(blend_pixels, pixels, blend_mask); \ |
| 2490 | } \ |
| 2491 | |
| 2492 | #define blend_blocks_blended_combine_untextured() \ |
| 2493 | |
| 2494 | |
| 2495 | #define blend_blocks_body_blend(blend_mode, texturing) \ |
| 2496 | { \ |
| 2497 | blend_blocks_##blend_mode(); \ |
| 2498 | blend_blocks_blended_combine_##texturing(); \ |
| 2499 | } \ |
| 2500 | |
| 2501 | #define blend_blocks_body_average(texturing) \ |
| 2502 | blend_blocks_body_blend(average, texturing) \ |
| 2503 | |
| 2504 | #define blend_blocks_body_add(texturing) \ |
| 2505 | blend_blocks_body_blend(add, texturing) \ |
| 2506 | |
| 2507 | #define blend_blocks_body_subtract(texturing) \ |
| 2508 | blend_blocks_body_blend(subtract, texturing) \ |
| 2509 | |
| 2510 | #define blend_blocks_body_add_fourth(texturing) \ |
| 2511 | blend_blocks_body_blend(add_fourth, texturing) \ |
| 2512 | |
| 2513 | #define blend_blocks_body_unblended(texturing) \ |
| 2514 | blend_pixels = pixels \ |
| 2515 | |
| 2516 | |
| 2517 | #define blend_blocks_builder(texturing, blend_mode, mask_evaluate) \ |
| 2518 | void \ |
| 2519 | blend_blocks_##texturing##_##blend_mode##_##mask_evaluate(psx_gpu_struct \ |
| 2520 | *psx_gpu) \ |
| 2521 | { \ |
| 2522 | block_struct *block = psx_gpu->blocks; \ |
| 2523 | u32 num_blocks = psx_gpu->num_blocks; \ |
| 2524 | vec_8x16u draw_mask; \ |
| 2525 | vec_8x16u pixels; \ |
| 2526 | vec_8x16u blend_pixels; \ |
| 2527 | vec_8x16u framebuffer_pixels; \ |
| 2528 | vec_8x16u msb_mask; \ |
| 2529 | \ |
| 2530 | u16 *fb_ptr; \ |
| 2531 | \ |
| 2532 | dup_8x16b(msb_mask, psx_gpu->mask_msb); \ |
| 2533 | \ |
| 2534 | while(num_blocks) \ |
| 2535 | { \ |
| 2536 | pixels = block->pixels; \ |
| 2537 | draw_mask = block->draw_mask; \ |
| 2538 | fb_ptr = block->fb_ptr; \ |
| 2539 | \ |
| 2540 | load_8x16b(framebuffer_pixels, fb_ptr); \ |
| 2541 | \ |
| 2542 | blend_blocks_mask_evaluate_##mask_evaluate(); \ |
| 2543 | blend_blocks_body_##blend_mode(texturing); \ |
| 2544 | \ |
| 2545 | or_8x16b(blend_pixels, blend_pixels, msb_mask); \ |
| 2546 | bif_8x16b(framebuffer_pixels, blend_pixels, draw_mask); \ |
| 2547 | store_8x16b(framebuffer_pixels, fb_ptr); \ |
| 2548 | \ |
| 2549 | blend_blocks++; \ |
| 2550 | num_blocks--; \ |
| 2551 | block++; \ |
| 2552 | } \ |
| 2553 | } \ |
| 2554 | |
| 2555 | #ifndef NEON_BUILD |
| 2556 | |
| 2557 | void blend_blocks_textured_unblended_off(psx_gpu_struct *psx_gpu) |
| 2558 | { |
| 2559 | } |
| 2560 | |
| 2561 | blend_blocks_builder(textured, average, off); |
| 2562 | blend_blocks_builder(textured, average, on); |
| 2563 | blend_blocks_builder(textured, add, off); |
| 2564 | blend_blocks_builder(textured, add, on); |
| 2565 | blend_blocks_builder(textured, subtract, off); |
| 2566 | blend_blocks_builder(textured, subtract, on); |
| 2567 | blend_blocks_builder(textured, add_fourth, off); |
| 2568 | blend_blocks_builder(textured, add_fourth, on); |
| 2569 | |
| 2570 | blend_blocks_builder(untextured, average, off); |
| 2571 | blend_blocks_builder(untextured, average, on); |
| 2572 | blend_blocks_builder(untextured, add, off); |
| 2573 | blend_blocks_builder(untextured, add, on); |
| 2574 | blend_blocks_builder(untextured, subtract, off); |
| 2575 | blend_blocks_builder(untextured, subtract, on); |
| 2576 | blend_blocks_builder(untextured, add_fourth, off); |
| 2577 | blend_blocks_builder(untextured, add_fourth, on); |
| 2578 | |
| 2579 | blend_blocks_builder(textured, unblended, on); |
| 2580 | |
| 2581 | #endif |
| 2582 | |
| 2583 | |
| 2584 | #define vertex_swap(_a, _b) \ |
| 2585 | { \ |
| 2586 | vertex_struct *temp_vertex = _a; \ |
| 2587 | _a = _b; \ |
| 2588 | _b = temp_vertex; \ |
| 2589 | triangle_winding ^= 1; \ |
| 2590 | } \ |
| 2591 | |
| 2592 | |
| 2593 | // Setup blocks parametric-variables: |
| 2594 | // SHADE TEXTURE_MAP SWIZZLING |
| 2595 | // 0 0 x |
| 2596 | // 0 1 0 |
| 2597 | // 0 1 1 |
| 2598 | // 1 0 x |
| 2599 | // 1 1 0 |
| 2600 | // 1 1 1 |
| 2601 | // 8 inputs, 6 combinations |
| 2602 | |
| 2603 | #define setup_blocks_switch_untextured_unshaded(dithering, target) \ |
| 2604 | setup_blocks_unshaded_untextured_undithered_unswizzled_##target \ |
| 2605 | |
| 2606 | #define setup_blocks_switch_untextured_shaded(dithering, target) \ |
| 2607 | setup_blocks_shaded_untextured_##dithering##_unswizzled_##target \ |
| 2608 | |
| 2609 | #define setup_blocks_switch_untextured(shading, texture_mode, dithering, \ |
| 2610 | target) \ |
| 2611 | setup_blocks_switch_untextured_##shading(dithering, target) \ |
| 2612 | |
| 2613 | #define setup_blocks_switch_texture_mode_4bpp(shading) \ |
| 2614 | setup_blocks_##shading##_textured_dithered_swizzled_indirect \ |
| 2615 | |
| 2616 | #define setup_blocks_switch_texture_mode_8bpp(shading) \ |
| 2617 | setup_blocks_##shading##_textured_dithered_swizzled_indirect \ |
| 2618 | |
| 2619 | #define setup_blocks_switch_texture_mode_16bpp(shading) \ |
| 2620 | setup_blocks_##shading##_textured_dithered_unswizzled_indirect \ |
| 2621 | |
| 2622 | #define setup_blocks_switch_textured(shading, texture_mode, dithering, target) \ |
| 2623 | setup_blocks_switch_texture_mode_##texture_mode(shading) \ |
| 2624 | |
| 2625 | #define setup_blocks_switch_blended(shading, texturing, texture_mode, \ |
| 2626 | dithering, mask_evaluate) \ |
| 2627 | setup_blocks_switch_##texturing(shading, texture_mode, dithering, indirect) \ |
| 2628 | |
| 2629 | #define setup_blocks_switch_unblended_on(shading, texturing, texture_mode, \ |
| 2630 | dithering) \ |
| 2631 | setup_blocks_switch_##texturing(shading, texture_mode, dithering, indirect) \ |
| 2632 | |
| 2633 | #define setup_blocks_switch_unblended_off(shading, texturing, texture_mode, \ |
| 2634 | dithering) \ |
| 2635 | setup_blocks_switch_##texturing(shading, texture_mode, dithering, direct) \ |
| 2636 | |
| 2637 | #define setup_blocks_switch_unblended(shading, texturing, texture_mode, \ |
| 2638 | dithering, mask_evaluate) \ |
| 2639 | setup_blocks_switch_unblended_##mask_evaluate(shading, texturing, \ |
| 2640 | texture_mode, dithering) \ |
| 2641 | |
| 2642 | #define setup_blocks_switch(shading, texturing, texture_mode, dithering, \ |
| 2643 | blending, mask_evaluate) \ |
| 2644 | setup_blocks_switch_##blending(shading, texturing, texture_mode, \ |
| 2645 | dithering, mask_evaluate) \ |
| 2646 | |
| 2647 | |
| 2648 | // Texture blocks: |
| 2649 | |
| 2650 | #define texture_blocks_switch_untextured(texture_mode) \ |
| 2651 | texture_blocks_untextured \ |
| 2652 | |
| 2653 | #define texture_blocks_switch_textured(texture_mode) \ |
| 2654 | texture_blocks_##texture_mode \ |
| 2655 | |
| 2656 | #define texture_blocks_switch(texturing, texture_mode) \ |
| 2657 | texture_blocks_switch_##texturing(texture_mode) \ |
| 2658 | |
| 2659 | |
| 2660 | // Shade blocks parametric-variables: |
| 2661 | // SHADE TEXTURE_MAP MODULATE_TEXELS dither_mode |
| 2662 | // 0 0 x x |
| 2663 | // 0 1 0 0 |
| 2664 | // 0 1 0 1 |
| 2665 | // x 1 1 x |
| 2666 | // 1 0 x 0 |
| 2667 | // 1 0 x 1 |
| 2668 | // 1 1 0 0 |
| 2669 | // 1 1 0 1 |
| 2670 | // 16 inputs, 8 combinations |
| 2671 | |
| 2672 | #define shade_blocks_switch_unshaded_untextured(modulation, dithering, target) \ |
| 2673 | shade_blocks_unshaded_untextured_##target \ |
| 2674 | |
| 2675 | #define shade_blocks_switch_unshaded_textured_unmodulated(dithering, target) \ |
| 2676 | shade_blocks_textured_unmodulated_##target \ |
| 2677 | |
| 2678 | #define shade_blocks_switch_unshaded_textured_modulated(dithering, target) \ |
| 2679 | shade_blocks_unshaded_textured_modulated_##dithering##_##target \ |
| 2680 | |
| 2681 | #define shade_blocks_switch_unshaded_textured(modulation, dithering, target) \ |
| 2682 | shade_blocks_switch_unshaded_textured_##modulation(dithering, target) \ |
| 2683 | |
| 2684 | #define shade_blocks_switch_unshaded(texturing, modulation, dithering, target) \ |
| 2685 | shade_blocks_switch_unshaded_##texturing(modulation, dithering, target) \ |
| 2686 | |
| 2687 | #define shade_blocks_switch_shaded_untextured(modulation, dithering, target) \ |
| 2688 | shade_blocks_shaded_untextured \ |
| 2689 | |
| 2690 | #define shade_blocks_switch_shaded_textured_unmodulated(dithering, target) \ |
| 2691 | shade_blocks_textured_unmodulated_##target \ |
| 2692 | |
| 2693 | #define shade_blocks_switch_shaded_textured_modulated(dithering, target) \ |
| 2694 | shade_blocks_shaded_textured_modulated_##dithering##_##target \ |
| 2695 | |
| 2696 | #define shade_blocks_switch_shaded_textured(modulation, dithering, target) \ |
| 2697 | shade_blocks_switch_shaded_textured_##modulation(dithering, target) \ |
| 2698 | |
| 2699 | #define shade_blocks_switch_shaded(texturing, modulation, dithering, target) \ |
| 2700 | shade_blocks_switch_shaded_##texturing(modulation, dithering, target) \ |
| 2701 | |
| 2702 | #define shade_blocks_switch_mask_off(shading, texturing, modulation, \ |
| 2703 | dithering) \ |
| 2704 | shade_blocks_switch_##shading(texturing, modulation, dithering, direct) \ |
| 2705 | |
| 2706 | #define shade_blocks_switch_mask_on(shading, texturing, modulation, \ |
| 2707 | dithering) \ |
| 2708 | shade_blocks_switch_##shading(texturing, modulation, dithering, indirect) \ |
| 2709 | |
| 2710 | #define shade_blocks_switch_blended(shading, texturing, modulation, dithering, \ |
| 2711 | mask_evaluate) \ |
| 2712 | shade_blocks_switch_##shading(texturing, modulation, dithering, indirect) \ |
| 2713 | |
| 2714 | #define shade_blocks_switch_unblended(shading, texturing, modulation, \ |
| 2715 | dithering, mask_evaluate) \ |
| 2716 | shade_blocks_switch_mask_##mask_evaluate(shading, texturing, modulation, \ |
| 2717 | dithering) \ |
| 2718 | |
| 2719 | #define shade_blocks_switch(shading, texturing, modulation, dithering, \ |
| 2720 | blending, mask_evaluate) \ |
| 2721 | shade_blocks_switch_##blending(shading, texturing, modulation, dithering, \ |
| 2722 | mask_evaluate) \ |
| 2723 | |
| 2724 | |
| 2725 | // Blend blocks parametric-variables: |
| 2726 | // TEXTURE_MAP BLEND BM_A BM_B mask_evaluate |
| 2727 | // x 0 x x 0 |
| 2728 | // x 0 x x 1 |
| 2729 | // 0 1 0 0 0 |
| 2730 | // 0 1 0 0 1 |
| 2731 | // 0 1 0 1 0 |
| 2732 | // 0 1 0 1 1 |
| 2733 | // 0 1 1 0 0 |
| 2734 | // 0 1 1 0 1 |
| 2735 | // 0 1 1 1 0 |
| 2736 | // 0 1 1 1 1 |
| 2737 | // 1 1 0 0 0 |
| 2738 | // 1 1 0 0 1 |
| 2739 | // 1 1 0 1 0 |
| 2740 | // 1 1 0 1 1 |
| 2741 | // 1 1 1 0 0 |
| 2742 | // 1 1 1 0 1 |
| 2743 | // 1 1 1 1 0 |
| 2744 | // 1 1 1 1 1 |
| 2745 | // 32 inputs, 18 combinations |
| 2746 | |
| 2747 | #define blend_blocks_switch_unblended(texturing, blend_mode, mask_evaluate) \ |
| 2748 | blend_blocks_textured_unblended_##mask_evaluate \ |
| 2749 | |
| 2750 | #define blend_blocks_switch_blended(texturing, blend_mode, mask_evaluate) \ |
| 2751 | blend_blocks_##texturing##_##blend_mode##_##mask_evaluate \ |
| 2752 | |
| 2753 | #define blend_blocks_switch(texturing, blending, blend_mode, mask_evaluate) \ |
| 2754 | blend_blocks_switch_##blending(texturing, blend_mode, mask_evaluate) \ |
| 2755 | |
| 2756 | |
| 2757 | #define render_blocks_switch_block_modulation(texture_mode, blend_mode, \ |
| 2758 | mask_evaluate, shading, dithering, texturing, blending, modulation) \ |
| 2759 | { \ |
| 2760 | setup_blocks_switch(shading, texturing, texture_mode, dithering, blending, \ |
| 2761 | mask_evaluate), \ |
| 2762 | texture_blocks_switch(texturing, texture_mode), \ |
| 2763 | shade_blocks_switch(shading, texturing, modulation, dithering, blending, \ |
| 2764 | mask_evaluate), \ |
| 2765 | blend_blocks_switch(texturing, blending, blend_mode, mask_evaluate) \ |
| 2766 | } \ |
| 2767 | |
| 2768 | #define render_blocks_switch_block_blending(texture_mode, blend_mode, \ |
| 2769 | mask_evaluate, shading, dithering, texturing, blending) \ |
| 2770 | render_blocks_switch_block_modulation(texture_mode, blend_mode, \ |
| 2771 | mask_evaluate, shading, dithering, texturing, blending, modulated), \ |
| 2772 | render_blocks_switch_block_modulation(texture_mode, blend_mode, \ |
| 2773 | mask_evaluate, shading, dithering, texturing, blending, unmodulated) \ |
| 2774 | |
| 2775 | #define render_blocks_switch_block_texturing(texture_mode, blend_mode, \ |
| 2776 | mask_evaluate, shading, dithering, texturing) \ |
| 2777 | render_blocks_switch_block_blending(texture_mode, blend_mode, \ |
| 2778 | mask_evaluate, shading, dithering, texturing, unblended), \ |
| 2779 | render_blocks_switch_block_blending(texture_mode, blend_mode, \ |
| 2780 | mask_evaluate, shading, dithering, texturing, blended) \ |
| 2781 | |
| 2782 | #define render_blocks_switch_block_dithering(texture_mode, blend_mode, \ |
| 2783 | mask_evaluate, shading, dithering) \ |
| 2784 | render_blocks_switch_block_texturing(texture_mode, blend_mode, \ |
| 2785 | mask_evaluate, shading, dithering, untextured), \ |
| 2786 | render_blocks_switch_block_texturing(texture_mode, blend_mode, \ |
| 2787 | mask_evaluate, shading, dithering, textured) \ |
| 2788 | |
| 2789 | #define render_blocks_switch_block_shading(texture_mode, blend_mode, \ |
| 2790 | mask_evaluate, shading) \ |
| 2791 | render_blocks_switch_block_dithering(texture_mode, blend_mode, \ |
| 2792 | mask_evaluate, shading, undithered), \ |
| 2793 | render_blocks_switch_block_dithering(texture_mode, blend_mode, \ |
| 2794 | mask_evaluate, shading, dithered) \ |
| 2795 | |
| 2796 | #define render_blocks_switch_block_mask_evaluate(texture_mode, blend_mode, \ |
| 2797 | mask_evaluate) \ |
| 2798 | render_blocks_switch_block_shading(texture_mode, blend_mode, mask_evaluate, \ |
| 2799 | unshaded), \ |
| 2800 | render_blocks_switch_block_shading(texture_mode, blend_mode, mask_evaluate, \ |
| 2801 | shaded) \ |
| 2802 | |
| 2803 | #define render_blocks_switch_block_blend_mode(texture_mode, blend_mode) \ |
| 2804 | render_blocks_switch_block_mask_evaluate(texture_mode, blend_mode, off), \ |
| 2805 | render_blocks_switch_block_mask_evaluate(texture_mode, blend_mode, on) \ |
| 2806 | |
| 2807 | #define render_blocks_switch_block_texture_mode(texture_mode) \ |
| 2808 | render_blocks_switch_block_blend_mode(texture_mode, average), \ |
| 2809 | render_blocks_switch_block_blend_mode(texture_mode, add), \ |
| 2810 | render_blocks_switch_block_blend_mode(texture_mode, subtract), \ |
| 2811 | render_blocks_switch_block_blend_mode(texture_mode, add_fourth) \ |
| 2812 | |
| 2813 | #define render_blocks_switch_block() \ |
| 2814 | render_blocks_switch_block_texture_mode(4bpp), \ |
| 2815 | render_blocks_switch_block_texture_mode(8bpp), \ |
| 2816 | render_blocks_switch_block_texture_mode(16bpp), \ |
| 2817 | render_blocks_switch_block_texture_mode(16bpp) \ |
| 2818 | |
| 2819 | |
| 2820 | render_block_handler_struct render_triangle_block_handlers[] = |
| 2821 | { |
| 2822 | render_blocks_switch_block() |
| 2823 | }; |
| 2824 | |
| 2825 | #undef render_blocks_switch_block_modulation |
| 2826 | |
| 2827 | #define render_blocks_switch_block_modulation(texture_mode, blend_mode, \ |
| 2828 | mask_evaluate, shading, dithering, texturing, blending, modulation) \ |
| 2829 | "render flags:\n" \ |
| 2830 | "texture mode: " #texture_mode "\n" \ |
| 2831 | "blend mode: " #blend_mode "\n" \ |
| 2832 | "mask evaluation: " #mask_evaluate "\n" \ |
| 2833 | #shading "\n" \ |
| 2834 | #dithering "\n" \ |
| 2835 | #texturing "\n" \ |
| 2836 | #blending "\n" \ |
| 2837 | #modulation "\n" \ |
| 2838 | |
| 2839 | char *render_block_flag_strings[] = |
| 2840 | { |
| 2841 | render_blocks_switch_block() |
| 2842 | }; |
| 2843 | |
| 2844 | |
| 2845 | #define triangle_y_direction_up 1 |
| 2846 | #define triangle_y_direction_flat 2 |
| 2847 | #define triangle_y_direction_down 0 |
| 2848 | |
| 2849 | #define triangle_winding_positive 0 |
| 2850 | #define triangle_winding_negative 1 |
| 2851 | |
| 2852 | #define triangle_set_direction(direction_variable, value) \ |
| 2853 | u32 direction_variable = (u32)(value) >> 31; \ |
| 2854 | if(value == 0) \ |
| 2855 | direction_variable = 2 \ |
| 2856 | |
| 2857 | #define triangle_case(direction_a, direction_b, direction_c, winding) \ |
| 2858 | case (triangle_y_direction_##direction_a | \ |
| 2859 | (triangle_y_direction_##direction_b << 2) | \ |
| 2860 | (triangle_y_direction_##direction_c << 4) | \ |
| 2861 | (triangle_winding_##winding << 6)) \ |
| 2862 | |
| 2863 | static int prepare_triangle(psx_gpu_struct *psx_gpu, vertex_struct *vertexes, |
| 2864 | vertex_struct *vertexes_out[3]) |
| 2865 | { |
| 2866 | s32 y_top, y_bottom; |
| 2867 | s32 triangle_area; |
| 2868 | u32 triangle_winding = 0; |
| 2869 | |
| 2870 | vertex_struct *a = &(vertexes[0]); |
| 2871 | vertex_struct *b = &(vertexes[1]); |
| 2872 | vertex_struct *c = &(vertexes[2]); |
| 2873 | |
| 2874 | triangle_area = triangle_signed_area_x2(a->x, a->y, b->x, b->y, c->x, c->y); |
| 2875 | |
| 2876 | #ifdef PROFILE |
| 2877 | triangles++; |
| 2878 | #endif |
| 2879 | |
| 2880 | if(triangle_area == 0) |
| 2881 | { |
| 2882 | #ifdef PROFILE |
| 2883 | trivial_rejects++; |
| 2884 | #endif |
| 2885 | return 0; |
| 2886 | } |
| 2887 | |
| 2888 | if(b->y < a->y) |
| 2889 | vertex_swap(a, b); |
| 2890 | |
| 2891 | if(c->y < b->y) |
| 2892 | { |
| 2893 | vertex_swap(b, c); |
| 2894 | |
| 2895 | if(b->y < a->y) |
| 2896 | vertex_swap(a, b); |
| 2897 | } |
| 2898 | |
| 2899 | y_bottom = c->y; |
| 2900 | y_top = a->y; |
| 2901 | |
| 2902 | if((y_bottom - y_top) >= 512) |
| 2903 | { |
| 2904 | #ifdef PROFILE |
| 2905 | trivial_rejects++; |
| 2906 | #endif |
| 2907 | return 0; |
| 2908 | } |
| 2909 | |
| 2910 | if(triangle_area < 0) |
| 2911 | { |
| 2912 | triangle_area = -triangle_area; |
| 2913 | triangle_winding ^= 1; |
| 2914 | vertex_swap(a, c); |
| 2915 | } |
| 2916 | |
| 2917 | if(b->x < a->x) |
| 2918 | vertex_swap(a, b); |
| 2919 | |
| 2920 | if(c->x < b->x) |
| 2921 | { |
| 2922 | vertex_swap(b, c); |
| 2923 | |
| 2924 | if(b->x < a->x) |
| 2925 | vertex_swap(a, b); |
| 2926 | } |
| 2927 | |
| 2928 | if((c->x - psx_gpu->offset_x) >= 1024 || (c->x - a->x) >= 1024) |
| 2929 | { |
| 2930 | #ifdef PROFILE |
| 2931 | trivial_rejects++; |
| 2932 | #endif |
| 2933 | return 0; |
| 2934 | } |
| 2935 | |
| 2936 | if(invalidate_texture_cache_region_viewport(psx_gpu, a->x, y_top, c->x, |
| 2937 | y_bottom) == 0) |
| 2938 | { |
| 2939 | #ifdef PROFILE |
| 2940 | trivial_rejects++; |
| 2941 | #endif |
| 2942 | return 0; |
| 2943 | } |
| 2944 | |
| 2945 | psx_gpu->triangle_area = triangle_area; |
| 2946 | psx_gpu->triangle_winding = triangle_winding; |
| 2947 | |
| 2948 | vertexes_out[0] = a; |
| 2949 | vertexes_out[1] = b; |
| 2950 | vertexes_out[2] = c; |
| 2951 | |
| 2952 | return 1; |
| 2953 | } |
| 2954 | |
| 2955 | static void render_triangle_p(psx_gpu_struct *psx_gpu, |
| 2956 | vertex_struct *vertex_ptrs[3], u32 flags) |
| 2957 | { |
| 2958 | psx_gpu->num_spans = 0; |
| 2959 | |
| 2960 | vertex_struct *a = vertex_ptrs[0]; |
| 2961 | vertex_struct *b = vertex_ptrs[1]; |
| 2962 | vertex_struct *c = vertex_ptrs[2]; |
| 2963 | |
| 2964 | s32 y_delta_a = b->y - a->y; |
| 2965 | s32 y_delta_b = c->y - b->y; |
| 2966 | s32 y_delta_c = c->y - a->y; |
| 2967 | |
| 2968 | triangle_set_direction(y_direction_a, y_delta_a); |
| 2969 | triangle_set_direction(y_direction_b, y_delta_b); |
| 2970 | triangle_set_direction(y_direction_c, y_delta_c); |
| 2971 | |
| 2972 | compute_all_gradients(psx_gpu, a, b, c); |
| 2973 | |
| 2974 | switch(y_direction_a | (y_direction_b << 2) | (y_direction_c << 4) | |
| 2975 | (psx_gpu->triangle_winding << 6)) |
| 2976 | { |
| 2977 | triangle_case(up, up, up, negative): |
| 2978 | triangle_case(up, up, flat, negative): |
| 2979 | triangle_case(up, up, down, negative): |
| 2980 | setup_spans_up_right(psx_gpu, a, b, c); |
| 2981 | break; |
| 2982 | |
| 2983 | triangle_case(flat, up, up, negative): |
| 2984 | triangle_case(flat, up, flat, negative): |
| 2985 | triangle_case(flat, up, down, negative): |
| 2986 | setup_spans_up_a(psx_gpu, a, b, c); |
| 2987 | break; |
| 2988 | |
| 2989 | triangle_case(down, up, up, negative): |
| 2990 | setup_spans_up_down(psx_gpu, a, c, b); |
| 2991 | break; |
| 2992 | |
| 2993 | triangle_case(down, up, flat, negative): |
| 2994 | setup_spans_down_a(psx_gpu, a, c, b); |
| 2995 | break; |
| 2996 | |
| 2997 | triangle_case(down, up, down, negative): |
| 2998 | setup_spans_down_right(psx_gpu, a, c, b); |
| 2999 | break; |
| 3000 | |
| 3001 | triangle_case(down, flat, up, negative): |
| 3002 | triangle_case(down, flat, flat, negative): |
| 3003 | triangle_case(down, flat, down, negative): |
| 3004 | setup_spans_down_b(psx_gpu, a, b, c); |
| 3005 | break; |
| 3006 | |
| 3007 | triangle_case(down, down, up, negative): |
| 3008 | triangle_case(down, down, flat, negative): |
| 3009 | triangle_case(down, down, down, negative): |
| 3010 | setup_spans_down_left(psx_gpu, a, b, c); |
| 3011 | break; |
| 3012 | |
| 3013 | triangle_case(up, up, up, positive): |
| 3014 | triangle_case(up, up, flat, positive): |
| 3015 | triangle_case(up, up, down, positive): |
| 3016 | setup_spans_up_left(psx_gpu, a, b, c); |
| 3017 | break; |
| 3018 | |
| 3019 | triangle_case(up, flat, up, positive): |
| 3020 | triangle_case(up, flat, flat, positive): |
| 3021 | triangle_case(up, flat, down, positive): |
| 3022 | setup_spans_up_b(psx_gpu, a, b, c); |
| 3023 | break; |
| 3024 | |
| 3025 | triangle_case(up, down, up, positive): |
| 3026 | setup_spans_up_right(psx_gpu, a, c, b); |
| 3027 | break; |
| 3028 | |
| 3029 | triangle_case(up, down, flat, positive): |
| 3030 | setup_spans_up_a(psx_gpu, a, c, b); |
| 3031 | break; |
| 3032 | |
| 3033 | triangle_case(up, down, down, positive): |
| 3034 | setup_spans_up_down(psx_gpu, a, b, c); |
| 3035 | break; |
| 3036 | |
| 3037 | triangle_case(flat, down, up, positive): |
| 3038 | triangle_case(flat, down, flat, positive): |
| 3039 | triangle_case(flat, down, down, positive): |
| 3040 | setup_spans_down_a(psx_gpu, a, b, c); |
| 3041 | break; |
| 3042 | |
| 3043 | triangle_case(down, down, up, positive): |
| 3044 | triangle_case(down, down, flat, positive): |
| 3045 | triangle_case(down, down, down, positive): |
| 3046 | setup_spans_down_right(psx_gpu, a, b, c); |
| 3047 | break; |
| 3048 | } |
| 3049 | |
| 3050 | #ifdef PROFILE |
| 3051 | spans += psx_gpu->num_spans; |
| 3052 | #endif |
| 3053 | |
| 3054 | if(unlikely(psx_gpu->render_mode & RENDER_INTERLACE_ENABLED)) |
| 3055 | { |
| 3056 | u32 i; |
| 3057 | |
| 3058 | if(psx_gpu->render_mode & RENDER_INTERLACE_ODD) |
| 3059 | { |
| 3060 | for(i = 0; i < psx_gpu->num_spans; i++) |
| 3061 | { |
| 3062 | if((psx_gpu->span_edge_data[i].y & 1) == 0) |
| 3063 | psx_gpu->span_edge_data[i].num_blocks = 0; |
| 3064 | } |
| 3065 | } |
| 3066 | else |
| 3067 | { |
| 3068 | for(i = 0; i < psx_gpu->num_spans; i++) |
| 3069 | { |
| 3070 | if(psx_gpu->span_edge_data[i].y & 1) |
| 3071 | psx_gpu->span_edge_data[i].num_blocks = 0; |
| 3072 | } |
| 3073 | } |
| 3074 | } |
| 3075 | assert(psx_gpu->span_edge_data[0].y < 1024u); |
| 3076 | |
| 3077 | u32 render_state = flags & |
| 3078 | (RENDER_FLAGS_MODULATE_TEXELS | RENDER_FLAGS_BLEND | |
| 3079 | RENDER_FLAGS_TEXTURE_MAP | RENDER_FLAGS_SHADE); |
| 3080 | render_state |= psx_gpu->render_state_base; |
| 3081 | |
| 3082 | if((psx_gpu->render_state != render_state) || |
| 3083 | (psx_gpu->primitive_type != PRIMITIVE_TYPE_TRIANGLE)) |
| 3084 | { |
| 3085 | psx_gpu->render_state = render_state; |
| 3086 | flush_render_block_buffer(psx_gpu); |
| 3087 | #ifdef PROFILE |
| 3088 | state_changes++; |
| 3089 | #endif |
| 3090 | } |
| 3091 | |
| 3092 | psx_gpu->primitive_type = PRIMITIVE_TYPE_TRIANGLE; |
| 3093 | |
| 3094 | psx_gpu->render_block_handler = |
| 3095 | &(render_triangle_block_handlers[render_state]); |
| 3096 | ((setup_blocks_function_type *)psx_gpu->render_block_handler->setup_blocks) |
| 3097 | (psx_gpu); |
| 3098 | } |
| 3099 | |
| 3100 | void render_triangle(psx_gpu_struct *psx_gpu, vertex_struct *vertexes, |
| 3101 | u32 flags) |
| 3102 | { |
| 3103 | vertex_struct *vertex_ptrs[3]; |
| 3104 | if (prepare_triangle(psx_gpu, vertexes, vertex_ptrs)) |
| 3105 | render_triangle_p(psx_gpu, vertex_ptrs, flags); |
| 3106 | } |
| 3107 | |
| 3108 | #if !defined(NEON_BUILD) || defined(SIMD_BUILD) |
| 3109 | |
| 3110 | void texture_sprite_blocks_8bpp(psx_gpu_struct *psx_gpu) |
| 3111 | { |
| 3112 | block_struct *block = psx_gpu->blocks; |
| 3113 | u32 num_blocks = psx_gpu->num_blocks; |
| 3114 | |
| 3115 | vec_8x16u texels; |
| 3116 | vec_8x8u texel_indexes; |
| 3117 | |
| 3118 | u16 *clut_ptr = psx_gpu->clut_ptr; |
| 3119 | u32 i; |
| 3120 | |
| 3121 | while(num_blocks) |
| 3122 | { |
| 3123 | texel_indexes = block->r; |
| 3124 | |
| 3125 | for(i = 0; i < 8; i++) |
| 3126 | { |
| 3127 | texels.e[i] = clut_ptr[texel_indexes.e[i]]; |
| 3128 | } |
| 3129 | |
| 3130 | block->texels = texels; |
| 3131 | |
| 3132 | num_blocks--; |
| 3133 | block++; |
| 3134 | } |
| 3135 | } |
| 3136 | |
| 3137 | #endif |
| 3138 | |
| 3139 | |
| 3140 | #define setup_sprite_tiled_initialize_4bpp_clut() \ |
| 3141 | u16 *clut_ptr = psx_gpu->clut_ptr; \ |
| 3142 | vec_8x16u clut_a, clut_b; \ |
| 3143 | vec_16x8u clut_low, clut_high; \ |
| 3144 | \ |
| 3145 | load_8x16b(clut_a, clut_ptr); \ |
| 3146 | load_8x16b(clut_b, clut_ptr + 8); \ |
| 3147 | unzip_16x8b(clut_low, clut_high, clut_a, clut_b) \ |
| 3148 | |
| 3149 | #define setup_sprite_tiled_initialize_4bpp() \ |
| 3150 | setup_sprite_tiled_initialize_4bpp_clut(); \ |
| 3151 | \ |
| 3152 | if(psx_gpu->current_texture_mask & psx_gpu->dirty_textures_4bpp_mask) \ |
| 3153 | update_texture_4bpp_cache(psx_gpu) \ |
| 3154 | |
| 3155 | #define setup_sprite_tiled_initialize_8bpp() \ |
| 3156 | if(psx_gpu->current_texture_mask & psx_gpu->dirty_textures_8bpp_mask) \ |
| 3157 | update_texture_8bpp_cache(psx_gpu) \ |
| 3158 | |
| 3159 | |
| 3160 | #define setup_sprite_tile_fetch_texel_block_8bpp(offset) \ |
| 3161 | texture_block_ptr = (u8 *)psx_gpu->texture_page_ptr + \ |
| 3162 | ((texture_offset + offset) & texture_mask); \ |
| 3163 | \ |
| 3164 | load_64b(texels, texture_block_ptr) \ |
| 3165 | |
| 3166 | |
| 3167 | #define setup_sprite_tile_add_blocks(tile_num_blocks) \ |
| 3168 | num_blocks += tile_num_blocks; \ |
| 3169 | sprite_blocks += tile_num_blocks; \ |
| 3170 | \ |
| 3171 | if(num_blocks > MAX_BLOCKS) \ |
| 3172 | { \ |
| 3173 | flush_render_block_buffer(psx_gpu); \ |
| 3174 | num_blocks = tile_num_blocks; \ |
| 3175 | block = psx_gpu->blocks; \ |
| 3176 | } \ |
| 3177 | |
| 3178 | #define setup_sprite_tile_full_4bpp(edge) \ |
| 3179 | { \ |
| 3180 | vec_8x8u texels_low, texels_high; \ |
| 3181 | vec_8x16u pixels; \ |
| 3182 | setup_sprite_tile_add_blocks(sub_tile_height * 2); \ |
| 3183 | \ |
| 3184 | while(sub_tile_height) \ |
| 3185 | { \ |
| 3186 | setup_sprite_tile_fetch_texel_block_8bpp(0); \ |
| 3187 | tbl_16(texels_low, texels, clut_low); \ |
| 3188 | tbl_16(texels_high, texels, clut_high); \ |
| 3189 | zip_8x16b(pixels, texels_low, texels_high); \ |
| 3190 | \ |
| 3191 | block->texels = pixels; \ |
| 3192 | block->draw_mask_bits = left_mask_bits; \ |
| 3193 | block->fb_ptr = fb_ptr; \ |
| 3194 | block++; \ |
| 3195 | \ |
| 3196 | setup_sprite_tile_fetch_texel_block_8bpp(8); \ |
| 3197 | tbl_16(texels_low, texels, clut_low); \ |
| 3198 | tbl_16(texels_high, texels, clut_high); \ |
| 3199 | zip_8x16b(pixels, texels_low, texels_high); \ |
| 3200 | \ |
| 3201 | block->texels = pixels; \ |
| 3202 | block->draw_mask_bits = right_mask_bits; \ |
| 3203 | block->fb_ptr = fb_ptr + 8; \ |
| 3204 | block++; \ |
| 3205 | \ |
| 3206 | fb_ptr += 1024; \ |
| 3207 | texture_offset += 0x10; \ |
| 3208 | sub_tile_height--; \ |
| 3209 | } \ |
| 3210 | texture_offset += 0xF00; \ |
| 3211 | psx_gpu->num_blocks = num_blocks; \ |
| 3212 | } \ |
| 3213 | |
| 3214 | #define setup_sprite_tile_half_4bpp(edge) \ |
| 3215 | { \ |
| 3216 | vec_8x8u texels_low, texels_high; \ |
| 3217 | vec_8x16u pixels; \ |
| 3218 | setup_sprite_tile_add_blocks(sub_tile_height); \ |
| 3219 | \ |
| 3220 | while(sub_tile_height) \ |
| 3221 | { \ |
| 3222 | setup_sprite_tile_fetch_texel_block_8bpp(0); \ |
| 3223 | tbl_16(texels_low, texels, clut_low); \ |
| 3224 | tbl_16(texels_high, texels, clut_high); \ |
| 3225 | zip_8x16b(pixels, texels_low, texels_high); \ |
| 3226 | \ |
| 3227 | block->texels = pixels; \ |
| 3228 | block->draw_mask_bits = edge##_mask_bits; \ |
| 3229 | block->fb_ptr = fb_ptr; \ |
| 3230 | block++; \ |
| 3231 | \ |
| 3232 | fb_ptr += 1024; \ |
| 3233 | texture_offset += 0x10; \ |
| 3234 | sub_tile_height--; \ |
| 3235 | } \ |
| 3236 | texture_offset += 0xF00; \ |
| 3237 | psx_gpu->num_blocks = num_blocks; \ |
| 3238 | } \ |
| 3239 | |
| 3240 | |
| 3241 | #define setup_sprite_tile_full_8bpp(edge) \ |
| 3242 | { \ |
| 3243 | setup_sprite_tile_add_blocks(sub_tile_height * 2); \ |
| 3244 | \ |
| 3245 | while(sub_tile_height) \ |
| 3246 | { \ |
| 3247 | setup_sprite_tile_fetch_texel_block_8bpp(0); \ |
| 3248 | block->r = texels; \ |
| 3249 | block->draw_mask_bits = left_mask_bits; \ |
| 3250 | block->fb_ptr = fb_ptr; \ |
| 3251 | block++; \ |
| 3252 | \ |
| 3253 | setup_sprite_tile_fetch_texel_block_8bpp(8); \ |
| 3254 | block->r = texels; \ |
| 3255 | block->draw_mask_bits = right_mask_bits; \ |
| 3256 | block->fb_ptr = fb_ptr + 8; \ |
| 3257 | block++; \ |
| 3258 | \ |
| 3259 | fb_ptr += 1024; \ |
| 3260 | texture_offset += 0x10; \ |
| 3261 | sub_tile_height--; \ |
| 3262 | } \ |
| 3263 | texture_offset += 0xF00; \ |
| 3264 | psx_gpu->num_blocks = num_blocks; \ |
| 3265 | } \ |
| 3266 | |
| 3267 | #define setup_sprite_tile_half_8bpp(edge) \ |
| 3268 | { \ |
| 3269 | setup_sprite_tile_add_blocks(sub_tile_height); \ |
| 3270 | \ |
| 3271 | while(sub_tile_height) \ |
| 3272 | { \ |
| 3273 | setup_sprite_tile_fetch_texel_block_8bpp(0); \ |
| 3274 | block->r = texels; \ |
| 3275 | block->draw_mask_bits = edge##_mask_bits; \ |
| 3276 | block->fb_ptr = fb_ptr; \ |
| 3277 | block++; \ |
| 3278 | \ |
| 3279 | fb_ptr += 1024; \ |
| 3280 | texture_offset += 0x10; \ |
| 3281 | sub_tile_height--; \ |
| 3282 | } \ |
| 3283 | texture_offset += 0xF00; \ |
| 3284 | psx_gpu->num_blocks = num_blocks; \ |
| 3285 | } \ |
| 3286 | |
| 3287 | |
| 3288 | #define setup_sprite_tile_column_edge_pre_adjust_half_right() \ |
| 3289 | texture_offset = texture_offset_base + 8; \ |
| 3290 | fb_ptr += 8 \ |
| 3291 | |
| 3292 | #define setup_sprite_tile_column_edge_pre_adjust_half_left() \ |
| 3293 | texture_offset = texture_offset_base \ |
| 3294 | |
| 3295 | #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \ |
| 3296 | setup_sprite_tile_column_edge_pre_adjust_half_##edge() \ |
| 3297 | |
| 3298 | #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \ |
| 3299 | texture_offset = texture_offset_base \ |
| 3300 | |
| 3301 | #define setup_sprite_tile_column_edge_post_adjust_half_right() \ |
| 3302 | fb_ptr -= 8 \ |
| 3303 | |
| 3304 | #define setup_sprite_tile_column_edge_post_adjust_half_left() \ |
| 3305 | |
| 3306 | #define setup_sprite_tile_column_edge_post_adjust_half(edge) \ |
| 3307 | setup_sprite_tile_column_edge_post_adjust_half_##edge() \ |
| 3308 | |
| 3309 | #define setup_sprite_tile_column_edge_post_adjust_full(edge) \ |
| 3310 | |
| 3311 | |
| 3312 | #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode, \ |
| 3313 | x4mode) \ |
| 3314 | do \ |
| 3315 | { \ |
| 3316 | sub_tile_height = column_data; \ |
| 3317 | setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \ |
| 3318 | setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \ |
| 3319 | setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge); \ |
| 3320 | } while(0) \ |
| 3321 | |
| 3322 | #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode, \ |
| 3323 | x4mode) \ |
| 3324 | do \ |
| 3325 | { \ |
| 3326 | u32 tiles_remaining = column_data >> 16; \ |
| 3327 | sub_tile_height = column_data & 0xFF; \ |
| 3328 | setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \ |
| 3329 | setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \ |
| 3330 | tiles_remaining -= 1; \ |
| 3331 | \ |
| 3332 | while(tiles_remaining) \ |
| 3333 | { \ |
| 3334 | sub_tile_height = 16; \ |
| 3335 | setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \ |
| 3336 | tiles_remaining--; \ |
| 3337 | } \ |
| 3338 | \ |
| 3339 | sub_tile_height = (column_data >> 8) & 0xFF; \ |
| 3340 | setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \ |
| 3341 | setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge); \ |
| 3342 | } while(0) \ |
| 3343 | |
| 3344 | |
| 3345 | #define setup_sprite_column_data_single() \ |
| 3346 | column_data = height \ |
| 3347 | |
| 3348 | #define setup_sprite_column_data_multi() \ |
| 3349 | column_data = 16 - offset_v; \ |
| 3350 | column_data |= ((height_rounded & 0xF) + 1) << 8; \ |
| 3351 | column_data |= (tile_height - 1) << 16 \ |
| 3352 | |
| 3353 | |
| 3354 | #define RIGHT_MASK_BIT_SHIFT 8 |
| 3355 | #define RIGHT_MASK_BIT_SHIFT_4x 16 |
| 3356 | |
| 3357 | #define setup_sprite_tile_column_width_single(texture_mode, multi_height, \ |
| 3358 | edge_mode, edge, x4mode) \ |
| 3359 | { \ |
| 3360 | setup_sprite_column_data_##multi_height(); \ |
| 3361 | left_mask_bits = left_block_mask | right_block_mask; \ |
| 3362 | right_mask_bits = left_mask_bits >> RIGHT_MASK_BIT_SHIFT##x4mode; \ |
| 3363 | \ |
| 3364 | setup_sprite_tile_column_height_##multi_height(edge_mode, edge, \ |
| 3365 | texture_mode, x4mode); \ |
| 3366 | } \ |
| 3367 | |
| 3368 | #define setup_sprite_tiled_advance_column() \ |
| 3369 | texture_offset_base += 0x100; \ |
| 3370 | if((texture_offset_base & 0xF00) == 0) \ |
| 3371 | texture_offset_base -= (0x100 + 0xF00) \ |
| 3372 | |
| 3373 | #define FB_PTR_MULTIPLIER 1 |
| 3374 | #define FB_PTR_MULTIPLIER_4x 2 |
| 3375 | |
| 3376 | #define setup_sprite_tile_column_width_multi(texture_mode, multi_height, \ |
| 3377 | left_mode, right_mode, x4mode) \ |
| 3378 | { \ |
| 3379 | setup_sprite_column_data_##multi_height(); \ |
| 3380 | s32 fb_ptr_advance_column = (16 - (1024 * height)) \ |
| 3381 | * FB_PTR_MULTIPLIER##x4mode; \ |
| 3382 | \ |
| 3383 | tile_width -= 2; \ |
| 3384 | left_mask_bits = left_block_mask; \ |
| 3385 | right_mask_bits = left_mask_bits >> RIGHT_MASK_BIT_SHIFT##x4mode; \ |
| 3386 | \ |
| 3387 | setup_sprite_tile_column_height_##multi_height(left_mode, right, \ |
| 3388 | texture_mode, x4mode); \ |
| 3389 | fb_ptr += fb_ptr_advance_column; \ |
| 3390 | \ |
| 3391 | left_mask_bits = 0x00; \ |
| 3392 | right_mask_bits = 0x00; \ |
| 3393 | \ |
| 3394 | while(tile_width) \ |
| 3395 | { \ |
| 3396 | setup_sprite_tiled_advance_column(); \ |
| 3397 | setup_sprite_tile_column_height_##multi_height(full, none, \ |
| 3398 | texture_mode, x4mode); \ |
| 3399 | fb_ptr += fb_ptr_advance_column; \ |
| 3400 | tile_width--; \ |
| 3401 | } \ |
| 3402 | \ |
| 3403 | left_mask_bits = right_block_mask; \ |
| 3404 | right_mask_bits = left_mask_bits >> RIGHT_MASK_BIT_SHIFT##x4mode; \ |
| 3405 | \ |
| 3406 | setup_sprite_tiled_advance_column(); \ |
| 3407 | setup_sprite_tile_column_height_##multi_height(right_mode, left, \ |
| 3408 | texture_mode, x4mode); \ |
| 3409 | } \ |
| 3410 | |
| 3411 | |
| 3412 | /* 4x stuff */ |
| 3413 | #define setup_sprite_tiled_initialize_4bpp_4x() \ |
| 3414 | setup_sprite_tiled_initialize_4bpp_clut() \ |
| 3415 | |
| 3416 | #define setup_sprite_tiled_initialize_8bpp_4x() \ |
| 3417 | |
| 3418 | |
| 3419 | #define setup_sprite_tile_full_4bpp_4x(edge) \ |
| 3420 | { \ |
| 3421 | vec_8x8u texels_low, texels_high; \ |
| 3422 | vec_8x16u pixels, pixels_wide; \ |
| 3423 | setup_sprite_tile_add_blocks(sub_tile_height * 2 * 4); \ |
| 3424 | u32 left_mask_bits_a = left_mask_bits & 0xFF; \ |
| 3425 | u32 left_mask_bits_b = left_mask_bits >> 8; \ |
| 3426 | u32 right_mask_bits_a = right_mask_bits & 0xFF; \ |
| 3427 | u32 right_mask_bits_b = right_mask_bits >> 8; \ |
| 3428 | \ |
| 3429 | while(sub_tile_height) \ |
| 3430 | { \ |
| 3431 | setup_sprite_tile_fetch_texel_block_8bpp(0); \ |
| 3432 | tbl_16(texels_low, texels, clut_low); \ |
| 3433 | tbl_16(texels_high, texels, clut_high); \ |
| 3434 | zip_8x16b(pixels, texels_low, texels_high); \ |
| 3435 | \ |
| 3436 | zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.low, pixels.low); \ |
| 3437 | block->texels = pixels_wide; \ |
| 3438 | block->draw_mask_bits = left_mask_bits_a; \ |
| 3439 | block->fb_ptr = fb_ptr; \ |
| 3440 | block++; \ |
| 3441 | \ |
| 3442 | block->texels = pixels_wide; \ |
| 3443 | block->draw_mask_bits = left_mask_bits_a; \ |
| 3444 | block->fb_ptr = fb_ptr + 1024; \ |
| 3445 | block++; \ |
| 3446 | \ |
| 3447 | zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.high, pixels.high); \ |
| 3448 | block->texels = pixels_wide; \ |
| 3449 | block->draw_mask_bits = left_mask_bits_b; \ |
| 3450 | block->fb_ptr = fb_ptr + 8; \ |
| 3451 | block++; \ |
| 3452 | \ |
| 3453 | block->texels = pixels_wide; \ |
| 3454 | block->draw_mask_bits = left_mask_bits_b; \ |
| 3455 | block->fb_ptr = fb_ptr + 1024 + 8; \ |
| 3456 | block++; \ |
| 3457 | \ |
| 3458 | setup_sprite_tile_fetch_texel_block_8bpp(8); \ |
| 3459 | tbl_16(texels_low, texels, clut_low); \ |
| 3460 | tbl_16(texels_high, texels, clut_high); \ |
| 3461 | zip_8x16b(pixels, texels_low, texels_high); \ |
| 3462 | \ |
| 3463 | zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.low, pixels.low); \ |
| 3464 | block->texels = pixels_wide; \ |
| 3465 | block->draw_mask_bits = right_mask_bits_a; \ |
| 3466 | block->fb_ptr = fb_ptr + 16; \ |
| 3467 | block++; \ |
| 3468 | \ |
| 3469 | block->texels = pixels_wide; \ |
| 3470 | block->draw_mask_bits = right_mask_bits_a; \ |
| 3471 | block->fb_ptr = fb_ptr + 1024 + 16; \ |
| 3472 | block++; \ |
| 3473 | \ |
| 3474 | zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.high, pixels.high); \ |
| 3475 | block->texels = pixels_wide; \ |
| 3476 | block->draw_mask_bits = right_mask_bits_b; \ |
| 3477 | block->fb_ptr = fb_ptr + 24; \ |
| 3478 | block++; \ |
| 3479 | \ |
| 3480 | block->texels = pixels_wide; \ |
| 3481 | block->draw_mask_bits = right_mask_bits_b; \ |
| 3482 | block->fb_ptr = fb_ptr + 1024 + 24; \ |
| 3483 | block++; \ |
| 3484 | \ |
| 3485 | fb_ptr += 2048; \ |
| 3486 | texture_offset += 0x10; \ |
| 3487 | sub_tile_height--; \ |
| 3488 | } \ |
| 3489 | texture_offset += 0xF00; \ |
| 3490 | psx_gpu->num_blocks = num_blocks; \ |
| 3491 | } \ |
| 3492 | |
| 3493 | #define setup_sprite_tile_half_4bpp_4x(edge) \ |
| 3494 | { \ |
| 3495 | vec_8x8u texels_low, texels_high; \ |
| 3496 | vec_8x16u pixels, pixels_wide; \ |
| 3497 | setup_sprite_tile_add_blocks(sub_tile_height * 4); \ |
| 3498 | u32 edge##_mask_bits_a = edge##_mask_bits & 0xFF; \ |
| 3499 | u32 edge##_mask_bits_b = edge##_mask_bits >> 8; \ |
| 3500 | \ |
| 3501 | while(sub_tile_height) \ |
| 3502 | { \ |
| 3503 | setup_sprite_tile_fetch_texel_block_8bpp(0); \ |
| 3504 | tbl_16(texels_low, texels, clut_low); \ |
| 3505 | tbl_16(texels_high, texels, clut_high); \ |
| 3506 | zip_8x16b(pixels, texels_low, texels_high); \ |
| 3507 | \ |
| 3508 | zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.low, pixels.low); \ |
| 3509 | block->texels = pixels_wide; \ |
| 3510 | block->draw_mask_bits = edge##_mask_bits_a; \ |
| 3511 | block->fb_ptr = fb_ptr; \ |
| 3512 | block++; \ |
| 3513 | \ |
| 3514 | block->texels = pixels_wide; \ |
| 3515 | block->draw_mask_bits = edge##_mask_bits_a; \ |
| 3516 | block->fb_ptr = fb_ptr + 1024; \ |
| 3517 | block++; \ |
| 3518 | \ |
| 3519 | zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.high, pixels.high); \ |
| 3520 | block->texels = pixels_wide; \ |
| 3521 | block->draw_mask_bits = edge##_mask_bits_b; \ |
| 3522 | block->fb_ptr = fb_ptr + 8; \ |
| 3523 | block++; \ |
| 3524 | \ |
| 3525 | block->texels = pixels_wide; \ |
| 3526 | block->draw_mask_bits = edge##_mask_bits_b; \ |
| 3527 | block->fb_ptr = fb_ptr + 1024 + 8; \ |
| 3528 | block++; \ |
| 3529 | \ |
| 3530 | fb_ptr += 2048; \ |
| 3531 | texture_offset += 0x10; \ |
| 3532 | sub_tile_height--; \ |
| 3533 | } \ |
| 3534 | texture_offset += 0xF00; \ |
| 3535 | psx_gpu->num_blocks = num_blocks; \ |
| 3536 | } \ |
| 3537 | |
| 3538 | |
| 3539 | #define setup_sprite_tile_full_8bpp_4x(edge) \ |
| 3540 | { \ |
| 3541 | setup_sprite_tile_add_blocks(sub_tile_height * 2 * 4); \ |
| 3542 | vec_16x8u texels_wide; \ |
| 3543 | u32 left_mask_bits_a = left_mask_bits & 0xFF; \ |
| 3544 | u32 left_mask_bits_b = left_mask_bits >> 8; \ |
| 3545 | u32 right_mask_bits_a = right_mask_bits & 0xFF; \ |
| 3546 | u32 right_mask_bits_b = right_mask_bits >> 8; \ |
| 3547 | \ |
| 3548 | while(sub_tile_height) \ |
| 3549 | { \ |
| 3550 | setup_sprite_tile_fetch_texel_block_8bpp(0); \ |
| 3551 | zip_8x16b(vector_cast(vec_8x16u, texels_wide), texels, texels); \ |
| 3552 | block->r = texels_wide.low; \ |
| 3553 | block->draw_mask_bits = left_mask_bits_a; \ |
| 3554 | block->fb_ptr = fb_ptr; \ |
| 3555 | block++; \ |
| 3556 | \ |
| 3557 | block->r = texels_wide.low; \ |
| 3558 | block->draw_mask_bits = left_mask_bits_a; \ |
| 3559 | block->fb_ptr = fb_ptr + 1024; \ |
| 3560 | block++; \ |
| 3561 | \ |
| 3562 | block->r = texels_wide.high; \ |
| 3563 | block->draw_mask_bits = left_mask_bits_b; \ |
| 3564 | block->fb_ptr = fb_ptr + 8; \ |
| 3565 | block++; \ |
| 3566 | \ |
| 3567 | block->r = texels_wide.high; \ |
| 3568 | block->draw_mask_bits = left_mask_bits_b; \ |
| 3569 | block->fb_ptr = fb_ptr + 1024 + 8; \ |
| 3570 | block++; \ |
| 3571 | \ |
| 3572 | setup_sprite_tile_fetch_texel_block_8bpp(8); \ |
| 3573 | zip_8x16b(vector_cast(vec_8x16u, texels_wide), texels, texels); \ |
| 3574 | block->r = texels_wide.low; \ |
| 3575 | block->draw_mask_bits = right_mask_bits_a; \ |
| 3576 | block->fb_ptr = fb_ptr + 16; \ |
| 3577 | block++; \ |
| 3578 | \ |
| 3579 | block->r = texels_wide.low; \ |
| 3580 | block->draw_mask_bits = right_mask_bits_a; \ |
| 3581 | block->fb_ptr = fb_ptr + 1024 + 16; \ |
| 3582 | block++; \ |
| 3583 | \ |
| 3584 | block->r = texels_wide.high; \ |
| 3585 | block->draw_mask_bits = right_mask_bits_b; \ |
| 3586 | block->fb_ptr = fb_ptr + 24; \ |
| 3587 | block++; \ |
| 3588 | \ |
| 3589 | block->r = texels_wide.high; \ |
| 3590 | block->draw_mask_bits = right_mask_bits_b; \ |
| 3591 | block->fb_ptr = fb_ptr + 24 + 1024; \ |
| 3592 | block++; \ |
| 3593 | \ |
| 3594 | fb_ptr += 2048; \ |
| 3595 | texture_offset += 0x10; \ |
| 3596 | sub_tile_height--; \ |
| 3597 | } \ |
| 3598 | texture_offset += 0xF00; \ |
| 3599 | psx_gpu->num_blocks = num_blocks; \ |
| 3600 | } \ |
| 3601 | |
| 3602 | #define setup_sprite_tile_half_8bpp_4x(edge) \ |
| 3603 | { \ |
| 3604 | setup_sprite_tile_add_blocks(sub_tile_height * 4); \ |
| 3605 | vec_16x8u texels_wide; \ |
| 3606 | u32 edge##_mask_bits_a = edge##_mask_bits & 0xFF; \ |
| 3607 | u32 edge##_mask_bits_b = edge##_mask_bits >> 8; \ |
| 3608 | \ |
| 3609 | while(sub_tile_height) \ |
| 3610 | { \ |
| 3611 | setup_sprite_tile_fetch_texel_block_8bpp(0); \ |
| 3612 | zip_8x16b(vector_cast(vec_8x16u, texels_wide), texels, texels); \ |
| 3613 | block->r = texels_wide.low; \ |
| 3614 | block->draw_mask_bits = edge##_mask_bits_a; \ |
| 3615 | block->fb_ptr = fb_ptr; \ |
| 3616 | block++; \ |
| 3617 | \ |
| 3618 | block->r = texels_wide.low; \ |
| 3619 | block->draw_mask_bits = edge##_mask_bits_a; \ |
| 3620 | block->fb_ptr = fb_ptr + 1024; \ |
| 3621 | block++; \ |
| 3622 | \ |
| 3623 | block->r = texels_wide.high; \ |
| 3624 | block->draw_mask_bits = edge##_mask_bits_b; \ |
| 3625 | block->fb_ptr = fb_ptr + 8; \ |
| 3626 | block++; \ |
| 3627 | \ |
| 3628 | block->r = texels_wide.high; \ |
| 3629 | block->draw_mask_bits = edge##_mask_bits_b; \ |
| 3630 | block->fb_ptr = fb_ptr + 8 + 1024; \ |
| 3631 | block++; \ |
| 3632 | \ |
| 3633 | fb_ptr += 2048; \ |
| 3634 | texture_offset += 0x10; \ |
| 3635 | sub_tile_height--; \ |
| 3636 | } \ |
| 3637 | texture_offset += 0xF00; \ |
| 3638 | psx_gpu->num_blocks = num_blocks; \ |
| 3639 | } \ |
| 3640 | |
| 3641 | |
| 3642 | #define setup_sprite_tile_column_edge_pre_adjust_half_right_4x() \ |
| 3643 | texture_offset = texture_offset_base + 8; \ |
| 3644 | fb_ptr += 16 \ |
| 3645 | |
| 3646 | #define setup_sprite_tile_column_edge_pre_adjust_half_left_4x() \ |
| 3647 | texture_offset = texture_offset_base \ |
| 3648 | |
| 3649 | #define setup_sprite_tile_column_edge_pre_adjust_half_4x(edge) \ |
| 3650 | setup_sprite_tile_column_edge_pre_adjust_half_##edge##_4x() \ |
| 3651 | |
| 3652 | #define setup_sprite_tile_column_edge_pre_adjust_full_4x(edge) \ |
| 3653 | texture_offset = texture_offset_base \ |
| 3654 | |
| 3655 | #define setup_sprite_tile_column_edge_post_adjust_half_right_4x() \ |
| 3656 | fb_ptr -= 16 \ |
| 3657 | |
| 3658 | #define setup_sprite_tile_column_edge_post_adjust_half_left_4x() \ |
| 3659 | |
| 3660 | #define setup_sprite_tile_column_edge_post_adjust_half_4x(edge) \ |
| 3661 | setup_sprite_tile_column_edge_post_adjust_half_##edge##_4x() \ |
| 3662 | |
| 3663 | #define setup_sprite_tile_column_edge_post_adjust_full_4x(edge) \ |
| 3664 | |
| 3665 | |
| 3666 | #define setup_sprite_offset_u_adjust() \ |
| 3667 | |
| 3668 | #define setup_sprite_comapre_left_block_mask() \ |
| 3669 | ((left_block_mask & 0xFF) == 0xFF) \ |
| 3670 | |
| 3671 | #define setup_sprite_comapre_right_block_mask() \ |
| 3672 | (((right_block_mask >> 8) & 0xFF) == 0xFF) \ |
| 3673 | |
| 3674 | |
| 3675 | #define setup_sprite_offset_u_adjust_4x() \ |
| 3676 | offset_u *= 2; \ |
| 3677 | offset_u_right = offset_u_right * 2 + 1 \ |
| 3678 | |
| 3679 | #define setup_sprite_comapre_left_block_mask_4x() \ |
| 3680 | ((left_block_mask & 0xFFFF) == 0xFFFF) \ |
| 3681 | |
| 3682 | #define setup_sprite_comapre_right_block_mask_4x() \ |
| 3683 | (((right_block_mask >> 16) & 0xFFFF) == 0xFFFF) \ |
| 3684 | |
| 3685 | |
| 3686 | #define setup_sprite_tiled_builder(texture_mode, x4mode) \ |
| 3687 | void setup_sprite_##texture_mode##x4mode(psx_gpu_struct *psx_gpu, s32 x, s32 y,\ |
| 3688 | s32 u, s32 v, s32 width, s32 height, u32 color) \ |
| 3689 | { \ |
| 3690 | s32 offset_u = u & 0xF; \ |
| 3691 | s32 offset_v = v & 0xF; \ |
| 3692 | \ |
| 3693 | s32 width_rounded = offset_u + width + 15; \ |
| 3694 | s32 height_rounded = offset_v + height + 15; \ |
| 3695 | s32 tile_height = height_rounded / 16; \ |
| 3696 | s32 tile_width = width_rounded / 16; \ |
| 3697 | u32 offset_u_right = width_rounded & 0xF; \ |
| 3698 | \ |
| 3699 | setup_sprite_offset_u_adjust##x4mode(); \ |
| 3700 | \ |
| 3701 | u32 left_block_mask = ~(0xFFFFFFFF << offset_u); \ |
| 3702 | u32 right_block_mask = 0xFFFFFFFE << offset_u_right; \ |
| 3703 | \ |
| 3704 | u32 left_mask_bits; \ |
| 3705 | u32 right_mask_bits; \ |
| 3706 | \ |
| 3707 | u32 sub_tile_height; \ |
| 3708 | u32 column_data; \ |
| 3709 | \ |
| 3710 | u32 texture_mask = (psx_gpu->texture_mask_width & 0xF) | \ |
| 3711 | ((psx_gpu->texture_mask_height & 0xF) << 4) | \ |
| 3712 | ((psx_gpu->texture_mask_width >> 4) << 8) | \ |
| 3713 | ((psx_gpu->texture_mask_height >> 4) << 12); \ |
| 3714 | u32 texture_offset = ((v & 0xF) << 4) | ((u & 0xF0) << 4) | \ |
| 3715 | ((v & 0xF0) << 8); \ |
| 3716 | u32 texture_offset_base = texture_offset; \ |
| 3717 | u32 control_mask; \ |
| 3718 | \ |
| 3719 | u16 *fb_ptr = psx_gpu->vram_out_ptr + (y * 1024) + (x - offset_u); \ |
| 3720 | u32 num_blocks = psx_gpu->num_blocks; \ |
| 3721 | block_struct *block = psx_gpu->blocks + num_blocks; \ |
| 3722 | \ |
| 3723 | u8 *texture_block_ptr; \ |
| 3724 | vec_8x8u texels; \ |
| 3725 | \ |
| 3726 | setup_sprite_tiled_initialize_##texture_mode##x4mode(); \ |
| 3727 | \ |
| 3728 | control_mask = tile_width == 1; \ |
| 3729 | control_mask |= (tile_height == 1) << 1; \ |
| 3730 | control_mask |= setup_sprite_comapre_left_block_mask##x4mode() << 2; \ |
| 3731 | control_mask |= setup_sprite_comapre_right_block_mask##x4mode() << 3; \ |
| 3732 | \ |
| 3733 | sprites_##texture_mode++; \ |
| 3734 | \ |
| 3735 | switch(control_mask) \ |
| 3736 | { \ |
| 3737 | default: \ |
| 3738 | case 0x0: \ |
| 3739 | setup_sprite_tile_column_width_multi(texture_mode, multi, full, full, \ |
| 3740 | x4mode); \ |
| 3741 | break; \ |
| 3742 | \ |
| 3743 | case 0x1: \ |
| 3744 | setup_sprite_tile_column_width_single(texture_mode, multi, full, none, \ |
| 3745 | x4mode); \ |
| 3746 | break; \ |
| 3747 | \ |
| 3748 | case 0x2: \ |
| 3749 | setup_sprite_tile_column_width_multi(texture_mode, single, full, full, \ |
| 3750 | x4mode); \ |
| 3751 | break; \ |
| 3752 | \ |
| 3753 | case 0x3: \ |
| 3754 | setup_sprite_tile_column_width_single(texture_mode, single, full, none, \ |
| 3755 | x4mode); \ |
| 3756 | break; \ |
| 3757 | \ |
| 3758 | case 0x4: \ |
| 3759 | setup_sprite_tile_column_width_multi(texture_mode, multi, half, full, \ |
| 3760 | x4mode); \ |
| 3761 | break; \ |
| 3762 | \ |
| 3763 | case 0x5: \ |
| 3764 | setup_sprite_tile_column_width_single(texture_mode, multi, half, right, \ |
| 3765 | x4mode); \ |
| 3766 | break; \ |
| 3767 | \ |
| 3768 | case 0x6: \ |
| 3769 | setup_sprite_tile_column_width_multi(texture_mode, single, half, full, \ |
| 3770 | x4mode); \ |
| 3771 | break; \ |
| 3772 | \ |
| 3773 | case 0x7: \ |
| 3774 | setup_sprite_tile_column_width_single(texture_mode, single, half, right, \ |
| 3775 | x4mode); \ |
| 3776 | break; \ |
| 3777 | \ |
| 3778 | case 0x8: \ |
| 3779 | setup_sprite_tile_column_width_multi(texture_mode, multi, full, half, \ |
| 3780 | x4mode); \ |
| 3781 | break; \ |
| 3782 | \ |
| 3783 | case 0x9: \ |
| 3784 | setup_sprite_tile_column_width_single(texture_mode, multi, half, left, \ |
| 3785 | x4mode); \ |
| 3786 | break; \ |
| 3787 | \ |
| 3788 | case 0xA: \ |
| 3789 | setup_sprite_tile_column_width_multi(texture_mode, single, full, half, \ |
| 3790 | x4mode); \ |
| 3791 | break; \ |
| 3792 | \ |
| 3793 | case 0xB: \ |
| 3794 | setup_sprite_tile_column_width_single(texture_mode, single, half, left, \ |
| 3795 | x4mode); \ |
| 3796 | break; \ |
| 3797 | \ |
| 3798 | case 0xC: \ |
| 3799 | setup_sprite_tile_column_width_multi(texture_mode, multi, half, half, \ |
| 3800 | x4mode); \ |
| 3801 | break; \ |
| 3802 | \ |
| 3803 | case 0xE: \ |
| 3804 | setup_sprite_tile_column_width_multi(texture_mode, single, half, half, \ |
| 3805 | x4mode); \ |
| 3806 | break; \ |
| 3807 | } \ |
| 3808 | } \ |
| 3809 | |
| 3810 | #ifndef NEON_BUILD |
| 3811 | setup_sprite_tiled_builder(4bpp,); |
| 3812 | setup_sprite_tiled_builder(8bpp,); |
| 3813 | |
| 3814 | setup_sprite_tiled_builder(4bpp,_4x); |
| 3815 | setup_sprite_tiled_builder(8bpp,_4x); |
| 3816 | #endif |
| 3817 | |
| 3818 | #if !defined(NEON_BUILD) || defined(SIMD_BUILD) |
| 3819 | |
| 3820 | void setup_sprite_16bpp(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u, |
| 3821 | s32 v, s32 width, s32 height, u32 color) |
| 3822 | { |
| 3823 | u32 left_offset = u & 0x7; |
| 3824 | u32 width_rounded = width + left_offset + 7; |
| 3825 | |
| 3826 | u16 *fb_ptr = psx_gpu->vram_out_ptr + (y * 1024) + (s32)(x - left_offset); |
| 3827 | u32 right_width = width_rounded & 0x7; |
| 3828 | u32 block_width = width_rounded / 8; |
| 3829 | u32 fb_ptr_pitch = (1024 + 8) - (block_width * 8); |
| 3830 | |
| 3831 | u32 left_mask_bits = ~(0xFF << left_offset); |
| 3832 | u32 right_mask_bits = 0xFE << right_width; |
| 3833 | |
| 3834 | u32 texture_offset_base = u + (v * 1024); |
| 3835 | u32 texture_mask = |
| 3836 | psx_gpu->texture_mask_width | (psx_gpu->texture_mask_height * 1024); |
| 3837 | |
| 3838 | u32 blocks_remaining; |
| 3839 | u32 num_blocks = psx_gpu->num_blocks; |
| 3840 | block_struct *block = psx_gpu->blocks + num_blocks; |
| 3841 | |
| 3842 | u16 *texture_page_ptr = psx_gpu->texture_page_ptr; |
| 3843 | u16 *texture_block_ptr; |
| 3844 | |
| 3845 | texture_offset_base &= ~0x7; |
| 3846 | |
| 3847 | stats_add(sprites_16bpp, 1); |
| 3848 | |
| 3849 | if(block_width == 1) |
| 3850 | { |
| 3851 | u32 mask_bits = left_mask_bits | right_mask_bits; |
| 3852 | |
| 3853 | while(height) |
| 3854 | { |
| 3855 | num_blocks++; |
| 3856 | sprite_blocks++; |
| 3857 | |
| 3858 | if(num_blocks > MAX_BLOCKS) |
| 3859 | { |
| 3860 | flush_render_block_buffer(psx_gpu); |
| 3861 | num_blocks = 1; |
| 3862 | block = psx_gpu->blocks; |
| 3863 | } |
| 3864 | |
| 3865 | texture_block_ptr = |
| 3866 | texture_page_ptr + (texture_offset_base & texture_mask); |
| 3867 | |
| 3868 | block->texels = *(vec_8x16u *)texture_block_ptr; |
| 3869 | block->draw_mask_bits = mask_bits; |
| 3870 | block->fb_ptr = fb_ptr; |
| 3871 | |
| 3872 | block++; |
| 3873 | |
| 3874 | texture_offset_base += 1024; |
| 3875 | fb_ptr += 1024; |
| 3876 | |
| 3877 | height--; |
| 3878 | psx_gpu->num_blocks = num_blocks; |
| 3879 | } |
| 3880 | } |
| 3881 | else |
| 3882 | { |
| 3883 | u32 texture_offset; |
| 3884 | |
| 3885 | while(height) |
| 3886 | { |
| 3887 | blocks_remaining = block_width - 2; |
| 3888 | num_blocks += block_width; |
| 3889 | sprite_blocks += block_width; |
| 3890 | |
| 3891 | if(num_blocks > MAX_BLOCKS) |
| 3892 | { |
| 3893 | flush_render_block_buffer(psx_gpu); |
| 3894 | num_blocks = block_width; |
| 3895 | block = psx_gpu->blocks; |
| 3896 | } |
| 3897 | |
| 3898 | texture_offset = texture_offset_base; |
| 3899 | texture_offset_base += 1024; |
| 3900 | |
| 3901 | texture_block_ptr = texture_page_ptr + (texture_offset & texture_mask); |
| 3902 | block->texels = *(vec_8x16u *)texture_block_ptr; |
| 3903 | |
| 3904 | block->draw_mask_bits = left_mask_bits; |
| 3905 | block->fb_ptr = fb_ptr; |
| 3906 | |
| 3907 | texture_offset += 8; |
| 3908 | fb_ptr += 8; |
| 3909 | block++; |
| 3910 | |
| 3911 | while(blocks_remaining) |
| 3912 | { |
| 3913 | texture_block_ptr = texture_page_ptr + (texture_offset & texture_mask); |
| 3914 | block->texels = *(vec_8x16u *)texture_block_ptr; |
| 3915 | |
| 3916 | block->draw_mask_bits = 0; |
| 3917 | block->fb_ptr = fb_ptr; |
| 3918 | |
| 3919 | texture_offset += 8; |
| 3920 | fb_ptr += 8; |
| 3921 | block++; |
| 3922 | |
| 3923 | blocks_remaining--; |
| 3924 | } |
| 3925 | |
| 3926 | texture_block_ptr = texture_page_ptr + (texture_offset & texture_mask); |
| 3927 | block->texels = *(vec_8x16u *)texture_block_ptr; |
| 3928 | |
| 3929 | block->draw_mask_bits = right_mask_bits; |
| 3930 | block->fb_ptr = fb_ptr; |
| 3931 | |
| 3932 | fb_ptr += fb_ptr_pitch; |
| 3933 | block++; |
| 3934 | |
| 3935 | height--; |
| 3936 | psx_gpu->num_blocks = num_blocks; |
| 3937 | } |
| 3938 | } |
| 3939 | } |
| 3940 | |
| 3941 | #endif |
| 3942 | |
| 3943 | #ifndef NEON_BUILD |
| 3944 | |
| 3945 | void setup_sprite_untextured_512(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u, |
| 3946 | s32 v, s32 width, s32 height, u32 color) |
| 3947 | { |
| 3948 | u32 right_width = ((width - 1) & 0x7) + 1; |
| 3949 | u32 right_mask_bits = (0xFF << right_width); |
| 3950 | u16 *fb_ptr = psx_gpu->vram_out_ptr + (y * 1024) + x; |
| 3951 | u32 block_width = (width + 7) / 8; |
| 3952 | u32 fb_ptr_pitch = 1024 - ((block_width - 1) * 8); |
| 3953 | u32 blocks_remaining; |
| 3954 | u32 num_blocks = psx_gpu->num_blocks; |
| 3955 | block_struct *block = psx_gpu->blocks + num_blocks; |
| 3956 | |
| 3957 | u32 color_r = color & 0xFF; |
| 3958 | u32 color_g = (color >> 8) & 0xFF; |
| 3959 | u32 color_b = (color >> 16) & 0xFF; |
| 3960 | vec_8x16u colors; |
| 3961 | vec_8x16u right_mask; |
| 3962 | vec_8x16u test_mask = psx_gpu->test_mask; |
| 3963 | vec_8x16u zero_mask; |
| 3964 | |
| 3965 | sprites_untextured++; |
| 3966 | |
| 3967 | color = (color_r >> 3) | ((color_g >> 3) << 5) | ((color_b >> 3) << 10); |
| 3968 | |
| 3969 | dup_8x16b(colors, color); |
| 3970 | dup_8x16b(zero_mask, 0x00); |
| 3971 | dup_8x16b(right_mask, right_mask_bits); |
| 3972 | tst_8x16b(right_mask, right_mask, test_mask); |
| 3973 | |
| 3974 | while(height) |
| 3975 | { |
| 3976 | blocks_remaining = block_width - 1; |
| 3977 | num_blocks += block_width; |
| 3978 | |
| 3979 | #ifdef PROFILE |
| 3980 | sprite_blocks += block_width; |
| 3981 | #endif |
| 3982 | |
| 3983 | if(num_blocks > MAX_BLOCKS) |
| 3984 | { |
| 3985 | flush_render_block_buffer(psx_gpu); |
| 3986 | num_blocks = block_width; |
| 3987 | block = psx_gpu->blocks; |
| 3988 | } |
| 3989 | |
| 3990 | while(blocks_remaining) |
| 3991 | { |
| 3992 | block->pixels = colors; |
| 3993 | block->draw_mask = zero_mask; |
| 3994 | block->fb_ptr = fb_ptr; |
| 3995 | |
| 3996 | fb_ptr += 8; |
| 3997 | block++; |
| 3998 | blocks_remaining--; |
| 3999 | } |
| 4000 | |
| 4001 | block->pixels = colors; |
| 4002 | block->draw_mask = right_mask; |
| 4003 | block->fb_ptr = fb_ptr; |
| 4004 | |
| 4005 | block++; |
| 4006 | fb_ptr += fb_ptr_pitch; |
| 4007 | |
| 4008 | height--; |
| 4009 | psx_gpu->num_blocks = num_blocks; |
| 4010 | } |
| 4011 | } |
| 4012 | |
| 4013 | #endif |
| 4014 | |
| 4015 | static void __attribute__((noinline)) |
| 4016 | setup_sprite_untextured_simple(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u, |
| 4017 | s32 v, s32 width, s32 height, u32 color) |
| 4018 | { |
| 4019 | u32 r = color & 0xFF; |
| 4020 | u32 g = (color >> 8) & 0xFF; |
| 4021 | u32 b = (color >> 16) & 0xFF; |
| 4022 | u32 color_16bpp = (r >> 3) | ((g >> 3) << 5) | ((b >> 3) << 10) | |
| 4023 | psx_gpu->mask_msb; |
| 4024 | u32 color_32bpp = color_16bpp | (color_16bpp << 16); |
| 4025 | |
| 4026 | u16 *vram_ptr16 = psx_gpu->vram_out_ptr + x + (y * 1024); |
| 4027 | u32 *vram_ptr; |
| 4028 | |
| 4029 | u32 num_width; |
| 4030 | |
| 4031 | if(psx_gpu->num_blocks) |
| 4032 | { |
| 4033 | flush_render_block_buffer(psx_gpu); |
| 4034 | } |
| 4035 | |
| 4036 | while(height) |
| 4037 | { |
| 4038 | num_width = width; |
| 4039 | |
| 4040 | vram_ptr = (void *)vram_ptr16; |
| 4041 | if((uintptr_t)vram_ptr16 & 2) |
| 4042 | { |
| 4043 | *vram_ptr16 = color_32bpp; |
| 4044 | vram_ptr = (void *)(vram_ptr16 + 1); |
| 4045 | num_width--; |
| 4046 | } |
| 4047 | |
| 4048 | while(num_width >= 4 * 2) |
| 4049 | { |
| 4050 | vram_ptr[0] = color_32bpp; |
| 4051 | vram_ptr[1] = color_32bpp; |
| 4052 | vram_ptr[2] = color_32bpp; |
| 4053 | vram_ptr[3] = color_32bpp; |
| 4054 | |
| 4055 | vram_ptr += 4; |
| 4056 | num_width -= 4 * 2; |
| 4057 | } |
| 4058 | |
| 4059 | while(num_width >= 2) |
| 4060 | { |
| 4061 | *vram_ptr++ = color_32bpp; |
| 4062 | num_width -= 2; |
| 4063 | } |
| 4064 | |
| 4065 | if(num_width > 0) |
| 4066 | { |
| 4067 | *(u16 *)vram_ptr = color_32bpp; |
| 4068 | } |
| 4069 | |
| 4070 | vram_ptr16 += 1024; |
| 4071 | height--; |
| 4072 | } |
| 4073 | } |
| 4074 | |
| 4075 | void setup_sprite_untextured_512(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u, |
| 4076 | s32 v, s32 width, s32 height, u32 color); |
| 4077 | |
| 4078 | void setup_sprite_untextured(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u, |
| 4079 | s32 v, s32 width, s32 height, u32 color) |
| 4080 | { |
| 4081 | if((psx_gpu->render_state & (RENDER_STATE_MASK_EVALUATE | |
| 4082 | RENDER_FLAGS_MODULATE_TEXELS | RENDER_FLAGS_BLEND)) == 0 && |
| 4083 | (psx_gpu->render_mode & RENDER_INTERLACE_ENABLED) == 0) |
| 4084 | { |
| 4085 | setup_sprite_untextured_simple(psx_gpu, x, y, u, v, width, height, color); |
| 4086 | return; |
| 4087 | } |
| 4088 | |
| 4089 | while (width > 0) |
| 4090 | { |
| 4091 | s32 w1 = width > 512 ? 512 : width; |
| 4092 | setup_sprite_untextured_512(psx_gpu, x, y, 0, 0, w1, height, color); |
| 4093 | x += 512; |
| 4094 | width -= 512; |
| 4095 | } |
| 4096 | } |
| 4097 | |
| 4098 | |
| 4099 | #define setup_sprite_blocks_switch_textured(texture_mode) \ |
| 4100 | setup_sprite_##texture_mode \ |
| 4101 | |
| 4102 | #define setup_sprite_blocks_switch_untextured(texture_mode) \ |
| 4103 | setup_sprite_untextured \ |
| 4104 | |
| 4105 | #define setup_sprite_blocks_switch(texturing, texture_mode) \ |
| 4106 | setup_sprite_blocks_switch_##texturing(texture_mode) \ |
| 4107 | |
| 4108 | |
| 4109 | #define texture_sprite_blocks_switch_4bpp() \ |
| 4110 | texture_blocks_untextured \ |
| 4111 | |
| 4112 | #define texture_sprite_blocks_switch_8bpp() \ |
| 4113 | texture_sprite_blocks_8bpp \ |
| 4114 | |
| 4115 | #define texture_sprite_blocks_switch_16bpp() \ |
| 4116 | texture_blocks_untextured \ |
| 4117 | |
| 4118 | #define texture_sprite_blocks_switch_untextured(texture_mode) \ |
| 4119 | texture_blocks_untextured \ |
| 4120 | |
| 4121 | #define texture_sprite_blocks_switch_textured(texture_mode) \ |
| 4122 | texture_sprite_blocks_switch_##texture_mode() \ |
| 4123 | |
| 4124 | #define render_sprite_blocks_switch_block_modulation(texture_mode, blend_mode, \ |
| 4125 | mask_evaluate, shading, dithering, texturing, blending, modulation) \ |
| 4126 | { \ |
| 4127 | setup_sprite_blocks_switch(texturing, texture_mode), \ |
| 4128 | texture_sprite_blocks_switch_##texturing(texture_mode), \ |
| 4129 | shade_blocks_switch(unshaded, texturing, modulation, undithered, blending, \ |
| 4130 | mask_evaluate), \ |
| 4131 | blend_blocks_switch(texturing, blending, blend_mode, mask_evaluate) \ |
| 4132 | } \ |
| 4133 | |
| 4134 | #define render_sprite_blocks_switch_block_blending(texture_mode, blend_mode, \ |
| 4135 | mask_evaluate, shading, dithering, texturing, blending) \ |
| 4136 | render_sprite_blocks_switch_block_modulation(texture_mode, blend_mode, \ |
| 4137 | mask_evaluate, shading, dithering, texturing, blending, modulated), \ |
| 4138 | render_sprite_blocks_switch_block_modulation(texture_mode, blend_mode, \ |
| 4139 | mask_evaluate, shading, dithering, texturing, blending, unmodulated) \ |
| 4140 | |
| 4141 | #define render_sprite_blocks_switch_block_texturing(texture_mode, blend_mode, \ |
| 4142 | mask_evaluate, shading, dithering, texturing) \ |
| 4143 | render_sprite_blocks_switch_block_blending(texture_mode, blend_mode, \ |
| 4144 | mask_evaluate, shading, dithering, texturing, unblended), \ |
| 4145 | render_sprite_blocks_switch_block_blending(texture_mode, blend_mode, \ |
| 4146 | mask_evaluate, shading, dithering, texturing, blended) \ |
| 4147 | |
| 4148 | #define render_sprite_blocks_switch_block_dithering(texture_mode, blend_mode, \ |
| 4149 | mask_evaluate, shading, dithering) \ |
| 4150 | render_sprite_blocks_switch_block_texturing(texture_mode, blend_mode, \ |
| 4151 | mask_evaluate, shading, dithering, untextured), \ |
| 4152 | render_sprite_blocks_switch_block_texturing(texture_mode, blend_mode, \ |
| 4153 | mask_evaluate, shading, dithering, textured) \ |
| 4154 | |
| 4155 | #define render_sprite_blocks_switch_block_shading(texture_mode, blend_mode, \ |
| 4156 | mask_evaluate, shading) \ |
| 4157 | render_sprite_blocks_switch_block_dithering(texture_mode, blend_mode, \ |
| 4158 | mask_evaluate, shading, undithered), \ |
| 4159 | render_sprite_blocks_switch_block_dithering(texture_mode, blend_mode, \ |
| 4160 | mask_evaluate, shading, dithered) \ |
| 4161 | |
| 4162 | #define render_sprite_blocks_switch_block_mask_evaluate(texture_mode, \ |
| 4163 | blend_mode, mask_evaluate) \ |
| 4164 | render_sprite_blocks_switch_block_shading(texture_mode, blend_mode, \ |
| 4165 | mask_evaluate, unshaded), \ |
| 4166 | render_sprite_blocks_switch_block_shading(texture_mode, blend_mode, \ |
| 4167 | mask_evaluate, shaded) \ |
| 4168 | |
| 4169 | #define render_sprite_blocks_switch_block_blend_mode(texture_mode, blend_mode) \ |
| 4170 | render_sprite_blocks_switch_block_mask_evaluate(texture_mode, blend_mode, \ |
| 4171 | off), \ |
| 4172 | render_sprite_blocks_switch_block_mask_evaluate(texture_mode, blend_mode, \ |
| 4173 | on) \ |
| 4174 | |
| 4175 | #define render_sprite_blocks_switch_block_texture_mode(texture_mode) \ |
| 4176 | render_sprite_blocks_switch_block_blend_mode(texture_mode, average), \ |
| 4177 | render_sprite_blocks_switch_block_blend_mode(texture_mode, add), \ |
| 4178 | render_sprite_blocks_switch_block_blend_mode(texture_mode, subtract), \ |
| 4179 | render_sprite_blocks_switch_block_blend_mode(texture_mode, add_fourth) \ |
| 4180 | |
| 4181 | #define render_sprite_blocks_switch_block() \ |
| 4182 | render_sprite_blocks_switch_block_texture_mode(4bpp), \ |
| 4183 | render_sprite_blocks_switch_block_texture_mode(8bpp), \ |
| 4184 | render_sprite_blocks_switch_block_texture_mode(16bpp), \ |
| 4185 | render_sprite_blocks_switch_block_texture_mode(16bpp) \ |
| 4186 | |
| 4187 | |
| 4188 | render_block_handler_struct render_sprite_block_handlers[] = |
| 4189 | { |
| 4190 | render_sprite_blocks_switch_block() |
| 4191 | }; |
| 4192 | |
| 4193 | |
| 4194 | void render_sprite(psx_gpu_struct *psx_gpu, s32 x, s32 y, u32 u, u32 v, |
| 4195 | s32 width, s32 height, u32 flags, u32 color) |
| 4196 | { |
| 4197 | s32 x_right = x + width - 1; |
| 4198 | s32 y_bottom = y + height - 1; |
| 4199 | |
| 4200 | #ifdef PROFILE |
| 4201 | sprites++; |
| 4202 | #endif |
| 4203 | |
| 4204 | if(invalidate_texture_cache_region_viewport(psx_gpu, x, y, x_right, |
| 4205 | y_bottom) == 0) |
| 4206 | { |
| 4207 | return; |
| 4208 | } |
| 4209 | |
| 4210 | if(x < psx_gpu->viewport_start_x) |
| 4211 | { |
| 4212 | u32 clip = psx_gpu->viewport_start_x - x; |
| 4213 | x += clip; |
| 4214 | u += clip; |
| 4215 | width -= clip; |
| 4216 | } |
| 4217 | |
| 4218 | if(y < psx_gpu->viewport_start_y) |
| 4219 | { |
| 4220 | s32 clip = psx_gpu->viewport_start_y - y; |
| 4221 | y += clip; |
| 4222 | v += clip; |
| 4223 | height -= clip; |
| 4224 | } |
| 4225 | |
| 4226 | if(x_right > psx_gpu->viewport_end_x) |
| 4227 | width -= x_right - psx_gpu->viewport_end_x; |
| 4228 | |
| 4229 | if(y_bottom > psx_gpu->viewport_end_y) |
| 4230 | height -= y_bottom - psx_gpu->viewport_end_y; |
| 4231 | |
| 4232 | if((width <= 0) || (height <= 0)) |
| 4233 | return; |
| 4234 | |
| 4235 | #ifdef PROFILE |
| 4236 | span_pixels += width * height; |
| 4237 | spans += height; |
| 4238 | #endif |
| 4239 | |
| 4240 | u32 render_state = flags & |
| 4241 | (RENDER_FLAGS_MODULATE_TEXELS | RENDER_FLAGS_BLEND | |
| 4242 | RENDER_FLAGS_TEXTURE_MAP); |
| 4243 | render_state |= |
| 4244 | (psx_gpu->render_state_base & ~RENDER_STATE_DITHER); |
| 4245 | |
| 4246 | if((psx_gpu->render_state != render_state) || |
| 4247 | (psx_gpu->primitive_type != PRIMITIVE_TYPE_SPRITE)) |
| 4248 | { |
| 4249 | psx_gpu->render_state = render_state; |
| 4250 | flush_render_block_buffer(psx_gpu); |
| 4251 | #ifdef PROFILE |
| 4252 | state_changes++; |
| 4253 | #endif |
| 4254 | } |
| 4255 | |
| 4256 | psx_gpu->primitive_type = PRIMITIVE_TYPE_SPRITE; |
| 4257 | |
| 4258 | color &= 0xFFFFFF; |
| 4259 | |
| 4260 | if(psx_gpu->triangle_color != color) |
| 4261 | { |
| 4262 | flush_render_block_buffer(psx_gpu); |
| 4263 | psx_gpu->triangle_color = color; |
| 4264 | } |
| 4265 | |
| 4266 | if(color == 0x808080) |
| 4267 | render_state |= RENDER_FLAGS_MODULATE_TEXELS; |
| 4268 | |
| 4269 | render_block_handler_struct *render_block_handler = |
| 4270 | &(render_sprite_block_handlers[render_state]); |
| 4271 | psx_gpu->render_block_handler = render_block_handler; |
| 4272 | |
| 4273 | ((setup_sprite_function_type *)render_block_handler->setup_blocks) |
| 4274 | (psx_gpu, x, y, u, v, width, height, color); |
| 4275 | } |
| 4276 | |
| 4277 | #define draw_pixel_line_mask_evaluate_yes() \ |
| 4278 | if((*vram_ptr & 0x8000) == 0) \ |
| 4279 | |
| 4280 | #define draw_pixel_line_mask_evaluate_no() \ |
| 4281 | |
| 4282 | |
| 4283 | #define draw_pixel_line_shaded() \ |
| 4284 | { \ |
| 4285 | color_r = fixed_to_int(current_r); \ |
| 4286 | color_g = fixed_to_int(current_g); \ |
| 4287 | color_b = fixed_to_int(current_b); \ |
| 4288 | \ |
| 4289 | current_r += gradient_r; \ |
| 4290 | current_g += gradient_g; \ |
| 4291 | current_b += gradient_b; \ |
| 4292 | } \ |
| 4293 | |
| 4294 | #define draw_pixel_line_unshaded() \ |
| 4295 | { \ |
| 4296 | color_r = color & 0xFF; \ |
| 4297 | color_g = (color >> 8) & 0xFF; \ |
| 4298 | color_b = (color >> 16) & 0xFF; \ |
| 4299 | } \ |
| 4300 | |
| 4301 | |
| 4302 | #define draw_pixel_line_dithered(_x, _y) \ |
| 4303 | { \ |
| 4304 | u32 dither_xor = _x ^ _y; \ |
| 4305 | s32 dither_offset = (dither_xor >> 1) & 0x1; \ |
| 4306 | dither_offset |= (_y & 0x1) << 1; \ |
| 4307 | dither_offset |= (dither_xor & 0x1) << 2; \ |
| 4308 | dither_offset -= 4; \ |
| 4309 | \ |
| 4310 | color_r += dither_offset; \ |
| 4311 | color_g += dither_offset; \ |
| 4312 | color_b += dither_offset; \ |
| 4313 | \ |
| 4314 | if(color_r < 0) \ |
| 4315 | color_r = 0; \ |
| 4316 | \ |
| 4317 | if(color_g < 0) \ |
| 4318 | color_g = 0; \ |
| 4319 | \ |
| 4320 | if(color_b < 0) \ |
| 4321 | color_b = 0; \ |
| 4322 | \ |
| 4323 | if(color_r > 255) \ |
| 4324 | color_r = 255; \ |
| 4325 | \ |
| 4326 | if(color_g > 255) \ |
| 4327 | color_g = 255; \ |
| 4328 | \ |
| 4329 | if(color_b > 255) \ |
| 4330 | color_b = 255; \ |
| 4331 | } \ |
| 4332 | |
| 4333 | #define draw_pixel_line_undithered(_x, _y) \ |
| 4334 | |
| 4335 | |
| 4336 | #define draw_pixel_line_average() \ |
| 4337 | color_r = (color_r + fb_r) / 2; \ |
| 4338 | color_g = (color_g + fb_g) / 2; \ |
| 4339 | color_b = (color_b + fb_b) / 2 \ |
| 4340 | |
| 4341 | #define draw_pixel_line_add() \ |
| 4342 | color_r += fb_r; \ |
| 4343 | color_g += fb_g; \ |
| 4344 | color_b += fb_b; \ |
| 4345 | \ |
| 4346 | if(color_r > 31) \ |
| 4347 | color_r = 31; \ |
| 4348 | \ |
| 4349 | if(color_g > 31) \ |
| 4350 | color_g = 31; \ |
| 4351 | \ |
| 4352 | if(color_b > 31) \ |
| 4353 | color_b = 31 \ |
| 4354 | \ |
| 4355 | |
| 4356 | #define draw_pixel_line_subtract() \ |
| 4357 | color_r = fb_r - color_r; \ |
| 4358 | color_g = fb_g - color_g; \ |
| 4359 | color_b = fb_b - color_b; \ |
| 4360 | \ |
| 4361 | if(color_r < 0) \ |
| 4362 | color_r = 0; \ |
| 4363 | \ |
| 4364 | if(color_g < 0) \ |
| 4365 | color_g = 0; \ |
| 4366 | \ |
| 4367 | if(color_b < 0) \ |
| 4368 | color_b = 0 \ |
| 4369 | |
| 4370 | #define draw_pixel_line_add_fourth() \ |
| 4371 | color_r = fb_r + (color_r / 4); \ |
| 4372 | color_g = fb_g + (color_g / 4); \ |
| 4373 | color_b = fb_b + (color_b / 4); \ |
| 4374 | \ |
| 4375 | if(color_r > 31) \ |
| 4376 | color_r = 31; \ |
| 4377 | \ |
| 4378 | if(color_g > 31) \ |
| 4379 | color_g = 31; \ |
| 4380 | \ |
| 4381 | if(color_b > 31) \ |
| 4382 | color_b = 31 \ |
| 4383 | |
| 4384 | |
| 4385 | #define draw_pixel_line_blended(blend_mode) \ |
| 4386 | s32 fb_pixel = *vram_ptr; \ |
| 4387 | s32 fb_r = fb_pixel & 0x1F; \ |
| 4388 | s32 fb_g = (fb_pixel >> 5) & 0x1F; \ |
| 4389 | s32 fb_b = (fb_pixel >> 10) & 0x1F; \ |
| 4390 | \ |
| 4391 | draw_pixel_line_##blend_mode() \ |
| 4392 | |
| 4393 | #define draw_pixel_line_unblended(blend_mode) \ |
| 4394 | |
| 4395 | |
| 4396 | #define draw_pixel_line(_x, _y, shading, blending, dithering, mask_evaluate, \ |
| 4397 | blend_mode) \ |
| 4398 | if((_x >= psx_gpu->viewport_start_x) && (_y >= psx_gpu->viewport_start_y) && \ |
| 4399 | (_x <= psx_gpu->viewport_end_x) && (_y <= psx_gpu->viewport_end_y)) \ |
| 4400 | { \ |
| 4401 | draw_pixel_line_mask_evaluate_##mask_evaluate() \ |
| 4402 | { \ |
| 4403 | draw_pixel_line_##shading(); \ |
| 4404 | draw_pixel_line_##dithering(_x, _y); \ |
| 4405 | \ |
| 4406 | color_r >>= 3; \ |
| 4407 | color_g >>= 3; \ |
| 4408 | color_b >>= 3; \ |
| 4409 | \ |
| 4410 | draw_pixel_line_##blending(blend_mode); \ |
| 4411 | \ |
| 4412 | *vram_ptr = color_r | (color_g << 5) | (color_b << 10) | \ |
| 4413 | psx_gpu->mask_msb; \ |
| 4414 | } \ |
| 4415 | } \ |
| 4416 | |
| 4417 | #define update_increment(value) \ |
| 4418 | value++ \ |
| 4419 | |
| 4420 | #define update_decrement(value) \ |
| 4421 | value-- \ |
| 4422 | |
| 4423 | #define update_vram_row_increment(value) \ |
| 4424 | vram_ptr += 1024 \ |
| 4425 | |
| 4426 | #define update_vram_row_decrement(value) \ |
| 4427 | vram_ptr -= 1024 \ |
| 4428 | |
| 4429 | #define compare_increment(a, b) \ |
| 4430 | (a <= b) \ |
| 4431 | |
| 4432 | #define compare_decrement(a, b) \ |
| 4433 | (a >= b) \ |
| 4434 | |
| 4435 | #define set_line_gradients(minor) \ |
| 4436 | { \ |
| 4437 | s32 gradient_divisor = delta_##minor; \ |
| 4438 | if(gradient_divisor != 0) \ |
| 4439 | { \ |
| 4440 | gradient_r = int_to_fixed(vertex_b->r - vertex_a->r) / gradient_divisor; \ |
| 4441 | gradient_g = int_to_fixed(vertex_b->g - vertex_a->g) / gradient_divisor; \ |
| 4442 | gradient_b = int_to_fixed(vertex_b->b - vertex_a->b) / gradient_divisor; \ |
| 4443 | } \ |
| 4444 | else \ |
| 4445 | { \ |
| 4446 | gradient_r = 0; \ |
| 4447 | gradient_g = 0; \ |
| 4448 | gradient_b = 0; \ |
| 4449 | } \ |
| 4450 | current_r = fixed_center(vertex_a->r); \ |
| 4451 | current_g = fixed_center(vertex_a->g); \ |
| 4452 | current_b = fixed_center(vertex_a->b); \ |
| 4453 | } |
| 4454 | |
| 4455 | #define draw_line_span_horizontal(direction, shading, blending, dithering, \ |
| 4456 | mask_evaluate, blend_mode) \ |
| 4457 | do \ |
| 4458 | { \ |
| 4459 | error_step = delta_y * 2; \ |
| 4460 | error_wrap = delta_x * 2; \ |
| 4461 | error = delta_x; \ |
| 4462 | \ |
| 4463 | current_y = y_a; \ |
| 4464 | set_line_gradients(x); \ |
| 4465 | \ |
| 4466 | for(current_x = x_a; current_x <= x_b; current_x++) \ |
| 4467 | { \ |
| 4468 | draw_pixel_line(current_x, current_y, shading, blending, dithering, \ |
| 4469 | mask_evaluate, blend_mode); \ |
| 4470 | error += error_step; \ |
| 4471 | vram_ptr++; \ |
| 4472 | \ |
| 4473 | if(error >= error_wrap) \ |
| 4474 | { \ |
| 4475 | update_##direction(current_y); \ |
| 4476 | update_vram_row_##direction(); \ |
| 4477 | error -= error_wrap; \ |
| 4478 | } \ |
| 4479 | } \ |
| 4480 | } while(0) \ |
| 4481 | |
| 4482 | #define draw_line_span_vertical(direction, shading, blending, dithering, \ |
| 4483 | mask_evaluate, blend_mode) \ |
| 4484 | do \ |
| 4485 | { \ |
| 4486 | error_step = delta_x * 2; \ |
| 4487 | error_wrap = delta_y * 2; \ |
| 4488 | error = delta_y; \ |
| 4489 | \ |
| 4490 | current_x = x_a; \ |
| 4491 | set_line_gradients(y); \ |
| 4492 | \ |
| 4493 | for(current_y = y_a; compare_##direction(current_y, y_b); \ |
| 4494 | update_##direction(current_y)) \ |
| 4495 | { \ |
| 4496 | draw_pixel_line(current_x, current_y, shading, blending, dithering, \ |
| 4497 | mask_evaluate, blend_mode); \ |
| 4498 | error += error_step; \ |
| 4499 | update_vram_row_##direction(); \ |
| 4500 | \ |
| 4501 | if(error > error_wrap) \ |
| 4502 | { \ |
| 4503 | vram_ptr++; \ |
| 4504 | current_x++; \ |
| 4505 | error -= error_wrap; \ |
| 4506 | } \ |
| 4507 | } \ |
| 4508 | } while(0) \ |
| 4509 | |
| 4510 | |
| 4511 | #define render_line_body(shading, blending, dithering, mask_evaluate, \ |
| 4512 | blend_mode) \ |
| 4513 | if(delta_y < 0) \ |
| 4514 | { \ |
| 4515 | delta_y *= -1; \ |
| 4516 | \ |
| 4517 | if(delta_x > delta_y) \ |
| 4518 | { \ |
| 4519 | draw_line_span_horizontal(decrement, shading, blending, dithering, \ |
| 4520 | mask_evaluate, blend_mode); \ |
| 4521 | } \ |
| 4522 | else \ |
| 4523 | { \ |
| 4524 | draw_line_span_vertical(decrement, shading, blending, dithering, \ |
| 4525 | mask_evaluate, blend_mode); \ |
| 4526 | } \ |
| 4527 | } \ |
| 4528 | else \ |
| 4529 | { \ |
| 4530 | if(delta_x > delta_y) \ |
| 4531 | { \ |
| 4532 | draw_line_span_horizontal(increment, shading, blending, dithering, \ |
| 4533 | mask_evaluate, blend_mode); \ |
| 4534 | } \ |
| 4535 | else \ |
| 4536 | { \ |
| 4537 | draw_line_span_vertical(increment, shading, blending, dithering, \ |
| 4538 | mask_evaluate, blend_mode); \ |
| 4539 | } \ |
| 4540 | } \ |
| 4541 | |
| 4542 | |
| 4543 | void render_line(psx_gpu_struct *psx_gpu, vertex_struct *vertexes, u32 flags, |
| 4544 | u32 color, int double_resolution) |
| 4545 | { |
| 4546 | s32 color_r, color_g, color_b; |
| 4547 | u32 triangle_winding = 0; |
| 4548 | |
| 4549 | fixed_type gradient_r = 0; |
| 4550 | fixed_type gradient_g = 0; |
| 4551 | fixed_type gradient_b = 0; |
| 4552 | fixed_type current_r = 0; |
| 4553 | fixed_type current_g = 0; |
| 4554 | fixed_type current_b = 0; |
| 4555 | |
| 4556 | s32 y_a, y_b; |
| 4557 | s32 x_a, x_b; |
| 4558 | |
| 4559 | s32 delta_x, delta_y; |
| 4560 | |
| 4561 | s32 current_x; |
| 4562 | s32 current_y; |
| 4563 | |
| 4564 | u32 error_step; |
| 4565 | u32 error; |
| 4566 | u32 error_wrap; |
| 4567 | |
| 4568 | u16 *vram_ptr; |
| 4569 | |
| 4570 | flush_render_block_buffer(psx_gpu); |
| 4571 | psx_gpu->primitive_type = PRIMITIVE_TYPE_LINE; |
| 4572 | |
| 4573 | vertex_struct *vertex_a = &(vertexes[0]); |
| 4574 | vertex_struct *vertex_b = &(vertexes[1]); |
| 4575 | |
| 4576 | u32 control_mask; |
| 4577 | |
| 4578 | #ifdef PROFILE |
| 4579 | lines++; |
| 4580 | #endif |
| 4581 | |
| 4582 | if(vertex_a->x >= vertex_b->x) |
| 4583 | { |
| 4584 | vertex_swap(vertex_a, vertex_b); |
| 4585 | (void)triangle_winding; |
| 4586 | } |
| 4587 | |
| 4588 | x_a = vertex_a->x; |
| 4589 | x_b = vertex_b->x; |
| 4590 | |
| 4591 | y_a = vertex_a->y; |
| 4592 | y_b = vertex_b->y; |
| 4593 | |
| 4594 | delta_x = x_b - x_a; |
| 4595 | delta_y = y_b - y_a; |
| 4596 | |
| 4597 | if(delta_x >= 1024 || delta_y >= 512 || delta_y <= -512) |
| 4598 | return; |
| 4599 | |
| 4600 | if(double_resolution) |
| 4601 | { |
| 4602 | x_a *= 2; |
| 4603 | x_b *= 2; |
| 4604 | y_a *= 2; |
| 4605 | y_b *= 2; |
| 4606 | delta_x *= 2; |
| 4607 | delta_y *= 2; |
| 4608 | } |
| 4609 | |
| 4610 | flags &= ~RENDER_FLAGS_TEXTURE_MAP; |
| 4611 | |
| 4612 | vram_ptr = psx_gpu->vram_out_ptr + (y_a * 1024) + x_a; |
| 4613 | |
| 4614 | control_mask = 0x0; |
| 4615 | |
| 4616 | if(flags & RENDER_FLAGS_SHADE) |
| 4617 | control_mask |= 0x1; |
| 4618 | |
| 4619 | if(flags & RENDER_FLAGS_BLEND) |
| 4620 | { |
| 4621 | control_mask |= 0x2; |
| 4622 | control_mask |= ((psx_gpu->render_state_base >> 6) & 0x3) << 4; |
| 4623 | } |
| 4624 | |
| 4625 | if(psx_gpu->render_state_base & RENDER_STATE_DITHER) |
| 4626 | control_mask |= 0x4; |
| 4627 | |
| 4628 | if(psx_gpu->render_state_base & RENDER_STATE_MASK_EVALUATE) |
| 4629 | control_mask |= 0x8; |
| 4630 | |
| 4631 | switch(control_mask) |
| 4632 | { |
| 4633 | case 0x0: |
| 4634 | render_line_body(unshaded, unblended, undithered, no, none); |
| 4635 | break; |
| 4636 | |
| 4637 | case 0x1: |
| 4638 | render_line_body(shaded, unblended, undithered, no, none); |
| 4639 | break; |
| 4640 | |
| 4641 | case 0x2: |
| 4642 | render_line_body(unshaded, blended, undithered, no, average); |
| 4643 | break; |
| 4644 | |
| 4645 | case 0x3: |
| 4646 | render_line_body(shaded, blended, undithered, no, average); |
| 4647 | break; |
| 4648 | |
| 4649 | case 0x4: |
| 4650 | render_line_body(unshaded, unblended, dithered, no, none); |
| 4651 | break; |
| 4652 | |
| 4653 | case 0x5: |
| 4654 | render_line_body(shaded, unblended, dithered, no, none); |
| 4655 | break; |
| 4656 | |
| 4657 | case 0x6: |
| 4658 | render_line_body(unshaded, blended, dithered, no, average); |
| 4659 | break; |
| 4660 | |
| 4661 | case 0x7: |
| 4662 | render_line_body(shaded, blended, dithered, no, average); |
| 4663 | break; |
| 4664 | |
| 4665 | case 0x8: |
| 4666 | render_line_body(unshaded, unblended, undithered, yes, none); |
| 4667 | break; |
| 4668 | |
| 4669 | case 0x9: |
| 4670 | render_line_body(shaded, unblended, undithered, yes, none); |
| 4671 | break; |
| 4672 | |
| 4673 | case 0xA: |
| 4674 | render_line_body(unshaded, blended, undithered, yes, average); |
| 4675 | break; |
| 4676 | |
| 4677 | case 0xB: |
| 4678 | render_line_body(shaded, blended, undithered, yes, average); |
| 4679 | break; |
| 4680 | |
| 4681 | case 0xC: |
| 4682 | render_line_body(unshaded, unblended, dithered, yes, none); |
| 4683 | break; |
| 4684 | |
| 4685 | case 0xD: |
| 4686 | render_line_body(shaded, unblended, dithered, yes, none); |
| 4687 | break; |
| 4688 | |
| 4689 | case 0xE: |
| 4690 | render_line_body(unshaded, blended, dithered, yes, average); |
| 4691 | break; |
| 4692 | |
| 4693 | case 0xF: |
| 4694 | render_line_body(shaded, blended, dithered, yes, average); |
| 4695 | break; |
| 4696 | |
| 4697 | case 0x12: |
| 4698 | render_line_body(unshaded, blended, undithered, no, add); |
| 4699 | break; |
| 4700 | |
| 4701 | case 0x13: |
| 4702 | render_line_body(shaded, blended, undithered, no, add); |
| 4703 | break; |
| 4704 | |
| 4705 | case 0x16: |
| 4706 | render_line_body(unshaded, blended, dithered, no, add); |
| 4707 | break; |
| 4708 | |
| 4709 | case 0x17: |
| 4710 | render_line_body(shaded, blended, dithered, no, add); |
| 4711 | break; |
| 4712 | |
| 4713 | case 0x1A: |
| 4714 | render_line_body(unshaded, blended, undithered, yes, add); |
| 4715 | break; |
| 4716 | |
| 4717 | case 0x1B: |
| 4718 | render_line_body(shaded, blended, undithered, yes, add); |
| 4719 | break; |
| 4720 | |
| 4721 | case 0x1E: |
| 4722 | render_line_body(unshaded, blended, dithered, yes, add); |
| 4723 | break; |
| 4724 | |
| 4725 | case 0x1F: |
| 4726 | render_line_body(shaded, blended, dithered, yes, add); |
| 4727 | break; |
| 4728 | |
| 4729 | case 0x22: |
| 4730 | render_line_body(unshaded, blended, undithered, no, subtract); |
| 4731 | break; |
| 4732 | |
| 4733 | case 0x23: |
| 4734 | render_line_body(shaded, blended, undithered, no, subtract); |
| 4735 | break; |
| 4736 | |
| 4737 | case 0x26: |
| 4738 | render_line_body(unshaded, blended, dithered, no, subtract); |
| 4739 | break; |
| 4740 | |
| 4741 | case 0x27: |
| 4742 | render_line_body(shaded, blended, dithered, no, subtract); |
| 4743 | break; |
| 4744 | |
| 4745 | case 0x2A: |
| 4746 | render_line_body(unshaded, blended, undithered, yes, subtract); |
| 4747 | break; |
| 4748 | |
| 4749 | case 0x2B: |
| 4750 | render_line_body(shaded, blended, undithered, yes, subtract); |
| 4751 | break; |
| 4752 | |
| 4753 | case 0x2E: |
| 4754 | render_line_body(unshaded, blended, dithered, yes, subtract); |
| 4755 | break; |
| 4756 | |
| 4757 | case 0x2F: |
| 4758 | render_line_body(shaded, blended, dithered, yes, subtract); |
| 4759 | break; |
| 4760 | |
| 4761 | case 0x32: |
| 4762 | render_line_body(unshaded, blended, undithered, no, add_fourth); |
| 4763 | break; |
| 4764 | |
| 4765 | case 0x33: |
| 4766 | render_line_body(shaded, blended, undithered, no, add_fourth); |
| 4767 | break; |
| 4768 | |
| 4769 | case 0x36: |
| 4770 | render_line_body(unshaded, blended, dithered, no, add_fourth); |
| 4771 | break; |
| 4772 | |
| 4773 | case 0x37: |
| 4774 | render_line_body(shaded, blended, dithered, no, add_fourth); |
| 4775 | break; |
| 4776 | |
| 4777 | case 0x3A: |
| 4778 | render_line_body(unshaded, blended, undithered, yes, add_fourth); |
| 4779 | break; |
| 4780 | |
| 4781 | case 0x3B: |
| 4782 | render_line_body(shaded, blended, undithered, yes, add_fourth); |
| 4783 | break; |
| 4784 | |
| 4785 | case 0x3E: |
| 4786 | render_line_body(unshaded, blended, dithered, yes, add_fourth); |
| 4787 | break; |
| 4788 | |
| 4789 | case 0x3F: |
| 4790 | render_line_body(shaded, blended, dithered, yes, add_fourth); |
| 4791 | break; |
| 4792 | } |
| 4793 | } |
| 4794 | |
| 4795 | |
| 4796 | void render_block_fill(psx_gpu_struct *psx_gpu, u32 color, u32 x, u32 y, |
| 4797 | u32 width, u32 height) |
| 4798 | { |
| 4799 | if((width == 0) || (height == 0)) |
| 4800 | return; |
| 4801 | |
| 4802 | invalidate_texture_cache_region(psx_gpu, x, y, x + width - 1, y + height - 1); |
| 4803 | |
| 4804 | u32 r = color & 0xFF; |
| 4805 | u32 g = (color >> 8) & 0xFF; |
| 4806 | u32 b = (color >> 16) & 0xFF; |
| 4807 | u32 color_16bpp = (r >> 3) | ((g >> 3) << 5) | ((b >> 3) << 10) | |
| 4808 | psx_gpu->mask_msb; |
| 4809 | u32 color_32bpp = color_16bpp | (color_16bpp << 16); |
| 4810 | |
| 4811 | u32 *vram_ptr = (u32 *)(psx_gpu->vram_out_ptr + x + (y * 1024)); |
| 4812 | |
| 4813 | u32 pitch = 512 - (width / 2); |
| 4814 | u32 num_width; |
| 4815 | |
| 4816 | if(psx_gpu->render_mode & RENDER_INTERLACE_ENABLED) |
| 4817 | { |
| 4818 | pitch += 512; |
| 4819 | height /= 2; |
| 4820 | |
| 4821 | if(psx_gpu->render_mode & RENDER_INTERLACE_ODD) |
| 4822 | vram_ptr += 512; |
| 4823 | } |
| 4824 | |
| 4825 | while(height) |
| 4826 | { |
| 4827 | num_width = width; |
| 4828 | while(num_width) |
| 4829 | { |
| 4830 | vram_ptr[0] = color_32bpp; |
| 4831 | vram_ptr[1] = color_32bpp; |
| 4832 | vram_ptr[2] = color_32bpp; |
| 4833 | vram_ptr[3] = color_32bpp; |
| 4834 | vram_ptr[4] = color_32bpp; |
| 4835 | vram_ptr[5] = color_32bpp; |
| 4836 | vram_ptr[6] = color_32bpp; |
| 4837 | vram_ptr[7] = color_32bpp; |
| 4838 | |
| 4839 | vram_ptr += 8; |
| 4840 | num_width -= 16; |
| 4841 | } |
| 4842 | |
| 4843 | vram_ptr += pitch; |
| 4844 | height--; |
| 4845 | } |
| 4846 | } |
| 4847 | |
| 4848 | void render_block_fill_enh(psx_gpu_struct *psx_gpu, u32 color, u32 x, u32 y, |
| 4849 | u32 width, u32 height) |
| 4850 | { |
| 4851 | if((width == 0) || (height == 0)) |
| 4852 | return; |
| 4853 | |
| 4854 | if(width > 1024) |
| 4855 | width = 1024; |
| 4856 | |
| 4857 | u32 r = color & 0xFF; |
| 4858 | u32 g = (color >> 8) & 0xFF; |
| 4859 | u32 b = (color >> 16) & 0xFF; |
| 4860 | u32 color_16bpp = (r >> 3) | ((g >> 3) << 5) | ((b >> 3) << 10) | |
| 4861 | psx_gpu->mask_msb; |
| 4862 | u32 color_32bpp = color_16bpp | (color_16bpp << 16); |
| 4863 | |
| 4864 | u32 *vram_ptr = (u32 *)(psx_gpu->vram_out_ptr + x + (y * 1024)); |
| 4865 | |
| 4866 | u32 pitch = 1024 / 2 - (width / 2); |
| 4867 | u32 num_width; |
| 4868 | |
| 4869 | while(height) |
| 4870 | { |
| 4871 | num_width = width; |
| 4872 | while(num_width) |
| 4873 | { |
| 4874 | vram_ptr[0] = color_32bpp; |
| 4875 | vram_ptr[1] = color_32bpp; |
| 4876 | vram_ptr[2] = color_32bpp; |
| 4877 | vram_ptr[3] = color_32bpp; |
| 4878 | vram_ptr[4] = color_32bpp; |
| 4879 | vram_ptr[5] = color_32bpp; |
| 4880 | vram_ptr[6] = color_32bpp; |
| 4881 | vram_ptr[7] = color_32bpp; |
| 4882 | |
| 4883 | vram_ptr += 8; |
| 4884 | num_width -= 16; |
| 4885 | } |
| 4886 | |
| 4887 | vram_ptr += pitch; |
| 4888 | height--; |
| 4889 | } |
| 4890 | } |
| 4891 | |
| 4892 | #ifndef PCSX |
| 4893 | void render_block_copy(psx_gpu_struct *psx_gpu, u16 *source, u32 x, u32 y, |
| 4894 | u32 width, u32 height, u32 pitch) |
| 4895 | { |
| 4896 | u16 *vram_ptr = psx_gpu->vram_ptr + x + (y * 1024); |
| 4897 | u32 draw_x, draw_y; |
| 4898 | u32 mask_msb = psx_gpu->mask_msb; |
| 4899 | |
| 4900 | if((width == 0) || (height == 0)) |
| 4901 | return; |
| 4902 | |
| 4903 | flush_render_block_buffer(psx_gpu); |
| 4904 | invalidate_texture_cache_region(psx_gpu, x, y, x + width - 1, y + height - 1); |
| 4905 | |
| 4906 | for(draw_y = 0; draw_y < height; draw_y++) |
| 4907 | { |
| 4908 | for(draw_x = 0; draw_x < width; draw_x++) |
| 4909 | { |
| 4910 | vram_ptr[draw_x] = source[draw_x] | mask_msb; |
| 4911 | } |
| 4912 | |
| 4913 | source += pitch; |
| 4914 | vram_ptr += 1024; |
| 4915 | } |
| 4916 | } |
| 4917 | |
| 4918 | void render_block_move(psx_gpu_struct *psx_gpu, u32 source_x, u32 source_y, |
| 4919 | u32 dest_x, u32 dest_y, u32 width, u32 height) |
| 4920 | { |
| 4921 | render_block_copy(psx_gpu, psx_gpu->vram_ptr + source_x + (source_y * 1024), |
| 4922 | dest_x, dest_y, width, height, 1024); |
| 4923 | } |
| 4924 | #endif |
| 4925 | |
| 4926 | void initialize_reciprocal_table(void) |
| 4927 | { |
| 4928 | u32 height; |
| 4929 | u32 height_normalized; |
| 4930 | u32 height_reciprocal; |
| 4931 | s32 shift; |
| 4932 | |
| 4933 | for(height = 1; height < sizeof(reciprocal_table) |
| 4934 | / sizeof(reciprocal_table[0]); height++) |
| 4935 | { |
| 4936 | shift = __builtin_clz(height); |
| 4937 | height_normalized = height << shift; |
| 4938 | height_reciprocal = ((1ULL << 51) + (height_normalized - 1)) / |
| 4939 | height_normalized; |
| 4940 | |
| 4941 | shift = 32 - (51 - shift); |
| 4942 | |
| 4943 | reciprocal_table[height] = (height_reciprocal << 10) | shift; |
| 4944 | } |
| 4945 | } |
| 4946 | |
| 4947 | |
| 4948 | #define dither_table_row(a, b, c, d) \ |
| 4949 | ((a & 0xFF) | ((b & 0xFF) << 8) | ((c & 0xFF) << 16) | ((d & 0xFF) << 24)) \ |
| 4950 | |
| 4951 | void initialize_psx_gpu(psx_gpu_struct *psx_gpu, u16 *vram) |
| 4952 | { |
| 4953 | vec_8x16u test_mask = |
| 4954 | { { { 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80 } } }; |
| 4955 | |
| 4956 | psx_gpu->test_mask = test_mask; |
| 4957 | |
| 4958 | psx_gpu->dirty_textures_4bpp_mask = 0xFFFFFFFF; |
| 4959 | psx_gpu->dirty_textures_8bpp_mask = 0xFFFFFFFF; |
| 4960 | psx_gpu->dirty_textures_8bpp_alternate_mask = 0xFFFFFFFF; |
| 4961 | psx_gpu->viewport_mask = 0; |
| 4962 | psx_gpu->current_texture_page = 0; |
| 4963 | psx_gpu->current_texture_mask = 0; |
| 4964 | psx_gpu->last_8bpp_texture_page = 0; |
| 4965 | |
| 4966 | psx_gpu->clut_settings = 0; |
| 4967 | psx_gpu->texture_settings = 0; |
| 4968 | psx_gpu->render_state = 0; |
| 4969 | psx_gpu->render_state_base = 0; |
| 4970 | psx_gpu->num_blocks = 0; |
| 4971 | psx_gpu->uvrgb_phase = 0x8000; |
| 4972 | |
| 4973 | psx_gpu->vram_ptr = vram; |
| 4974 | psx_gpu->vram_out_ptr = vram; |
| 4975 | |
| 4976 | psx_gpu->texture_page_base = psx_gpu->vram_ptr; |
| 4977 | psx_gpu->texture_page_ptr = psx_gpu->vram_ptr; |
| 4978 | psx_gpu->clut_ptr = psx_gpu->vram_ptr; |
| 4979 | |
| 4980 | psx_gpu->viewport_start_x = psx_gpu->viewport_start_y = 0; |
| 4981 | psx_gpu->viewport_end_x = psx_gpu->viewport_end_y = 0; |
| 4982 | psx_gpu->mask_msb = 0; |
| 4983 | |
| 4984 | psx_gpu->texture_window_x = 0; |
| 4985 | psx_gpu->texture_window_y = 0; |
| 4986 | psx_gpu->texture_mask_width = 0xFF; |
| 4987 | psx_gpu->texture_mask_height = 0xFF; |
| 4988 | |
| 4989 | psx_gpu->render_mode = 0; |
| 4990 | |
| 4991 | memset(psx_gpu->vram_ptr, 0, sizeof(u16) * 1024 * 512); |
| 4992 | |
| 4993 | initialize_reciprocal_table(); |
| 4994 | psx_gpu->reciprocal_table_ptr = reciprocal_table; |
| 4995 | |
| 4996 | // 00 01 10 11 |
| 4997 | // 00 0 4 1 5 |
| 4998 | // 01 6 2 7 3 |
| 4999 | // 10 1 5 0 4 |
| 5000 | // 11 7 3 6 2 |
| 5001 | // (minus ones(4) * 4) |
| 5002 | |
| 5003 | // d0: (1 3 5 7): x1 ^ y1 |
| 5004 | // d1: (2 3 6 7): y0 |
| 5005 | // d2: (4 5 6 7): x0 ^ y0 |
| 5006 | |
| 5007 | psx_gpu->dither_table[0] = dither_table_row(-4, 0, -3, 1); |
| 5008 | psx_gpu->dither_table[1] = dither_table_row(2, -2, 3, -1); |
| 5009 | psx_gpu->dither_table[2] = dither_table_row(-3, 1, -4, 0); |
| 5010 | psx_gpu->dither_table[3] = dither_table_row(3, -1, 2, -2); |
| 5011 | |
| 5012 | psx_gpu->primitive_type = PRIMITIVE_TYPE_UNKNOWN; |
| 5013 | |
| 5014 | psx_gpu->saved_hres = 256; |
| 5015 | } |
| 5016 | |
| 5017 | u64 get_us(void) |
| 5018 | { |
| 5019 | struct timeval tv; |
| 5020 | gettimeofday(&tv, NULL); |
| 5021 | |
| 5022 | return (tv.tv_sec * 1000000ULL) + tv.tv_usec; |
| 5023 | } |
| 5024 | |
| 5025 | #if 0 //def NEON_BUILD |
| 5026 | |
| 5027 | u32 get_counter() |
| 5028 | { |
| 5029 | u32 counter; |
| 5030 | __asm__ volatile("mrc p15, 0, %0, c9, c13, 0" : "=r"(counter)); |
| 5031 | |
| 5032 | return counter; |
| 5033 | } |
| 5034 | |
| 5035 | void init_counter(void) |
| 5036 | { |
| 5037 | u32 value; |
| 5038 | asm volatile("mrc p15, 0, %0, c9, c12, 0" : "=r"(value)); |
| 5039 | value |= 5; // master enable, ccnt reset |
| 5040 | value &= ~8; // ccnt divider 0 |
| 5041 | asm volatile("mcr p15, 0, %0, c9, c12, 0" :: "r"(value)); |
| 5042 | // enable cycle counter |
| 5043 | asm volatile("mcr p15, 0, %0, c9, c12, 1" :: "r"(1 << 31)); |
| 5044 | } |
| 5045 | |
| 5046 | void triangle_benchmark(psx_gpu_struct *psx_gpu) |
| 5047 | { |
| 5048 | u32 i; |
| 5049 | |
| 5050 | u32 ticks; |
| 5051 | u32 ticks_elapsed; |
| 5052 | |
| 5053 | const u32 iterations = 500000; |
| 5054 | |
| 5055 | psx_gpu->num_blocks = 64; |
| 5056 | psx_gpu->clut_ptr = psx_gpu->vram_ptr; |
| 5057 | |
| 5058 | for(i = 0; i < 64; i++) |
| 5059 | { |
| 5060 | memset(&(psx_gpu->blocks[i].r), 0, 16); |
| 5061 | } |
| 5062 | |
| 5063 | init_counter(); |
| 5064 | |
| 5065 | ticks = get_counter(); |
| 5066 | |
| 5067 | for(i = 0; i < iterations; i++) |
| 5068 | { |
| 5069 | texture_sprite_blocks_8bpp(psx_gpu); |
| 5070 | } |
| 5071 | |
| 5072 | ticks_elapsed = get_counter() - ticks; |
| 5073 | |
| 5074 | printf("benchmark: %lf cycles\n", (double)ticks_elapsed / (iterations * 64)); |
| 5075 | } |
| 5076 | |
| 5077 | #endif |
| 5078 | |
| 5079 | #include "psx_gpu_4x.c" |
| 5080 | |
| 5081 | // vim:ts=2:sw=2:expandtab |